BEHAVIORAL AND BRAIN SCIENCES (1995) 18, 403-414 Printed in the United States of America Future directions for rhodopsin structure and function studies Paul A. Hargrave Department of Ophthalmology, and Department of Biochemistry and Molecular Biology, School of Medicine, University of Florida, Gainesville, FL 32610 Electronic mail: hargrave@eyel.eye.ufl.edu Abstract: To understand how the photoreceptor protein rhodopsin performs in its role as a receptor, its structure needs to be determined at the atomic level. Upon receiving a photon of light, rhodopsin undergoes a change in conformation that allows it to bind and activate the C-protein, transducin. An important future goal should be to determine the structure of both the inactive and the photoactivated state of rhodopsin, R*. This should provide the groundwork necessary for experiments on how rhodopsin achieves its signaling state R*, and how R* functions to activate transducin. To do this, the crystal structure of both rhodopsin and R* must be determined. Few membrane proteins have been successfully crystallized, so this is not a trivial undertaking. Two- or threedimensional crystals of rhodopsin must be prepared that are well ordered, to produce a high-resolution structure. Rhodopsin must be purified to homogeneity and the appropriate detergent(s) selected for crystallization experiments. Long-term thermal stability of the rhodopsin-detergent complex must be achieved in the presence of a precipitant. Two-dimensional crystals may offer advantages in investigating the structure of R*, but the structure obtained may be limited in resolution. It is necessary to work with rhodopsin in the dark, unless suitable light-stable retinal derivatives are developed. Protein engineering of rhodopsin offers attractive opportunities to improve its ability to crystallize, but is presently hindered by the absence of a high-yielding expression system. Knowledge of the structure of rhodopsin will have general importance. Because rhodopsin is a member of the family of G-protein-coupled receptors, knowledge of the structure and the mechanism of action of rhodopsin suggests by analogy how other members of the receptor family may function. Keywords: detergent; C-protein-coupled receptor; membrane protein; membrane protein crystallization; photoreceptor; protein crystallization; protein structure; retinal; rhodopsin; X-ray structure 1. Introduction1 What does rhodopsin do as a photoreceptor protein, and how does it do it? The answer to these questions will provide the details of rhodopsin's structure, and how it acts to carry out the functions of a photoreceptor protein. We know the details in rough outline - the folding and insertion of the opsin polypeptide chain into the membrane, its binding of 11-cis retinal, the vectorial transport to the outer segment, absorption of a photon by the retinal, the change in conformation of the protein, and the subsequent binding and activation of transducin. But how is this all actually achieved at the molecular level? How does the protein primary structure adopt the appropriate conformation to accomplish these results? To answer this it will be necessary first to know the details of rhodopsin's structure in its inactive state in the dark, and its structure when activated by light, in its signaling state. It would be important to determine the structure of Metarhodopsin I (which contains all-trans retinal but cannot activate transducin) and the difference between Metarhodopsin I and Metarhodopsin II (R*). These are necessary and achievable goals that will provide the baseline for understanding rhodopsin function. Rhodopsin is more easily obtained in the amounts needed for physical studies than other members of this large class of G-protein-coupled receptors. By analogy, insights into the function of rhodopsin should be © J995 Cambridge University Press 0140-525X195 $9.00+.10 valuable for understanding the signaling mechanism of the entire class of receptors. Mutants of rhodopsin, naturally occurring and otherwise, have proven useful in demonstrating what particular regions or amino acids in rhodopsin are most functionally significant (see Berson 1993; Dryja 1992; Khorana 1992; Nathans 1992; Oprian 1992). This important area must be left for consideration elsewhere. In the present target article I will discuss the prospects for obtaining a highresolution atomic structure for rhodopsin, which I believe will be critical for understanding the function of Gprotein-linked receptors generally. 2. What do we think we know about the structure of rhodopsin? There has been a virtual explosion of information about rhodopsin during the past twenty years. Original references to the literature are to be found in several recent reviews (Applebury 1991; Applebury & Hargrave 1986; Chabre 1985; Chabre & Deterre 1989; Findlay 1986; Hargrave & McDowell 1992; 1993; Khorana 1992; Liebman et al. 1987; Nathans 1992). Although many gaps in our knowledge have been filled, we are still a long way from the goals outlined above. We view rhodopsin as an intrinsic membrane protein, with one-half of its mass 403 Hargrave: Rhodopsin structure and function Ac-N- Figure 1. A topographic model for bovine rhodopsin in the rod cell disk membrane. Rhodopsin's polypeptide chain is shown traversing the lipid bilayer 7 times yielding 7 hydrophobic helical segments (I to VII) that are separated from each other by hydrophilic segments. Loops il-t4 and the carboxyl-terminal sequence face the cytoplasniic surface; loops el-e3 and the aminoterminal sequence are sequestered in the disk membrane lumen. (Based on Dratz & Hargrave 1983; Ovchinnikov et al. 1988a). embedded in the lipid bilayer and the remaining half approximately equally distributed between the two hydrophilic surfaces facing the cytoplasm and the intradiscal environment. We believe it to be topographically organized so that the polypeptide chain traverses the lipid bilayer seven times, alternately exposing hydrophilic sequences at membrane surfaces and burying hydrophobic sequences in the lipid bilayer (Fig. 1). The amino terminus, to which carbohydrate is attached at two sites, is located in the intradiscal environment; the carboxyl terminus, containing serine and threonine residues that may be phosphorylated by rhodopsin kinase, is exposed to the rod cell cytoplasm. The transmembrane segments are thought to be largely helical in structure, although several of them may be irregular due to presence of a proline in their sequence. Rhodopsin's chromophore, ll-cis retinal, is attached by Schiff base linkage to the side chain of lysine 296 (Lys296), located midway in the membrane-spanning seventh helix. The retinal is enveloped in a pocket formed by the amino acids lining the interior surfaces of rhodopsin's helices. Further characteristic features emerge when the sequences of many vertebrate rhodopsins are compared. Present are two highly conserved cysteines that are involved in a stabilizing disulfide bond linking helix III with loop el. One or more cysteines are generally present in the C-terminal sequence following helix VII. These cysteines become palmitoylated and presumably insert into the lipid bilayer forming a fourth cytoplasmic loop region. Additional amino acids are found to be conserved in all rhodopsins studied to date. These conserved residues are good candidates for structurally-functionally essential residues that are required components of what it means to be 404 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 a rhodopsin. Some of these residues are also conserved in other members of the class of G-protein-coupled receptors (Probst et al. 1992) and may be assumed to be required for features that are common to members of the receptor class, such as membrane translocation or signal transmission. 3. What do we think we know about the function of rhodopsin? Within a millisecond of absorbing light and isomerizing its ll-cis retinal to all-trans, rhodopsin forms Metarhodopsin I (MI). This intermediate is not able to bind and activate transducin, but Metarhodopsin II (Mil) is (Fig. 2). Deprotonation of the retinylidene-Schiffbase accompanies the conversion of MI to Mil (reviewed by Hofmann 1986). This deprotonation event is essential for the activation of transducin. The protonated retinylidene-Schiffbase is stabilized by an ion pair with the carboxyl group of glutamic acid 113 (Glull3) (Nathans 1990; Sakmar et al. 1989; Zhukovsky & Oprian 1989). Studies with mutant rhodopsins show that the maintenance of the interaction between Lys296 and Glull3 is essential in keeping rhodopsin inactive. Opsins lacking either of these charged groups constitutively activate transducin (Robinson et al. 1992). Thus it appears, in part, that when MI loses the proton from its Schiff base, this breaks a key ionic interaction that allows it to assume the more open protein structure, MIL The cytoplasmic surface of rhodopsin and MI is unable to bind and activate transducin. But the conformational change that accompanies the formation of Mil causes a Hargrave: Rhodopsin structure and function 7 ATP Rhodopsin Kinase all-trans retinal 7ADP Figure 2. The rhodopsin cycle. Rhodopsin (R) is activated by light (hv) and thereby forms Metarhodopsin I (MI), which exists in equilibrium with Metarhodopsin II (Mil, which is equivalent to R*). R* causes transducin (T) to become activated (to T*) by CDP-» GTP exchange. R* becomes phosphorylated (R*(P,)7), allowing it to bind arrestin, thereby blocking the ability of R* to continue to activate transducin. Upon loss of all-trans retinal, phosphorylated opsin {OiP^) is dephosphorylated and rebinds 11-cw retinal, regenerating rhodopsin. rearrangement of surface residues that provides the appropriate surface array for binding transducin. Amino acids in loops t2, t'3, and t4 are involved in this proteinprotein recognition event (Konig et al. 1989; reviewed in Hargrave & McDowell 1993). Further molecular details concerning how rhodopsin achieves all of these transformations, remain to be elucidated. 4. What will most advance our knowledge of how rhodopsin acts as a photoreceptor? To better understand how rhodopsin acts to transduce the reception of a photon into activation of a G-protein, it will be necessary to have an understanding of rhodopsin structure at the level of atomic resolution. This will require both a knowledge of where rhodopsin's amino acids are located before absorption of a photon, and where rhodopsin's amino acids become relocated to following absorption of a photon, when rhodopsin becomes transformed to MI and then to its signaling state, MIL All proteins are stabilized by a variety of individual interactions of their amino acids, involving disulfide bonds, hydrogen bonds, ionic bonds, and Van der Waals's forces. In rhodopsin's inactive state these forces contribute to maintain a stable three-dimensional structure in which the protein is constrained from binding and activating transducin. All proteins have a certain range of dynamic movement at any temperature, but this must be quite small for vertebrate rhodopsin, since "noise" (signaling in the absence of photon absorption) is extremely low (Baylor 1987). The juxtaposition of amino acids in rhodopsin's signaling state serves to define the binding site for transducin. The particular three-dimensional array that permits this binding is not present in MI but is formed when rhodopsin adopts the Mil conformation. Thus, a knowledge of the structure of Mil will show what amino acid interactions have been broken in rhodopsin and have been formed to stabilize the receptor signaling state, MIL Such information would provide the boundary conditions and would be essential in beginning to understand how the isomerization of 11-cts retinal leads to the structural reorganization that allows binding and activation of transducin. 5. What are the prospects for obtaining a highresolution structure for rhodopsin? There are basically three methods for obtaining atomic level structural data for proteins: (1) X-ray diffraction of three-dimensional crystals (Kiihlbrandt 1988); (2) electron diffraction, electron microscopy, and image processing of two-dimensional crystals (Amos et al. 1982; Kiihlbrandt 1992); and (3) nuclear magnetic resonance (NMR). Rhodopsin, at 40 kDa, is too large for analysis by currently available NMR techniques and cannot be examined in the membrane by liquid-state NMR. That leaves crystallography. What are the prospects for getting a high-resolution structure for any membrane protein or protein complex? At present there are atomic resolution structures for more than 1,700 proteins, but only four classes of membrane proteins are represented: bacteriorhodopsin from Halobacterium halobium (Henderson et al. 1990); the bacterial photosynthetic reaction centers (Arnoux et al. 1989; Deisenhofer & Michel 1989; Feher et al. 1989) bacterial porins (Cowan et al. 1992; Weiss et al. 1991); and light harvesting complex from green plants (Kiihlbrandt et al. 1994). _ 6. What are the chances of getting crystals of rhodopsin? The chances are excellent! Several investigators have already devoted many years to this task and crystals have been obtained. Both two-dimensional (Corless et al. 1982; Demin et al. 1987; Dratz et al. 1985; Schertler et al. 1993; Schertler & Hargrave 1995) and three-dimensional crystals have been obtained (de Grip et al. 1992; Demin et al. 1987; Yurkova et al. 1990; E. Dratz, personal communication). Two-dimensional crystals, presumably of rhodopsin, have been induced to form in the frog disk membrane following extraction of some of the lipid with the detergent Tween 80 (Corless et al. 1982). The molecules in these crystals appear to be 20-25A in width and 70-80A in length and to have a cross-sectional area similar to that of bacteriorhodopsin. When observed, crystals made up ~5%-10% of the membrane. Although they were not definitively demonstrated to be composed of rhodopsin, it seems unlikely that they would be composed of minor membrane proteins. Two-dimensional (2-D) crystals of bovine rhodopsin that diffract to a resolution of about 25A have also been formed, using the Tween 80 extraction method (Demin et al. 1987). In a different study, twodimensional crystals of bovine rhodopsin were induced to form within the disk membrane in the presence of ammonium sulfate and an amphiphilic compound, chlorhexidine (Dratz et al. 1985). Analysis of these crystals by electron microscopy also showed a surface area for rhodopsin similar to that of bacteriorhodopsin, further supBEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 405 Hargrave: Rhodopsin structure and function porting the presence of seven transmembrane segments in rhodopsin. However, none of the crystal preparations obtained by these investigators were of sufficient quality to give higher-resolution data; for example, data that would allow individual helices of the protein to be resolved or the position and identity of individual amino acids to be defined. Visualization of the helices of bovine rhodopsin has been obtained by cryoelectron microscopy of twodimensional crystals formed from purified rhodopsin in phosphatidyl choline (Schertler et al. 1993). Rhodopsin was solubilized in n-octyl tetraoxy-ethylene (C8E4) and crystals were formed as the detergent was removed by dialysis. Analysis of the crystals as frozen hydrated specimens allowed collection of data adequate for calculation of a 9A resolution map. The map shows an elongated arcshaped feature flanked by four resolved peaks of density. Orientation of the helices is clearly different from that of bacteriorhodopsin. More recently, tubular structures containing rhodopsin crystals have been formed in good quantity by extracting frog disk membranes with Tween 80 (Schertler & Hargrave 1995). Electron micrographs of the frozen hydrated crystals allowed a projection structure to be calculated to 6A resolution. This showed an arrangement of helices similar to that in the map obtained previously for bovine rhodopsin (Schertler et al. 1993). Helices 4, 6, and 7 are nearly perpendicular to the membrane plane, but helix 5 is more tilted or bent than anticipated from the 9A map. The rhodopsin molecule can be described as a bundle of four tilted helices alongside three perpendicular helices that are arranged in a straight line. The next step will be to collect data from tilted specimens to allow calculation of a three-dimensional map so that helix tilt angles can be determined. Three-dimensional crystals of rhodopsin have been obtained over a pH range from 5.5 to 7.0, using five different detergents and two different precipitants (de Grip et al. 1992). Crystals have also been obtained from octyl polyoxyethylene (o-POE), using ammonium sulfate, sodium phosphate, or sodium citrate as precipitants (Yurkova et al. 1990). Unfortunately, the crystals formed to date have been too small, fragile, and disordered to allow high-resolution diffraction analysis. This also has been the experience of other investigators who have worked with rhodopsin (E. Dratz, personal communication). The crystals obtained from o-POE were needle shaped with maximum dimension 70 u,m X 70 |x?n X 1000 |im; too small for X-ray analysis (Yurkova etal. 1990). Such observations are certainly not unique to rhodopsin and appear to be experienced with the majority of membrane proteins. It is often difficult enough to obtain ordered crystals from well-behaved soluble, globular proteins, but these difficulties are compounded when dealing with non-water-soluble proteins that have to be handled in detergent solutions. The conditions that will eventually succeed in producing large, stable, and highly diffracting crystals of rhodopsin will probably be unique for rhodopsin and will have to be determined empirically. The particular properties and behavior of rhodopsin as a protein will dictate what conditions will eventually succeed. For that reason, detailed knowledge of rhodopsin's properties as a protein are a prerequisite to such an undertaking. The body of experience obtained from the crystallization of all pre406 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 vious proteins provides only guidelines for procedures to be used as the studies with rhodopsin proceed. We will consider the steps in the process of obtaining protein crystals and look for the most productive approaches that might be utilized with respect to rhodopsin. 7. Preparation of rhodopsin-containing membranes Cattle retinas are conveniently obtained in quantity from animals slaughtered under conditions minimizing brightlight exposure. From each retina it is often possible to obtain more than 700 fxg of rhodopsin. Purified rod cell outer segments that contain membrane-bound rhodopsin are conveniently obtained by a number of protocols that include homogenization, differential centrifugation, and density gradient centrifugation (McDowell 1993; Papermaster & Dreyer 1974). Following hypotonic lysis of the rod cell outer segments and washing at low ionic strength, membranes are prepared that contain nearly 95% of their protein content as rhodopsin plus opsin. The components other than rhodopsin are opsin, the "rim protein," peripherin/rds, ROM-1, remaining membrane-associated proteins, plasma membrane proteins, and phospholipids (reviewed in Hargrave & McDowell 1993; Molday 1989). These membrane preparations can be solubilized in any of a variety of detergents and submitted to chromatographic steps that lead to purified rhodopsin. 8. Purification of detergent-solubilized rhodopsin For purposes of crystallization it seems most desirable that a reproducible well-defined homogeneous preparation of rhodopsin is obtained, free of opsin and other components. There are two main methods in use for the preparative chromatography of rhodopsin; hydroxyapatite chromatography and lectin affinity chromatography. When rhodopsin is purified by calcium phosphate chromatography using the commercial detergent preparation Emulphogene, it shows a single protein band by electrophoresis and a spectral ratio of 280nm/498nm of 1.75 (Shichi et al. 1969). However, rhodopsin purified by this method contains variable amounts of lipid (Papermaster & Dreyer 1974). When chromatography on hydroxyapatite is carried out using the cationic detergent tridecyltrimethylammonium bromide (TrTAB), the protein binds to the column and can be conveniently washed free of lipid (Hong & Hubbell 1973). Rhodopsin of the same high-spectral purity is then eluted using a salt gradient and contains from 0.2 to 0.8 moles of phosphate (phospholipid) per mole of protein. One disadvantage of this method is that TrTAB is not commercially available and must be synthesized by the investigator. However, it is dialyzable, and rhodopsin purified in TrTAB can be conveniently prepared in another detergent by addition of a nondialyzable detergent to the rhodopsin TrTAB solution followed by dialysis (Hong & Hubbell 1973). Probably the most widely used method to prepare rhodopsin employs chromatography on concanavalin A Sepharose. Several methods have been described (de Grip 1982a; Litman 1982). Rhodopsin, in a variety of detergents, is applied to the affinity matrix in a buffer containing salts of Mg+2, Ca +2 , and Mn +2 (to stabilize the Hargrave: Rhodopsin structure and function bound concanavalin A tetramer). Washing with detergentbuffer removes lipids and contaminating proteins. Rhodopsin is eluted with a-methyl mannoside in detergentbuffer. Dialysis removes the sugar and presumably the heavy metal ions. One must be alert to possible contamination by variable amounts of divalent metal ions, especially in light of a report that zinc becomes tightly associated with rhodopsin (Shuster et al. 1992). Another contaminant that can be introduced by this purification method is concanavalin A itself. It can be removed by passing the purified rhodopsin over a column containing an antibody to concanavalin A (de Grip 1982a) or by mannose agarose affinity chromatography. However, a more attractive approach is to prepare rhodopsin in a mild detergent, such as octyl or nonyl glucoside, that does not cause concanavalin A to be removed from the column (Litman 1982). 9. Homogeneity of rhodopsin Purified rhodopsin should be assessed for homogeneity by sodium dodecylsulphate (SDS) polyacrylamide gel electrophoresis (SDS-PAGE). The stained protein band on SDS-PAGE is always broader and less sharp than that of most other proteins. This appears to be due to heterogeneity of glycosylation. Vertebrate rhodopsins contain two sites of glycosylation, but not all molecules are identical in their carbohydrate content. About 70% of the oligosaccharides on bovine rhodopsin contain Man3GlcNAc3, 10% Man4GlcNAc3, and 20% Man5GlcNAc3 (Fukuda et al. 1979; Margrave et al. 1984; Liang et al. 1979). The distribution of these three components between the two sites in rhodopsin is unknown. This heterogeneity could make a difference in the ability of differently glycosylated molecules to pack and interact in crystals and represents a potential source of crystal disorder. One approach to dealing with this source of heterogeneity would be removal of the oligosaccharides by endoglycosidase digestion (Plantner et al. 1991). Peptide-N-glycosidase F is capable of removing both oligosaccharide chains completely, leaving deglycosylated rhodopsin as the sole product. The rhodopsin species with 0, 1, and 2 oligosaccharide chains are easily distinguished by SDS-PAGE (Plantner et al. 1991). If required, reaction mixtures might be further purified by passing through a column of concanavalin A Sepharose, allowing passage of successfully deglycosylated rhodopsin molecules. When rhodopsin is separated by SDS-PAGE, a series of molecular weight bands is often generated corresponding to dimer, trimer, and higher molecular weight oligomers. This generation of a multiplicity of bands can be considered a unique characteristic of the protein, but is also a nuisance when analyzing the protein for homogeneity, since the presence of contaminating proteins can be obscured. Rhodopsin is thought to be a monomer in disk membranes (Cone 1972; Downer 1985). The production of oligomers by SDS-PAGE is an artifact. Oligomers are probably produced after solubilization of rhodopsin by conditions that promote rhodopsin-rhodopsin collisions that alter detergent-lipid interaction and promote interaction of rhodopsin monomers that leads to aggregation. Oligomer formation may be eliminated by incubating rhodopsin solutions or rhodopsin-containing membranes at low protein concentration (< 1 mg/mL) in the presence of high concentrations of SDS (5%) and high concentrations of reducing agent, avoiding elevated temperatures (using room temperature or 37°C), and avoiding the use of stacking gels that concentrate rhodopsin during PAGE (Papermaster & Dreyer 1974). Since cattle are rarely completely dark adapted, we must consider the possibility that rhodopsin prepared from such light-exposed retinas could contain phosphorylated opsin as a contaminant. The various species of phosphorylated rhodopsin can be conveniently detected by isoelectric focusing. Rhodopsin itself has a pi of 6.0, and the different phosphorylated species have progressively more acidic pis (Aton et al. 1984; Kiihn & McDowell 1977). It would be important to assess whether a rhodopsin sample, prepared for purposes of crystallization, contained species other than that of the unphosphorylated protein with a pi of 6.0. If such phosphorylated rhodopsins are found, they can be conveniently removed by passing the detergent-solubilized rhodopsin sample over a Fe +3 -Chelex column (Andersson & Porath 1986; J. H. McDowell, personal communication) or by anion-exchange chromatography of rhodopsin at neutral pH (W. de Grip, personal communication). Rhodopsin is also posttranslationally modified by palmitoylation of two adjacent carboxyl-terminal cysteine residues (Ovchinnikov et al. 1988a; Papac et al. 1992). Partial or complete loss of palmitate would result in a significant change in the hydrophobicity and probably also the organization of the carboxyl-terminal region of rhodopsin. Disorganization of the carboxyl-terminal 40 amino acids of rhodopsin could result in a major problem for crystallization. It is known that the cysteine thioester linkage is labile to hydroxylamine, alkali, and to reducing agents (O'Brien & Zatz 1984). It has been shown by O'Brien and colleagues that simply storing solubilized rhodopsin at 4°C for four days led to a decrease in fatty acid content from 2.26 moles to 0.83 moles/mole rhodopsin (O'Brien et al. 1987). The rate of loss is temperaturedependent and is accelerated with increase in temperature. This suggests that in crystallization experiments that take several weeks to complete, rhodopsin will be heterogeneous with respect to palmitate content and may lose one or both of its palmitates during the course of the experiment. In experiments reported by de Grip, it was found necessary to include 5 mM dithioerythritol in the rhodopsin buffer in order to prevent damage to rhodopsin due to detergent impurities (de Grip et al. 1992). Presence of this reducing agent and the long times for crystallization would be expected to lead to loss of palmitate from its thioester linkage to rhodopsin. 10. What are the characteristics of an ideal detergent for the crystallization of rhodopsin? An ideal detergent would provide an environment for rhodopsin that simulated its membrane environment so well that its properties in detergent solution would be the same as those measured in the disk membrane. Rhodopsin would be as thermally stable in this detergent as in the membrane. It would bleach with the same kinetics, form a stable opsin, and the opsin would be fully regenerable to BEHAVIORAL AND BRAIN SCIENCES (1995) 18.3 407 Hargrave: Rhodopsin structure and function formation. Too large a micelle may interfere with proteinprotein interactions. An unstable micelle leads to nonspecific hydrophobic interactions, aggregation, and precipitation. 11. Measurement of properties of rhodopsindetergent complexes CHO Figure 3. Model for a rhodopsin-detergent complex. Upper left: Cross-sectional diagram of a rhodopsin molecule encircled by a detergent micelle. Polar (P) surfaces of rhodopsin, including its carbohydrate chains (CHO) interact with the aqueous environment and with detergent head groups (DH). Buried hydrophobic regions interact with hydrophobic detergent tails (DT). Lower right: Two rhodopsin-detergent complexes are shown forming interactions between their polar surfaces that presumably are required for crystal formation. Larger detergent micelles would make these protein-protein interactions less effective. Figures based on (Michel 1991). rhodopsin upon binding 11-cts retinal. No such detergent has yet been found. In addition, such an ideal detergent must have other important properties. An appropriate detergent for protein crystallization must be a defined chemical substance that is available commercially in pure form or that can be synthesized in quantity economically and relatively easily. It must be pure and chemically stable. The detergent should have good solubility at room temperature and at 4°C, and over a reasonable range of pH and ionic strength. Although the ability to solubilize efficiently membrane-bound rhodopsin would be helpful, this is not required, since a detergent can usually be exchanged. It is generally assumed that detergent molecules form a ring around solubilized membrane proteins, binding to hydrophobic surfaces previously occupied by lipid fatty acyl chains (Fig. 3). This leaves the hydrophilic ends of the protein exposed to the aqueous environment. The detergent should produce small monodisperse micelles and create the smallest stable protein-detergent complex possible (Garavito 1991). Such a small complex might be expected to enhance ionic interactions between hydrophilic protein surfaces that may promote good crystal 408 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 Numerous detergents have been used in the study of the properties of rhodopsins (de Grip 1982b; de Grip et al. 1992; Fong et al. 1982) and a variety of parameters have been measured to assess the characteristics of the rhodopsin-detergent complexes. Since crystallization trials may require two to four weeks, it is the stability, in particular the long-term stability of the rhodopsindetergent complex, that is most pertinent to the consideration of its crystallization. Thermal stability is the primary parameter that has been measured to assess the stability of the rhodopsindetergent complex. In one study the rhodopsin-detergent complexes were heated at various temperatures and the denaturation of rhodopsin followed by measurement of decrease in absorption at 500 nm (de Grip 1982b; de Grip et al. 1992). Temperatures at which the half-time of denaturation is 10 min for each detergent were determined (Table 1). Among the detergents tested, the one imparting the greatest stability to rhodopsin is Pdodecylmaltose. Table 1. Thermal stability of bovine rhodopsin and opsin in detergents i CMC" 5O Detergent" (mM) rhodopsin (°C) P-l-octylglucose a-1-octylglucose a-1-octylmannose C 8 P0E C8HESO P-1-nonylglucose C9POE C9-N-methylglucamide P-1-decylmaltose C1OPOE C10-N-methylglucamide P- 1-dodecylmaltose p-1-dodecylphosphomaltose C 12 E 8 C 12 E 10 CHAPS CHAPSO 23.0 13.0 1.5 8.5 18.0 6.5 2.2 16.5 1.7 0.7 3.5 0.2 3.5 0.06 0.02 6.5 7.0 50 53 52 42 42 53 49 51 56 52 51 60 48 48 50 56 56 ^1/2 opsin (min) =10 ND« 40 ND ND 20 ND ND 90 15 85 >40h ND ND 140 35 h 20 h "Detergent concentration was 30 mM in PIPES buffer pH 6.5. ^Critical micelle concentration. °^5o presents the temperature where the half time of rhodopsin denaturation, measured as 500 nm decrease, amounts to 10 min. ''Half life of opsin (measured as regenerability with 11-cts retinal) at 20°C. C ND = not determined. Note: Reproduced from de Grip et al., 1992. Hargrave: Rhodopsin structure and function Table 2. Calorimetric parameters of thermal denaturation of rhodopsin and opsin in detergent solutions Unbleached ROS disk membranes Bleached ROS disk membranes AHca. Detergent Concentration (mM) AHcaI (kcal/mole) CO disk membranes digitonin dodecyl maltoside octylglucoside TrTAB 10 6 90 200 172 155 158 122 109 71.7 62.5 61.8 56.0 39.0 ± ± ± ± ± 3 12 4 1 13 ± ± ± ± ± .2 .2 .1 .3 .1 (kcal/mole) T,n (°C) 129 ± 3 50 ± 3 54 ± 8 ND° ND 55.8 ± .4 44.5 ± .1 39.1 ± .1 ND ND "ND = not detectable. Thermal stability of rhodopsin has also been measured by differential scanning calorimetry (Khan et al. 1991; McDowell et al. 1992; Miljanich et al. 1985; Shnyrov & Berman 1988). By programmed heating of rhodopsin in the membrane and in detergent solution, temperatures of denaturation are obtained (Table 2). Such data show that even the mildest detergent tested, digitonin, falls far short of offering the protective environment that is available to rhodopsin in its native membrane. Long-term stability of the rhodopsin-detergent complex under conditions used in crystallization has been examined by de Grip et al. (1992). They examined rhodopsin for maintenance of spectral integrity, structural stability (avoidance of formation of lower molecular weight fragments measured by SDS-PAGE and immunoblotting), and formation of crystals when rhodopsin/ detergent/precipitant solutions were kept at 20°C for two to three months. Some fragmentation of rhodopsin occurred when detergents or precipitants contained oxyethylene units. This suggested that the degradation was caused by peroxidation, which seems quite likely because it was subsequently eliminated by inclusion of a reducing agent. One wonders whether scrupulous purification of the detergents/precipitants to remove peroxidants, the choice of more stable detergents, or inclusion of vitamin E or butyl-hydroxytoluene as an antioxidant, and an argon atmosphere (Farnsworth & Dratz 1976) might prove helpful and eliminate the need for inclusion of a reducing agent (which may lead to depalmitoylation, as discussed above). 12. Small amphiphiles are often helpful in crystallization of membrane proteins Considerable success has been achieved in crystallization of membrane proteins by the addition of l%-5% of small amphiphilic compounds to the protein-detergent complex. Mixed micelles are formed in which the amphiphiles intercolate into positions that would be occupied by larger detergent molecules. They reduce the size of the protein-detergent complex that is thought to maximize the hydrophilic protein-protein interactions needed for good crystal lattice formation (Michel 1991). Of the more than 100 different compounds investigated, threo1,2,3-heptane triol and benzamidine have been the most successful in trials with bacteriorhodopsin and the lightharvesting complex. However, inclusion of the am- phiphiles is not essential, since the proteins have been successfully crystallized under other conditions in their absence. Other investigators have found that amphiphiles improved the solubility of their membrane protein and influenced the type of crystal formed (Allen & Feher 1991; Garavito 1991). Crystals of a reaction center complex that showed a resolution of 10A were improved to 7A when 1,6-hexanediamine was added. In this instance it has been suggested that the diamine acted as a bridge stabilizing the protein micelles in the crystal lattice (Welte & Wacker 1991). Although the addition of small amphiphiles has been applied to the formation of 3-D crystals of rhodopsin, conditions have not yet been found in which they yield an improvement (de Grip et al. 1992; Demin et al. 1987). However, the quality of 2-D rhodopsin crystals in the membrane was improved by the inclusion of the surface-active agent, chlorhexidine (Dratz et al. 1985). 13. What are the prospects for obtaining twodimensional crystals of rhodopsin that will yield a high-resolution structure? Currently available methods are capable of yielding highresolution structures from well-ordered two-dimensional crystals of membrane proteins. Given the perfect microscope and the perfect membrane specimen, atomic resolution can be achieved. Bacteriorhodopsin is the classic example where a structure has been obtained with a resolution of 3.5 A in the plane parallel to the membrane (Henderson et al. 1990). Interpretation of the structure has allowed assignment of 21 amino acids from all 7 helices that contribute directly to the environment of the retinal, allowing a proposal to be made for the location of amino acids that constitute the proton channel. This is the type of information that would be very useful for vertebrate rhodopsin. A wide variety of examples of membrane proteins have been examined by electron microscopy and image analysis. The two-dimensional crystals studied thus far fall into three categories: (1) proteins that occur naturally in a semicrystalline array; bacteriorhodopsin in purple membrane, gap junctions from hepatocytes (Unwin & Zampighi 1980); (2) crystalline arrays that can be produced from membranes in vitro; cytochrome oxidase (Vanderkooi 1974), rhodopsin (Corless et al. 1982; Dratz et al. BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 409 Hargrave: Rhodopsin structure and function 1985), acetylcholine receptor (Toyoshima & Unwin 1990), H,K-ATPase (Hebert et al. 1992); and (3) proteins crystallized from protein-detergent complexes or protein-lipiddetergent complexes, as two-dimensional sheets; bacterial porin (Dorset et al. 1983), cytochrome reductase, cytochrome b-cl complex (Hovmoller et al. 1983), lightharvesting complex (Kiihlbrandt 1984), photosynthetic reaction center (Miller & Jacobs 1983), and NADH dehydrogenase (Boekema et al. 1986). Vertebrate rhodopsin exists in a highly fluid membrane and does not naturally form a crystalline array. Because the lamellar part of the rod cell disk membrane is nearly pure rhodopsin in lipid, it is attractive to attempt to induce rhodopsin to form a crystalline array within its native membrane (Corless etal. 1982; Demin etal. 1987; Dratz et al. 1985). However, the level of resolution obtained by that approach has not come close to what is required. Improvements have been made by examining unstained membranes and by use of improved methods for data analysis (Schertler et al. 1993; Schertler & Hargrave 1994). However, substantial improvements in resolution require that the crystals be larger, which may require that they be formed by a different approach. The most versatile approach to the formation of 2-D crystals is to reconstitute detergent-solubilized proteins into a lipid environment formed during the slow dialysis of detergent. With this approach there is no control over the type of detergent, the amount and type of lipid, and the method and rate of removal of detergent. Rhodopsin's physical and photochemical properties have been studied in a variety of different lipids (Deese et al. 1981; de Grip et al. 1983; Mitchell et al. 1992; Ryba & Marsh 1992). Based on these and other studies it may be possible to choose a lipid environment that will be more conducive to the formation and stabilization of rhodopsin in a crystalline array. 14. It may be easier to study all forms of rhodopsin in two-dimensional crystals It is not only possible to examine 2-D crystals of membrane proteins at low temperature, it may be desirable. The best electron microscopic image yet obtained from bacteriorhodopsin was from a sample at the temperature of liquid helium (Henderson etal. 1990). Electron microscopic data is frequently taken from frozen or refrigerated samples. Studies of the neutron diffraction of an intermediate in the photocycle of bacteriorhodopsin involved measuring it at -180°C (Dencher et al. 1989). Low temperatures have been used to isolate selectively the spectral intermediates in bleaching of rhodopsin. Various intermediates can be produced by irradiation at low temperature followed by warming to the temperature at which the intermediate is stable. MI can be produced by warming to a temperature between — 40°C and - 15°C, and Mil can be formed over the range - 15°C to 0°C. Thus, by photolyzing the rhodopsin 2-D crystal and subjecting it to the appropriate temperature it should be possible to form these structural intermediates. By immediately freezing in liquid ethane and maintaining the crystals at liquid nitrogen temperature, the intermediates should be preserved for examination by electron crystallography. Difference maps between two rhodopsin forms 410 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 should indicate the areas where the rhodopsin polypeptide chain has undergone a change in conformation in proceeding from rhodopsin to MI and from MI to Mil. Such an approach has been successfully applied to detection of structural changes between the ground state and the M intermediate of bacteriorhodopsin (Subramaniam et al. 1993). 15. The objective: Three-dimensional crystals of rhodopsin X-ray crystallography of three-dimensional crystals has produced the highest resolution structural data for proteins. Only four classes of membrane proteins have thus far yielded crystals of sufficient order to produce reasonably high-resolution structures. The successful methods for production of crystals from membrane proteins have been adaptations of methods used for soluble globular proteins. Concentrated solutions of purified membrane proteins in detergent may form crystals in the presence of increasing concentrations of precipitants such as ammonium sulfate or polyethylene glycol. Critical parameters such as the choice of detergent have been discussed earlier. Here I wish to discuss the experimental problems introduced by the intrinsic nature of rhodopsin itself: its extreme sensitivity to light. 16. Working with rhodopsin in the dark To produce and analyze crystals of rhodopsin demands that all aspects of the process be conducted under dim red light (light of > 620 nm) or by infrared image converter. This means that all of the crystallization trials, selection and examination of crystals, mounting, and X-ray analysis be conducted under darkroom conditions. Such work is not only tedious but has been reported to lead to failure in identifying crystals (de Grip et al. 1992). Eventually another problem will arise when the crystals are examined by X-ray analysis. Interaction of the measuring beam with water in the membrane sample causes fluorescence - light that would be expected to bleach the rhodopsin being examined. Such considerations have led investigators to attempt to produce light-insensitive rhodopsins. 17. The search for light-stable rhodopsin: The nonisomerizable chromophore Locking retinal into place so that it would not be photosensitive is one approach to handling rhodopsin more conveniently for crystallization and analysis. What is desired is a rhodopsin that will not be subject to a change in structure upon light exposure. Such a rhodopsin might also have greater long-term stability, enhancing its ability to withstand multi-week crystallization times. Opsin will combine with a number of isomers and derivatives of retinal to yield visual pigments of varying stability (Derguini & Nakanishi 1986). To be optimally useful for crystallization trials, any visual pigment formed by combination of a retinal derivative with opsin must have the following characteristics: (1) minimal perturbation of its native structure; (2) long-term stability in detergent; (3) insensitivity to light, as measured by main- Hargrave: Rhodopsin structure and function tenance of spectral integrity; and (4) insensitivity to attack by hydroxylamine. (This is a measure of the native conformation of the protein-retinal complex. Hydroxylamine attacks the retinylidene-Schiff base linkage when it becomes accessible to the aqueous environment. The retinylidene-Schiff base is unavailable for attack in native rhodopsin.) Ring-locked derivatives of retinal designed to eliminate photoisomerization about the 11-12 double bond have been investigated as possible chromophores that would yield a rhodopsin insensitive to photoisomerization. Such compounds have been produced with 5-membered (Ito et al. 1982), 6-membered (Bhattacharya et al. 1992; van der Steen et al. 1989), 7-membered (Akita et al. 1980), 8- and 9-membered rings (Hu et al. 1994) built around the retinal 11-12 double bond (Fig. 4). Each of these compounds (except the 9-membered one) has been successfully recombined with opsin, and the pigments formed (Rh5, Rh6, Rh7, and Rh8 respectively) have been the subject of many interesting studies (de Grip et al. 1990; Fukada et al. 1984; Hu et al. 1994; Zankel et al. 1990). However, pigments Rh5 and Rh7 have been subject to attack by hydroxylamine and are unstable in digitonin solution. By contrast, Rh6 is stable to hydroxylamine in the dark in detergent solution and is nearly as thermally stable as rhodopsin itself. Although it has unusual photostability, it does bleach in detergent solution at a rate 0.6% that of rhodopsin (Bhattacharya et al. 1992; de Grip et al. 1990). The rhodopsin analog has greatly reduced yet measurable activity in stimulating enzymes in the phototransduction pathway and in serving as a substrate for rhodopsin kinase. It is probable that all of the 11-cis ring-locked chromophores photoisomerize to a limited extent about the 7- and 9-double bonds, and that during this photoisomerization the Schiff base becomes transiently accessible for hydrolysis. To restrict further the potential for photoisomerization, a second ring be- Figure 4. Structures of different retinals of interest for rhodopsin function and crystallography. (1) 11-cis retinal; (2) all-trans retinal; (3) Ret-5, 5-membered ring-locked retinal; (4) Ret-6, 6-membered ring-locked retinal; (5) Ret-7, 7-membered ring-locked retinal; (6) 6-membered ring-locked retinoyl fluoride; (7) Ret-6 with an additional ring-linking carbons 9 and 11. tween the retinal carbons 9 and 11 has been introduced (Fig. 4, compound 7). Unfortunately the photosensitivity of the rhodopsin pigment formed with this retinal is not further reduced when compared to Rh6 (W. de Grip, personal communication). 18. Truly light-stable rhodopsin? There have been two different approaches taken to capitalizing on the greatly enhanced light stability of rhodopsin containing the 6-membered ring-locked retinal. One approach has been chemical - to strengthen further the association of the retinal to rhodopsin by covalent attachment (van der Steen et al. 1989). The other approach has been biological - to engineer the protein so that it is less susceptible to bleaching (Ridge et al. 1992). A promising compound for locking rhodopsin in a photo-insensitive state is the acid fluoride of the 6-membered ring-locked retinal (van der Steen et al. 1989). It reacts with opsin to yield a blue-shifted rhodopsin analog with 390nm absorbance maximum that is thermally stable, nonbleachable, and is completely inactive in signal transduction (de Grip et al. 1990). It can be heated in detergent at 60°C (under conditions that denature rhodopsin with t1/2 < 1 min.) with complete stability. The rhodopsin analog can be illuminated for 1 hr. under conditions that bleach rhodopsin with t1/2 = 15 sec, without loss of absorbance (van der Steen et al. 1989). Such photo- and thermal stability appears to come not only from a good fit to the retinal binding site and stabilization toward photoisomerization, but from the irreversible amide bond that replaces the normally hydrolyzable Schiff base. However, this rhodopsin analog has been difficult to purify (W. de Grip, personal communication): the acylation reaction is slow and incomplete, and the acylation must be performed on the completely reductively methylated protein in which all lysines except Lys296 are methylated. This leads to heterogeneity and difficulty in purification. Another disadvantage is that in formation of the retinyl-amide linkage, the protonated Schiff base is destroyed; thus the ion-pairing that is an important part of rhodopsin structure and function cannot be observed. Although crystals of this derivative might be useful in examining many overall features of the rhodopsin molecule, it would differ in important details in the region of the retinal attachment site. A novel alternative approach to stabilizing further the Rh6 rhodopsin has been to engineer the protein to make it less photosensitive (Ridge et al. 1992). Previous studies had identified a number of amino acids that were located in the retinal binding pocket. Twelve mutant rhodopsins that exhibited spectral differences in their combination with 11-cis retinal in the ground or excited state were recombined with the cyclohexene-locked retinal. One such mutant rhodopsin, Trp265Phe, was completely stable to illumination. The mutant Rh6 rhodopsin was stable to hydroxylamine and was inactive in phototransduction or as a substrate for rhodopsin kinase (Ridge et al. 1992). The lack of photosensitivity of this mutant resulting from a single amino acid mutation points to an important role of Trp265 in the signal transduction mechanism. Mutant Trp265Phe Rh6 would appear to be an ideal candidate for crystallization trials. However, one must not overlook the BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 411 Hargrave: Rhodopsin structure and function need to produce sufficient quantities of the mutant rhodopsin in a large-scale expression system and the requirement for the synthetic retinal for regeneration of rhodopsin. Although expensive and laborious, the approach appears quite promising. Although many of the aforementioned approaches to the crystallization of rhodopsin appear promising, additional approaches still need to be considered. 19. Rhodopsins of different species may have advantages for crystallization There are many instances in which a protein from one species will fail to yield crystals of good quality and order, but the protein that differs in only a few amino acids from a closely related species will crystallize beautifully. For many of the glycolytic enzymes, the protein from yeast or from rabbit muscle has been quite satisfactory, but for others it has been necessary to seek alternative sources such as crayfish or chicken (Campbell et al. 1971). Although the difficulty in obtaining good crystals of bovine rhodopsin is probably related to its characteristics as a membrane protein, it would be good to consider that rhodopsins from pig, horse, or sheep (Findlay 1986) could have small differences that are important in successful crystal formation. The thermal stability of rhodopsin may be an important characteristic in successful crystal formation, as we have mentioned previously. Rhodopsins from organisms that experience high body temperatures could offer this advantage. Examples include the desert iguana lizard Dipsosaurus dorsalis that has the highest known body temperature for a vertebrate, at 47°C (McFall-Nagi & Horwitz 1990), and the Saharan silver ant Cataglyphis bombycina, whose body temperature can reach 53.6°C (Wehner et al. 1992). Gaining information about the properties of rhodopsins from these exotic species offers special challenges. However, if study of these rhodopsins enhances our understanding of rhodopsin's stability, this would be a valuable addition to our knowledge of rhodopsin structure-function relationships. Invertebrate rhodopsins may have advantages for crystallization. The primary structures of octopus (Ovchinnikov et al. 1988b) and squid (Hall et al. 1991) rhodopsins have been determined, and many of their biochemical properties have been investigated. The proteins can be obtained in sufficient quantity. Invertebrate rhodopsins are more photostable than vertebrate rhodopsins inasmuch as their chromophore, 11-cw retinal, does not dissociate from its binding site following photoisomerization. A stable metarhodopsin is formed that is then converted back to rhodopsin upon reception of another photon of the correct wavelength. This ability of invertebrate rhodopsins to form a stable metarhodopsin makes them attractive candidates for more conveniently obtaining a structure for the signaling state of the receptor, metarhodopsin. One disadvantage, however, is their much lower thermal stability in detergents. The octopus or squid rhodopsins may have an additional characteristic that will be of importance in crystal formation. They are larger proteins, 50-51 kDa, due to a large C-terminal extension of about 150 amino acids (Ovchinnikov et al. 1988b; Hall et al. 1991). Although the 412 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 significance of this region for rhodopsin function is not clear, the presence of such a large globular addition to a predominantly membrane-embedded protein, may promote protein-protein association in the crystal lattice and may reduce some of the problems associated with crystallization from detergents. 20. Can rhodopsin be made to look more like a soluble protein, for purposes of crystallization? Based upon the membrane proteins that have formed the most well-ordered crystals to date, it would appear that a membrane protein that has the maximum amount of hydrophilic surface area would have the best chances of ordered crystal formation. One way to increase the amount of hydrophilic protein on rhodopsin's surface is to form a complex with a protein with which it interacts. Complexes of antibody-Fab fragments with their protein antigens have been successfully crystallized and have yielded high-resolution crystal structures (Mariuzza et al. 1987). There are a number of high-affinity monoclonal antibodies of well-defined specificity for rhodopsin that have been described (Adamus et al. 1991; Molday 1989). IgG class antibodies specific for rhodopsin's carboxylterminal sequence are easily accessible to their binding site on the opsin molecule either in its rhodopsin form (in the dark) or following light exposure. Such antibodies would be good choices to investigate the potential of this approach. This approach is not trivial since it is necessary to determine individually the proteolysis conditions suitable for production of a Fab fragment from each monoclonal antibody. Homogeneous Fab fragments must be produced that are suitable for crystallization purposes. Yields can be low, requiring many tens of milligrams of purified antibody as starting material. The Fab-rhodopsin complex should be formed and then purified to homogeneity. The complex must be sufficiently stable to remain intact during the purification and crystallization process. Although the validity of this approach has not yet been demonstrated with membrane proteins, the idea is appealing. The lectins concanavalin A and wheat germ agglutinin bind to rhodopsin's oligosaccharide chains. These lectins have been used for affinity purification of rhodopsin and for labeling rhodopsin for microscopic analysis. They are potential candidates for protein ligands that would impart a larger hydrophilic surface to the rhodopsin molecule. The lectins would bind to the intradiscal surface of rhodopsin and might serve to mask the heterogeneity of rhodopsin's oligosaccharide chains. However, concanavalin A is a tetramer whose tendency for subunit dissociation is likely to complicate purification and crystallization procedures. Protein engineering is one route that is free from the problem of dissociation of a bound protein ligand. By fusing the gene for rhodopsin with that for a desired protein, the resulting larger fusion protein is a single covalent entity. De Grip is exploring application of this method to rhodopsin (de Grip et al. 1992). The protein for fusion could be chosen from a list of many, such as lysozyme, that are models often used for testing new methods for protein crystallization. In theory, the poten- Hargrave: Rhodopsin structure and function tial of this approach is great, especially if one has good intuition or guidance concerning what fusion protein construct(s) will be most valuable to investigate. At present this approach suffers from the lack of an expression system capable of producing hundreds of milligrams of the required protein. Expression systems for rhodopsin have been described for COS-1 monkey kidney cells (Oprian et al. 1987), human embryonic kidney cells (Nathans et al. 1989), and an insect cell line (Janssen et al. 1991), but the levels of expression still need to be improved. Overexpression of membrane proteins generally has proven to be difficult (Schertler 1992). 21. An approach to the crystallization of Metarhodopsin II (R*) Although we suggested earlier that 2-D crystals might be made by freezing out Mil using conditions under which it is stable, this technique is unlikely to be practical for 3-D crystals. Crystallization as the complex with transducin may offer a more viable approach to the problem. This route is another variation on the attempt to crystallize rhodopsin in the presence of a bound protein ligand. fi* may be stabilized in its complex with transducin, if guanosine triphosphate (GTP) is absent, and if all traces of GDP are removed (Kiihn 1980). This leads to a complex fl* • Tc in which the T a nucleotide site is empty. This complex is stable "almost indefinitely" under conditions of physiological ionic strength (Bornancin et al. 1989). R* continues to bind retinal and remains spectroscopically in the Mil state, stabilized by the binding of transducin. However, for purposes of obtaining a crystal structure of this complex it would be necessary for the complex to remain stable in detergent under conditions suitable for purification and long-term crystallization. It is doubtful whether these stringent conditions could be met. The R* • Tc complex may be treated with hydroxylamine to form the Rc • Tc complex in which rhodopsin has lost its retinal and in which transducin is free of nucleotide (Bornancin et al. 1989). Any dissociation of this complex would be irreversible, leading to decay of the products and denaturation of the proteins. Thus the complex would have to maintain its strong association for periods of weeks under conditions suitable for crystallization. 22. Another approach to obtaining the crystal structure of R* The activated state of rhodopsin, R*, is that state of rhodopsin that binds to and causes the activation of transducin. It is possible that this state may be adequately simulated by more than one mutant form of rhodopsin. Disruption of the ion pair between the protonated Schiff base of Lys296 and the carboxyl group of Glull3 leads to formation of R*. When Glull3 is replaced by glutamine, the resulting rhodopsin is constitutively active once ll-cis retinal is removed from its binding pocket. Similarly, when Lys296 is replaced by site-specific mutagenesis, the.resultingopsin becomes constitutively active (Robinson et al. 1992). These observations suggest that if one of these mutant opsins were to be crystallized, the resulting structure might simulate the signaling state of rhodopsin. However, opsins are much less stable in detergents than rhodopsins, and the stability of the detergent complexes of these mutants may prove to be inadequate. In the above examples, key mutations in single amino acids in rhodopsin were enough to shift the conformation of R to that of R*. It is possible that mutations in other regions of rhodopsin might also result in formation of a constitutive mutant. It is known, for example, that rhodopsin loops t2, t3, and t4 participate in binding of transducin (Konig et al. 1989; reviewed in Hargrave et al. 1993) and that loops 12 and t3 participate in activating it (Franke et al. 1990; 1992). Recently, a single amino acid mutation in the carboxyl end of loop i3 of the a 1B adrenergic receptor was found to produce the constitutively active receptor (Kjelsberg et al. 1992). This shows that the adrenergic receptor is delicately poised and rather easily pushed over the energy barrier toward activation. It is likely that the activation of rhodopsin will not be so easily achieved. Nonetheless, our knowledge of the mechanism of activation of rhodopsin is in its early stages, and we should be alert to the possibilities of other participating residues whose alteration may readily lead to the formation of R*. Our present interest is in routes that may aid our understanding of rhodopsin structurefunction relationships. 23. Conclusion An essential basis for understanding how a protein functions is the determination of its primary, secondary and tertiary structures. A structure at the level of atomic resolution provides the framework on which the underStanding of function may develop. Rhodopsin is a particularly important and attractive target for study. From rhodopsin we will not only learn about the mechanism of phototransduction but also, by analogy, about the mechanism of action of receptors that couple to G-proteins. There is a large body of information about rhodopsin that will be of value in helping construct and evaluate experimental strategies. At present, the strategies leading to formation of 2-D crystals appear most versatile and most likely to succeed in production of structures at low to moderate levels of resolution. This will be extremely valuable in elucidating general features of rhodopsin's molecular architecture but may not yield the required level of atomic resolution to help in elucidating function. The techniques of mutagenesis are powerful and may be required to tailor a rhodopsin molecule more amenable to stable 3-D crystal formation. Full utilization of these techniques awaits development of a high-yielding expression system capable of producing the hundred milligram amounts of material that will be needed for innumerable trials. Whatever methods eventually succeed will triumph from application of a high degree of creativity coupled with detailed knowledge of the system, and an intense concentration of effort and resources over an extended period of time. ACKNOWLEDGMENTS I would like to thank numerous colleagues who have read this manuscript in draft form and have made many valuable suggestions, most of which I have adopted. During preparation of this manuscript Dr. Hargrave's research on rhodopsin was BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 413 Hargrave: Rhodopsin structure and function funded by grants from the National Eye Institute of the National Institutes of Health (EY06225 and EY06226), a grant from the International Human Frontier Science Program, and an unrestricted departmental award from Research to Prevent Blindness, Inc. Dr. Hargrave is Francis N. Bullard Professor of Ophthalmology. 414 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 NOTE 1. This manuscript was delivered at the conference Controversies in Neuroscience III: Signal Transduction in the Retina and Brain, at the Robert S. Dow Neurological Sciences Institute and Good Samaritan Hospital & Medical Center, Portland, Oregon, October 31-November 1, 1992. BEHAVIORAL AND BRAIN SCIENCES (1995) 18, 415-424 Printed in the United States of America What are the mechanisms of photoreceptor adaptation? M. Deric Bownds Laboratory of Molecular Biology, University of Wisconsin, Madison, Wl 53706. Electronic mail: bownds@macc.wisc.edu Vadim Y. Arshavsky1 Harvard Medical School and The Massachusetts Eye and Ear Infirmary, Boston, MA 02114. Electronic mall: vadim@macc.wisc.edu Abstract: This article evaluates each of the reactions known to be involved in visual transduction as a potential site for the regulation of light adaptation. Extensive evidence suggests that calcium acts as a feedback messenger at several different points and recent work suggests a role for cGMP in regulating the primary excitatory pathway. A conclusion is that adaptation is likely to be regulated by multiple and redundant mechanisms. The goal of future experimentation will be to determine the relative importance of each of these. Keywords: calcium; cyclic GMP; cyclic GMP phosphodiesterase; guanylate cyclase; light adaptation; phosphorylation; photoreceptor; rhodopsin; rod outer segments; transducin 1. Photoreceptor outer segments are a favorable preparation for studying adaptation processes Vertebrate photoreceptors show striking powers of adaptation, adjusting their gain to remain responsive to light transients as ambient light intensity increases over 3-5 log units. A primary locus of these adaptation processes is the outer segment portion of the photoreceptor, on which this target article focuses. It is clear that these structures determine the basic parameters that define subsequent stages of visual information processing in the retina (cf. Shapley & Enroth-Cugell 1984). The enzymes that regulate excitation and adaptation have been characterized mainly by studies on cattle and frog rod outer segments (ROS), and are described in several recent reviews (Chabre & Deterre 1989; Detwiler & Gray-Keller 1992; Hargrave & McDowell 1992; Hurley 1992; Koutalos & Yau 1993; Lagnado & Baylor 1992; Pfister et al. 1993; Pugh & Lamb 1990; Stryer 1991; Yarfitz & Hurley 1994; Yau 1994). The milligram to gram quantities of ROS necessary for purification and analysis of transduction enzymes are more easily obtained from cattle retinas. Larger amphibian ROS, although more difficult to prepare in quantity, have the advantage that pure and physiologically active suspensions can be obtained. This permits study of both the electrophysiological and biochemical correlates of adaptation. The composition of these structures is better characterized than that of any other primary sensory organelle: the mass is half protein and half lipid (Fliesler & Anderson 1983), 70% of the protein mass is rhodopsin, 17% is the G-protein transucin (Gt), and most of the © J995 Cambridge University Press 0140-525X195 $9.00+.10 balance is accounted for by proteins involved in cyclic nucleotide metabolism and protein phosphorylations (Hamm & Bownds 1986). Rhodopsin is composed of the protein opsin covalently linked to 11-cis retinal, and the photochemistry of this complex has been well studied. Thus the system allows precise quantitation of input (the number of rhodopsin molecules hit by light). Sophisticated modelling and kinetic analysis not yet achieved in other internal messenger systems is possible (Bownds & Thomson 1988; Dawis 1991; Forti et al. 1989; Lamb & Pugh 1992; Sneyd & Tranchina 1989). An outline of how excitation works is largely in place, and interest focuses now on adaptation. 2. What must be explained? Excitation and adaptation A single photon hitting one of the 3 X 109 rhodopsin molecules present in the disk membrane system of a dark frog ROS causes the closing of several hundred channels in the plasma membrane. This transiently halts the inward movement of several million ions. In the frog, the duration of the dark-adapted response is approximately four seconds with a time to peak of 900 msec (Baylor et al. 1979a). This excitation process is very stereotyped and reliable (Lagnado & Baylor 1992). If a step of background illumination is turned on, the response relaxes to a plateau level within seconds after an initial peak. This process is usually called background adaptation. Superimposed responses to test flashes become smaller and 415 Bownds & Arshavsky: Photoreceptor adaptation more rapid as this background light is increased over 3-5 log units. At high backgrounds, the amphibian rod response peaks at approximately 300 msec and turns off two to three times faster (Baylor et al. 1980; Fain 1976; Nicol & Bownds 1989). Theflashsensitivity (Sf), defined as the change in current per photon absorbed declines in a linear fashion as a function of the log of the background intensity. Until recently it had been supposed that background adaptation was a characteristic of the rods of lower vertebrates, and not displayed by mammalian rods. Yau and his collaborators, however, have now shown that adaptation behavior is observed in several warm-blooded animals, including rats, rabbits, cows, and monkeys (Nakatani et al. 1991; Tamura et al. 1991). If a significant amount of rhodopsin is bleached (greater than ~5%), the photoreceptor s sensitivity is at first abolished. In a subsequent dark period, sensitivity begins to recover as opsin regenerates to form rhodopsin but remains low (bleaching adaptation) as long as opsin is still present (cf. Kahlertetal. 1990). Recent work suggests that the underlying mechanism may be similar to that of background adaptation (Clack & Pepperberg 1982; Cornwall & Fain 1992). The major differences between bleaching and background adaptation may be that in the former the large amount of opsin present has residual excitatory activity and the efficiency with which the ROS captures photons is lowered because less rhodopsin is present. The residual excitatory activity can be inhibited by adding retinal or some of its analogs (Jin et al. 1993). Fain and Lisman (1993) have suggested that residual excitatory activity may underlie photoreceptor degeneration associated with vitamin A deficiency or rhodopsin mutations that are constitutively active (Robinson et al. 1992). 3. The basic excitation pathway Other articles in this review series have described reactions of the excitation cascade: hv -• Rh* -> G* -* PDE* -» cGMP i -• gNa Ca 1 They are a variation on the ubiquitous pattern seen in many other G-protein systems. Recent work has shown that olfactory receptors use an analogous system, except that cyclic AMP and/or triphosphoinositol are the relevant channel regulators (Breer & Boekhoff 1992; Firestein 1991; Ronnett & Synder 1992). The interaction of excited rhodopsin (Rh*) with many G-protein molecules (Gt, transducin) causes each to release GDP and bind GTP. Gt-GTP then activates cGMP phosphodiesterase (PDE) which hydrolyzes cGMP to 5'-GMP. The drop in cGMP caused by PDE activation causes closing of channels held open by cGMP in the darkness, halting a continuous entry of Na+ and Ca ++ ions. The system inactivates as Rh* is quenched (see below) and the GTP of Gt-GTP is hydrolyzed to GDP, stopping PDE activation. The decrease in Ca ++ entry activates a guanylate cyclase that enhances cGMP recovery. The excitation process thus increases GDP and 5'-GMP concentrations, and decreases GTP, cGMP, Na+, and Ca ++ concentrations. All of these products are putative feedback regulators. Most interest, however, has focused on Ca ++ and more recently cGMP. Because the rod metabolism efficiently buffers high energy phosphates (Groskoph et al. 1992), levels of GTP and ATP are maintained well above the 416 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 binding constants of the reactions that utilize them. Thus variations in their concentration occur only at very bright intensities and are unlikely to be regulatory in background adaptation (Biernbaum & Bownds 1985). Because we think it plausible that adaptation might occur at almost any step of the photoreceptor enzymatic cascade, this discussion outlines the activation and inactivation of the primary reactions, considering each as a possible locus for adaptational controls. We will emphasize, where possible, measurements made on more intact and concentrated ROS preparation at the low light levels (between 1 and 105 Rh*/ROS/sec) at which rods normally function and discuss technical problems that arise. It is important to point out that the majority of biochemical studies have used very unphysiological conditions: diluted suspensions of disrupted ROS (or purified enzymes) and illumination bleaching a substantial fraction of the rhodopsin present. 4. Excited rhodopsin catalyzes the activation of a G-protein (Gt, transducin) It is instructive to visualize rhodopsin excitation and subsequent transduction reactions with respect to the surfaces of single disk membranes. A typical amphibian ROS contains a stack of approximately 2,000 disk membranes that arise as evaginations of the ciliary plasma membrane during outer segment morphogenesis (Steinberg et al. 1980; Williams et al. 1988) and then become self enclosed (thus there are 4,000 disk membrane surfaces). The structure is continuously renewed throughout its lifetime. The disks occupy ~50% of the volume of the ROS; the balance is aqueous (Korenbrot et al. 1973). The surrounding plasma membrane contains the channels that ultimately are gated as a consequence of bleaching rhodopsin in the disk membranes. The ROS contains approximately 3 X 109 rhodopsin molecules (Liebman & Entine 1968), or roughly 106 rhodopsins/disk membrane surface. This rhodopsin accounts for 15-20% of the disk surface, with individual rhodopsin molecules being approximately 2 nm apart, and colliding every 1-10 (xsec. Several recent reviews provide relevant detailed information about rhodopsin structure (Khorana 1992; Nathans 1992). The concentrations of rhodopsin, Gt, and PDE in the ROS are approximately 6,000, 600, and 22 u,M, respectively, calculated with respect to the aqueous volume of the ROS (Dumke et al. 1994; Hamm & Bownds 1986). Figure 1 visualizes the relative abundance of the components. The square contains 1000 rhodopsin molecules (the smallest dots), 100 G, (grey dots), 4 PDE molecules (large dots) and one free cGMP molecule (the small black dot shown between disk and plasma membrane). Phospholipids of the bilayer, ~50 for each rhodopsin molecule, are highly unsaturated and provide a fluid environment for lateral interactions of the protein components (Fliesler & Anderson 1983). The number of channels (the large dot in the plasma membrane) is approximately 1 for every 10 of these frames. When one of the rhodopsin molecules in an ROS absorbs a photon, there is 50% probability that isomerization of the 11-cis retinal chromophore to the all-trans configuration will occur, and lead to the formation, on a Bownds & Arshavsky: Photoreceptor adaptation L Figure 1. Relative abundance of phototransduction cascade components in rod outer segments. millisecond time scale, of excited rhodopsin metarhodopsin II (abbreviated as Rh*) and generation of the photoresponse (Baylor etal. 1979b). Photometric measurement of the absorption shift caused by photon absorption can yield the number of rhodopsin molecules that form the metarhodopsin II intermediate that interacts directly with G t . In no other system can receptor activation be so accurately specified. Knowing the number of excited rhodopsin molecules permits us to calculate the gains of subsequent steps in the excitation pathway with respect to the initial input. We can then ask whether this gain changes during adaptation processes. It is unlikely that the initial excitation step is a site of adaptation. This would be the equivalent of saying that the quantum efficiency of bleaching changes during adaptation, and electrophysiological measurements suggest that the probability that photon absorption will generate a photoresponse is not altered by dim background illumination (Baylor et al. 1979b). It is possible, however, that control of the lifetime of metarhodopsin II (Rh*) could be a factor in adaptation and this will be considered below with the reactions that terminate its activation. Excited rhodopsin exposes within milliseconds a protein surface that binds to the G,-GDP, permitting the exchange of GTP for GDP on the a subunit. The crystal structures of both the GTP- and GDP-bound forms of the a subunit recently have been determined (Lambright et al. 1994; Noel et al. 1993). G ta -GTP (G*) is released from Rh*, accompanied by dissociation of the (i-"y subunit complex, and the process repeats many times (Fig. 2). Interaction between Rh* (an intrinsic membrane protein) and G, occurs via translational diffusion and collision on the disk surface (which contains 106 rhodopsin and 105 Gt molecules, Hainm & Bownds 1986). Because very little G t -GTP formation occurs in the dark, radioactive GTPot-32P or GTP-7-35S can be used to monitor its generation over physiological ranges of illumination that bleach 102 to 10^ rhodopsin molecules/sec. High gains (~40,000 Gf/Rh*) have been observed in electropermeabilized ROS that retain most of their complement of G, (GrayKeller et al. 1990). In this preparation rates of 200-400 Gt*/Rh*/sec have been obtained, while measurements of the light scattering changes which accompany G t activation have been interpreted to indicate rates as high as 1000 Gt*/Rh*/sec (Uhl 1990; Vuong et al. 1984). This discrepancy needs to be resolved. A curious feature is that G,-GTP can be generated not only by rhodopsin bleaching but also by mechanical disruption of ROS (Gray-Keller et al. 1990), or freezing and thawing (Klenchin & Bownds, unpublished data). The mechanism for this activation is not clear. Might the rate of G, activation be altered by adaptation chemistry? From a functional perspective it would make little sense at moderate background levels of light to have the onset of the photoresponse slowed by adaptation, for a change in light levels needs to be sensed just as rapidly as the response to the initial onset of illumination. What is relevant in the presence of background light is that the amplitude of the response per absorbed photon be smaller and that it terminate more rapidly. Thus one would expect mechanisms that determine inactivation times are more likely loci of adaptational control than those that determine initial kinetics. Two lines of evidence suggest that the initial kinetics of Gt activation may not be sensitive to the previous history of illumination. Kahlert et al. (1990) have shown that the presence of bleached rhodopsin does not alter the sensitivity or gain of a light-scattering signal recorded from a living bovine retina that is taken to monitor G, activation. For reasons that are not understood, similar signals cannot be recorded from amphibian retinas. Earlier reports also indicated that background illumination has very little effect on the initial kinetics of a superimposed flash response (Baylor & Hodgkin 1974; Lamb 1984). However, more recently Lagnado and Baylor (1994) have shown in recordings from truncated amphibian rods that reproducing the C a + + concentration decrease that occurs during background illumination lowers the initial slope of the flash response. Direct chemical measurements of the initial kinetics of Gt* production after a dim flash, and then after a dim flash superimposed on background illumination, might resolve the issue. Gt, as well as PDE and several other transduction enzymes, is covalently modified by lipid attachment. Farnesylation of the 7 subunit of G t enhances its activation, as does further methylation (Fukada et al. 1990; Ohguro et al. 1991; Perez-Sala et al. 1991). Neubert et al. (1992) have characterized four different N-terminal acylations of G ta . Anant et al. (1992) have demonstrated differential prenylation in vivo of the a and (3 subunits of PDE by farnesylation and geranylgeranylation, respectively. Disk Membrane Plasma Membrane GTP Figure 2. Activation of transducin by photoexcited rhodopsin. BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 417 Bownds & Arshavsky: Photoreceptor adaptation They also report in vivo farnesylation of rhodopsin kinase (Anant & Fung 1992). It has been suggested that these reactions might regulate enzyme activity, but it seems unlikely that they act on the time scale of excitation and adaptation. They are more likely to be long-term structural modifications whose purpose is to enhance membraneprotein association so that reactants are favorably poised to interact with each other in the two dimensional surface of the disk membrane. The Gt cycle completes when Gf-GTP converts to G ta GDP (see below) and reassociates with Gjj r Phosphoproteins are found in both frog ROS (Components I and II, Hamm 1990; Polans et al. 1979; Suh & Hamm 1988) and bovine ROS (Phosducin, Lee et al. 1987; 1990a) that are dephosphorylated after bright illumination and form complexes with Gp 7 . If this complex formation competes with reformation of G ta p 7 , a possible adaptation mechanism is suggested: depletion of the pool of G ta p 7 because its reassembly has been blocked. The phosphoproteins are present in an amount comparable with Gt. However, their subcellular distribution is not clear. Large amounts are found in the inner segment portions of rod receptor cells (Lee et al. 1988; 1990b). 5. Does adaptation modulate the duration of G, activation? This brings us to consider the processes that terminate the generation of Gf. Reversal of the PDE activation caused by G* in vitro is greatly delayed by the removal of ATP (Liebman & Pugh 1979; 1980), and the photocurrent in whole cell clamp recordings (Sather & Detwiler 1987) or truncated ROS (Nakatani & Yau 1988a) approaches physiological time scales (seconds) only if ATP is present. It is commonly assumed that the main function of ATP is to quench Gt* generation by serving as substrate in the phosphorylation of metarhodopsin II by rhodopsin kinase (Fig. 3). This light-dependent reaction leads to incorporation of up to nine phosphates per Rh* molecule (Wilden & Kuhn 1982). The idea that the effect of ATP is mediated by Rh* phosphorylation is confirmed by two independent lines of evidence. First, the effect of ATP can be observed only in the presence of rhodopsin kinase (Sitaramayya 1986; Sitaramayya & Liebman 1983a). Second, the ability of phos- 'GDP Figure 3. Inactivation of photoexcited rhodopsin by rhodopsin kinase and arrestin. 418 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 phorylated Rh* to activate G, is reduced (Arshavsky et al. 1985; Miller et al. 1986; Wilden et al. 1986). Arshavsky et al. (1987) have demonstrated that the rate of Gt activation decreases 1.5 to 3.3-fold as phosphate incorporation increases from 2 to 6 phosphates/rhodopsin, and that the inhibitory mechanism involves an increase in the time necessary for the activation of each bound G, rather than a reduction in the binding affinity of G t for phosphorylated Rh*. Rhodopsin phosphorylation is followed by binding of a 48kD protein, arrestin, a "capping" reaction that blocks access of G, so that no further excitation can occur (Wilden et al. 1986). Palczewski et al. (1992) have demonstrated that sangivamycin (an inhibitor of rhodopsin kinase) and phytic acid (an inhibitor of arrestin binding to phosphorylated rhodopsin) slow recovery of the photoresponse when they are dialyzed into gecko ROS. It is interesting to note they show also that high concentrations of sangivamycin cause changes in the light response that cannot be explained by selective inhibition of rhodopsin kinase, and suggest that other protein kinases, for example protein kinase C, may be needed for normal rod function. Hofmann et al. (1992) have recently shown that recycling of rhodopsin back to the active pool requires reduction of the retinal chromophore to retinol. Only then does arrestin dissociate, opsin becomes dephosphorylated, and opsin combines with 11-cis retinal to form rhodopsin again. The shut-off of rhodopsin activation must be very reliable and irreversible, so that only one burst of activity is set off by each photon absorption. Lagnado and Baylor (1992) make the point that quenching of rhodopsin by binding of arrestin to rhodopsin with one or two phosphates would not explain the constancy of the single photon responses observed. They suggest that reproducibility of the single photon response might be due in part to feedback control of rhodopsin's active lifetime. Can rhodopsin phosphorylation be measured in fact to occur before, or on the same time scale as, the turning off of the photo response? A positive answer to this question was obtained by Sitaramayya and Liebman (1983b), demonstrating in bovine ROS that bleaching approximately one rhodopsin in each photoreceptor disk causes incorporation of about 20 phosphates per bleached rhodopsin in the first two seconds after the flash. Since each Rh* cannot incorporate more than nine phosphate groups, this suggests that not only Rh* but also some nonactivated rhodopsin can be phosphorylated, confirming the initial report of Bownds et al. (1972). A slower rate of Rh* phosphorylation, about one phosphate per bleached rhodopsin in the first 1-2 seconds after a light flash, was documented for frog ROS by Binder et al. (1990). This lower stoichiometry does not necessarily reflect a difference between the two species because higher bleaching levels of about 1000 Rh* per disk were used in the frog study. It would appear then that incorporation of one or two phosphates per mole of rhodopsin is adequate for photoresponse quenching. Experiments of Bennett and Sitaramayya (1988) suggest that only one or two phosphates/Rh* may be sufficient to promote arrestin binding and maximally quench the interaction between rhodopsin and Gt. The residues initially phosphorylated have recently been identified as serines 338 and 343 and threonine 336 (McDowell et al. 1993; Ohguro et al. 1993; Papac Bownds & Arshavsky: Photoreceptor adaptation et al. 1993). The data are consistent with a role for rhodopsin phosphorylation in quenching Gt* generation, but a determination of the kinetic correlations between the time course of rhodopsin phosphorylation and the decay of G* generation has not been accomplished. An alternative view is that the crucial event in quenching Rh* is the binding of kinase, rather than the subsequent phosphorylation that is catalyzed by the kinase (Pulvermuller et al. 1993). In this model the phosphorylation would serve the function of insuring that the initial quenching was made permanent. However, the fact that ATP is required for quenching the cascade means that kinase binding by itself does not insure Rh* inactivation. Is modulation of rhodopsin phosphorylation a possible site of adaptation? Lagnado and Baylor (1992) suggest that calcium might influence Rh* shutoff, making it more stereotyped because it would be determined by the fall in calcium concentration that follows channel closure (see below). Further, increasing the rate of Rh* phosphorylation during background illumination could potentially cause acceleration of photoresponse recovery of the sort observed during background adaptation. Recently Kawamura and his collaborators have provided convincing evidence that just such a scenario may be appropriate, by showing that decreasing C a + + concentration shortens the duration of the photoresponse (Kawamura & Murakami 1991). (It should be noted that the physiological data on Ca + + regulation of Rh* lifetime has thus far been obtained in broken truncated ROS whose interior is exposed to a bathing medium.) Analysis ofphotoresponse recovery from bright flashes in intact cells leads Pepperberg et al. (1992; 1994) to argue that this lifetime does not change during adaptation, but rather that a later stage is involved. Kawamura's group has identified a protein, named recoverin binds and imparts C a + + sensitivity to rhodopsin kinase. The situation is not as clear with respect to cGMP as a putative feedback regulator of rhodopsin phosphorylation. Hermolin et al. (1982) and Shuster and Farber (1984) noted inhibition of rhodopsin phosphorylation by cGMP, but more recently Palczewski et al. (1988b) and Binder et al. (1989) found no effect. Another possible role for Rh* phosphorylation in light adaptation can be suggested: if a substantial portion of nonbleached rhodopsin were to become phosphorylated during dim background illumination, its excitation should result in photoresponses with lowered amplitude and accelerated recovery. The amplitude reduction would be expected from the reduced ability of phosphorylated Rh* to activate G t , while turnoff acceleration would result from faster arrestin binding to pre-phosphorylated rhodopsin. A hint that this might take place has come from measurements of rhodopsin phosphorylation in electropermeabilized amphibian ROS. At light levels in the operating range of the rod, bleaching one rhodopsin molecule can result in the incorporation of phosphate groups into several hundred dark rhodopsin molecules (Binder et al. 1990). The gain diminishes rapidly as each interdiskal space begins to receive more than one photon, and the high gain reaction is lost if outer segments are fragmented into smaller pieces. This high gain reaction is sensitive to calcium, being more active at 10 nM than at \xM calcium levels (Calvert and Bownds, unpublished observations). In permeabilized ROS, however, the highgain phosphorylation reaction saturates when it has acted on less than 1% of the total rhodopsin present. This would not significantly alter the probability of an ROS generating a response upon photon absorption. It: is possible, S-modulin (M. W. 23 kDa), that confers C a + + sensitivity however, that measurements on the living retina might on rhodopsin phosphorylation (Kawamura 1993; Kawamura et al. 1992). This protein turns out to be the frog homolog of bovine recoverin (Dizhoor et al. 1991), a Ca + + binding protein originally proposed to regulate guanylate cyclase (see below). The three-dimensional structure of recoverin recently has been determined (Flaherty et al. 1993). Gray-Keller et al. (1993) have shown that internal perfusion of gecko rods with gecko S-modulin as well as recoverin delays the recovery of the photoresponse. This correlates with the observation that at high C a + + concentration these proteins inhibit rhodopsin phosphorylation, perhaps by acting on rhodopsin kinase. The inhibition is relieved if Ca + + is lowered to the level caused by illumination (< 100 nM). This suggests that the lowering of C a + + that occurs during background adaptation should decrease the lifetime of Rh* and thus the duration of PDE activation, causing smaller and faster responses to superimposed flashes. Both earlier (Kawamura & Bownds 1981; Robinson et al. 1980) and more recent (Kawamura & Murakami 1991) experiments had shown in fact such an effect of C a + + on PDE activation. The data of Wagner et al. (1989) also suggest C a + + regulation of rhodopsin phosphorylation.'A light-scattering transient that they interpret as rhodopsin deactivation in ROS suspensions is accelerated as calcium is lowered from 1 to 0.1 u-M. Recoverin appears to be essential for the C a + + regulation, for Palczewski et al. (1988a) have found that C a + + does not act directly on rhodopsin kinase. Most recently Chen and Hurley (1994) have reported that show that the high-gain rhodopsin phosphorylation reaction alters a more significant fraction of the visual pigment present. Such measurements have not yet been done. The mechanism of the high-gain phosphorylation remains to be determined. Recent studies (Buczylko et al. 1991; Fowles et al. 1988; Palczewski et al. 1991) have shown that rhodopsin kinase, on binding to bleached rhodopsin, becomes able to phosphorylate rhodopsin C-terminal peptide fragments. Perhaps it also phosphorylates the C-terminus of some dark rhodopsin molecules. An alternative possibility is that another kinase acts on rhodopsin. Newton and Williams (1991) suggest a possible analogy with the (3-adrenergic receptor system, where both cAMP dependent kinase and (3-adrenergic receptor kinase phosphorylate the receptor (Roth et al. 1991), with the cAMP dependent kinase more important at low receptor occupancy and the (3-adrenergic receptor kinase acting at higher levels of receptor occupancy. The observation (Newton & Williams 1993) that rhodopsin is the major substrate of protein kinase C in ROS raises the possibility that this enzyme may play a role corresponding to the cAMP dependent kinase of the (3-adrenergic receptor system. Another possible mechanism leading to the accumulation of nonbleached phosphorylated rhodopsin in photoreceptor cells is an inhibition of the dephosphorylation reaction without blocking the regeneration of bleached rhodopsin by 11-cis-retinal. This possibility is discussed by Biernbaum et al. (1991), who demonstrated that a dim BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 419 Bownds & Arshavsky: Photoreceptor adaptation background illumination prevents rhodopsin dephosphorylation in a preparation of intact frog rods. 6. Excited G, activates PDE The next step in the phototransduction cascade is the action of Gt* on its effector, PDE. The PDE holoenzyme is an aP"y2 tetramer. The a and (3 subunits each contain one catalytic and one, or possibly two, noncatalytic cGMP binding sites (Charbonneau et al. 1990; Cote & Brunnock 1993; Gillespie & Beavo 1988; 1989; Li et al. 1990). The two identical 7 subunits serve as inhibitors of the enzyme (Deterre et al. 1988; Fung et al. 1990; Hurley & Stryer 1982). Full activation requires the binding of one Gf molecule to each of the two inhibitory 7 subunits. Each frog disk face contains ~3,700 PDE molecules, anchored by prenylation to the membrane surface. An excited Gf contacts one of these PDE molecules rapidly enough to activate it within milliseconds (Heck & Hoffman 1993). This reaction is occurring by lateral diffusion on the surface of the disk. Given that the binding constant between PDE and Gf is approximately 100-600 nM (Bennett & Clerc 1989), lower than the concentration of Gt* at the disk surface (which, from above, is > 1 u,M), essentially all of the Gt* has activated PDE within milliseconds. (In diluted fragmented ROS, a smaller fraction of Gf is successful in activating PDE [Liebman et al. 1987].) Pugh and Lamb (1993) point out that each of the activation reactions in the transduction pathway can be treated as a short first order delay stage, so that the time course of PDE activation is a delayed ramp, with slope proportional to light intensity; the initial delay is about 10-20 msec (Lamb & Pugh 1992). The PDE activation-inactivation sequence occurs in a fundamentally different context than the Rh* — Gt sequence. In the latter case the reactants are poised for reaction but silent in the dark ROS, awaiting the discharge induced by illumination. PDE activation, on the other hand, modulates an ongoing flux: PDE Cyclase GTP-•cGMP- GMP Both biochemical and electrophysiological studies (Dawis et al. 1988; Hodgkin & Nunn 1988) suggest flux rates on the order of one turnover of the entire pool of unbound cGMP per second, and Pugh and Lamb (1990) estimate dark levels of free cytoplasmic cGMP to be 4 u,M. This means that each interdiskal space has about 650 molecules of cGMP, all turning over once a second. Direct measurement of the dark PDE activity that underlies this flux is complicated by the dilution of reactants that occurs when ROS are disrupted to assay the enzyme. This disruption causes some dissociation of the inhibitory 7 subunit of PDE and thus increases dark background activity (cf. Arshavsky et al. 1992). Measurements performed under conditions preventing PDE^ loss from PDE catalytic subunits indicates that the PDE activity in the dark is at least 300-fold less than the activity of light-activated enzyme (Arshavsky et al. 1992). Taking this fraction of the light activity, the cGMP concentration in the dark as constant and equal to 4 JJLM, PDE concentrations as —22 jxM (see above), Vmax (in the light) —4700 420 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 s"1 and Km as - 9 5 u,M (Bownds et al. 1992; Dumke et al. 1994) Michaelis-Menten calculations yield a dark flux estimate for cGMP of less than 15u.M/sec. This number is reasonably close to the 4 u,M/sec estimated in vivo (Pugh & Lamb 1990). The effect of the large increase in PDE activity caused by light can be visualized by making a simple calculation of the amount of cGMP hydrolyzed after absorption of a single photon (i.e., 1 Rh*) in a frog rod. More precise calculations are reported in Lamb and Pugh (1992). Let us suppose that the rate of Gt activation is 1000 s"1 and that this rate is constant during the time of the photoresponse. Therefore, an average of 500 Gf molecules are present during the rising phase of the response (— 1 sec). Since the complete activation of a PDE molecule demands two Gf, the average number of PDE molecules activated during this time will be about 250 molecules. Taking Vmax of activated PDE as about 4700 turnovers/sec and Km as 95 u-M (Dumke et al. 1994), we can calculate that each activated PDE will hydrolyze about 200 cGMP molecules. The total amount of hydrolyzed cGMP then will be on the order of 50,000 molecules. This corresponds to the total cGMP content of approximately 80 interdiskal spaces or 4% of the whole pool of unbound rod cGMP. This is reasonably close to the length of ROS whose conductance is suppressed following a single photon absorption (see below). It is likely that a considerably smaller change in cGMP would be sufficient to suppress this conductance. Because gating of the channel by cGMP is cooperative, hydrolysis of only a fraction of the cGMP present might be sufficient to cause a large conductance change. Overall, it would seem more efficient for cGMP to drop by a small amount over a longer length of the ROS than to undergo a large change over a small length (cf. Lamb & Pugh 1992; Rispoli et al. 1993). Pugh and Lamb (1993) have pointed out that these kinetic parameters can generate very different relative changes in cGMP concentration if ROS volume is decreased, as in mammalian rods and cones. Equivalent cascade activation in a smaller ROS causes a larger fractional change in the cGMP pool. The PDE enzyme is so active that it depletes cGMP in the interdiskal space very rapidly. As a result, the diffusion of cGMP into the interior of the disk stacks can become rate limiting in conventional biochemical measurement of cGMP hydrolysis in ROS fragments. This can have the effect of raising the apparent Km of PDE as well as limiting its apparent maximal velocity. Recent studies show that true values for Km and kcat of PDE (—95 u,M and 4700 sec"1) are approached only when ROS membranes are completely disrupted into small vesicles (Dumke et al. 1994). Techniques for directly measuring these parameters in intact cells are not currently available. Perhaps the most accurate estimates of PDE activation kinetics in intact ROS come from reverse calculations based on the kinetics of the photoresponse (cf. Hodgkin & Nunn 1988; Pugh & Lamb 1993). 7. The turnoff of activated PDE is a putative point of light adaptation The action of Gf, like that of other G-proteins, has been thought to be terminated as its bound GTP is hydrolyzed Bownds & Arshavsky: Photoreceptor adaptation by an "intrinsic GTPase activity" (cf. Bourne et al. 1990; 1991). This activity measured in dilute suspensions has been much too slow (Baehretal. 1982; Fung etal. 1981) to explain the rapid turnoif of PDE in suspensions of bovine ROS (Sitaramayya & Liebman 1983b). More recent work has indicated that transducin GTPase under more physiological conditions is much faster than in reconstituted systems (Arshavsky et al. 1989; 1991; Dratz et al. 1987; Wagner et al. 1988). A number of factors appear to be responsible for G t GTPase regulation in photoreceptors. One is the activation of G, GTPase by PDE (Arshavsky & Bownds 1992; Arshavsky et al. 1991; Pag<§s et al. 1992; 1993), and more specifically its 7 subunit acting as a GAP (GTPase activating protein) (Arshavsky & Bownds 1992; Arshavsky et al. 1994). Acceleration of G t GTPase by PDE 7 requires the presence of ROS membranes. Antonny et al. (1993) have made the point that it is not observed in the soluble complex of G,-GTP with P D E r A second factor has been suggested by recent observations that the effect of PDE 7 becomes more pronounced as ROS membranes are concentrated (Angleson & Wensel 1994; Arshavsky et al. 1994). The nature of this second mechanism remains unclear. Two reports suggest that G, GTPase can be faster than PDE turnoff. Microcalorimetric measurements on ROS have shown GTP and light dependent rapid transients, interpreted as reflecting the time course of G, activation and hydrolysis (Vuong & Chabre 1990; 1991). A problem is that the heat responses measured cannot be clearly assigned to the hydrolysis of G,-GTP, rather than dissociation-association reactions of Rh*-Gt or G t -PDE, as well as some other light-dependent processes. Ting and Ho (1991) report GTPase activity more rapid than PDE inactivation, but do not prove that the burst of GTPase activity observed in their experiments is associated with transducin, nor do they measure GTPase and PDE inactivation under the same conditions. An explicit critique of this work is provided by Antonny et al. (1993). It remains to be seen whether mechanisms in addition to the transducin GTPase might be involved in PDE inactivation. The rates of G, GTPase and PDE turnoff measured in the same suspension of bovine ROS coincide (Angleson & Wensel 1993; Arshavsky et al. 1989). In contrast, Erickson et al. (1992) have recently asserted that PDE inactivation can occur in the absence of GTP hydrolysis when G t is activated by a nonhydrolyzable analog of GTP. They propose that ROS contain a pool of PDE inhibitor, presumably PDE -y-subunits that inhibit activated PDE prior to the G r GTPase reaction. However, their measurements conducted in the presence of GTP, making the G,-GTPase reaction possible, reveal faster and more complete PDE turnoff when compared with that observed with the nonhydrolyzable analog. The next experiments needed are an examination of the rates of transducin GTPase and PDE turnoff in suspensions of permeabilized, or broken and very concentrated, ROS. The step of PDE inactivation would seem to be a good locus for light adaptation. A mechanism that increased the rate of PDE turnoff would be expected to lead to a reduction in both photoresponse duration and amplitude during light adaptation. In fact, a regulation of PDE turnoff by cGMP binding to noncatalytic cGMP binding sites on PDE a and (3 subunits has been demonstrated in recent work (Arshavsky et al. 1991; Arshavsky & Bownds 1992). When these sites are occupied, transducin GTPase and PDE turnoff are slowed. As cGMP concentration falls over its physiological range, from ~ 5 to below 1 u,M, the sites empty and GTPase and PDE turnoff are accelerated several fold. Further work by Arshavsky et al. (1992) has provided evidence that cGMP binding in the PDE noncatalytic sites determines the fashion in which PDE is activated by Gt*. If the noncatalytic sites are occupied by cGMP, G t activates PDE and remains bound to the PDE heterotetramer. Alternatively, when the sites are empty, G* physically removes PDE 7 from PDE a p upon activation. These observations are summarized in the scheme (Fig. 4) shown below. Stage 1 shows PDE as aheterotetramer in the dark photoreceptor with cGMP bound to its noncatalytic sites. Activation by Gf removes inhibition of the catalytic sites by PDE 7 (stage 2). As long as cGMP remains bound to the noncatalytic sites, G* remains in a complex with PDE, and the hydrolysis of GTP that causes PDE inactivation is slowed (upper solid arrow). In stage 3, as illumination continues, bound cGMP dissociates and is hydrolyzed. This dissociation might occur from both activated and nonactivated PDE. PDE formed with the noncatalytic sites empty (stages 3 and 4) undergoes a different activation-inactivation sequence. G* physically removes PDE 7 , GTPase activity becomes more rapid, and therefore inactivation occurs faster (lower solid arrow). To complete the cycle of G t , G ta -GDP must dissociate from PDE 7 and recombine with G , P r This process requires the mediation of Gp 7 (Yamazaki et al. 1990). How might this process be involved in adaptation? In the dark cGMP levels are high, the noncatalytic cGMP sites on PDE are occupied (see Fig. 4). This slows GTPase activity so that PDE stays on longer, making the light response bigger and longer. When cGMP is lowered by light, cGMP dissociates from noncatalytic sites of activated and possibly nonactivated PDE molecules. Those PDE molecules that have lost their bound cGMP are inactivated more rapidly upon subsequent activation, making light responses smaller and faster. At first glance, one might suppose that the detachment of PDE 7 shown in Figure 4. A scheme for the putative role of PDE noncatalytic cGMP-binding sites in the background adaptation of photoreceptors (reproduced with permission from Arshavsky et al. 1992). BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 421 Bownds & Arshavsky: Photoreceptor adaptation stage 3 would slow inactivation of PDE, because the reattachment of PDE 7 to PDE a p might be rate limiting. However, direct measurements show that GTP hydrolysis, rather than PDE Y reattachment, is rate limiting under these conditions (Arshavsky et al., in preparation). Validation of a model of this sort requires that both the binding constants and off kinetics of the non-catalytic cGMP binding sites be determined. Cote et al. (1994) have recently accomplished this, finding that activation of frog PDE lowers the binding affinity and accelerates the dissociation kinetics of these sites, so that the cGMP drop occurring upon illumination favors release of bound cGMP. The time scale for dissociation, tens of seconds, corresponds to the period during continuous bright illumination of photoreceptors when an increase in the response recovery rate is being observed (Cervetto et al. 1984; Coles & Yamane 1975; McCarthy & Owen, personal communication). The mechanism may be limited to lower vertebrates, because bovine PDE does not readily lose bound cGMP in vitro (Gillespie & Beavo 1989). The noncatalytic cGMP binding sites of vertebrate cone PDEs have binding constants intermediate between the amphibian and vertebrate rod PDEs (Gillespie & Beavo 1988), and so might potentially use this mechanism to achieve their faster turnoff times. In both cases, it is possible that dissociation occurs more readily for the activated enzyme. The situation regarding PDE 7 may become even more complicated if its recently demonstrated phosphorylation by different protein kinases (Hayashi 1994; Hayashi et al. 1991;Tsuboietal. 1994a; 1994b; Udovichenkoetal. 1994) is shown to change during the activation-inactivation cycle. Tsuboi et al. (1994a; 1994b) report that a cGMP inhibited kinase can phosphorylate either free PDE 7 or PDE 7 in complex with G ta -GTP. The phosphorylated forms lose their ability to bind G ta -GTP, and thus render PDE refractory to activation. While Tsuboi et al. suggest that this PDE 7 phosphorylation is a mechanism of rapid PDE turnoff independent of GTPase activity of G ta , we think it more likely that it is a means of removing some PDE from the active pool during light adaptation. Experiments are needed to demonstrate that this phosphorylationdephosphorylation actually occurs during either flash responses or light adaptation. 8. The drop in cytoplasmic cGMP resulting from PDE activation causes cGMP-gated channels to close The last step of the phototransduction cascade is the closure of cGMP-gated cationic channels in the ROS plasma membrane. Because their properties are discussed in detail in several recent reviews (Kaupp & Koch 1992; McNaughton 1990; Molday & Hsu, this issue; Yau & Baylor 1989) we will provide only a brief account. The amphibian ROS contains approximately 5 X 105 channels, and only about 2% of these are open in the dark-adapted cell. Electrophysiological measurements show that absorption of a single photon leads to a 1-5% reduction in the dark conductance of the plasma membrane, caused by almost complete closure of channels over a length of the ROS of not less than ljjun (Matthews 1986), or more than 6 (xm (Lamb et al. 1981). These numbers are in reasonable correspondence with the calculations above indicating that ~ 4 % of rod cGMP is hydrolyzed during a single photon photoresponse. Is the channel a site of modification during adaptation? Recent observations of Gordon et al. (1992) suggest that channel sensitivity to cGMP may be regulated by phosphorylation. Plasma membrane patches from frog ROS become more sensitive to cGMP over a period of minutes after their excision, and this sensitivity increase is delayed if protein phosphatase inhibitors are added. Hsu and Molday (1993) have found that the cGMP-gated channel from bovine rods can be modulated by calmodulin. Lowering C a + + concentration over the range it falls during illumination increases the affinity of the channel for cGMP. Thus less cGMP might be required to reopen the channel during recovery of the photoresponse in the light-adapted cell than is required to keep it open in the dark. The physiological relevance of this calmodulin effect is discussed in detail by Molday and Hsu in a companion article in this issue. 9. The drop in cytoplasmic C a + + caused by channel closing activates a negative feedback loop by stimulating guanylate cyclase and cGMP recovery C a + + ions continually enter the dark ROS through cGMP-gated channels and are extruded by a Na + / C a + + , K + exchanger (see Fig. 5). The decrease in cytoplasmic calcium that occurs upon channel closing has been shown in numerous studies to be crucial in regulating recovery and adaptation of the photoresponse. Numerous reviews can be consulted on this topic (Detwiler & Gray-Keller 1992; Kaupp & Koch 1992; Koch 1992; Lagnado & Baylor 1992; McNaughton 1990; Pugh & Lamb 1990; Stryer 1991). The central electrophysiological observations are that clamping internal C a + + concentration blocks the gain reduction associated with adaptation (Matthews et al. 1988; Nakatani & Yau 1988b), and that buffering internal calcium slows the recovery phase of the flash response (Lamb et al. 1986). Biochemical studies on bovine ROS show that guanylate cyclase activity is stimulated by lowering C a + + from 200 nM to 50 nM (Koch & Stryer 1988). This C a + + sensitivity was first thought to be mediated by the 26 kDa C a + + binding protein, recoverin (Dizhoor et al. 1991; Lambrecht & Koch 1991). Recov- TV Na+ K+.Ca^ Figure 5. 422 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 T Plasma membrane Ca++ Na+ ' Regulation of guanylate cyclase by calcium ions. Bownds & Arshavsky: Photoreceptor adaptation erin, however, has since been shown to be a regulator of rhodopsin phosphorylation rather than guanylate cyclase (Hurley et al. 1993; see above). Two groups have now isolated small protein(s) that confer C a + + sensitivity on cyclase (Baehr et al. 1994; Dizhoor et al. 1994; Gorczyca et al. 1994). These observations are all consistent with the idea that activation of cyclase caused by the lowering of calcium can accelerate cGMP resynthesis and the restoration of dark conductance. Several quantitative models of the photoresponse have included the calcium feedback loop and reproduce electrophysiological recordings very closely (Forti et al. 1989; Sneyd & Tranchina 1989; Tamura et al. 1991; Tranchina et al. 1991). These models, however, do not exclude the possibility of further regulatory loops, and in fact Koutalos, Nakatani, and Yau (personal communication) have established the importance of both the cyclase feedback loop and the PDE attenuation that occurs as a fall in Ca + + levels accelerates rhodopsin phosphorylation and inactivation mentioned above. They used the truncated rod outer segment preparation to measure the Ca + + -dependence of the cGMP-gated channel, the guanylate cyclase, and the steady-state PDE activity elicited by a step of light. They were able to account for the steady-state response and sensitivity of the intact rod as a function of background light intensity. In terms of relative contribution to adaptation to background light, they found that the C a + + modulation of the cyclase is most important at low background light intensities and modulation of light-activated PDE contributes at higher light levels. The C a + + modulation of the channel was found to add relatively little to light adaptation. Several groups have used fluorescent probes (GrayKeller & Detwiler 1994; McCarthy et al. 1993; Ratto et al. 1988; Younger et al. 1992) or aequorin (Lagnado et al. 1992) to measure C a + + concentration changes that occur on illumination. The most recent estimates judge dark C a + + concentration to be approximately 500 nM, falling to ~50 nM upon saturating illumination. The dark level corresponds to ~80 C a + + ions per disk face, falling below ~ 8 upon illumination. This represents a very small fraction of the capacity of the Na + /Ca + + , K + exchanger (McNaughton 1990). The fall occurs in two phases, the first in less than one second and the next with a half-time of ~ 5 seconds. Changes in cytosolic free C a + + , sensitivity and fractional current suppression follow a broadly similar form during steady state exposure to different background light intensities. It is unlikely that calcium is the sole regulator of adaptation processes. Photoresponses with normal kinetics can be measured in C a + + depleted ROS, incubated in 10 nM C a + + and calcium ionophore (Nicol et al. 1987), and there would appear to be mechanisms capable of reducing the gain of the photoresponse and speeding its recovery at higher background intensities in the absence of significant changes in cytoplasmic calcium (Nicol & Bownds 1989). The cGMP feedback mentioned above would be one candidate for such a mechanism. Rispoli and Detwiler (1991) have demonstrated that increasing internal C a + + concentration speeds, rather than slows, response recovery. In an earlier report (Rispoli & Detwiler 1989) they also noted that the recovery kinetics of the light response were independent of the dark current level that presumably regulates internal C a + + . 10. Can the most important sites of light adaptation be specified? In summary, recent experiments have provided evidence for pathways by which the primary transduction messenger, cGMP, can exert feedback control over both its synthesis and light-induced degradation. The lightinduced decrease in cGMP can damp further decreases by leading to stimulation of guanylate cyclase and accelerating the inactivation of light-activated PDE: on* Cyclase PDE GTP GMP • cGMPA central question is whether one of the mechanisms we have mentioned plays a predominant role in adaptation, or whether equally important multiple attenuations occur at each step of the transduction cascade. The fact that all of the turnoff reactions of the cascade have now been implicated as sites of C a + + or cGMP regulation suggests that the latter possibility is more likely. The data suggest that light adaptation consists of enhancement of all of the turnoff and restoration processes accompanying the photoresponse, starting with rhodopsin phosphorylation and ending with the putative modulation of the plasma membrane channel by C a + + or phosphorylation. We think it likely that a sequence of reactions are recruited as background light intensities increase: 1. C a + + feedback on guanylate cyclase; 2. C a + + feedback on rhodopsin phosphorylation; 3. cGMP feedback on the lifetime of PDE; 4. phosphorylation of unbleached rhodopsin. Three reactions appear to be most central for turnoff of the photoresponse: rhodopsin quenching, PDE inactivation, and guanylate cyclase activation. The data presently available do not permit us to argue strongly for one of these being most central. The crucial balance between the enzymes that set cGMP concentration during recovery of the photoresponse might be considered in several ways: Several reviews emphasize cyclase activation rather than PDE inactivation as central in cGMP and conductance restoration after a flash of light. We think it most likely that the function of cyclase acceleration is to avoid a delay between PDE turnoff and cGMP concentration restoration. Such a delay is most likely the cause of the lengthening of the flash response that occurs when calcium buffers are injected into the ROS cytoplasm. Is cyclase activation sufficient to contribute to adaptation during background light? The > 300-fold activation of PDE that occurs upon illumination (Arshavsky et al. 1992) needs to be compensated by cyclase activation. Although the maximum cyclase activation measured in vitro has been < 10-fold (reviewed in Stryer 1991) this activation (mediated by diffusible Ca + + ) could be expected to spread much further than the disks containing activated PDE molecules. This suggests that at lower levels of illumination, where cyclase is activated over a significantly larger cell volume than PDE, cyclase activation may be sufficient to be part of the mechanism that counters PDE activation during light adaptation. This suggestion is compatible with the finding of Koutalos, BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 423 Bownds & Arshavsky: Photoreceptor adaptation Nakatani and Yau, mentioned above, that modulation of cyclase activity by C a + + is most important at lower background light intensities. Another concern is the extent to which Rh* turnoff versus PDE inactivation is rate limiting for terminating cyclic GMP hydrolysis. PDE inactivation would be rate limiting if Rh* turnoff were significantly faster than the rate of G,-GTPase that causes termination of PDE activation. Conversely, Rh* decay would become limiting if its rate were slower than that of the GTPase. Current biochemical data suggest that these processes are occurring at similar times, so that each might be modulated during light adaptation. Further progress in understanding which of these reactions is most important requires the development of techniques that measure them on the time scale of the photoresponse at dim light intensities. It would not be surprising to find controls that are redun- 424 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 dant and increase the reliability of the system. Studies on this system are repeating the history of the discoveries of basic catabolic and anabolic pathways in the 1950s and 1960s. After establishing a basic pathway, even more effort is being directed toward describing the relevant array of feedback and other regulatory controls. ACKNOWLEDGMENT This manuscript was prepared with support from N.I.H. grants EY-00463 and EY-10336. We are grateful for comments on the manuscript of Mark Gray-Keller, Peter Detwiler, MarcChabre, and several anonymous reviewers. NOTE 1. Arshavsky's present address is Howe Laboratories, Harvard Medical School and the Massachusetts Eye and Ear Infirmary, 243 Charles St., Boston, MA 02114. BEHAVIORAL AND BRAIN SCIENCES (1995) 18, 425-428 Printed in the United States ol America Recoverin and Ca 2+ in vertebrate phototransduction James B. Hurley Department of Biochemistry and Howard Hughes Medical Institute, SL-15 University of Washington, Seattle, WA 98195 Electronic mail: jbhhh@u.washington.edu Abstract: Recoverin is a 23 kDa Ca2+-binding protein that has been detected primarily in vertebrate photoreceptors. The role of recoverin in phototransduction has been investigated using a variety of biochemical methods. Initial reports suggesting that recoverin regulates photoreceptor guanylyl cyclase have not been confirmed. Instead, recoverin appears to determine the lifetime of lightstimulated phosphodiesterase activity, perhaps by regulating rhodopsin phosphorylation. Retinal recoverin is heterogeneously fatty acylated at its ammo-terminus. The amino-terminal fatty acid appears to be involved in the interaction of recoverin with photoreceptor membranes. Keywords: cyclic CMP; guanylyl cyclase; phosphodiesterase; phosphorylation; phototransduction; recoverin; retina; rhodopsin kinase 1. Introduction Recoverin, visinin, and S-modulin were first identified as retina-specific Ca2+-binding proteins. They are relatively abundant in photoreceptor cells, but their functions there remain controversial. Several highly conserved recoverin homologues have been identified in retina and other tissues, suggesting that this family of proteins is important for cellular function. This target article reviews recent studies of recoverin and related proteins with an emphasis on their possible role in photoreceptors. During the excitation phase of the vertebrate rod photoresponse light stimulates an enzymatic cascade that culminates in the hydrolysis of cyclic GMP (cGMP) (Lagnado & Baylor 1992; Stryer 1991). Photoconversion of rhodopsin to metarhodopsin II within the rod disk membranes stimulates GTP-binding to the a subunit of the heterotrimeric G-protein, transducin. Light-activated transducin a (Ta-GTP) dissociates from the T$y complex, then stimulates cGMP phosphodiesterase (PDE) activity by relieving inhibition imposed on the PDE catalytic subunits by a small inhibitory subunit, PDE*y. The mechanism of PDE activation appears to involve formation of a complex between Tot-GTP and PDE-y. The ensuing hydrolysis of intracellular cGMP reduces cGMP-gated cation channel activity in the rod outer segment plasma membrane. Several processes contribute to the recovery phase of the photoresponse. First, photolyzed rhodopsin is phosphorylated by rhodopsin kinase. This reduces the ability of the photolyzed rhodopsin to stimulate transducin. It also stimulates binding of arrestin to the photolyzed rhodopsin. Arrestin binding further quenches the ability of the photolyzed rhodopsin to activate transducin. Activated transducin a subunits (Tot) already formed by pho- © 1995 Cambridge University Press 0140-525X195 S9.00+.10 tolyzed rhodopsin deactivate when their bound GTP is hydrolyzed to GDP. Ta-GDP then reassociates with TB-y and releases the PDE-y subunit, which reinhibits PDE activity. Light-stimulated hydrolysis of cGMP within rod photoreceptors reduces the activity of cGMP-gated cation channels in the photoreceptor plasma membrane. Because these channels are the major route by which Ca2+ enters the photoreceptor, the intracellular concentration of Ca2+ falls. Lowered Ca2+ levels stimulate a photoreceptor, guanylyl cyclase, to resynthesize cGMP (Koch & Stryer 1988). In addition, lowered Ca2+ speeds the rate of deactivation of cGMP PDE (Kawamura & Murakami 1991). Recently, it has also been reported that Ca2+ may reduce the affinity of the cGMP-gated plasma membrane cation channel for cGMP (Hsu & Molday 1993). Each of these effects of Ca2+ may help promote recovery of the photoreceptor following photoexcitation. The mechanisms by which Ca2+ regulates photoreceptor guanylyl cyclase and cGMP PDE are not completely understood. However, a 23 kDa Ca2+-binding protein specifically found in photoreceptors has been implicated in both of these regulatory processes (Dizhoor et al. 1991; Kawamura & Murakami 1991; Lambrecht & Koch 1991). Initial reports suggested that one form of this protein, recoverin, isolated from bovine retinas, is a regulator of photoreceptor guanylyl cyclase (Dizhoor et al. 1991; Lambrecht & Koch 1991). However, it now appears that those initial reports were incorrect (Hurley et al. 1993). Other observations have upheld the view that this protein imparts Ca2+-sensitivity to the cGMP PDE (Kawamura 1993; Kawamura & Murakami 1991). In the next section, I describe some characteristics of this protein and its homologs. The protein has been given several different names by the laboratories that discovered it. Isolates from 425 Hurley: Vertebrate phototransduction chicken were referred to as visinin, isolates from bovine retinas were referred to either as recoverin or p26 and isolates from frog retinas were referred to as S-modulin. 2. Members of the recoverin family 2.1. Visinin. Visinin was identified as a 24 kDa cone protein from chicken retinas (Yamagata et al. 1990). Its cDNA was isolated and used to produce visinin-specific antibodies. These antibodies were used to localize visinin to chicken cone photoreceptors. The nucleotide sequence of visinin cDNA was found to encode a protein with three potential Ca2+-binding sites known as "EFhands." Ca2+-binding to visinin was confirmed using visinin expressed in E. coli. Although the function of visinin has not been specifically investigated, it has been reported that visinin dialyzed into rod photoreceptors affects photoresponse lifetime in the same way as recoverin. (Gray-Keller et al. 1993). 2.2. Recoverin. Recoverin was originally identified as a protein with an apparent molecular weight of 26 kDa, that bound to a column made from detergent solubilized rhodopsin immobilized onto concanavalin A-Sepharose (Dizhoor et al. 1991). Recoverin immunoreactivity has been detected in rod and cone photoreceptors and in a class of bipolar cells in the retina (Milam et al. 1992). The amino acid sequence of recoverin was determined by Edman degradation of proteolytic fragments (Dizhoor et al. 1991). The sequence included Ca2+-binding sites and strong homology with visinin. A cDNA encoding recoverin was isolated and it has been used to express recoverin in E. coli (Ray et al. 1992). Both recoverin isolated from bovine retinas and recoverin expressed in E. coli bind Ca 2+ . Recently, the crystal structure of recoverin was determined by X-ray analysis. Recoverin is a compact molecule with a concave hydrophobic cleft between two Ca2+-binding domains (Flaherty et al. 1993). Initial reports suggested that recoverin is a Ca2+sensitive stimulator of guanylyl cyclase. However, more recent findings have not confirmed this conclusion (Hurley et al. 1993). The relationship between recoverin and guanylyl cyclase will be described in a later section of this paper. 2.3. S-modulin. S-modulin was identified as a protein that binds in a Ca2+-dependent manner to frog photoreceptor membranes (Kawamura & Murakami 1991). This property was used to purify S-modulin from frog photoreceptors. The activity of cGMP PDE in frog photoreceptors is sensitive to Ca2+. Kawamura and Murakami found that a factor responsible for this Ca2+-sensitivity eluted from truncated photoreceptors when the free Ca2+ concentration was lowered to 20 nM (Kawamura & Murakami 1991). Reconstituting truncated photoreceptors or photoreceptor homogenates with purified S-modulin restored Ca2+sensitivity to the PDE. Recent reports (Kawamura 1993; Kawamura et al. 1993) have clarified the effect of S-modulin and recoverin on photoreceptor cGMP PDE. An efficient method for purifying S-modulin/recoverin allowed reconstitution experiments using concentrations of S-modulin/recoverin that 426 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 were significantly higher than were used in earlier experiments. The results demonstrated clearly that S-modulin/recoverin prolongs the lifetime of activated PDE following a light flash. The up to four-fold effect occurred at free Ca2+ concentrations in the submicromolar range. Further analyses revealed that Ca2+ and S-modulin/ recoverin partially suppressed light-stimulated phosphorylation of rhodopsin. More recently, it has been reported that recoverin forms a Ca2+-dependent complex with rhodopsin kinase (Chen & Hurley 1994; Subbaraya et al. 1994). There have been no reports of recoverin interactions with other soluble photoreceptor-specific proteins such as phosphodiesterase subunits. Although the detection of a recoverin/rhodopsin kinase complex suggests that recoverin may regulate rhodopsin kinase directly, thesefindingsdo not rule out other interactions of recoverin, perhaps with membranes or membrane proteins, that might also occur and influence rhodopsin phosphorylation. 2.4. Molecule p26 (CAR-antigen). Molecule p26 was identified in screens for proteins that reacted with antisera from human patients with retinas degenerating from cancer-associated retinopathy (CAR) (Polans et al. 1991; Thirkill et al. 1992). In retinas of CAR patients, photoreceptors are degraded through an autoimmune reaction presumably stimulated by a tumor in a nonretinal tissue. Sera from some CAR patients showed immunoreactivity with a retina-specific protein of apparent size 26 kDa. The protein was purified and a partial amino acid sequence was determined. The sequence included three potential Ca2+-binding sites and it was found to be homologous to chicken visinin and identical to bovine recoverin. The mechanisms by which some tumors induce production of antibodies that recognize recoverin have not yet been seriously addressed. However, the existence of homologs of recoverin in other tissues raises the possibility that proteins expressed in some tumors could induce the production of antibodies that cross-react with retinal recoverin. 2.5. Recoverin homologs in other tissues. Recoverin immunoreactivity has been detected in both rods and cones (Dizhoor etal. 1991; Milam etal. 1992). However, immunological studies using a variety of antibodies raised against recoverin and against homologs from different animal species have not yet shown whether or not rods and cones express the same or different forms of recoverin (Polans et al. 1993). A variety of recoverin homologs including neurocalcin, hippocalcin, and VILIP from nonretinal tissues have also been identified. Neurocalcin has been detected in bovine brain (Okazaki et al. 1992) and in retinal amacrine cells and ganglion cells (Nakano et al. 1992). VILIP was detected in chicken brain and retina (Lenz et al. 1992) and hippocalcin was detected in rat hippocampus (Kobayashi et al. 1992). Functional studies for these recoverin homologs have not been reported. 3. Post-translational modifications of recoverin Recoverin expressed in E. coli was found to have properties that are quite different from the properties of recov- Hurley: Vertebrate phototransduction erin purified from bovine retinas (Ray et al. 1992). For example, the fluorescence emission spectrum of bovine retinal recoverin red shifts when the concentration of Ca 2+ is raised from 70 nM to 10 JJIM while the emission of recombinant recoverin undergoes only a minor change in intensity. Furthermore, the mobilities of retinal and recombinant recoverins are quite different in native gel electrophoresis and in isoelectric focusing (Lambrecht & Koch 1992; Ray et al. 1992). To investigate the cause of these differences, Dizhoor et al. (Dizhoor et al. 1992) used electro-spray mass spectrometry to examine recoverin purified from bovine retinas for post-translational modifications. The mass of retinal recoverin was found to be 207.7 atomic mass units larger than the mass of recombinant recoverin. This difference is consistent with the amino-terminal methionine of recoverin being replaced by a short chain fatty acyl residue such as a myristoyl group. This finding is consistent with the consensus sequence for N-terminal myristoylation present at the recoverin amino-terminus. Closer examination of the recoverin amino-terminus by mass spectrometric analyses of proteolytic fragments revealed four different types of ainino-termini. The N-terminal glycine of recoverin was found linked to either 12:0; 14:2 cis, cis A5A8; 14: lcis A5 or 14:0 fatty acyl residues. The identities of the fatty acyl residues and the positions and configurations of the double bonds were confirmed by a variety of methods including gas phase chromatography, ozonolysis, and tandem mass spectrometry. Similar heterogeneity was also detected at the amino-terminus of another photoreceptor protein, the transducin a subunit (Neubert et al. 1992). The functional significance of photoreceptor protein heterogeneity is unknown but it is currently being investigated. The functional significance of the fatty acyl residue on the recoverin amino-terminus was investigated by comparing the biochemical properties of myristoylated and nonacylated recoverin expressed in E. coli (Ray et al. 1992). Myristoylated recoverin was produced in E. coli by coexpressing recoverin with yeast N-myristoyl transferase and myristic acid. In addition to differences in fluorescence emission and electrophoretic mobility, nonacylated recoverin has a significantly higher affinity for Ca 2 + than acylated recoverin. Affinities of acylated and nonacylated recoverins for membranes were also investigated. Kawamura and Murakami (1991) had shown previously that S-modulin, a recoverin homolog from frog photoreceptors, binds to photoreceptor membranes only in the presence of >lu,M Ca 2+ . Ca 2+ also promotes binding of recoverin to rod outer segment (ROS) membranes (Dizhoor et al. 1993; Zozulya & Stryer 1992). Myristoylated recombinant recoverin was found to bind to ROS membranes more efficiently than nonacylated recombinant recoverin. These findings suggested that the fatty acid on the recoverin amino-terminus may serve as a Ca 2+ -dependent membrane anchor. To test this model, recoverin was prepared with a [3H] myristic acid residue at its amino-terminus. The labeled recoverin, either in its Ca 2+ -liganded form or its Ca 2+ -free form, was then exposed to trypsin under nondenaturing conditions. The labelled amino-terminus of Ca 2+ -free recoverin was not cleaved by trypsin. The labelled amino-terminus of Ca 2+ recoverin was rapidly cleaved suggesting that Ca 2+ bind- ing to recoverin induces a conformational change that exposes the amino-terminus. Recoverin lacking a tetrapeptide from its amino-terminus failed to bind to ROS membranes. These findings suggest that Ca 2+ binding frees the hydrophobic acylated amino-fatty acid residue to make it accessible for membrane interactions. The site on the membranes with which recoverin interacts has not been identified. 4. Recoverin and guanylyl cyclase Guanylyl cyclase in rod outer segment homogenates is active only at concentrations of free Ca 2+ less than approximately - 2 0 0 u,M. Koch and Stryer (1988) found that a soluble factor required for the Ca 2+ -sensitivity of guanylyl cyclase could be eluted from rod outer segment membranes. When recoverin was first isolated and characterized (Dizhoor et al. 1991) it was apparent that it had properties suggesting that it might regulate guanylyl cyclase. Preparations of purified recoverin were assayed and were found to stimulate guanylyl cyclase in a Ca 2+ sensitive manner. These results together with findings from another laboratory (Lambrecht & Koch 1991) led to the conclusion that recoverin was the soluble factor that imparted Ca 2+ -sensitivity to guanylyl cyclase. However, the following recent findings suggest that recoverin is not the soluble factor that stimulated guanylyl cyclase in the original recoverin preparations (Hurley et al. 1993). 1. Reconstitution experiments using purified recombinant recoverin showed that neither myristoylated nor nonmyristoylated recombinant recoverin stimulate guanylyl cyclase. 2. Highly purified recoverin isolated from bovine retinas by several fractionation methods does not stimulate photoreceptor guanylyl cyclase. 3. Recoverin dissociates from photoreceptor membranes under conditions that stimulate membrane-associated guanylyl cyclase activity. The Ca 2+ -dependent membranebinding properties of recoverin appear to be inconsistent with a role in guanylyl cyclase regulation. 4. Recoverin separates from other fractions containing guanylyl cyclase stimulatory activity during several fractionation procedures. The most highly purified fractions of guanylyl cyclase stimulatory activity contain nearly undetectable amounts of recoverin. Recent observations from another laboratory also suggest that recoverin does not stimulate guanylyl cyclase activity. Gray-Keller et al. (1993) have shown that purified recoverin dialyzed into rod cells slows recovery. These findings are inconsistent with stimulation of guanylyl cyclase by recoverin. 5. Summary Recoverin is a 23 kDa Ca 2+ -binding protein specifically found in vertebrate photoreceptor cells. The role of recoverin in phototransduction may be to increase photosensitivity by prolonging the lifetime of photolyzed rhodopsin in darkness when intracellular Ca 2+ concentrations are high. The Ca 2+ -bound form of recoverin appears to inhibit deactivation of photolyzed rhodopsin by phos- BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 427 Hurley: Vertebrate phototransduction phorylation. Following illumination, lowered intracellular Ca 2 + levels may promote recovery by allowing more rapid phosphorylation and inactivation of rhodopsin. Recoverin homologs in other tissues may impart Ca 2 + sensitivity to other signalling mechanisms in a similar " 428 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 ACKNOWLEDGMENTS I wish to thank all of the members of my laboratory for many stimulating discussions. I wish to particularly thank Alexander M. Dizhoor and Jason Chen, who are primarily responsible for man y °luthe f m ^ s f r o m "iy laboratory discussed in this review. Their work was supported by grant ROl EYO6641 from the National Eye Institute. BEHAVIORAL AND BRAIN SCIENCES (1995) 18, 429-440 Printed in the United States of America Do the calmodulin-stimulated adenylyl cyclases play a role in neuroplasticity? Zhengui Xia, Eui-Ju Choi, and Daniel R. Storm Department of Pharmacology, University of Washington, Seattle, WA 98195 Electronic mall: dstorm@u.washington.edu Christine Blazynski Department of Biochemistry and Molecular Biophysics, Washington University, School of Medicine, St. Louis, MO 63110 Abstract: Evidence from invertebrate systems including Aplysia and Drosophila, as well as studies carried out with mammalian brain, suggests that Ca 2+ -sensitive adenylyl cyclases may be important for long-term synaptic changes and learning and memory. Furthermore, some forms of long-term potentiation (LTP) in the hippocampus elevate cyclic AMP (cAMP) signals, and activation of adenylyl cyclases and cAMP-dependent protein kinase may be required for late stages of LTP. We propose that long-term changes in neurons and at synapses may require synergism between the cAMP and Ca 2 + signal transduction systems which regulates transcription and synthesis of specific proteins required for long-term synaptic changes. During LTP, protein kinase C is activated and intraccllular Ca 2 + increases. We hypothesize that the calmodulin (CaM)-regulated adenylyl cyclases may be activated during LTP because of increases in intracellular Ca 2 + , release of free CaM from neuromodulin, activation by protein kinase C, release of neurotransmitters, or a combination of these events. Synergistic activation of CaM-sensitive adenylyl cyclases may produce a robust or prolonged cAMP signal required for transcriptional control. Furthermore, the coupling of the Ca 2 + and cAMP systems may provide positive feedback regulation of Ca 2 + channels by cAMP-dependent protein kinase. Keywords: adenylyl cyclase; calcium; calmodulin; cAMP; long-term potentiation; neuroplasticity 1. Introduction An area of intense interest in neurobiology is the molecular mechanism underlying short-term and long-term adaptive changes in synaptic function. Several regulatory systems have been implicated in neuroplasticity including the cyclic AMP (cAMP) and the Ca 2+ signal transduction systems. Ca 2 + , calmodulin (CaM), CaM-sensitive protein kinases, adenylyl cyclases have all been implicated in long-term adaptive responses in neurons and synaptic plasticity. The purpose of this target article is to present a hypothesis that the CaM-sensitive adenylyl cyclases may play a crucial role in long-term adaptive responses in brain and to present molecular models to explain the role of these enzymes in these processes. Evidence supporting this hypothesis is derived from studies of the invertebrate and mammalian CaM-sensitive adenylyl cyclases, as well as data concerning the role of the cAMP signal transduction system for LTP in brain. In mammalian brain, adenylyl cyclase activity is regulated by neurotransmitter and hormone receptors coupled to the enzyme through the G regulatory proteins, Gs and Gf (reviewed by Ross & Gilman 1980), as well as by intracellular free calcium (reviewed by Cheung & Storm 1982). Protein phosphorylations catalyzed by the © 199S Cambridge University Press 0140-S25X/95 S9.00+.10 cAMP-dependent protein kinase regulate several important aspects of neuronal function including ion channel activity, gene expression, and neurotransmitter synthesis (reviewed by Krebs & Beavo 1979; Nairn et al. 1985; Nestler & Greengard 1983). Furthermore, cAMP has been implicated in the regulation of synaptic plasticity and may play an important role in mechanisms underlying learning and memory (reviewed by Dudai 1988; Kandel & Schwartz 1982). Elucidation of molecular mechanisms for neuroplasticity requires a multidisciplinary approach with electrophysiology as the bridging discipline between signal transduction molecular biology and the behavioral sciences. A useful electrophysiological model for neuroplasticity is LTP which occurs in many areas of brain including the hippocampus. The relationship between LTP and behavior is still controversial. However, both spatial learning in rats and LTP in the hippocampus are blocked by N-methyl-D-aspartate (NMDA) antagonists (Davis et al. 1992) demonstrating a possible relationship between these two phenomena. Other evidence suggesting a correlation between LTP and learning comes from gene disruption studies which have demonstrated that mice lacking the alpha form of CaM kinase II (Silva et al. 1992a; 1992b) or the fyn-tyrosine kinase (Grant et al. 429 Xia et al.: Neuroplasticity 1992) are deficient in LTP and spatial learning. On the other hand, PKC-gamma mutant mice showed impaired LTP in the hippocampus but can perform the hidden platform test of the Morris water task indicating that LTP induced by conventional tetanic stimulation is not essential for this specific type of learning (Abeliovich 1993a; 1993b). These mice were, however, deficient in the transfer test of the Morris water task which may be a more reliable measure of spatial learning. Although many forms of LTP decay over several hours, long-lasting LTP (L-LTP) which is produced by multiple trains of high frequency stimulation can last hours or even days depending upon the stability of the preparation (reviewed in Matthies et al. 1990) and is sensitive to inhibitors of transcription and translation (Frey et al. 1988; Grecksch & Matthies 1980; Krug et al. 1984; Squire & Davis 1981). Various forms of LTP in the hippocampus require entry of Ca2+ through NMDA receptors or voltage-gated Ca2+ channels (reviewed in Kauer et al. 1988; Nicoll et al. 1988) and an increase in intracellular Ca2+ can induce LTP (Malenka et al. 1988). Consequently, Ca2+-sensitive adenylyl cyclases may be stimulated when Ca+ channels are activated during LTP. 2. The Aplysia adenylyl cyclase system and the gill withdrawal reflex One of the most interesting systems demonstrating the possible importance of Ca2+-sensitive adenylyl cyclases for learning and memory is the gill withdrawal reflex of the invertebrate Aplysia (Byrne 1987; Castellucci & Kandel 1976). In this system, the conditioned stimulus, a weak touch to the siphon, elicits a minimal defensive response by withdrawing the gill. In contrast, the unconditioned stimulus, a shock to the tail, produces a strong withdrawal reflex. If the conditioned stimulus is paired with the unconditioned stimulus the animal learns to associate touch with the shock and shows a strong withdrawal response to touch alone. The length of the memory for this response depends upon the number of stimuli applied (Frost et al. 1985) and can last for days when 4 or 5 noxious signals are administered (Castellucci et al. 1989). An extensive analysis of this system has led to the conclusion that the conditioned stimulus results in an increase in intracellular Ca2+ within the sensory neuron, and that the unconditioned stimulus results in the release of serotonin into the same cell (Abrams et al. 1991). It has been proposed that paired activation of the Aplysia adenylyl cyclase by Ca2+/CaM and serotonin results in a synergistic activation of adenylyl cyclase which may be required for long-term memory. Since long-term facilitation can be induced by the injection of cAMP into the presynaptic sensory neuron (Schacher et al. 1988; Scholz & Byrne 1988) and blocked by inhibitors of protein kinase A (Ghirardi et al. 1992), it is very likely that cAMP protein kinase plays a central role in long-term facilitation in Aplysia. Furthermore, long-term facilitation in this system is sensitive to inhibitors of RNA and protein synthesis (Castellucci et al. 1989; Montarolo et al. 1986) suggesting that cAMP-mediated transcription may be required for long-term facilitation in Aplysia (Kaang et al. 1993). 430 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 The gene(s) for the Aplysia adenylyl cyclase(s) have not been cloned and their relationship to the cloned mammalian adenylyl cyclases has not been determined. However, Ca2+/CaM-sensitive adenylyl cyclase activity is present in Aplysia membranes and Yovell et al. (1992) have demonstrated synergism for activation of Aplysia adenylyl cyclase by Ca2+ and serotonin when brief pulses of Ca2+ followed by serotonin were paired. Synergistic activation of this adenylyl cyclase by Ca2+ and Gs is quite novel in that it was only observed with sequential applications of Ca2+ and serotonin and is not observable in steady-state assays. Although this synergism was relatively weak and its physiological significance remains to be established, the small increment in cAMP signal may be functionally important because of the signal amplification that is inherent in the cAMP signal transduction pathway. Synergistic activation of the type I adenylyl cyclases by Ca2+ and Gs in vivo is observable in steadystate assays and does not require ordered administration of Ca2+ and activators of Gs (Wayman et al. 1994). Consequently, the Aplysia adenylyl cyclase may be quite distinct from the mammalian type I and type III CaMsensitive adenylyl cyclases. If cAMP-regulated transcription plays a major role in long term facilitation in Aplysia neurons, then cAMP signals generated at synapses must either diffuse to the cell body in order to affect transcription in the nucleus, or a secondary signal arising from the initial synaptic cAMP signal must reach the nucleus. This is of particular concern in neurons because of substantial distances required for transport or diffusion of cAMP along neurites. Recently, this question has been directly addressed in Aplysia neurons by monitoring gradients of cAMP produced in response to serotonin stimulation of adenylyl cyclase (Bacskai et al. 1993). Bath application of serotonin produced an extensive cAMP gradient between the processes and the cell body which was consistent with the diffusion of cAMP from neurite tips to the cell body of the neuron. Nuclear translocation of the catalytic subunit of cAMP-dependent protein kinase was relatively slow and probably only occurs after prolonged or enhanced cAMP signals. In contrast to other cAMPregulated phenomena such as regulation of metabolism and ion channel function, which are rapid short term responses, cAMP regulation of transcription in neurons may require strong or persistent activation of adenylyl cyclase activities. 3. The Drosophila rutabaga learning mutant Other evidence that Ca2+-sensitive adenylyl cyclases may be important for synaptic plasticity in invertebrates has come from studies of the Drosophila learning mutant, rutabaga. Rutabaga is an X-linked recessive mutant that is deficient in associative learning (Dudai & Zvi 1984, 1985; Livingston 1985; Livingston et al. 1984). In contrast to wild type Drosophila, the rutabaga fly lacks Ca2+/CaM-sensitive adenylyl cyclase activity. The gene for an adenylyl cyclase similar to the type I adenylyl cyclase maps within a region on the X chromosome that includes the rut locus and a single point mutation in this gene is sufficient to destroy all enzyme activity (Levin et al. 1992). Feany (1990) has proposed that calcium re- Xia et al.: Neuroplasticity sponsiveness, rather than the overall cAMP synthesis may be the crucial component of adenylyl cyclase activity required for associative learning in Drosophila. In rutabaga larvae, voltage clamp analysis of neuromuscular transmission indicated deficient synaptic facilitation and post-tetanic potentiation (Zhong & Wu 1991). The rutabaga mutant has provided convincing evidence that the type I adenylyl cyclase is important for associative learning in invertebrates and synaptic facilitation. 4. The family of mammalian adenylyl cyclases The existence of distinct CaM-stimulated and CaMinsensitive adenylyl cyclases in brain was first demonstrated by the separation of these two forms of the enzyme from bovine brain using CaM-Sepharose affinity chromatography (Westcott et al. 1979). The CaM-sensitive adenylyl cyclase activity absorbed to CaM-Sepharose in the presence of Ca 2+ and was stimulated by Ca 2+ when reconstituted with CaM. Half-maximal stimulation of the enzyme occurred at 80 nM free Ca 2+ . The CaMinsensitive forms of adenylyl cyclase present in brain did not absorb to CaM-Sepharose in the presence or absence of Ca 2 + , and were not stimulated by Ca 2+ . Furthermore, polyclonal and monoclonal antibodies have been isolated that distinguish between the CaM-sensitive and CaMinsensitive adenylyl cyclase in brain providing further evidence for separate adenylyl cyclases (Mollner & Pfeuffer 1988; Mollner et al. 1991; Rosenberg & Storm 1987a). The catalytic subunit of a CaM-sensitive adenylyl cyclase was purified to homogeneity from bovine brain using CaM-Sepharose and Forskolin-Sepharose affinity chromatography (Minocherhomjee et al. 1987; Smigel et al. 1986; Yeager et al. 1985). Characterization of the purified CaM-sensitive adenylyl cyclase from brain indicated that it is a glycoprotein that interacts directly with CaM (Minocherhomjee et al. 1987). Furthermore this enzyme can couple to Gs and beta-adrenergic receptors (May et al. 1985; Rosenberg et al. 1987b) as well as Gf and muscarinic receptors (Dittman et al. 1994; Tota et al. 1990). The expression of the type I adenylyl cyclases in the insect/Bacculovirus system and the development of new strategies for its purification from these cells offers great promise for direct characterization of the protein (Taussig et al. 1993). cDNA clones for the type I adenylyl cyclase have been isolated from bovine brain (Krupinski et al. 1989) and human brain cDNA libraries (Villacres et al. 1992). To date, cDNA clones for eight distinct ACs have been published (Bakalyar & Reed 1990; Cali et al. 1994; Feinstein et al. 1991; Gao & Gilman 1991; Ishikawa et al. 1992; Krupinski et al. 1989; Yoshimura & Cooper 1992). Although these enzymes share sequence homology, they contain hypervariable regions and exhibit different regulatory properties. I-AC (Choi et al. 1992a; Tang et al. 1991), III-AC (Choi et al. 1992b), and VIII-AC (Cali et al. 1994) are stimulated by Ca 2+ and CaM in vitro whereas II-AC, IV-AC, V-AC and VI-AC are not. The diversity of this enzyme system undoubtedly reflects different mechanisms for regulation of cAMP levels in animal cells, and the variety of physiological processes that are regulated by intracellular cAMP. 5. Regulation of the type I and III adenylyl cyclases by calcium and CaM 5.1. TVpe III adenylyl cyclase. Modulation of adenylyl cyclase activity by Ca 2+ has been demonstrated in several tissues including brain and retina and it has been proposed that cAMP levels may be controlled by fluctuations in intracellular free Ca 2 + . Most mammalian tissues contain mixtures of adenylyl cyclases, and the Ca 2+ sensitivity of specific forms of adenylyl cyclase has only recently been addressed. The type III adenylyl cyclase was expressed in human kidney 293 cells to determine if it is stimulated by Ca 2+ and CaM (Choi et al. 1992b). In isolated membranes, the type III enzyme was not stimulated by Ca 2+ and CaM in the absence of other effectors. It was, however, stimulated by Ca 2+ through CaM when the enzyme was concomitantly activated by forskolin (Fig. IB). The concentrations of free Ca 2+ for half-maximal stimulation of type III adenylyl cyclases was 5.0 (JLM Ca 2+ (Fig. 2B). The sensitivity of the type III adenylyl cyclase to Ca 2+ in vivo has not been reported. Toscano et al. (1979) isolated a partially purified adenylyl cyclase from bovine brain that was not stimulated by GppNHp, NaF, or Ca 2 + /CaM. Sensitivity to these effectors was restored by incubation of the adenylyl cyclase preparation with detergent solubilized membranes from cerebral cortex. Reconstitution of Ca 2+ /CaM sensitivity required the presence of guanyl nucleotides. The properties of this adenylyl cyclase are similar to the type III adenylyl cyclase and these early studies showed synergistic activation of an adenylyl cyclase by Ca 2+ and Gs. Is Ca 2+ regulation of the type III adenylyl cyclase at concentrations of 5 JJLM Ca 2+ physiologically significant, and what role could it play in signal transduction pathways in olfactory sensory neurons, retina, or brain? Free Ca 2+ in many animal cells generally varies from less than 0.1 to 10 u,M. Local Ca 2+ concentrations at the membrane surface in neurons may increase to 100 |J,M or even higher during action potentials (Smith & Augustine 1988). Since the type III adenylyl cyclase is only activated at the higher end of the free Ca 2+ range, and when the enzyme is activated by other effectors, the enzyme may allow Ca 2+ amplification of cAMP signals. For example, the existence of cAMP-gated ion channels in neurons suggests that initial cAMP signals, generated through receptors coupled to adenylyl cyclase, may be further amplified by increases in intracellular Ca 2 + . Thus the type III adenylyl cyclase is able to integrate multiple signals and may function as a "coincidence detector" (Bourne & Nicoll 1993). 5.2. Type I adenylyl cyclase. CaM stimulates the type I adenylyl cyclase activity in membranes from CDM8(IAC)-transfected 293 cells at a half-maximal concentration of approximately 20 nM (Fig. 1). Similar CaM sensitivities have been reported for the CaM sensitive adenylyl cyclase purified from bovine brain (Minocherhomjee et al. 1987) and the type I adenylyl cyclase expressed in insect SF9 cells (Tang et al. 1991). Half-maximal stimulation of type I adenylyl cyclase occurred at approximately 50 nM free Ca 2+ (Fig. 2), which is consistent with the Ca 2+ sensitivity of the CaM-stimulated adenylyl cyclase isolated from bovine brain calculated from the data of Westcott et al. (1979). These Ca 2+ dependencies were deterBEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 431 Xia et al.: Neuroplasticity oe E o a. o a eu < A. < [CaM] (uM) [CaM] (nM) B 25 • forskolin rO » .X Id, 0.1 1 [CaM] (nM) 10 Figure 1. The effect of CaM on either type I (A) or type III (B) adenylyl cyclase activities transiently expressed in human 293 cells. Human 293 cells were transfected with control CDM8 vector, CDM8(I-AC) to express the type I adenylyl cyclase or CDM8(III-AC) to express the type III adenylyl cyclase. Cell membranes prepared from transfected cells were washed with buffer A containing 1 mM EGTA to remove endogenous CaM, and then assayed for adenylyl cyclase activity as a function of CaM concentration. A; the effect of CaM on control (O) or type I adenylyl cyclase activity (•). The assay was performed in the presence of 1.9 u,M free Ca 2 + and varying concentrations of CaM. Control activity was the adenylyl cyclase activity in cell membranes from CDM8-transfected 293 cells. B; the effect of CaM on type HI adenylyl cyclase activity in the absence (O) or in the presence of 10 u,M forskolin (•). The enzyme assay was performed in the presence of 30 |xM free Ca 2 + and varying concentrations of CaM. Adenylyl cyclase activities due to the type III enzyme were calculated from data shown in panel C by subtracting adenylyl cyclase activities in control cells from those of cells transfected with CDM8(III-AC). In panel C, cell membranes from control 293 cells (X, A) or cells transfected with CDM8(III-AC) (O, • ) were assayed for adenylyl cyclase activity with varying concentrations of CaM and 30 (xM Ca 2 + in the absence (X, O) or presence of 10 u,M forskolin (A, • ) . Data are from Choi et al. 1992a. mined using EGTA/Ca 2+ buffers. Because of the uncertainties associated with calculating free Ca 2+ by this method, these data are only estimates of the Ca 2+ dependence of the enzyme (Yovell et al. 1992). It has been generally assumed that CaM-sensitive adenylyl cyclase can function to couple increases in intracellular free Ca 2+ to elevations in cAMP in vivo. The purified type I adenylyl cyclase and the enzyme in membrane preparations is stimulated by Ca 2+ and CaM. However, it was important to demonstrate that increases in intracellular free Ca 2+ can actually stimulate the type I adenylyl cyclase in intact cells. Therefore, we expressed the enzyme in human 293 cells and examined the effect of a Ca 2+ ionophore, A23187, and extracellular Ca 2+ on intracellular cAMP (Choi et al. 1992a). Human 293 cells were selected for these studies because their endogenous adenylyl cyclase activity is quite low and insensitive to extracellular Ca 2 + and the Ca 2+ ionophore A23187 (Fig. 3A). In the presence of 2 mM extracellular Ca 2+ , the 432 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 intracellular cAMP levels of control cells, transfected with the CDM8 vector alone, were unaffected by addition of A23187. In contrast, intracellular cAMP in 293 cells stably expressing the type I adenylyl cyclase increased approximately 16-fold with addition of 10 JJLM A23187 (Fig. 3A). Under these conditions, the intracellular free Ca 2+ increased to 1.0 (xM. The increase in intracellular cAMP stimulated by A23187 depended upon the concentration of Ca 2+ applied (Fig. 3B). Elevated cAMP was detectable within a few minutes after addition of 10 u.M A23187 and 2 mM Ca 2+ (Fig. 3C). The data described above indicated that intracellular Ca 2+ can stimulate the type I adenylyl cyclase activity in vivo. Ca 2+ stimulation of the enzyme in vivo may be due to direct interactions of the enzyme with Ca 2 + and CaM, or indirect mechanism involving stimulation of the enzyme by Ca 2+ -activated protein kinases. We have made several point mutations within the calmodulin binding domain to determine if the Ca 2+ sensitivity of the enzyme Xia et al.: Neuroplasticity I OS e Q. 1 1 £ a. 20 40 60 80 [calcium ton] (u.M) 0.2 0.4 0.6 3 [calcium ion] (uM) 150 1 £ < 20 40 60 80 [calcium ion] (|iM) can be modified by mutagenesis (Wu et al. 1993). The catalytic activities of the mutant enzymes were comparable to wild-type type I adenylyl cyclase. Ca 2+ and CaM stimulation was abolished by substitution of Phe-503 with Arg-503 (FR-I-AC). Stimulation of type I adenylyl cyclase activity in vivo by intracellular Ca 2+ was also greatly diminished with the Arg-503 mutant indicating that Ca 2+ stimulation of the enzyme in vivo is due primarily to direct interactions with CaM and Ca 2+ . These data demonstrated that the Ca 2+ sensitivity of this enzyme can be modulated by point mutagenesis within the putative calmodulin binding domain, and indicate that the enzyme can be directly regulated by Ca 2+ and CaM, in vivo. Kidney 293 cells contain various receptors that can couple directly or indirectly to adenylyl cyclases, including muscarinic receptors. Therefore, we examined the influence of carbachol, a muscarinic agonist, on the intracellular cAMP levels of 293 cells expressing the type I Figure 2. The effect of Ca 2 + on type I (A) or type III (B) adenylyl cyclase activities expressed in human 293 cells. Cell membranes from human 293 cells transfected with control CDM8 vector, CDM8(I-AC) to express the type I adenylyl cyclase, or CDM8(HI-AC) to express the type III adenylyl cyclase were examined for adenylyl cyclase activity as a function of free Ca 2 + concentration by varying concentrations of CaCl 2 in the presence of 0.2 mM EGTA in the assay. A; the effect of Ca 2 + on control (O) or type I adenylyl cyclase activity (•). The enzyme assay was performed in the presence of 2.4 u,M CaM and various concentrations (0—3.7 |J.M) of free Ca 2 + . B; the effect of Ca 2 + on type III adenylyl cyclase activity in the absence (O) or in the presence of 100 p-M GppNHp (•). The enzyme assay was performed in the presence of 2.4 \xM CaM and various concentrations (0-74 u,M) of free Ca 2 + . Type III adenylyl cyclase activity was calculated froin data shown in panel C by subtracting adenylyl cyclase activities in control cells from those of cells transfected with CDM8(III-AC). In panel C, cell membranes from control cells (A, • ) or cells transfected with CDM8(III-AC) (O, • ) were assayed for adenylyl cyclase activity with varying concentrations of free Ca and 2.4 u,M CaM in the absence (A, O) or presence of 100 p.M GppNHp (A, • ) . Data are from Choi et al. 1992b. adenylyl cyclase (Choi et al. 1992a). Carbachol stimulated intracellular cAMP levels approximately 3-fold in 293 cells stably expressing type I adenylyl cyclase, but was without significant effect on cAMP in control cells (Fig. 4A). Maximal stimulation by carbachol occurred at approximately 100 u,M, and this increase in cAMP was inhibited by the muscarinic antagonist, atropine. No carbachol stimulation of intracellular cAMP was seen without forskolin, even when IBMX was present. The requirement for forskolin reflects the low sensitivity of this assay for cAMP and more sensitive assays using a CRE-beta-galactosidase reporter construct can detect Ca 2+ -stimulated cAMP signals in 293 cells without the presence of forskolin or phosphodiesterase inhibitors (Impey & Storm, unpublished observations). These data illustrated that the type I adenylyl cyclase can be regulated by muscarinic receptors in vivo either by direct coupling through a G regulatory protein or BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 433 Xia et al.: Neuroplasticity 0.01 0.1 10 100 [A23187J (n.M) indirectly by mobilization of intracellular Ca 2 + . The latter explanation seems most likely since carbachol had no effect on the basal, CaM-stimulated, or forskolinstimulated activities of type I adenylyl cyclase activity in isolated membranes. Furthermore, we analyzed the influence of carbachol on intracellular free Ca 2+ in 293 cells using fura-2. One mM carbachol increased free Ca 2+ from a baseline of 40 nM to 140 nM and this increase was blocked by 100 |xM BAPTA/AM, the intracellular Ca 2+ chelator. Furthermore, 100 u-M BAPTA/AM completely inhibited carbachol-stimulated increases in intracellular cAMP (Fig. 4B). We concluded that the type I adenylyl cyclase can function to couple increases in intracellular Ca 2+ to cAMP production in whole cells, and that the enzyme can also be indirectly stimulated through muscarinic receptors by mobilization of intracellular free Ca 2 + . 4000 1 10 100 1000 [carbachol] (p.M) I 800 r 5000 Time (min) 2+ Figure 3. Ca and A23187 stimulation of intracellular cAMP levels in human 293 cells expressing the type I adenylyl cyclase. A; cultured 293 cells expressing the type I adenylyl cyclase (•) or control cells transfected with the CDM8 vector (O) were treated with varying concentrations of A23187 in the presence of 2 mM CaCl 2 for 30 minutes. B; 293 cells expressing the type I adenylyl cyclase (•) or control cells (O) were treated with varying concentrations of CaCl2 for 30 minutes in the presence of 10 (JLM A23187. C; 293 cells expressing the type I adenylyl cyclase ( • , A) or control cells (O, X) were treated for various periods of time with 2 mM CaCl 2 and 10 |JLM A23187 in the presence ( • , O) or absence (A, X) of one mM IBMX. 434 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 - carbachol carbachol (lmM) carbachol BAPTA/AM (100 nM) Figure 4. Carbachol stimulation of intracellular cAMP levels in human 293 cells expressing the type I adenylyl cyclase. A; control cells ( • , Mj or human 293 cells expressing the type I adenylyl cyclase ( • , Wj were treated with varying concentrations of carbachol for 30 minutes in the absence ( • , • ) or presence (H, ^ ) of 1.0 p,M atropine. One mM IBMX was present in all assays. B; pretreatment of 293 cells for one hour with BAPTA/AM at 100 (JLM completely blocked the increase in intracellular cAMP caused by carbachol. Data are from Choi et al. 1992a. Xia et al.: Neuroplasticity 6. Synergistic regulation of type I adenylyl cyclase by Ca 2 + and isoproterenol in vivo Evidence from several studies has indicated that adenylyl cyclase activity from various areas of mammalian brain may be synergistically activated by CaM/Ca 2+ and Gs, or Gs-coupled receptors (Choi et al. 1992a; Gnegy & Treisman 1981; Natsukari et al. 1990; Toscano et al. 1979). However, other investigators have found that CaM and G protein stimulation of AC activities are additive and not synergistic (Piascik et al. 1981; Salter et al. 1981; Sano 1985). This apparent discrepancy reflects the regulatory diversity of the adenylyl cyclases, their distribution within brain, and the different preparations used in these studies. Although the type I adenylyl cyclase is stimulated by Ca 2 + in vivo, we have discovered that it is not stimulated by Gs-coupled receptors in vivo unless it is also activated by Ca 2+ /CaM (Wayman et al. 1994). We examined the sensitivity of I-AC expressed in HEK-293 cells to isoproterenol or glucagon when intracellular Ca 2+ was elevated. The cells used in this study were stable transfectants expressing type I adenylyl cyclase and glucagon receptors. The Ca 2+ ionophore A23187 stimulated the enzyme approximately three-fold, isoproterenol did not stimulate the enzyme, but the combination of the two stimulated adenylyl cyclase activity 13-fold in vivo. Similarly, glucagon did not stimulate the enzyme but the combination of A23187 and glucagon activated the enzyme 90-fold. This phenomenon was not observed with a mutant enzyme (FR-I-AC) that is insensitive to Ca 2+ and CaM. Therefore, I-AC may couple Ca 2+ and neurotransmitter signals to generate optimal cAMP levels, a property of the enzyme that may be important for learning and memory in mammals. We have also demonstrated that the expression of type I adenylyl cyclase in HEK-293 cells allows Ca 2+ to stimulate reporter gene activity mediated through the CRE response element (Impey et al. 1994). Simultaneous activation by Ca 2+ and isoproterenol caused synergistic stimulation of cyclic AMP response element (CRE)-mediated transcription in HEK-293 cells and cultured neurons. 7. Regulation of the CaM-regulated adenylyl cyclases by protein kinase C 7.1. Activation of the type I and type III adenylyl cyclases by activators of protein kinase C. Protein kinase C (PKC) is activated during many forms of LTP and phorbol esters and other activators of PKC can also affect intracellular cAMP levels in various tissues and cultured cells. Therefore, the effect of phorbol esters on the activity of the type I and type III adenylyl cyclases in whole cells has been examined using stably transfected 293 cells expressing either enzyme (Choi et al. 1993). TPA markedly enhanced the forskolin responsiveness of the type I and type III adenylyl cyclases expressed in kidney 293 cells. The effect of 12-O-tetradecanoylphorbol 13 acetate (TPA) on the activity of the CaM-sensitive adenylyl cyclases was not mediated through increases in intracellular free calcium. Jacobowitz et al. (1993) have also examined the sensitivities of various adenylyl cyclases to phorbol esters by transient expression of the enzymes in 293 cells. Their data indicated that the type II adenylyl cyclase was particularly sensitive to phorbol esters and that types IV, V, and VI showed modest stimulations upon phorbol myristic acid (PMA) treatment. Since PKC is activated during some forms of LTP, cAMP levels may be regulated during LTP. 7.2. Does protein kinase C regulate adenylyl cyclase activities in the brain by controlling the levels of free CaM? The activity of the CaM-sensitive adenylyl cyclases depends on the concentrations of free intracellular Ca 2+ and free CaM in neurons. We have described the biochemical properties of a neurospecific CaM binding protein, neuromodulin (GAP-43), that may regulate the levels of free CaM present in neurons and consequently may play a key role in CaM-regulated processes in neurons. The neurobiology of neuromodulin has recently been reviewed (Benowitz & Routtenerg 1987; Liu & Storm 1990; Skene 1989). Neuromodulin is neurospecific and bovine brain contains approximately 60 pmol neuromodulin/ mg of membrane protein, making it the most abundant CaM-binding protein in brain with a concentration comparable to CaM itself (Cimler et al. 1985). Neuromodulin is a prominent constituent of neuronal growth cone membranes, comprising up to one percent of the total growth cone membrane protein and it is transported by rapid axoplasmic transport (Pfenninger et al. 1983; Skene et al. 1986). Neuromodulin has been implicated in several neuromodulatory roles including axon growth and synaptic plasticity and it has been studied extensively as a potential mediator of synapse formation and modification. Phosphorylation of neuromodulin by PKC has been correlated with synaptic LTP (Nelson & Routtenberg 1985). On the basis of the affinity of neuromodulin for CaM under physiologically relevant ionic strength (Alexander et al. 1987) and the concentrations of these two abundant proteins in brain, we predict that the majority of CaM will be complexed to neuromodulin in vivo. Neuromodulin is phosphorylated by PKC with a phosphate: protein molar ratio of 1:1. CaM decreases the rate of phosphorylation of neuromodulin by PKC, and phosphorylation prevents neuromodulin binding to CaM (Alexander et al. 1987). Phosphorylation of neuromodulin by PKC inhibits CaM binding because the phosphorylation site is within the CaM binding domain of the protein (Apel et al. 1990). In addition, phosphoneuromodulin is an excellent substrate for calcineurin, the CaM-stimulated phosphatase which is particularly abundant in brain (Liu & Storm 1989). On the basis of these observations, we have proposed that neuromodulin may bind and localize CaM at specific sites within neurons and that PKC may regulate the levels of free CaM available in neurons for stimulation of various enzymes, including adenylyl cyclases and protein kinases (Andreasen et al. 1983; Alexander et al. 1987; Lieu & Storm 1990). Evidence that PKC may regulate the levels of free CaM in neurons or neuronal-like cells in vivo comes from several different studies. For example, activation of PKC in PC 12 cells increased the levels of free CaM, presumably because of the phosphorylation of neuromodulin or related proteins (MacNicol & Schulman 1992). Mangels & BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 435 Xia et al.: Neuroplasticity Gnegy (1990) demonstrated that the ratio of cytosolic to membrane-associated CaM increases in neuroblastoma cells when PKC is activated through muscarinic receptors or by phorbol esters. Because these cells contain membrane-associated neuromodulin, the redistribution of CaM may be due to PKC phosphorylation of neuromodulin. If this hypothesis is true, then CaM may be liberated during LTP by activation of PKC and phosphorylation of neuromodulin. 8. Tissue distribution of the type I and type II adenylyl cyclases 8.1. Tissue distribution of the enzymes. Although the type III adenylyl cyclase was originally cloned from an olfactory cDNA library and is greatly enriched in this tissue, type III adenylyl cyclase mRNA is also expressed in brain, spinal cord, adrenal medulla, adrenal cortex, heart atrium, aorta, lung, retina, 293 cells, and PC-12 cells (Xia et al. 1992). In contrast, the type I adenylyl cyclase is neurospecific (Xia et al. 1993). The only bovine tissues showing a positive signal for type I mRNA were brain, retina, and adrenal medulla (Fig. 5). The weak signal seen with whole adrenal was most likely due to adrenal medulla since the cortex gave a negative signal. In addition to the expected transcript seen at approximately 11.7 kb, retina also contained a second transcript at 6.5 kb. Several cultured cell lines including neuroblastoma cell l l l l l s 1 1 11 1 1 lk ll i j I I ! I I If I111 Aii Figure 5. Northern analysis of the type I-sensitive adenylyl cyclase using mRNA from various bovine tissues. Two micrograms of poly (A)+ selected RNA samples were electrophoresed to a 1.2% agarose/formaldehyde gel. The blots were hybridized with an alpha [32P]dCTP-Iabeled cDNA probe 3C that is specific for the bovine type I adenylyl cyclase. The 0.24-9.5 kb RNA ladder from BRL was used as the molecular weight standard. Poly (A)+ selected RNA was isolated from various bovine tissues. Data are from Xia et al. 1993. 436 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 N1E-115, neuroglio hybridoma cell NG-108, rat glioma 36B-10 cell, and PC-12 cells were also analyzed and found not to express mRNA for the type I adenylyl cyclase. The restricted expression of type I adenylyl cyclase mRNA in neural tissues contrasts sharply with most of the other mammalian enzymes which show fairly broad distribution in both neural and non-neuronal tissues. 8.2. Distribution of type I adenylyl cyclase mRNA in rat retina examined by in situ hybridization. The presence of type I adenylyl cyclase mRNA in retina is consistent with the Ca 2+ and CaM sensitivities of adenylyl cyclase activities reported for retina (Gnegy et al. 1984). The distribution of type I adenylyl cyclase mRNA in retina was examined in more detail by in situ hybridization. Retina cross sections from rat, rabbit, and bovine were analyzed with a ^S-UTP labeled bovine riboprobe specific for the type I adenylyl cyclase (Xia et al. 1991; 1993). Messenger RNA for the type I adenylyl cyclase was detected in the inner segment layer of the photoreceptors (IS), and in all three nuclear layers of the neural retina (Fig. 6). The outer nuclear layer (ONL) which contains rods and cones, the inner nuclear layer (INL) which contains horizontal cells, bipolar cells, and amacrine cells, and the ganglion cell layer (GCL) all contained mRNA for type I adenylyl cyclase. The intensity of the labeling was strongest in the IS (the cytoplasm of photoreceptors) and the ONL. 8.3. Distribution of type I adenylyl cyclase mRNA in rat brain examined by In situ hybridization. The distribution of mRNA encoding the type I adenylyl cyclase in rat brain was also examined by in situ hybridization (Xia et al. 1991). In situ hybridizations in adult rat brain revealed high levels of type I adenylyl cyclase mRNA in specific areas of brain including the hippocampal formation, neocortex, entorhinal cortex, cerebellum cortex, and the olfactory system (Figs. 7 and 8). The dentate gyms in the hippocampal formation showed very intense labeling which appeared to be associated with the granule cell layer. Moderately strong labeling was also evident in association with the pyramidal cells in CA1, CA2, and CA3 layers of the hippocampus. The data reported in Figure 7 shows expression of mRNA for the CaMsensitive adenylyl cyclase in granule cells of the dentate gyrus and in pyramidal cells of the hippocampus, providing the first evidence that the enzyme is expressed in neurons. These data illustrate that mRNA for the type I adenylyl cyclase is not generally distributed throughout the brain, suggesting that it does not play a general regulatory role (e.g., in regulation of cell metabolism) and that it may be important for specific neuronal functions. Messenger RNA for the type I adenylyl cyclase is highly localized to specific regions of brain, including those areas that show LTP and have been implicated in learning and memory. Although these data do not define the function of the type I CaM-sensitive adenylyl cyclase, its mRNA distribution is consistent with the proposal that this enzyme may be important for learning and memory. 8.4. Disruption of the gene for the type I adenylyl cyclase leads to deficiencies in spatial learning. Recently, we evaluated the role of the type I adenylyl cyclase for Xia et al.: Neuroplasticity Mutant and wild type mice were analyzed for spatial learning by the Morris water-maze task, a set of assays that has been used to examine spatial learning in other mutant mice deficient in specific genes. Both sets of animals showed decreased escape latencies with training, and there were no statistically significant differences in the ability of the mutant and wild type mice to find the visible or hidden platform. However, escape latencies in the hidden platform task are a poor indicator of spatial learning and even rodents with hippocampal lesions that affect other forms of spatial learning can learn to find the hidden platform in the Morris water-maze task (Davis et al. 1992; Morris et al. 1982; 1986; Morris 1990). A better indicator of spatial learning is the transfer test in which the animal is trained to find the hidden platform at a specific site in the pool. The platform is then removed, and the number of times that a mouse swims across the target area or the time in the target quadrant is quantitated. There were significant and reproducible differences in transfer test behavior between the mutant and wild type mice. Wild type mice crossed the target area 6 ± 0 . 4 times during a 60 sec trial whereas the mutant mice crossed only 4 ± 0.4 times (p < .002). The difference in transfer ability was also evident when the time in various quadrants was analyzed. Only wild type mice showed a bias for quadrant A. They spent 42% ± 3.0 of their time in quadrant A searching for the platform. The mutant mice showed no significant preference for quadrant A (27% ± 2.0) indicating an impaired ability in this specific task (p < .001). These data illustrate that the mutant mice have a significant and lasting place navigational impairment that was dissociated from visual, motivational, or motor requirements of the test. We conclude that I-AC may play an important role for signal transduction pathways underlying some forms of learning and memory. Figure 6. Distribution of the type I Ca2+/CaM-sensitive adenylyl cyclase mRNA within various layers of bovine neural retina examined by in situ hybridizations. Cross sections of bovine eyecups hybridized with 35S-UTP labeled bovine type I adenylyl cyclase specific riboprobe 3C were treated with NTB2 emulsion for two weeks, and counterstained with cresyl violet acetate. A; light microscope photomicrograph; B; phase contrast photomicrograph; C; dark field photomicrograph. Abbreviations: IS, inner segment layer of the photoreceptor cells; ONL, the outer nuclear layer which contains nuclei of rods and cones; INL, the inner nuclear layer which contains cell bodies of horizontal cells, bipolar cells, and ainacrine cells; CCL, the ganglion cell layer. Data are from Xia et al. 1993. learning and memory by disruption of the gene for the enzyme in mice (Wu & Storm, unpublished observations). Brain coronal sections showed no detectable anatomical differences in the hippocampus, neocortex, or cerebellum between wild type and mutant mice. There were also no differences in the arrangement of cell body layers of the hippocampus or cerebellum. The mutant mice had normal motor coordination, suckling behavior, weight gain, and reproduced with litter sizes comparable to wild type mice. An analysis of Ca 2+ -sensitive adenylyl cyclase activity in membranes from the cerebellum, neocortex, hippocampus, and brain stem revealed decreases of 62%, 38%, 46%, and 6%, respectively. 9. Role of adenylyl cyclases and cAMP in longterm potentiation The general hypothesis under consideration in this paper suggests that the Ca 2+ -sensitive adenylyl cyclases may play an important role in neuroplasticity and participate in signal transduction pathways underlying longterm adaptive responses in neurons such as LTP. Because LTP results in postsynaptic increases in Ca 2+ and various areas of the hippocampus including CA1, CA3, and the dentate gyrus contain type I adenylyl cyclase (Xia et al. 1991), one might expect that activation of NMDA receptors or other Ca 2+ channels during LTP may elevate cAMP in these regions. Indeed, activation of NMDA receptors gives increased cAMP in area CA1 of the hippocampus (Chetkovich et al. 1991). Furthermore, LTP in the dentate gyrus (Stanton & Sarvey 1985a) and the CA1 (Chetkovich & Sweatt 1993) have both been reported to produce increases in cAMP. Although the cAMP increase caused by LTP in the CA1 was very small (20-25%), it may be difficult to detect these increases against a background of cells in the preparation which are not potentiated. There is also evidence that adenylyl cyclases, cAMP, and cAMP-dependent protein kinases may play an important role in some forms of LTP, particularly in the hippocampus. For example, stimulation of adenylyl cyclase BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 437 Xia et al.: Neuroplasticity Riboprobe Control Cx IG SHi 4t:J- Pir Tu Oliaonucleotide Probe '*• ( Figure 7. In situ hybridizations for type I adenylyl cyclase in middle rat brain. Rat brain sections were hybridized with either ^Slabeled antisense riboprobe 3C (A and C) or oligonucleotide probe ZX4 (£). The specificity of hybridization was demonstrated by incubation of the respective 35S-labeled probe with a 1000-fold molar excess of the unlabeled probe (B, D, F). Exposure time: 3 days (A-D), and 7 days (E-F). Abbreviations: Cx, neocortex; DG, dentate gyrus; Hi, hippocampus; IG, indusium griseum; Pir, piriform cortex; SHi, septohippocampal nucleus; Tu, olfactory tubercle. Data are from Xia et al. 1991. activity in the dentate gyrus by norepinephrine produces LTP (Hopkins & Johnston 1988; Stanton & Sarvey 1985b). The early phase of LTP in the CA1 persists only 1 to 2 hours and is initiated by a single train of high-frequency stimulation, whereas the late phase requires three or more trains of high-frequency stimulation and lasts up to 10 hours. Since D : dopamine antagonists block L-LTP and Dj receptors are coupled to stimulation of adenylyl cyclase (Frey et al. 1991) it has been proposed that cAMPstimulated protein kinase may play a pivotal role in L-LTP. In fact, dibutyryl cAMP induces increases in synaptic efficacy in the CA1 region of the hippocampus (Slack & Pockett 1991) and L-LTP in the CA1 is blocked by Rp-cAMPS, an inhibitor of cAMP-dependent protein kinase (Frey et al. 1993). Furthermore, Sp-cAMPS, which activates cAMP-dependent kinase, produces L-LTP in the CA1. 438 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 10. Biochemical model for the role of the CaMregulated adenylyl cyclases in neuroplasticity The type I adenylyl cyclase is a neural specific adenylyl cyclase (Xia et al. 1993) with a highly restricted expression in mammalian brain which includes the dentate gyrus, CA1, CA2, and CA3 regions of the hippocampus (Xia et al. 1991). Furthermore, the type III and VIII adenylyl cyclases are also expressed in the hippocampus (Cali et al. 1994; Glatt & Snyder 1993). What is the function of these enzymes in neurons, and what possible role(s) might they have in signal transduction systems important for neuroplasticity? During LTP, PKC is activated (reviewed by Linden & Routtenberg 1989), intracellular Ca 2 + increases, and neuromodulin is phosphorylated by PKC (Routtenberg 1985). We hypothesize that phosphorylation of neuromodulin at Xia et al.: Neuroplasticity Cb BS B Figure 8. /n situ hybridizations for type I adenylyl cyclase in cerebellum. Rat brain cerebellum sections were hybridized with 35Slabeled antisense riboprobe 3C (A and B). Specificity of hybridization was demonstrated by incubation of the respective 3SS-labeled probe with a 1000-fold molar excess of unlabeled probe (C). Exposure time: 3 days (A, B) and 7 days (C). Abbreviations: Cb, cerebellum; BS, brain stem. Data are from Xia et al. 1991. specific sites in neurons may regulate the concentrations of free CaM available to activate the CaM-regulated adenylyl cyclases and other enzymes, including CaM kinases and NO synthetase (Fig. 9). Long-term changes in synaptic function may be due, at least in part, to cAMP control of transcription through cAMP-responsive DNA elements such as CRE (reviewed by Mitchell & Tjian 1989). We hypothesize that synergistic stimulation of adenylyl cyclases by Ca 2+ and neurotransmitters or PKC may produce exceptionally strong or prolonged cAMP signals required for stimulation of transcription. Stimulation of transcription by cAMP, which requires the nuclear translocation of PKA (Hagiwara et al. 1993; Nigg et al. 1985), requires higher or more persistent cAMP signals than other cAMP-regulated events, particularly in neurons. For example, stimulation of PKA nuclear translocation in Aplysia neurons (Bacskai et al. 1993) and serotonin stimulation of transcription through CRE (Kaang et al. 1993) are relatively slow processes that require multiple doses of serotonin for AC activation. Robust BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 439 Xia et al.: Neuroplasticity Neurotransmitter POTENT1ATION NO Synthase \ N 0 Cytoskeleton I Remodeling of the synapse CaM Kinase u + Neurotransmitter release Retrograde messenger Phosphorylation of transcription factors Protein biosynthesis Long term synaptlc change Figure 9. Hypothetical model for the role of neuromodulin and the type I adenylyl cyclase in neuroplasticity. it is hypothesized that neuromodulin binds and concentrates CaM at specific sites in neurons and that the levels of free CaM are regulated by phosphorylation of neuromodulin by protein kinase C (PKC). During LTP, PKC is activated and Ca 2 + is mobilized. We propose that the CaM-sensitive adenylyl cyclases may be activated during LTP by Ca 2 + /CaM, neurotransmitters, and/or activation of PKC. Activation of the CaM-sensitive adenylyl cyclase(s) results in coupling of the Ca 2 + and cAMP regulatory systems during LTP. Simultaneous or ordered activation of the Ca 2 + and cAMP regulatory systems may be important for amplified cAMP signals required for transcription, synergism between Ca 2 + and cAMP activated kinases, and/or positive feedback regulation of Ca 2 + channels by cAMP-dependent kinase. Abbreviations: CaM, calmodulin; NM, neuromodulin (GAP-43); PKC, protein kinase C; NM-P, neuromodulin phosphorylated on Ser-41; PKA, cAMP-dependent protein kinase. cAMP signals may be required for transcriptional control in neurons because a significant cAMP gradient must be established from the synapse to the cell body. Elevated cAMP signals arising from synergistic activation of the type I adenylyl cyclase by Ca 2+ and neurotransmitters (or other signals) may therefore play an important role in synaptic plasticity. Alternatively, the coupling of the Ca 2+ and cAMP systems may result in simultaneous or sequentially ordered activation of the Ca 2+ and cAMP stimulated protein kinases, or provide positive feedback regulation of 440 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 Ca 2+ channels by cAMP-dependent protein kinase. All of these mechanisms are dependent upon the unique property of the CaM-sensitive adenylyl cyclases to integrate multiple signals for modification of synaptic function. ACKNOWLEDGMENT This research was supported by NIH grant NS 20498. Choi was supported by a Washington Heart Association postdoctoral fellowship and Xia was supported by a Keck neuroscience fellowship. BEHAVIORAL AND BRAIN SCIENCES (1995) 18, 441-451 Printed in the United States of America The cGMP-gated channel of photoreceptor cells: Its structural properties and role in phototransduction Robert S. Molday Department of Biochemistry and Molecular Biology, University of British Columbia, Vancouver, B.C., Canada V6T 723 Yi-Te Hsu Department of Biochemistry and Molecular Biology, University of British Columbia, Vancouver, B.C., Canada V6T 1Z3 Electronic mail: molday@unixg.ubc.ca Abstract: The cyclic GMP-gated channel responds to changes in free intracellular cGMP, and as a result, it plays a central role in the phototransduction process in rod and cone photoreceptor cells. Recent biochemical, immunochemical, and molecular biology studies indicate that this channel consists of a complex of two distinct subunits and one or more associated proteins. Primary structural analysis indicates that the a and (3 subunits contain a cGMP-binding domain, an even number of membrane-spanning segments, a voltage sensor motif and a pore region. The latter two features are found in voltage-gated channels and suggest that these two classes of channels have evolved from the same ancestral channel protein. A working model for the membrane topography of the channel subunits is proposed based on immunogold labeling studies and sequence analysis. Recent studies also indicate that calmodulin binds to the 240 kDa protein of the channel complex and modulates the sensitivity of the channel for cGMP in a Ca2+-dependent manner. The molecular properties of the channel complex and the possible role of Ca2+-calmodulin modulation of the channel during photoactivation and photorecovery are discussed in relation to the current mechanism of phototransduction in photoreceptor cells. Keywords: calmodulin; cGMP-gated channel; photorecovery; phototransduction; rod photoreceptor cells 1. Introduction The cyclic GMP-gated channel of vertebrate rod photoreceptor cells plays a central role in the phototransduction process. In the dark a relatively high concentration of cGMP in the rod outer segment (ROS) maintains a significant number of channels in their open state by the direct, reversible binding of cGMP to the channels (Fesenko et al. 1985; Yau & Baylor 1989; Yau & Nakatani 1985a). This allows for a steady influx of Na + and Ca 2+ into the outer segment and maintains the cell in a partially depolarized state. The influx of Ca 2+ through the channel is balanced by an efflux of Ca 2+ by the Na + /Ca 2+ -K+ exchanger in the ROS plasma membrane, thereby maintaining the intracellular level of Ca 2+ at a relatively constant level (Yau & Nakatani 1984a). Low intracellular Na + concentration is maintained by the balanced extrusion of Na + by Na + -K + ATPase localized in the plasma membrane of the rod inner segment. Photobleaching of rhodopsin in the ROS disk membrane leads to an activation of the visual enzyme cascade system (Chabre & Deterre 1989; Stryer 1986; 1991) and a reduction in the concentration of cGMP by a phosphodiesterase catalyzed reaction (Fig. 1). This results in the closure of the cGMP-gated channels and a transient hyperpolarization of the cell. The closure of the channels © 1995 Cambridge University Press 0U0-525XI95S9.0O+.10 to Ca 2+ results in a reduction in intracellular Ca 2+ because the Na + /Ca 2 + -K + exchanger continues to extrude Ca 2+ from the outer segment. The reduction in Ca 2+ has been shown to be important in the recovery of the rod outer segment to its dark state and in light adaptation (Kaupp & Koch 1992; Koch & Stryer 1988; Matthews et al. 1988; Yau & Nakatani 1985b). This feedback mechanism is suggested to occur through the interaction of Ca 2+ -binding proteins with various proteins involved in the phototransduction process, and in particular, guanylate cyclase (Koch & Stryer 1988). Cone photoreceptors appear to have a similar visual cascade system and a related cGMP-gated channel (Bonigk et al. 1993; Haynes & Yau 1985; Picones and Korenbrot 1992; Watanabe & Murakami 1991). The physiological properties of the cGMP-gated channel of ROS have been extensively studied and recently reviewed (Kaupp 1991; Yau & Baylor 1989). Generally, cGM P directly and cooperatively binds to and activates the channel with a K1/2 of 10-50 u,M and a Hill coefficient (n) of 1.7-3.5. Analogues of cGMP with substituents at position 8, suchas8-bromocyclicGMP, are also effective activators of the channel in the (JLM range (Zimmerman et al. 1985). Although the channel is cation selective, it exhibits a low degree of specificity for various monovalent and divalent cations (Furman & Tanaka 1990; Hodgkin et al. 1985; 441 Molclay & Hsu: Photoreceptor cells Exchanger - 4 No' cGMP-gated Channel Phosphodiesterase I GTP Light Ca" Disk Membrane Plasma Membrane Figure 1. A diagram showing the basic reactions of the visual transduction pathway. In the dark, elevated cGMP concentrations maintain a relevant number of cGMP-gated channels in their open state. This allows the influx of Na+ and Ca2+ into the outer segment and maintains the photoreceptor cell in a partially depolarized state. Photobleaching of rhodopsin, involving the photoisoinerization of 1 l-cis retinal to its all-trans isomer, results in the formation of activated (meta II) rhodopsin which catalyzes the exchange of bound GDP for GTP on transducin. Transducin will in turn activate phosphodiesterase which catalyzes the hydrolysis of cGMP to 5'-GMP. The decrease in free cGMP concentration will cause the channel to close as cGMP dissociates from the channel. Closure of the channel to the influx of Na+ and Ca2+ will result in a transient hyperpolarization of the cell. The recovery of the rod outer segment to its dark state occurs through (1) the inactivation of rhodopsin by a rhodopsin kinase (RK) catalyzed phosphorylation reaction and the binding of arrestin (Ar); (2) inactivation of transducin by hydrolysis of bound GTP to GDP; (3) inhibition of phosphodiesterase by the rebinding of the inhibitory subunits to the catalytic subunits of phosphodiesterase; (4) resyn thesis of cGMP from GTP by guanylate cyclase (GC); and (5) the reopening of the cGMP-gated channel as cGMP levels increase. Schnetkamp 1990; Yau & Nakatani 1984b). The rod channel is permeable not only to the physiological ions, Na + , Ca 2+ , and Mg 2+ , but also to other monovalent and divalent cations including Li+, K+, Cs+, Rb + , Mn 2+ , andBa 2+ . Under physiological conditions the channel is relatively insensitive to voltage changes. Divalent cations have been shown to block the flow of ions through the channel, thereby reducing the effective conductance of the channel (Yau & Baylor 1989). L-cis diltiazem has been shown to be a stereoselective inhibitor of channel activity in the micromolar concentration range (Koch & Kaupp 1985; Stern et al. 1986). The channel has been localized to the plasma membrane of ROS and has a density of 200-500 channels per |xm2 (Bodoia & Detwiler 1985; Cook et al. 1989; Zimmerman & Baylor 1986). Recently, biochemical, immunochemical, and molecular biology studies have provided insight into the molecular properties of the cGMP-gated channel of rod and cone photoreceptors. In this target article, issues related to the molecular composition and structural features of the cGMPgated channel of photoreceptor cells are addressed, and the regulation of the cGMP-gated channel by ealmodulin is discussed in terms of its possible role in phototransduction. 442 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 2. Molecular composition of the cGMP-gated channel 2.1. Identification and purification of the a-subunit. In the latter half of the 1980s several laboratories had identified different polypeptides as candidates for the cGMP-gated channel subunit of ROS. Cook et al. (1987) isolated a 63 kDa polypeptide from CHAPS detergent-solubilized bovine ROS by ion exchange and red-dye affinity chromatography. This preparation exhibited cGMP-dependent ion flux activity when reconstituted into liposomes, but, in contrast to ROS membranes, the channel activity was not inhibited by micromolar concentrations of l-cis diltiazem. At about the same time, Matesic and Liebman (1987) described the partial purification of a 39 kDa polypeptide from cholate-treated bovine ROS. This 39 kDa protein, which was reported to exhibit immunochemical crossreactivity with rhodopsin, could be photoaffinity labeled with 8'-azido-cGMP and reconstituted into vesicles for measurements of cGMP-dependent channel activity. On this basis, they concluded that the ROS channel was composed of 39 kDa subunits. Molday & Hsu: Photoreceptor cells PMs 5E11 PMc 1D1 1.0 - [cGMP] (pM) Figure 2. Immunoaffinity purification and functional reconstitution of the cGMP-gated channel complex of bovine rod outer segments (ROS). (a) SDS gels of ROS membranes (lane a) and the iminunoaffinity-purified channel (lane b) stained with Coomassie Blue (CB) or transferred to Immobilon membranes and labeled with monoclonal antibodies PMs 5E11 against the 240 kDa channel protein and PMc 1D1 against the a-subunit of the channel, (b) cCMP-dependent Ca2+ efflux from liposomes reconstituted with the immunoaffinity purified channel. The sigmoidal curve was calculated usinga Km of33 (iM and a H ill coefficient of3.3 for cGMP. For these studies, CHAPS detergent-solubilized ROS membranes were added to a PMc 6E7 anti-channel monoclonal antibody-Sepharose column and eluted with a N-terminal peptide corresponding to the epitope for this antibody (Molday et al. 1991). The purified channel complex was reconstituted into lipid vesicles by the procedure of Cook et al. (1987). Shinozawa et al. (1987) suggested that a 250 kDa protein constituted the cGMP-gated channel subunit on the basis of photoaffinity labeling of frog ROS membranes with 3 HcGMP and channel activity measurements in planar lipid bilayers. Finally, Clack and Stein (1988) reported that purified opsin preparations exhibited cGMP-activated single-channel activity, suggesting that the photopigment protein rhodopsin itself may function as the cGMP-gated channel. Controversy surrounding the identity of this channel subunit was initially resolved when a series of immunochemical studies established the 63 kDa polypeptide as a major subunit of the channel. This subunit is now referred to as the a-subunit. A monoclonal antibody which specifically labeled the 63 kDa channel subunit in both ROS membranes and purified channel preparations (Cook et al. 1989) was shown to quantitatively immunoprecipitate both cGMP-gated channel activity and a complex consisting of a 63 kDa and a 240 kDa polypeptides (Molday et al. 1990). In contrast, an antirhodopsin monoclonal antibody which selectively immunoprecipitated rhodopsin did not immunoprecipitate cGMPdependent channel activity. More recently, a monoclonal antibody generated against a peptide corresponding to the N-terminal segment of the 63 kDa polypeptide (Molday et al. 1991) has been used to immunoaffinitypurify the channel complex from detergent-solubilized ROS in a functionally active form. As shown in Figure 2a, the immunoaffinity-purified channel consists of the 63 kDa a-subunit of the channel and an associated 240 kDa polypeptide. The 240 kDa polypeptide had been previously observed by Cook et al. (1987) in some of their earlier channel preparations. The immunoaffinity-purified channel complex, when reconstituted into liposomes, exhibits cGMP-dependent channel activity. The K1/2 of 33 (xm and Hill coefficient of 3.3 for cGMP (Fig. 2b) are in general agreement with the values obtained for red-dye purified channel preparation as reported by Cook et al. (1987). More recently, Brown et al. (1993) have labeled the 63 kDa a-subunit with 8-p-azidophenacylthio-cGMP, a photoaffinity derivative of cGMP, thus confirming the presence of a cGMP binding site. A cGMP-Sepharose column has been used by Hurwitz and Holcombe (1991) as an affinity matrix to purify the channel from detergent-solubilized ROS preparations. This channel preparation was reported to consist of an 80 kDa polypeptide which was sensitive to 1-cis diltiazem. Since this polypeptide binds a monoclonal antibody against the 63 kDa channel subunit, it is likely that this 80 kDa polypeptide is related to the 63 kDa a-subunit of the channel. The higher apparent molecular weight as determined by SDS gel electrophoresis may be due to anomalies in the migration of this channel subunit under different electrophoresis conditions. Alternatively, it may be the result of the isolation of the unprocessed form of the channel subunit. The observation that the channel isolated by cGMP-Sepharose affinity chromatography is inhibited by 1-cis diltiazem has been interpreted by Hurwitz and Holcombe (1991) to indicate that 1-cis diltiazem sensitivity is a property of the full length 80 kDa channel subunit, and not the truncated 63 kDa form. However, recent studies by Chen et al. (1993) indicate that the 1-cis diltiazem sensitivity of the channel as found in native ROS is not obtained when the full-length 80 kDa a-subunit is BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 443 Molday & Hsu: Photoreceptor cells expressed by itself, but instead requires the coexpression of the a and 3 subunits (see below). This suggests that the channel sensitivity to 1-cis diltiazem is inherent in the 3-subunit or the interaction of the 3 subunit with the a subunit. The differences in apparent molecular weights and 1-cis diltiazem sensitivities for the different isolated channel preparations still remain to be clarified. More recently, calmodulin-Sepharose chromatography has been used to obtain a highly enriched channel preparation (Hsu & Molday 1993). The channel isolated by this method was found to contain the 63 kDa and 240 kDa polypeptides as the major components and several other calmodulin binding proteins as minor contaminants. These studies taken together indicate that the a-subunit of apparent molecular mass (Mr) 63 kDa is a major component of the native channel complex. Previous reports indicating that a 39 kDa polypeptide and rhodopsin exhibit cGMP-dependent channel activity appear to be a result of contamination of these preparations identified a second subunit (P-subunit) by screening a human cDNA library with a partial clone of the a-subunit under low stringency conditions. Sequence analysis of this 3-subunit indicates that it exhibits a 30% overall sequence identity to the a-subunit and 50% identity within the cGMP binding domain. Alternative splicing results in two transcripts which differ in size. The shorter transcript codes for a polypeptide of 623 amino acids with a calculated molecular weight of 70,843 and the longer transcript codes for a polypeptide of 909 amino acids having a molecular weight of 102,330. The two polypeptides are identical in sequence except for an extended N-terminal segment of 286 amino acids for the latter. Patch-clamp studies of human kidney 293 cells transfected with the 3-subunit cDNA indicate that neither form of the 3-subunits when expressed alone is functionally active (Chen et al. 1993). However, when coexpressed with the a-subunit, cGMP-gated channel activity can be measured that more closely resembles the activity with this channel complex. An earlier report of Shinozawa found in the ROS membranes. In particular, coexpression et al. (1987) suggesting that the channel subunit is a 250 kDa polypeptide may be related to a recent finding that the channel contains a second subunit referred to as the 3-subunit (see below). of the a and 3-subunits results in rapid bursts of channel openings or a flickering response and u,M sensitivity of the channel to 1-cis diltiazem as found for the channel in ROS membranes. In contrast, the expressed'a-subunit, by itself, displays a more stable opening and closing behavior and is over an order of magnitude less sensitive to 1-cis diltiazem (Chen et al. 1993; Kaupp et al. 1989). Immunocytochemical studies have confirmed that the 3-subunit is present in human rod outer segments (Chen et al. 1993). An antibody raised against a peptide corresponding to the C-terminus of the 3-subunit was observed to label the outer segment layer of human rod photoreceptor cells, but not cone photoreceptor cells. However, an antibody generated against a peptide segment near the N-terminus of the polypeptide encoded by the longer 3 transcript and not present in the polypeptide encoded by the shorter transcript did not label photoreceptor outer segments. On the basis of these studies, Chen et al. (1993) have suggested that the 3-subunit in ROS is encoded by the shorter transcript. Direct studies showing that the shorter transcript is in fact the predominant form expressed in ROS, however, have not been carried out, leaving some question as to whether the long or short form of the 3-subunit is preferentially expressed in rod photoreceptors. Other issues that need to be addressed include the detection and identification of the 3-subunit in both isolated ROS membranes and isolated channel preparations and the stoichiometric relationship between the a and 3-subunits. 2.2. Cloning and expression of the a-subunit. Direct evidence indicating that the 63 kDa polypeptide is a major subunit of the cyclic GMP-gated channel was obtained from the cloning and expression studies of Kaupp et al. (1989). Oligonucleotide probes prepared from tryptic peptide sequences of the 63 kDa polypeptide were used to screen a bovine retinal cDNA library. A fulllength cDNA clone was obtained which encoded a 79.6 kDa polypeptide containing up to six putative transmembrane segments and a cGMP binding domain. Injection of Xenopus oocytes with mRNA derived from the cloned cDNA resulted in the functional expression of a cGMP-gated channel having many of the electrophysiological properties found for the channel in ROS membranes (Kaupp 1991; Kaupp et al. 1989). The expressed channel has been reported to be cooperatively activated by cGMP with a K1/2 of 52 |xM and a Hill coefficient of 1.75, to display a cation selectivity similar to that found in ROS, and to exhibit a single channel conductance of 20 pS. The expressed channel, however, is considerably less sensitive to 1-cis diltiazem than the native channel found in ROS. The a-subunit expressed in both monkey kidney COS-1 cells and Xenopus oocytes has an apparent Mr of 78 kDa indicating that the full length polypeptide is expressed in these cells (Molday et al. 1991). Dhallan et al. (1992) have cloned and functionally expressed the cDNA for the a-subunit of the human ROS channel in embryonic kidney 293 cells. 2.3. Identification and characterization of the p-subunit. The cGMP-gated channel has been generally considered to consist of four or five subunits (Kaupp 1991), each of which contains a cGMP binding site. This is based on the finding that cGMP cooperatively activates the channel with a Hill coefficient of 2-3.5. The Hill coefficient gives a minimum number of cGMP molecules required for activation of the channel. Until recently, it had been assumed that the cGMP-gated channel consisted of identical a-subunits. Chen et al. (1993), however, have recently 444 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 2.4. The 240 kDa channel-associated polypeptide. In addition to the 63 kDa a-subunit, cGMP-gated channel preparations from bovine ROS also contain another prominent polypeptide having an apparent Mr of 240 k by SDS gel electrophoresis under reducing conditions (Fig. 2a). Several monoclonal antibodies have been generated against the 240 kDa polypeptide (Molday et al. 1990; Wong & Molday 1986). One antibody was found to crossreact with red blood cell spectrin and brain fodrin. On the basis of this crossreactivity, its size, and its ability to bind calmodulin (see below), it had been initially suggested that the 240 kDa polypeptide may be a member of the spectrin family of cytoskeletal proteins. However, recent rotary shadowing studies have indicated that the 240 kDa Molday & Hsu: Photoreceptor cells polypeptide does not have the extended rod-shaped structure observed for spectrin (L. Molday & R. Molday, unpublished results). Furthermore, initial peptide sequence studies indicate that the 240 kDa polypeptide is not related to spectrin, but instead consists of two components, one of which is the (3-subunit of the channel (M. Illing, A. Williams & R. Molday, unpublished results). The finding that the P-subunit is contained within the 240 kDa polypeptide explains the reported findings that cGMP and 8-p-azidophenacylthio-cGMP photoaffinity label a high Mr component (240-250 kDa) in frog and bovine ROS and channel preparations (Shinozawa et al. 1987; Brown et al. 1993). On the basis of these recent studies the cGMP-gated channel appears to consist of at least two subunits. The a-subunit appears to be the major functional subunit since cGMP-gated activity can be obtained by expression of this subunit alone; the (J-subunit interacts with the a-subunit in rod photoreceptors to alter some of the properties of the channel and regulate the cGMPdependent activity of the channel (see below). The channel complex also contains at least one additional polypeptide which, together with the B-subunit, constitutes the 240 kDa polypeptide observed by SDS gel electrophoresis in purified channel preparations. Additional studies are needed to further identify, characterize, and quantify these components and to define their role in the structure and function of the channel in rod and cone photoreceptor membranes. 3. Primary structural features of the cGMP-gated channel subunits 3.1. Structure of the at-subunit. The primary structure of the bovine cGMP-gated channel a-subunit was first determined by cloning and sequence analysis of its cDNA (Kaupp et al. 1989). The full-length cDNA encodes a polypeptide chain of 690 amino acids with a calculated Mr of 79,601. A stretch of 80-100 amino acids close to the C-terminus of the channel (Fig. 3a) has been identified as the cGMP binding domain on the basis of its sequence similarity to the tandem cGMP binding domains of bovine lung cGMP-dependent protein kinase and to other cyclic nucleotide binding proteins (Kaupp et al. 1989; Kaupp 1991), and on the basis of site-directed mutagenesis studies (Altenhofen et al. 1991). The N-terminal region contains a hydrophilic stretch of about 60 amino acids with a lysine-rich segment. Hydropathy plots suggest the existence of six relatively hydrophobic segments designated H1-H6, which are of sufficient length to span the lipid bilayer. With the exception of H4 and H5 the hydrophobicity index of these segments is generally lower than that observed for transmembrane segments of well-characterized membrane proteins. Accordingly, it remains to be experimentally determined if all or only some of these segments span the membrane. Five consensus sequences (Asn-X-Thr or Asn-X-Ser) for N-linked glycosylation are present at asparagine residues 90, 91, 177, 327, and 423. Lectin binding and endoglycosidase studies have confirmed that the a-subunit is glycosylated at a single site (Wohlfart et al. 1989). The a subunit of the rod cGMP-gated channel has been cloned from several species. The mouse and human sequences are about 85% identical to bovine (Dhallan et al. 1992; Pittler et al. 1992), whereas the chicken rod subunit is 75.7% identical (Bonigk et al. 1993). The chicken cone a-subunit has also been cloned and shown to be 65% identical to the bovine rod subunit (Bonigk et al. 1993). The cone a-subunit has the same structural features as the rod channel and similar electrophysiological properties when expressed in Xenopus oocytes. The molecular weight of the bovine channel a-subunit calculated from the cDNA-derived protein sequence (79.6 kDa) is considerably larger than the apparent molecular weight of 63 kDa observed by SDS-PAGE. This difference is not due solely to anomalous migration of the channel subunit on SDS gels, because the channel asubunit expressed in mammalian cells and Xenopus oocytes migrates with an apparent Mr of 78 k (Molday et al. 1991). The lower molecular weight of the channel subunit in ROS is primarily due to the absence of the' first 92 amino acids as shown by N-terminal sequence analysis. This truncated form of the channel consisting of 598 amino acids has a calculated molecular weight of 69.4 k. The difference in molecular weight of the a-subunit as determined by SDS gel electrophoresis (63 kDa) and that calculated on the basis of the amino acid composition of the truncated channel (69.4 kDa) is attributed to the inaccuracy of molecular weight determinations of membrane glycoproteins by SDS gel electrophoresis (Molday et al. 1991). Immunochemical labeling studies for light microscopy and electron microscopy employing a monoclonal antibody specific for the N-terminus of the 63 kDa a-subunit has confirmed that this shortened form of the channel is the major form present in ROS and is not a result of nonspecific proteolysis during the preparation of ROS membranes. It is likely that a photoreceptor-cellspecific posttranslational cleavage reaction results in the truncated form of the channel subunit found in ROS, although this remains to be determined experimentally. The a-subunit of other mammalian species has also been observed to migrate on SDS gels with an apparent Mr of 63 k, suggesting that truncation of the a-subunit is a general characteristic of the mammalian rod channel (Molday et al. 1991). Comparison of the rod and cone a-subunits in chicken retinal cell extracts with that of the a-subunits transiently expressed in COS-1 kidney cells indicates that a similar cleavage reaction occurs in chicken rod and cone photoreceptor cells (Bonigk et al. 1993). The cleavage site between serine 92 and serine 93 in the bovine sequence N-N-S-S-N-K-E (N-Asn, S-Ser, K-Lys, E-Glu) is conserved in the human and mouse channel a-subunit (Fig. 3b). Related although nonidentical sequences are found in the chicken rod and cone channel subunits; it remains to be determined if this segment is also the site of cleavage for the chicken channel rod and cone a-subunits. Sequence analysis of the rod and cone channel a-subunit has also revealed two structural features in voltage-gated channels (Fig. 3b; Bonigk et al. 1993). A voltage sensor motif related to the S-4 segment of the voltage-gated shaker K+ channel is found in photoreceptor, as well as olfactory, cyclic nucleotide-gated channel subunits (Goulding et al. 1992; Jan & Jan 1990). This motif consists of a repeating sequence of a positively charged Arg (R) or Lys (K) residue separated by two predominantly hydrophobic amino acids. Voltage-gated channel BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 445 Molday & Hsu: Photoreceptor cells CYCLIC NUCLEOTIDE BINDING DOMAIN Bovine Rod (a) YSPGDYICKKGDIGREMYIIKEGKLAWAD-DGITQFWLSDGSYFGEISILNIKGBKAGNRRTANIKSIGYSDLFCLSKD : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : :: I : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : ; : i Human Rod (a) YSPGDYICKKGDIGREMYIIKEGKLAWAD-DGVTQFWLSDGSYFGEISIUIIKGSKAGIIRRTANIKSIGYBDLFCLBKD Mouse Rod (a) YSPGDYICKKGDIGREMYIIKEGKIAVVAD-DGITQFVVLSDGSYFGEISILNIKGSKAGNRRTANIKSIGYSDLFCLSKD Chicken Rod (a) YSPGDYICRKGDIGREMYIIKEGKLAWAD-DGVTQFWLSDGSYFGEISILNIKGSKAGNRRTANIRSIGYSDLFCLSKD Chicken Cone (a) FSPGDYICKKGDIGREMYIIKEGKLAWAD-DGITQFWLSDGSYFGEISILNIKGSKSGNRRTANIRBIGYSDLFCLSKD Bovine Olfactory FSPGDYICRKGDIGKEMYIIKEGKLAWAD-DGVTQYALLSAGSCFGEISILNIKGSKMGNRRTANIRSLGYSDLFCLSKO Human Rod (/3) YLPNDYVCKKGEIGREMYIIQAGQVQVLGGPDGKSVLVTLKAGSVFGEISLLAVGG—GNRRTANWAHGFTHLFILDKK (498-577) (496-575) (491-570) (452-531) (545-624) (475-554) (355-432) • VOLTAGE SENSOR MOTIF (S4) Shaker K+ Bovine Rod ( a ) I L R V I R L V R V F R I F K L B R H B K : t : : i : E I R L N R L L R I S R M F E F F Q R T B : : : : : : : : : Human Rod ( a ) Mouse Rod ( a ) Chicken Rod ( a ) Chicken Cone ( a ) E L R F N R L L R I A R L F E F F D R T B : : : : : (267-287) : : : : : : : : : : • E I R L N R L L R F B R M E I R L N R L L R I S R M : : : : : : : : : B L R I N R L L R V A R M : (360-380) F F : F E E : E F F : F F F : F : : : : : : Q Q : Q R R : R T B T B : i T B : : (265-285) (260-280) (221-241) (314-334) t Bovine Olfactory E V R F N R L L H F A R M F E F F D R T B (244-264) Human R o d (/3) : : : : : s : : t L L R L P R C L K Y M A F F E F N S R L E (132-152) * Shaker K + D A F W W A V V T M T T V Q Y O D M T P V (431-451) Bovine Rod (a) Y 8 L Y W S T L T L T T I 0 (349-367) Human Rod (a) Y 8 L Y W 8 T L T L T T I 0 - - E T P P P (347-365) Mouse Rod (a) Y 8 L Y W 8 T L T L T T I 0 - - E T P P P (342-360) Chicken Rod (a) Y 8 L Y W B T L T L T T I G (303-321) PUTATIVE PORE Chicken Cone (a) Bovine Olfactory Human Rod (0) Y B L Y H B T L T L T T I G E T P P P E T P P P :::':: E T P P P Y C L Y W B T L T L T T I G - - E T P P P : : : : : : : : R C Y Y F A V K T L I T I Q G L P D P (396-414) (326-344) (206-224) * N-TERMINAL CLEAVAGE SITE Bovine Rod (a) N N Human Rod (a) N N 8 8 N K D (89-95) Mouse Rod (a) N N S 8 N K D (84-90) Chicken Rod (a) N N N 8 N K D : > : 8 N N T N E D (53-59) Chicken Cone (a) s s NK B (90-96) (145-151) Figure 3. Sequence alignment of various structural domains of cyclic nucleotide-gated channels from different species, (a) Alignment of the potential cyclic nucleotide binding sites of the bovine rod a-subunit (Kaupp et al. 1989); human and mouse rod a-subunit (Pittler et al. 1992); the chicken rod and cone a-subunits (Bonigk et al. 1993); the bovine olfactory channel subunit (Ludwig et al. 1990); and the rod (3-subunit (Chen et al. 1993). A high degree of sequence identity within the nucleotide binding domain is observed, (b) Sequence alignment of the voltage-sensor-like motifs (S4), pore regions, and putative N-terminal cleavage sites. subunits generally have seven repeating sequences, whereas the nucleotide-gated channels have four or fewer (Fig. 3b). It has been suggested that in voltage-gated channels this sequence spans the membrane bilayer. A change in the voltage across the membrane is thought to induce a conformational change in this segment which in turn affects the ion translocating properties of the channel. In the photoreceptor and olfactory cyclic nucleotidegated channel this voltage-sensor-like motif is more limited and contains several negatively charged residues. 446 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 These negatively-charged residues or ineffective coupling of this segment to the pore region may be responsible for the insensitivity of these cyclic nucleotide-gated channels to physiological changes in voltage (Kaupp 1991; Yau & Baylor 1989). Photoreceptor and olfactory channel subunits also contain a segment that has considerable sequence similarity to the pore region of the voltage-gated K+ channel (Bonigk et al. 1993; Goulding et al. 1992; Heginbotham et al. 1992). For voltage-gated channels, this region consists Molday & Hsu: Photoreceptor cells of a stretch of about 21 amino acids, which is considered to extend into the lipid bilayer as two antiparallel (3 strands connected by a hairpin turn (Durell & Guy 1992; Yellen et al. 1991). Toxin binding studies and site-directed mutagenesis studies of the K + channel have provided support for the functional role of this segment in the translocation of ions across membranes. Yool and Schwartz (1991) have reported that mutations in this pore region of the K + alters the ion selectivity of the channel without affecting the gating properties. Heginbotham et al. (1992) have also shown that deletion of two amino acids (Tyr-Gly-) in the pore region of the K + channel that are absent in cyclic nucleotide-gated channels results in a loss of K + selectivity and introduction of a divalent ion blockage of the channel as found in cGMP-gated channels. In the case of cyclic nucleotide-gated channels, exchange of the pore sequence of the rod a-subunit with the pore sequence of the olfactory subunit has been shown to impart ionconducting properties of the olfactory channel onto the rod channel (Goulding et al. 1993). Site-directed mutagenesis studies have also defined the residue in the pore region which is responsible for the observed divalent cation block of conductance (Root & McKinnon 1993). In these studies mutation of glutamic acid 363 to a glutamine residue has been shown to effectively eliminate the blockage of the channel conductance by external Mg 2+ and Ca 2 + . These studies support the role of this segment of the channel as the primary structural domain involved in ion translocation through the channel. 3.2. Structure of the p-subunit. The P-subunit of the rod channel has many of the structural features of the a-subunit (Chen etal. 1993). In particular this subunit has a modified voltage-sensor-like motif and a pore region in positions similar to that of the a-subunit (Fig. 3b). The putative pore region of the p-subunit, however, unlike the pore region of the a-subunit, does not have a negatively charged residue. Hydropathy plots also show the presence of up to six hydrophobic segments, although some of these are less hydrophobic than corresponding segments of the a-subunit. Unlike the a-subunit, the P-subunit does not contain a consensus sequence for N-linked glycosylation, suggesting that this subunit is either not glycosylated or is O-glycosylated on serine or threonine residues. cated form of the a-subunit as found in bovine ROS membranes, there are three potential glycosylation sites at positions 177, 327, and 423. Position 177 is within the HI hydrophobic segment and is not considered as a likely site for glycosylation. Immunocytochemical and biochemical studies using antibodies generated against synthetic peptides encompassing glycosylation sites at 327 and 423 have confirmed that asparagine 327 contains the N-glycosylation site and is localized on the extracellular surface of the ROS plasma membrane. On the basis of labeling studies and sequence analysis, we have proposed a working model for the organization of the a-subunit of the channel within the membrane as shown in Figure 4. In this model the N-terminal and C-terminal segments are oriented on the cytoplasmic side and the glycosylation site at Asn 327 is exposed on the extracellular surface of the ROS plasma membrane. The voltage sensor motif S4 is found in voltage-gated channels and hydrophobic segments H1-H5 are viewed as transmembrane segments. The pore region located between H4 and H5 is viewed as extending into the membrane in an antiparallel P conformation as suggested for the K + channel (Durell & Guy 1992). The cGMP binding domain is located near the C-terminus and likely interacts with the pore region to control the ion translocation properties of the channel. This model has many general structural features common to voltage-gated channels and suggests that the two families of ion channels may have evolved from the same ancestral channel. Similar models have been developed independently by other groups on the basis of sequence similarities of the cyclic nucleotidegated channels to voltage-gated channels (Goulding et al. 1992; Heginbotham et al. 1992). The P-subunit is envisioned to have a similar topography to that of the a-subunit with both a voltage-sensormotif and a pore region (Chen et al. 1993). The pore Pore Extracellular CHO Intracellular 4. Proposed topological model for the channel subunits in the membrane Immunocytochemical labeling studies have led to some insight into the organization of the a-subunit of the channel in ROS membranes. Monoclonal antibodies directed against epitopes near the N-terminus and the C-terminus have been observed to label the cytoplasmic surface of the ROS plasma membrane (Cook et al. 1989; Molday et al. 1991; 1992). These studies indicate that the N and C termini of the a-subunit of the native channel are localized on the cytoplasmic side of the ROS plasma membrane and support the view that this subunit contains an even number of membrane-spanning segments. The identity and orientation of the glycosylation site of the a-subunit has been determined using peptidedirected antibodies (Wohlfart et al. 1992). In the trun- .- - -NH, •-'-NH, Figure 4. Working model for the organization of the a-subunit of bovine rod cGMP-gated channel within the lipid bilayer. The first 92 amino acids (dashed line) as predicted from cDNA sequence analysis are absent in the channel in ROS membranes possibly by a posttranslational cleavage reaction. Segments labeled H1-H5 and the S4 voltage-sensor-like motif are considered as transmembrane segments and the segment between H4 and H5 is the pore region. Immunogold labeling studies have established the N and C termini on the cytoplasmic side (Cook et al. 1989; Molday et al. 1991) and asparagine Asn-327 containing a N-linked oligosaccharide chain on the extracellular side (Wohlfart et al. 1992). The cGMP binding domain is located near the C-terminus. The P-subunit is suggested to have a similar topographical organization. BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 447 Molday & Hsu: Photoreceptor cells regions of both the a- and 3-subunits would likely line the central cavity of the channel and be responsible for many of the ion translocating properties of the channel. There are, however, some sequence differences in the a- and P-subunits. Of particular interest is the absence of a negatively charged glutamic acid residue in the (3-subunit. This may affect the binding affinity of divalent cations to the channel and alter divalent cation blockage properties found when the a-subunit alone is expressed for structure-function studies (Root & MacKinnon 1993). (Hsu & Molday 1993). When detergent solubilized ROS membrane proteins are passed through a calmodulinSepharose column in the presence of Ca 2+ and are subsequently eluted in Ca 2+ -free buffer containing EDTA, the cGMP-gated channel complex consisting of the 63 kDa a-subunit and the 240 kDa protein is observed as the major component as visualized by SDS gel electrophoresis. Western blots of the calmodulin-purified and immunoaffinity-purified channel complex labeled with 125 I-calmodulin indicate that calmodulin binds to the 240 kDa polypeptide, but not to the 63 kDa a-subunit. 5. The rod cGMP-gated channel is a member of a family of nucleotide-gated ion channels 6.2. Calmodulin modulation of cGMP-gated channel activity in ROS membranes. The effect of calmodulin on cGMPgated channel ion flux activity was measured in ROS vesicles loaded with the Ca 2+ -sensitive dye Arsenazo III (Hsu & Molday 1993). In the absence of calmodulin, the channel in ROS membrane vesicles was observed to be cooperatively activated by cGMP with a Km of 19 |xM and a Hill coefficient of 3.7 for cGMP (Fig. 5). In the presence of Ca 2+ -calmodulin the dose-response curve for cGMP was shifted to the right so that the Km increased to 33 (JLM, but the Hill coefficient and the maximum rate (Vmax) of ion influx essentially remained unchanged. Although this shift in Km in the presence of Ca 2+ -calmodulin is modest, the presence of calmodulin can result in a 5-6-fold decrease in channel activity at relatively low cGMP (~12 (JLM) as may be found under physiological conditions, because the cooperativity is high. This change in the apparent Km likely reflects an effect of Ca 2+ -calmodulin on the binding affinity of the channel for cGMP and not on ion translocation properties of the channel. This is supported by the finding that the calmodulin effect is observed for different cations and does not affect the maximum velocity of ion influx (Hsu & Molday 1993). Caretta et al. (1988) had earlier reported a Two major classes of ion channels include the voltagegated channels and the ligand-gated channels (Jan & Jan 1989; Miller 1989; 1991). Voltage-gated channels respond to changes in voltage across a membrane and are represented by K + selective channels, Na + selective channels, and Ca 2+ selective channels. Ligand-gated channels re- spond to the binding of extracellular ligands and are represented by such complexes as the acetylcholine receptor, GABA receptor, and glycine receptor. Cyclic nucleotide-gated channels constitute another family of channels that respond to intracellular cyclic nucleotides. In addition to cGMP-gated channels of rod and cone photoreceptor cells (Bonigketal. 1993;Kauppetal. 1989; Pittler et al. 1992), the olfactory neurons contain channels that respond to cAMP as well as cGMP (Nakamura & Gold 1987). The a-subunit of olfactory channels has been cloned from bovine (Ludwig et al. 1990), rat (Dhallan et al. 1990), and catfish (Goulding et al. 1992) olfactory neurons. They share a high degree of sequence identity (~57%) to the a-subunit of rod and cone photoreceptor channels and contain a cyclic nucleotide binding domain, a voltage-sensor-like motif, and a pore region (Figs. 3a and 3b). It is likely that the olfactory channel, like the rod and cone channels, contains more than one subunit, although this remains to be determined. Recent molecular biology and electrophysiology studies indicate that cyclic nucleotide gated channels are also present in other cells and tissues including pineal (Dryer & Henderson 1991), heart (Biel et al. 1993), kidney (Ahmad et al. 1990; Light et al. 1990), and bipolar cells (Nawy & Jahr 1990). Initial cloning studies indicate that the corresponding a-subunits from different tissues show a high degree of sequence similarity to the rod, cone, or olfactory channel a-subunit (Ahmad et al. 1992; Biel et al. 1993; Hundal et al. 1993). Detailed studies are needed to define the subunit composition, structure, location, and role of these channels in cell function. 6. Interaction of calmodulin with the cGMP-gated channel 6.1. Binding of calmodulin to the channel. Earlier studies had indicated that ROS contain significant amounts of the calcium binding protein calmodulin (Kohnken et al. 1981), but the identity of the target proteins for calmodulin.in ROS was not determined. Recently, we have shown by calmodulin affinity chromatography and Western blotting that the cGMP-gated channel complex is a major calmodulin binding protein of ROS membranes 448 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 cGMP Concentration Figure 5. The effect of calmodulin on the activation of the rod channel by cGMP. Calcium influx assays were carried out using ROS membrane vesicles containing Arsenazo III dye either in the absence (•) or presence (•) of calmodulin. Calmodulin is shown to increase the Km of the channel for cCMP from 19 |AM to 33 u.M without affecting its maximum velocity or cooperativity for cGMP. The solid lines were drawn from a sigmoidal binding isotherm using the indicated Km and a Hill coefRcient of 3.6 (Hsu & Molday 1993). Molday & Hsu: Photoreceptor cells similar effect of Ca 2+ on the binding of fluorescentlabeled cGMP to ROS membranes. It is likely that endogenous calmodulin was present in the ROS preparations used in this study and was responsible for this observed Ca 2+ -dependent decrease in cGMP-binding. Calmodulin modulation of the channel activity has been found to be dependent on Ca 2+ (Hsu & Molday 1993). In the absence of calmodulin, Ca 2+ has no effect on channel activity as measured by ion flux assays. However, in the presence of calmodulin, Ca 2+ suppresses the channel activity over a Ca 2+ concentration range of 50-300 nM. The calmodulin effect is also inhibited by excess masstoparen, an inhibitor of calmodulin (Hsu & Molday, unpublished results). These studies indicate that at the level of isolated ROS membranes, Ca 2+ -calmodulin alters the sensitivity of the channel for cGMP. Modulation of the affinity of the channel for cGMP by Ca 2+ calmodulin is analogous to modulation of the affinity of hemoglobin for oxygen by changes in pH as defined by the Bohr effect. Recently, patch clamp studies have indicated that the sensitivity of the olfactory channel for cyclic nucleotides is also modulated by Ca 2+ (Kramer & Siegelbaum 1992). The calcium binding protein that interacts with the olfactory channel has not yet been identified, however. It is possible that calmodulin may be involved in this modulation of the olfactory channel. A light-activated channel encoded by the trp gene in Drosophila photoreceptors has also been shown to contain a calmodulin binding domain (Hardie & Minke 1992). Its role in invertebrate phototransduction has not yet been defined, however. It remains to be determined if the Ca 2+ -calmodulin modulation is a feature of many cation channels and in particu- lar other members of the family of nucleotide-gated channels. It also remains to be determined if other endogeneous Ca 2+ -binding proteins interact with the channel and if the channel is regulated by other mechanisms such as phosphorylation, as suggested in the studies of Gordon et al. (1992). 7. Possible role of Ca2+-calmodulin regulation of the channel in phototransduction If Ca 2+ -calmodulin binds to and modulates the channel activity in intact photoreceptors under physiological conditions as suggested by the in vitro studies, then one may envision a mechanism in which the channel switches between a low affinity and high affinity state during photoactivation and recovery (Fig. 6). In the dark, free cGMP concentration of 4-10 p,M maintains a significant number of channels in their open state, allowing for the influx of Na + and Ca 2 + into the outer segment. The balanced efflux of Ca 2 + by the Na + /Ca 2 + -K + exchanger maintains Ca 2+ levels at about 300 nM. Under these conditions, Ca 2+ -calmodulin would bind to the channel and maintain it in its low-affinity state (Fig. 7a). Photobleaching of rhodopsin and activation of phosphodiesterase via the visual cascade system results in a decrease in cGMP and a closure of the cGMP-gated channels (Fig. 7b). This will result in a lowering of intracellular Ca 2+ concentration below 100 nM as Ca 2+ is extruded through the Na + /Ca 2 + -K + exchanger. Reduced Ca 2+ concentration will cause Ca 2+ to dissociate from calmodulin and cause the channel to switch from its low- Co" K cGMP cGMP tea" Hi K cGMP • •Co* High C a ( > 300 nM) (Dark) .2+ (<100nM) (Ught) LowCa" Low cGMP affinity (High K') High cGMP affinity (Law K) 2*. Figure 6. Schematic model depicting the interaction of calmodulin with the channel complex. In the presence of high intracellular Ca 2 + (~300 nM) as found in dark-adapted rod photoreceptors, Ca 2+ -calmodulin binds to the 240 kDa protein of the channel complex and maintains the channel in a low-affinity state (high K') for cGMP. Under conditions of low intracellular Ca 2 + (< 100 nM), as found after photoexcitation, Ca 2 + and possibly calmodulin dissociate from the channel complex and cause the channel to switch to its highaffinity state (low K) for cGMP. Calmodulin binds to either the P-subunit of the channel or to unidentified polypeptide(s) (X) making up the 240 kDa polypeptide. In this model, the channel is viewed as a complex consisting of three a-subunits (one subunit is omitted for simplicity) and two B-subunits. The actual number of subunits and associated polypeptides, however, remains to be determined. BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 449 Molday & Hsu: Photoreceptor cells Ca2+K 2 tCa ' PDE tcGMP (Inactive state) 2+ Ccf' - ' ' Ught PDE- Rho* M Ugh AflWty State GC (basaD — • GC W ~ (active) cGMPt GC (basal) Low Afflntty State Figure 7. Possible role for calmodulin modulation of the channel during the visual transduction process, (top) In the dark, an elevated level of cGMP maintains a significant number of cGMP-gated channels in their open state and allows the influx of Na+ and Ca2+ into the outer segment. The balanced influx of Ca2+ through the channel with the efflux of Ca2+ through the Na + /Ca 2+ -K + exchanger results in a relatively high level of Ca2+ (~300 nM) within the outer segment. Under these conditions calmodulin associates with the channel complex and maintains the channel in its low-affinity state for cGMP. The channel in its lowaffinity state will respond to a decrease in the level of cGMP during photoexcitation of the outer segment, (bottom) Photobleaching of rhodopsin (Rho*) results in the activation of phosphodiesterase (PDE) and a decrease in the level of cGMP ©. This causes the closure of the cGMP-gated channel ® and a decrease in intracellular Ca2+ (D due to the continuous extrusion of Ca2+ by Na + /Ca 2+ -K + exchanger. This drop in Ca2+ will cause calmodulin to dissociate from the channel complex and shift the channel from its low-affinity state to its high-affinity state for cGMP. The low level of Ca2+ will also activate guanylate cyclase through a Ca2+-dependent guanylate-cyclase-activating protein. The combined effect of resynthesizing cGMP © and making the channel more sensitive to cGMP levels would facilitate the recovery of the outer segment to its dark level. As the channels reopen, the Ca2+ level in the outer segment is restored <D. This results in the rebinding of calmodulin to the channel and conversion of the channel to its low-affinity state. Guanylate cyclase is also restored to its basal level of activity. 450 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 Molday & Hsu: Photoreceptor cells affinity state to its high-affinity state (Fig. 7 bottom). This will enable the channel to open at lower cGMP concentrations. Reduced intracellular Ca 2+ levels will also activate guanylate cyclase through a guanylate-cyclase-activating protein (Koch & Stryer 1988). These two processes, activation of guanylate cyclase and conversion of the channel from a low to a high-affinity state for cGMP, will facilitate photorecovery to dark levels (Fig. 7 bottom). An increased intracellular Ca 2+ concentration resulting from reopening of the channels will return guanylate cyclase to its basal level of activity and the channel to its low-affinity state by rebinding Ca 2+ -calmodulin (Fig. 7 top). The presence of several Ca 2+ -binding proteins in ROS suggests that changes in Ca 2+ levels during photoactivation and recovery may affect several reactions. These include regulation of guanylate cyclase activity by a putative guanylate cyclase activator protein (Koch & Stryer 1988), modulation of the affinity of the channel for cGMP by calmodulin (Hsu & Molday 1993) and regulation of the light activation of phosphodiesterase (PDE) through the effect of S-modulin (recoverin) on rhodopsin phosphorylation (Kawamura 1993). 8. Summary The cGMP-gated channel of rod and cone photoreceptor cells is a member of a family of cyclic-nucleotide-gated channels found in many different cells. Primary structural analysis, site-directed mutagenetic, biochemical, and immunochemical studies have indicated that the channel consists of two major siibunits and one or more associated proteins. The a- and ($-subunits have a cGMP-binding domain near the C-terminus, an even number of transmembrane segments (possibly as many as six), a voltagesensor-like motif, and a pore region. The latter two features are found in voltage-gated channels and suggest that the nucleotide-gated channels and the voltage-gated channels have evolved from the same ancestral channel. In rod and cone photoreceptor cells the a-subunit undergoes a posttranslational cleavage reaction involving the removal of a segment at the N-terminus. The ($-subunit appears to be associated with another polypeptide which together constitute a 240 kDa polypeptide observed in purified channel preparations by SDS gel electrophoresis. The basis for these modifications is not yet known. Calmodulin has also been shown to bind to the 240 kDa polypeptide and modulate the activity of the rod photoreceptor channel in a Ca 2+ -dependent manner in in vitro systems. This Ca 2+ -dependent interaction of the channel with calmodulin alters the affinity of the channel for cGMP. Together with other Ca 2+ -dependent processes, this Ca 2+ -dependent regulation of the channel is suggested to play a role in the recovery of the photoreceptor cell following photoactivation. Although significant progress has been made over the past several years on analysis of the molecular properties of cyclic-nucleotidegated channels, much remains to be understood about the molecular composition, structure, regulation, and localization of these channels in various cell systems and the role of these channels in cell processes. ACKNOWLEDGMENT This work was supported by grants from NIH (EY-02422), MRC (MT 9588), and the RP Foundation of Canada. BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 451 BEHAVIORAL AND BRAIN SCIENCES (1995) 18, 452-467 Printed in the United States ot America Correlation of phenotype with genotype in inherited retinal degeneration Stephen P. Daiger, Lori S. Sullivan, and Joseph A. Rodriguez Human Genetics Center, School of Public Health, The University of Texas Health Science Center, Houston, TX 77030 Electronic mail: daiger@gsbs21.uth Abstract: Diseases causing inherited retinal degeneration in humans, such as retinitis pigmentosa and macular dystrophy, are genetically heterogeneous and clinically diverse. More than 40 genes causing retinal degeneration have been mapped to specific chromosomal sites; of these, at least 10 have been cloned and characterized. Mutations in two proteins, rhodopsin and peripherin/RDS, account for approximately 35% of all cases of autosomal dominant retinitis pigmentosa and a lesser fraction of other retinal conditions. This target article reviews the genes and mutations causing retinal degeneration and proposes mechanisms whereby specific mutations lead to particular clinical consequences, that is, the relationship between genotype and phenotype. Retinitis pigmentosa and macular dystrophy are genetically heterogeneous diseases that cause retinal degeneration in humans and often result in severe visual impairment or blindness. Although many of the genes causing these diseases have not been identified, three photoreceptor-specific proteins have been implicated: rhodopsin, peripherin/RDS, and the P-subunit of rod phosphodiesterase. Mutations in the genes for these three proteins can cause either dominant retinitis pigmentosa, recessive retinitis pigmentosa, dominant congenital stationary night blindness, or dominant macular degeneration. Why this multiplicity of clinical phenotypes? Our target article summarizes the genetic and biochemical background to this question and proposes a number of possible explanations. Discussion focuses mainly on 73 distinct disease-causing mutations of rhodopsin. We feel that rhodopsin and other photoreceptor proteins can serve as model systems for unraveling the connection between genotype and phenotype, not only for inherited retinal diseases but for other degenerative disorders as well. Keywords: human genetic diseases; inherited retinal degeneration; macular dystrophy; peripherin/RDS; phosphodiesterase P-subunit; retinitis pigmentosa; rhodopsin; rod and cone photoreceptor cells 1. Introduction Within the past decade rapid progress has been made in identifying genes and mutations causing many forms of inherited retinal degeneration in humans and other animals. However, we are still ignorant of the biological mechanisms that underlie many of these diseases. The connection between genotype, that is, specific mutations, and phenotype, such as retinal degeneration, is often obscure or unknown. Although much has been learned about human genes, less than 2% of the human genome has been sequenced and fewer than 1% of human genes have been fully characterized (Burks et al. 1992). Also, gene products function within complex cellular environments, and the relationship between proteins, their substrates, organelles, and other cellular components is poorly understood. Furthermore, we know even less about the function of proteins as they become degraded with time. Finally, proteins may function differently in different tissues, or their function may change in the same tissue at different developmental stages. The vertebrate retina has served as a model system for investigating embryogenesis, cellular ultrastructure, signal transduction, signal transmission and processing, and 452 cellular turnover and death. We believe that the retina, the mammalian retina in particular, also serves as a useful model for investigating genotype-phenotype relationships in inherited degenerative diseases and disorders. Many retinal genes have been cloned, and many specific mutations of photoreceptor-specific genes are known to cause retinal degeneration in humans. Such mutations can contribute to our understanding of both abnormal and normal visual processes. There are more than 200 inherited diseases that lead to retinal degeneration in humans (probably an underestimate). At least 10% of inherited diseases involve the retina directly or indirectly (McKusick 1992). Over the pastfiveyears more than 40 genes causing retinal degeneration in humans have been mapped, mostly by linkage methods. Seven of these genes have been cloned. Thus there are a number of opportunities to correlate genotype with phenotype of the resulting disease. Our discussion focuses on the genes for rhodopsin, peripherin/RDS,1 and the rod phosphodiesterase P-subunit (PDEB) genes that are expressed specifically in photoreceptors. Rhodopsin is the rod photoreceptor visual pigment located in disc membranes and in the plasma membrane of rod outer segments (ROS). Rhodopsin constitutes approx© 7995 Cambridge University Press 0140-525X195 S9.00+.10 Daiger et al.: Inherited retinal degeneration imately 90% of the protein in ROS. Peripherin/RDS is a protein of uncertain function that is expressed in both rods and cones. It is found on the periphery of photoreceptor discs, where it may play a structural role. PDEB is a catalytic subunit of phosphodiesterase, the protein in the visual transduction cascade that hydrolyses cGMP to 5'-GMP when disinhibited by activated transducin (the rod-specific G-protein). The phosphodiesterase subunit that maps to 4p appears to be rod-specific, although expression in other retinal cells cannot be excluded. Of primary interest here are mutations in genes for rhodopsin, peripherin/RDS, and PDEB that result in phenotypes that cause progressive retinal degeneration, such as autosomal dominant retinitis pigmentosa, autosomal recessive retinitis pigmentosa, and autosomal dominant macular dystrophy.2 Several recent reviews provide additional details (Berson 1993; Heckenlively 1988; Humphries et al.1992; 1993; Musarella et al. 1992). Recently, mutations in the gene for rod cGMP-gated channel (CNCG) have been suggested as another cause of arRP (autosomal recessive retinitis pigmentosa; McGee et al. 1994). This supports the expectation that many of the 40 or more genes causing inherited retinal degeneration are likely to be photoreceptor-specific. 2. Background 2.1. Diseases causing Inherited retinal degeneration Diseases causing inherited retinal degeneration in humans can be classified broadly into those that first affect peripheral vision and the peripheral retina, such as retinitis pigmentosa, and those that primarily affect central vision and the macula, such as macular dystrophy. Although the macula, especially the fovea, has the highest concentration of cones and the peripheral retina is dominated by rods, factors contributing to whether a given degeneration is expressed as a peripheral disease or a macular disease are more complicated than simply the density of a particular type of photoreceptor. The clinical symptoms of retinitis pigmentosa include night blindness and loss of peripheral vision (Heckenlively 1988). With time, visual impairment progresses toward the center of the retina causing "tunnel-vision." The disease is usually accompanied by pigmentary deposition, although the appearance of pigment is thought to be a consequence of the disease and not its cause. Early symptoms usually occur within the first or second decades of life. For many patients visual fields continue to constrict over a period of years or decades, eventually resulting in blindness. Retinal degeneration is usually bilateral and is often preceded and accompanied by characteristic changes in the electroretinogram (ERG). Retinitis pigmentosa can be subdivided into several genetic categories: autosomal dominant (adRP), autosomal recessive (arRP) X-linked (xlRP) or syndromic. The overall incidence of retinitis pigmentosa is about 1/4000 without apparent ethnic or racial distinctions (Fishman 1978; Halloran 1985). AdRP (autosomal dominant retinitis pigmentosa) accounts for about 20% of cases; arRP for 15%; XlRP for 10%; and syndromic forms, such as the recessive disease Usher syndrome (which combines congenital deafness with retinitis pigmentosa), for 20%. The remaining 35% of cases are isolated or sporadic. Thus the mode of inheritance cannot be determined for all patients, but it is assumed that all cases have a genetic basis. There are also a number of clinical categories for retinitis pigmentosa. Limiting discussion to autosomal dominant forms of the disease, adRP families and patients can be classified by a number of clinical criteria such as age of onset of night blindness and visual impairment, rate of progression, whether pigmentary deposits are diffuse, regional, or sectorial, and whether ERG responses are diminished (or absent) from rods, from cones, or from both. These classes have been condensed into two broad categories. Type 1 retinitis pigmentosa is characterized by rapid progression and diffuse, severe pigmentation; type 2 retinitis pigmentosa has a slower progression and more regional, less severe pigmentation (Massof & Finkelstein 1981). In an alternate system of classification (Lyness et al. 1985), these two types correspond roughly to type D or "diffuse" and type R or "regional" retinitis pigmentosa. Some investigators propose "sectorial" retinitis pigmentosa as a.third classification, as well as additional types (Fishman et al. 1985; Heckenlively 1988). Classification of macular degeneration is more tenuous (Fishman 1990). Macular degeneration can have either a genetic basis or it may be an acquired disease. Approximately 10% of Americans over the age of 50 are afflicted with age-related macular degeneration (Bressler et al. 1988), an acquired form of disease. The inherited forms of macular degeneration are much less common but usually more severe. Inherited macular degeneration is characterized by early development of macular abnormalities such as yellowish deposits and atrophic or pigmented lesions, followed by progressive loss of central vision. The macular lesions may appear early in life and often precede the loss of vision by many years. Like retinitis pigmentosa, macular degeneration is usually bilateral. Unlike retinitis pigmentosa, only the central retina is affected. The relative incidence and importance of inherited versus noninherited factors is unclear. In addition, genetic factors are likely to play a role in acquired forms of macular degeneration. The high frequency of acquired forms makes it very difficult to determine the incidence of inherited forms. However, several distinct, heritable forms of the disease such as autosomal dominant North Carolina macular dystrophy, autosomal dominant Best macular dystrophy, and autosomal recessive Stargardt disease have been mapped. Genes causing the inherited forms of macular degeneration are very promising candidates for genetic factors that predispose to age-related macular degeneration. There are many other inherited diseases that cause retinal degeneration in humans. Among these are gyrate atrophy, Norrie disease, choroideremia and various conerod dystrophies. In addition there are numerous inherited, systemic diseases, such as Bardet-Biedl, CharcotMarie-Tooth, and Refsum disease, which include retinal degeneration among a multiplicity of other symptoms. 2.2. Mapped and cloned genes causing Inherited retinal degeneration Table 1 lists mapped and cloned genes causing retinal degeneration and related diseases in humans. Most of the diseases have been mapped to specific chromosomal sites BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 453 Daiger et al.: Inherited retinal degeneration Table 1. Cloned and/or mapped human genes causing inherited retinal diseases" (in chromosomal order) Symbol'' McKusick no. Location Diseases'1, protein (STDG1) 248200 Ip21-pl3 USH2 RP12 RHO PDEB 276901 180380 180072 iq Iq31-q32.1 3q21-q24 4pl6.3 CNCG RP7 123825 179605 4pl4-ql3 6p21.2-cen MCDR1 RCD1 RP9 DCMD RP10 BCP RP1 VMD1 OAT USH1C ROM1 136550 180020 180104 153880 180105 190900 180100 153840 258870 276904 180721 6ql3-ql6 6q25-q26 7p 7pl5-p21 7q 7q31.3-32 8qll-q21 8q24 10q26 Ilpl5-pl3 VMD2 EVR1 VRN1 USH1B RMCH USH1A (RP13) CORD1 CORD2 RP11 SFD RS RP6 DMD RP3 COD1 PRD NDP; XLFEVR 153700 133780 193235 276903 216900 276900 600059 120970 Ilql3 Ilql3-q23 Ilql3 Ilql4 136900 312700 312612 310200 312610 304020 312500 310600 18q21-q21.3 19ql3.1-ql3.2 19ql3.4 22ql3-qter Xp22.2-p22.1 Xp21.3-p21.2 Xp21.3-p21.1 Xp21.1 Xp21.1-pll.2 Xpll.3 Xpll.4-pll.3 CSNB1 RP2 AIED CHM RCP 310500 312600 300600 303100 303900 Xpll.4-pll.23 Xpll.4-pll.23 Xpll.4-q21 Xq21.1-q21.3 Xq28 GCP 303800 Xq28 recessive juvenile Stargardt's disease (fundus flavimaculatus) recessive Usher syndrome, type 2 recessive RP dominant RP, recessive RP and others; rhodopsin recessive RP and dominant CSNB; mouse rd, mouse r and Irish Setter rcdl; cGMP phosphodiesterase P recessive RP; rod cGMP-gated channel dominant RP, dominant MD, digenic RP with ROM1 and others; mouse rds; peripherin-RDS dominant North Carolina MD dominant retinal-cone dystrophy 1 dominant RP dominant cystoid MD dominant RP dominant tritanopia; blue cone pigment dominant RP dominant atypical vitelliform MD dominant gyrate atrophy; ornithine aminotransferase recessive Usher syndrome, Acadian "digenic" RP with RDS, possible dominant RP; rod outer segment membrane protein 1 dominant Best MD dominant familial exudative vitreoretinopathy dominant neovascular inflammatory vitreoretinopathy recessive Usher syndrome, type 1 recessive rod monochromacy recessive Usher syndrome, French dominant RP (RP12 in OMIM) dominant cone-rod dystrophy 1 dominant cone-rod dystrophy (2) dominant RP, locus distinct from CORD2 dominant Sorby's fundus dystrophy X-linked retinoschisis X-linked RP Oregon eye disease; may be dystrophin X-linked RP X-linked cone dystrophy 1 primary retinal dysplasia Norrie disease, familial exudative vitreoretinopathy; Norrie disease protein X-linked CSNB X-linked RP Aland island eye disease choroideremia; geranylgeranyl transferase A protanopia and rare retinal dystrophy in blue cone monochromacy; red cone pigment deuteranopia and rare retinal dystrophy in blue cone monochromacy; green cone pigment Ilql3 14 14q32 17p "References are in GDB and OMIM, or in the text. ^Symbols in parentheses are pending. C 'RP = retinitis pigmentosa. MD = macular dystrophy. CSNB = congenital stationary night blindness. Source: 454 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 Daiger et al.: Inherited retinal degeneration by linkage testing. For genes listed as "cloned," the causative gene has been isolated, sequenced, and characterized to a limited extent at least. The table is not exhaustive in that most syndromic and systemic causes of degeneration are not included. The list is certain to be outdated soon, because new disease genes are being mapped and cloned at a rapid pace. The table lists the GDB-approved symbol, if known, and the McKusick number. The McKusick number is the code assigned in OMIM (see n. 1). We use the GDB or OMIM symbols because they are the most commonly accepted terms, but we recognize that different symbols may be used with equal validity (McKusick 1992; Pearson et al. 1992). 2.2.1. Mapped genes causing retinitis pigmentosa and macular degeneration. The first mapped gene for retinitis pigmentosa was RP2, a gene for xlRP that maps to Xpll (Bhattacharya et al. 1984). Subsequently, one form of adRP was mapped to 3q (McWilliams et al. 1989). Shortly thereafter the 3q form of adRP was shown to be caused by mutations in rhodopsin (Dryja et al. 1990a; Farrar et al. 1990b). Since then additional genes causing adRP have been mapped to 6p, 7p, 7q, 8q, 17p, and 19q (AlMaghtheh, Inglehearn et al. 1994; Blanton et al. 1991; Farrar et al. 1991a; Greenberg et al. 1994; Inglehearn et al. 1993; Jordan et al. 1993). Of these, the gene on 6p has been identified as the gene for peripherin/RDS; the others have not been cloned as yet. Two additional genes for xlRP have also been reported (Muscarella et al. 1990; Ott et al. 1990). Finally, recessive retinitis pigmentosa may be caused by mutations either in genes for rhodopsin, PDEB (which maps to 4p), or CNCG (which also maps to 4p - McGee et al. 1994; McLaughlin et al. 1993; Rosenfeld et al. 1992). Recently, another form of arRP, RP12, was mapped to lp, but the causative gene is not known (van Soest et al. 1994). The first mapped gene for autosomal dominant macular dystrophy (adMD) was VMD1, which was assigned to 8q (Ferrell et al. 1983). Thereafter, the gene for North Carolina macular dystrophy was mapped to 6q (Small et al. 1992) and the gene for Best macular dystrophy was mapped to l l q (Stone et al. 1992). Additional genes causing adMD have been mapped to 6p and 7p (Jordan et al. 1992; Kremer et al. 1994). One of these has been identified as the gene for peripherin/RDS (Jordan et al. 1992; Kajiwara et al. 1991); the others have not been cloned as yet. In summary at least 3 genes can cause xlRP (RP2, RP3, and RP6), 7 can cause adRP (rhodopsin, peripherin/RDS, RP1, RP9, RP10, RP11, and RP13), 4 can cause arRP (rhodopsin, PDEB, CNCG, and RP12), and 5 can cause adMD (peripherin/RDS, MCDR1, DCMD, VMD1, and VMD2). In addition, genes for Usher syndrome have been mapped to lq, l i p , l l q , and 14q (Kaplan et al. 1991; Kimberling et al. 1990; Kimberling et al. 1992; Lewis et al. 1990; Smith etal. 1992). As indicated in Table 1, many other retinal conditions have been mapped. For most disease categories it is important to emphasize that there are affected families whose disease gene is excluded from all known loci (e.g., Kumar-Singh et al. 1993). 2.2.2. Cloned genes causing retinal degeneration and related conditions. From Table 1 and the preceding section we know that mutations in the genes for rhodopsin, peripherin/RDS, and PDEB can cause either adRP or arRP. About 30% of adRP cases are caused by rhodopsin gene mutations and 5% by mutations in the peripherin/RDS gene (Dryja 1992; Kajiwara et al. 1991). Mutations in the PDEB gene may account for 1% to 5% of arRP cases (McLaughlin et al. 1993); the percent of arRP cases caused by rhodopsin is unknown. It is known that mutations in the peripherin/RDS gene also can cause adMD. Furthermore, mutations in rhodopsin and, possibly, PDEB can cause dominant congenital stationary night blindness (adCSNB) (Dryja et al. 1993; Rao et al. 1994; Sieving et al. 1992). There is an additional form of retinitis pigmentosa with a mode of inheritance distinct from dominant or recessive: digenic RP (Kajiwara et al. 1994). Degeneration in these patients is the result of a compound of one mutation in peripherin/RDS and one in rod outer segment membrane protein 1 (ROM1); neither mutation alone in a heterozygote causes disease. One implication of this finding is that both peripherin/RDS and ROM1 should be screened for mutations in patients with retinitis pigmentosa for which the mode of inheritance is unclear. The genes for peripherin/RDS and PDEB were first found to be the cause of retinal degeneration in mice before their connection to human diseases was demonstrated. The recessive mouse disease retinal degeneration slow (rds) is caused by a 10 kb (kilobase) insertion in the gene for peripherin/RDS (Connell et al. 1991; Travis et al. 1991). The recessive disease retinal degeneration (rd) is caused by insertion of a retrovirus gene fragment (Bowes et al. 1990; 1993) or by a nonsense mutation (Pittler & Baeher 1991) in the gene for PDEB. The mouse rd nonsense mutation is identical to the mouse r mutation described more than 70 years ago (Pittler et al. 1993). Finally, the autosomal recessive disease rod/cone dysplasia 1 (rcdt) found in the Irish Setter is also caused by a nonsense mutation in PDEB (Farber et al. 1992; Suber et al. 1993). What these findings demonstrate most clearly is the exceptional heterogeneity of inherited retinal degeneration. Mutations in different genes can cause the same disease (genetic heterogeneity); different mutations within the same gene can cause different diseases and different clinical subtypes (allelic heterogeneity); or the same mutation can have different clinical consequences in different individuals (clinical heterogeneity). Thus, it is not possible to deduce the underlying causative gene or mutation based on the clinical phenotype of a patient or family. Eventually psychophysical or electrophysiological findings may distinguish one disease gene from another but this is not possible yet. 2.3. Retinal biology relevant to inherited degeneration Rhodopsin is synthesized in rod inner segments, posttranslationally modified by the addition of oligosaccharides and fatty acids, assembled into vesicles, transported to the connecting cilium, and inserted into nascent discs. Rhodopsin has 7 transmembrane domains, and is a member of the G-coupled receptor superfamily (Hargrave & McDowell 1992a; 1992b; Nathans 1992; Stryer 1986). (See Figure la.) The amino-terminus of rhodopsin is on the intradiscal or lumenal side of disc membranes, and the carboxyl-terminus is on the cytoplasmic side. BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 455 Daiger et al.: Inherited retinal degeneration cytoplasmic t r a n s m e Er a n e Figure la. Human rhodopsin Dark-adapted rhodopsin contains 11-cts retinal as its photosensitive chromophore, attached by a protonated Schiffbase to a lysine at codon 296. Light isomerizes the chromophore to an a\\-trans configuration and activates rhodopsin into an equilibrium between metarhodopsin I and metarhodopsin II. Metarhodopsin II binds transiently to transducin. While transducin is bound to rhodopsin, the ot-subunit of the G-protein exchanges a bound GDP for GTP. The a-subunit of transducin with bound GTP dissociates from the 0- and 8-subunits and subsequently binds to one of the two inhibitory 8-subunits of phosphodiesterase. The disinhibited membrane-bound a- and (3-subunits are then activated to hydrolyze cGMP to 5'-GMP. Lowering the cytosolic concentration of cGMP reduces the conductance of a cGMP-gated channel in the plasma membrane, which hyperpolarizes the cell and reduces the rate of transmitter release from the rod synapse. Inactivating this cascade requires phosphorylation of metarhodopsin II by rhodopsin kinase, blockage of further interaction of metarhodopsin II with transducin by binding of arrestin (also called S-antigen) to phosphorylated rhodopsin, release and recycling of all-trans retinal, regeneration of cGMP from GTP by guanylate cyclase and other related but poorly defined steps. Therefore, although many of the proteins (and their genes) involved in phototransduction are well characterized, there are certain to be other genes and proteins involved that are yet unknown. 456 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 Similar mechanisms and homologous proteins are involved in phototransduction within cones, but the details are different and different genes code for these proteins. Specifically, there are distinct cone opsins, transducin subunits, phosphodiesterase subunits, and cGMP-gated channels (Blatt et al. 1988; Levine et al. 1990; Peng et al. 1992). Thus, rod and cone degeneration may be caused by functionally similar proteins that are encoded by different genes. 3. General considerations How can mutations in the genes for three photoreceptor proteins - rhodopsin, peripherin/RDS, and PDEB account for several different diseases including adRP, arRP, adMD, and adCSNB? To address this question, some generalizations are in order. First, dominant diseases are often the result of positiveacting mutations, particularly in structural proteins, whereas recessive mutations are often the result of nullfunction mutations, usually in enzymes. Second, degenerative retinal diseases may be related to each other in several ways. Over most of the retina, rods and cones are closely interdigitated in a dense, twodimensional cellular matrix. They share many metabolites and cofactors and have at least some genes in common. Thus gene mutations responsible for retinitis pigmentosa and macular degeneration may affect com- Daiger et al.: Inherited retinal degeneration mon metabolic pathways and result in similar end effects upon photoreceptor physiology. In addition, autosomal dominant congenital stationary night blindness (adCSNB) is part of a continuum of rod dysfunction that may result from either a diminution of rod function or a loss of the rods themselves. Third, it is important to recognize inherent biases in the available data. The known associations between diseases and genes derive from linkage testing in families and from mutation screening in selected patients. Because many possible genes may cause similar diseases, absence of linkage in a particular family does not mean that the excluded gene cannot cause the disease in another family. Further, some families are too small to establish linkage, and recessive diseases in particular present special problems for detecting linkage because of heterogeneity, and other factors. Moreover, mutation testing has its own bias, namely, the selection of patients to be tested: if we have not guessed the correct gene to test then we will not find the underlying mutation. Also, the problem of many genes causing similar disorders means that an uncommon molecular cause may be missed in a modest sample of patients. Finally, there are technical difficulties with mutation screening. Several laboratories, our own included, have tested collections of patients for mutations in rhodopsin and peripherin/RDS (Dryja et al. 1991; Gannon et al. 1993; Ingleheam et al. 1992; Kajiwara et al. 1991; Rodriguez et al. 1993b; Sheffield et al. 1991). Less extensive surveys of PDEB and other photoreceptor proteins have also been conducted (Jacobson & Bascom 1993; McGee et al. 1992; Ringens et al. 1990). Most mutation screening is usually done using single-strand conformational analysis (SSCA), denaturing gradient gel electrophoresis (DGGE), GC-clamped DGGE, or heteroduplex analysis. At best, each of these methods is only 90% effective, thus some mutations are missed. In addition, since multiple polymerase chain reactions (PCR) are involved, the larger the gene the more cumbersome the test. For example, rhodopsin with 5 exons and peripherin/RDS with 3 exons are amenable to PCR analysis, but PDEB with 22 exons is considerably more challenging (Kajiwara et al. 1991; McLaughlin et al. 1993; Rodriguez et al. 1993a). Thus, because present methods for detecting mutations are very laborious, no current research group has tested a multiplicity of genes in a wide range of patients. We hope that as new techniques are realized, such as analysis of illegitimate transcripts and improvements in DNA sequencing, these limitations will be resolved. 3.1. Available data Despite the inherent limitations of linkage testing and mutation screening, a large number of disease-causing mutations have been reported. At least 73 mutations in the rhodopsin gene have been described. These are listed in Table 2, including unpublished mutations from our laboratory. At least 23 peripherin/RDS mutations and 5 PDEB mutations, plus many benign variants, have been described (Humphries et al. 1993; McLaughlin et al. 1993). These mutations constitute the raw material for beginning the elucidation of the connection between genotype and phenotype. In this review we focus on the rhodopsin mutations, because more rhodopsin-related diseases have been reported, because better clinical data are available, and because rhodopsin is better characterized than peripherin/RDS or PDEB. 3.2. Why do these proteins cause these diseases? 3.2.1. Rhodopsin. Rhodopsin is a membrane-spanning protein that comprises the bulk of ROS protein. Thus mutations that change its tertiary structure within the membrane should have a dominant effect since, in this context, rhodopsin has a structural role. Yet rhodopsin also has a catalytic role in phototransduction so that, in this context, mutations may be recessive. Still other mutations may simply perturb phototransduction without cell damage, resulting in congenital stationary night blindness in which the perturbation may act dominantly or recessively. We will show examples of each of these possibilities. Do mutations in cone opsins cause cone degeneration? Abnormalities in cone opsins are generally associated with color blindness, not degeneration (Nathans 1992; Nathans et al. 1992). X-linked deuteranopia is caused by absence of green function, X-linked protanopia is caused by absence of red function, and autosomal dominant tritanopia is caused by absence of blue function. X-linked blue cone monochromacy results from absence of both red and green function (Nathans et al. 1993). The molecular basis of these diseases is well established (Nathans 1992). The red and green cone opsins are X-linked; blue cone opsin maps to 7q. Deuteranopia, protanopia, and tritanopia can be caused by missense mutations, nonsense mutations, deletions, or rearrangements in each applicable gene. Tritanopia can be caused by multiple mutations, by rearrangements of the red and green opsins, or by deletion of a locus control region for these genes. However, none of the known mutations causes cone degeneration. Retinal degeneration is occasionally seen with blue cone monochromacy (mutations unknown) but never with dominant tritanopia. It may be that the density of each cone type taken separately is too low to cause recognizable degeneration, especially the blue cones which are absent from the central retina. Also, developmental pathways during embryogenesis may simply exclude production of cones with an aberrant opsin, compensating for the deficit with additional normal cones. In any case, this remains a challenging mystery. 3.2.2. Peripherin/RDS. Peripherin/RDS, like rhodopsin, is a membrane-spanning protein found in photoreceptor discs (Arikawa et al. 1992; Connell et al. 1991; Travis etal. 1991). Unlike rhodopsin, it is expressed both in rods and in cones. Peripherin/RDS has four transmembrane domains, so both the amino-terminus and carboxylterminus are on one side of the disc, the cytoplasmic side. Very little is known about the function of peripherin/RDS. In the disc membrane it forms a homodimer and is associated with another membrane-spanning protein. ROM1, which is about 30% similar in sequence to peripherin/RDS (Bascom et al. 1992). These two proteins are found predominantly around the rim of discs in rods where, it is speculated, they contribute to the formation and maintenance of the disc structure. BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 457 Daiger et al.: Inherited retinal degeneration Table 2. Rhodopsin mutations in humans Codon no. Codon change Codon location or function; sequence conservation" 4 ACA-> AAA 1st intradiscal Thr4Lys (T4K) 15 AAT-> AGT Asnl5Ser (N15S) 17 A C G ^ > ATG 1st intradiscal, glycosylated; conserved in vertebrates 1st intradiscal 23 C C C - >CAC 23 CCC -» CTC 28 Mutation 6 Clinical consequences; clinical phenotype; comment 0 References A. Disease-causing missense mutations dominant RP; affects glycosylation of Asn-2 (possibly not incorporated into membrane) dominant RP; type 2, regional Bunge et al. 1993 Thrl7Met (T17M) dominant RP; type 2, regional, retinal neovascularization in Japanese; class Ha, multiple families, C -» T in CpG 1st intradiscal; conserved in vertebrates Pro23His (P23H) Pro23Leu (P23L) C A G - • CAC 1st intradiscal; conserved in vertebrates 1st intradiscal dominant RP; type 2, regional, variable severity (some with normal function); class Ha, 10% of USA Caucasians dominant RP; class Ha Bunge et al. 1993; Dryja et al. 1991; Fishman et al. 1992b; Fujiki et al. 1992; Hayakawa et al. 1993; Sheffield et al. 1991 Berson et al. 1991a; Dryja et al. 1990b; Heckenlively et al. 1990; Stone et al. 1991 Dryja et al. 1991 dominant RP Bunge et al. 1993 40 C T G - > CGG 1st transmembrane GIn28His (Q28H) Leu40Arg (L40R) dominant RP; early onset, severe 45 T T T - * CTT 1st transmembrane dominant RP; class I 46 C T G - • CGG 1st transmembrane 51 G G C - » GCC 1st transmembrane 51 GGC -> CGC 1st transmembrane 51 GGC GTC 1st transmembrane 53 C C C -»CGC 1st transmembrane 58 A C G - > AGG 1st transmembrane Phe45Leu (F45L) Leu46Arg (L46R) GIy51Ala (G51A) GIy51Arg (G51R) Gly51Val (G51V) Pro53Arg (P53R) Thr58Arg (T58R) Al-Maghtheh et al. 1994; Kim et al. 1993 Sung et al. 1991a 87 G T C - > GAC 2nd transmembrane dominant RP; class lib 89 G G T - »GAT 2nd transmembrane 90 G G C - •* GAC 2nd transmembrane VaI87Asp (V87D) Gly89Asp (G89D) Gly90Asp (G90D) -H» Kranich et al. 1993; Sullivan et al. 1993a dominant RP; type 1, rapid progression probable dominant RP Rodriguez et al. 1993a dominant RP Dryja et al. 1992 dominant RP; class I Dryja et al. 1991 dominant RP; class lib Inglehearn et al. 1992 dominant RP; type 2, regional; class lib, multiple families, reduced transducin activation Bunge et al. 1993; Dryja et al. 1990a; Fishman et al. 1991; Min et al. 1993; Moore et al. 1992; Richards et al. 1991 Sung et al. 1991a dominant RP; class lib dominant congenital stationary night blindness; constitutively activates transducin w/o chromophore Macke et al. 1993 Dryja et al. 1991; Sung et al. 1991a Rao et al. 1994; Sieving et al. 1992 (continued) 458 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 Daiger et al.: Inherited retinal degeneration Table 2. (Continued) Codon location or function; sequence conservation" Codon no. Codon change 106 CGG -» AGC 1st intradiscal loop; conserved in vertebrates GlylO6Arg (G106R) dominant RP; mild, regional; class lib, multiple families 106 GGC - * TGG GlylO6Trp (G106W) dominant RP; class lib 110 TGC -> TAC CysllOTyr (C110Y) dominant RP Dryja et al. 1992 125 CTG -> CGG 1st intradiscal loop; conserved in vertebrates 1st intradiscal loop, disulf. w/187; conserved in G-coupled receptors 3rd transmembrane Fishman et al. 1992; Inglehearn et al. 1992; Macke et al. 1993 Sung et al. 1991a dominant RP; class lib Dryja et al. 1991 135 CGG -* GGG Leul25Arg (L125R) Argl35Gly (R135G) dominant RP; class lib Bunge et al. 1993; Macke et al. 1993 135 CGC -» CTG Argl35Leu (R135L) Andr6asson et al. 1992 135 CCG -» CTT 135 CGC -» CCG dominant RP; rapid progression, severe; Swedish family, distinct from Argl35Leu (CGC -» CTT) below dominant RP; severe; class lib, absent transducin activation; unusual GG —> TT mutation dominant RP 135 CCC -> TGG Argl35Trp (R135W) dominant RP; severe; class lib, absent transducin activation Sung et al. 1991a 140 TGT -» TCT Cysl40Ser (C140S) dominant RP Macke et al. 1993 167 TGC -> CGC dominant RP; class lib Dryja et al. 1991 171 CCA -» CTA Cysl67Arg (C167R) Prol71Leu (P171L) dominant RP; class Ha Dryja et al. 1991 171 CCA -» TCA Prol71Ser (P171S) dominant RP Stone et al. 1993; Vaithinathan et al. 1993 178 TAC -» TGC Tyrl78Cys (Y178C) Bell et al. 1992; Farrar et al. 1991b 180 CCC -» CCC dominant RP; type 1, diffuse, early onset; class Ha, multiple families from British Isles dominant RP 3rd transm., transducin activation; conserved in G-coupled receptors 3rd transm., transducin activation; conserved in G-coupled receptors 3rd transm., transducin activation; conserved in G-coupled receptors 3rd transm., transducin activation; conserved in G-coupled receptors 3rd transm., transducin activation; conserved in G-coupled receptors 2nd cyto. loop, transducin interaction; conserved in vertebrates 4th transmembrane 4th transmembrane; conserved in vertebrates and invertebrates 4th transmembrane; conserved in vertebrates and invertebrates 2nd intradiscal loop 2nd intradiscal loop; conserved in vertebrates and invertebrates Mutation 6 Argl35Leu (R135L) Argl35Pro (R135P) Prol80Ala (P180A) Clinical consequences; clinical phenotype; comment c References Jacobson et al. 1991; Min et al. 1993; Sung et al. 1991a Rodriguez et al. 1993a Rodriguez et al., unpublished (continued) BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 459 Daiger et al.: Inherited retinal degeneration Table 2. (Continued) Codon no. Codon change Codon location or function; sequence conservation" 181 GAG -» AAG 182 GGC -» AGC 186 TCG -^ CCG 187 TGT -> TAT 188 GGA -+ AGA 188 GGA -> GAA 190 Mutation^ Clinical consequences; clinical phenotype; comment0 2nd intradiscal loop Glul81Lys (E181K) dominant RP; type 1; class Ha, multiple families 2nd intradiscal loop; conserved in vertebrates and invertebrates 2nd intradiscal loop; conserved in vertebrates 2nd intradiscal loop, disulfide w/110; cons, in vet. and invet. Glyl82Ser (G182S) dominant RP; mild, regional; class Ila Bunge et al. 1993; Dryja et.al. 1991; Rodriguez et al. unpub.; Saga et al. 1993 Fishman et al. 1992b; Sheffield et al. 1991 Serl86Pro (S186P) dominant RP; class Ila Dryja et al. 1991 Cysl87Tyr (C187Y) Nathans et al. 1993; Scott et al. 1993 Glyl88Arg (G188R) Glyl88Glu (G188E) dominant RP; class Ila Macke et al. 1993 GAC -» AAC 2nd intradiscal loop; conserved in vertebrates 2nd interdiscal loop; conserved in vertebrates 2nd intradiscal loop dominant RP; severe; polymorphic C203R mutation at homologous site in green opsin causes inactivation but not degeneration dominant RP; class Ila Aspl90Asn (D190N) dominant RP; mild, regional; class Ila, multiple families 190 GAC -» GGC 2nd intradiscal loop dominant RP; class Ila 190 GAC -»• TAC 2nd intradiscal loop 207 ATG -+ AGG 5th transmembrane 209 GTG -» ATG 5th transmembrane 211 CAC -»• CGC 5th transmembrane, strongly stabilizes inetarhodopsin II Aspl90Gly (D190G) Aspl90Tyr (D190Y) Met207Arg (M207R) Val209Met (V209M) His211Arg (H211R) Bunge et al. 1993; Dryja et al. 1991; Keen et al. 1991; Rodriguez et al. 1993b Dryja et al. 1991; Sung et al. 1991a Fishman et al. 1992c 211 CAC -> CCC His211Pro (H211P) dominant RP; class Ila 216 ATG - * AAG Met216Lys (M216K) dominant RP 220 TTT^TGT 5th transmembrane, strongly stabilizes inetarhodopsin II 5th transmembrane; Leu found in all other opsins 5th transmembrane dominant RP Bunge et al. 1993 222 TGC -» CGC 5th transmembrane dominant RP Bunge et al. 1993 267 CCC -» CTC 6th transmembrane; conserved in vertebrates Phe220Cys (F220C) Cys222Arg (C222R) Pro267Leu (P267L) dominant RP; mild, regional; class Ha, P264S mutation in homologous site in blue opsin causes dominant tritanopia Fishman et al. 1992c; Sheffield et al. 1991; Weitz et al. 1992b dominant RP; severe, diffuse dominant RP; type 1, severe; first Irish family probable dominant RP dominant RP; type 1-type 2; class Ila, multiple families References Bunge et al. 1993; Dryja et al. 1991 Farrar et al. 1992 Macke et al. 1993 Macke et al. 1993; Reig et al. 1994; Rodriguez et al. 1993a; Weitz & Nathans 1992a Keen et al. 1991; Weitz & Nathans 1992a Al-Maghtheh et al. 1994a (continued) 460 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 Daiger et al.: Inherited retinal degeneration Table 2. (Continued) Codon no. Codon change Codon location or function; sequence conservation" 292 CCC -> CAG 7th transmembrane Ala292Glu (A292E) 296 AAC -» CAG Lys296Glu (K296E) 296 AAG -> ATC 328 CTG -+ CCG 7tli transmembrane, 11-cis retinal attachment; cons, in vet. & invet. 7th transmembrane, 11-cts retinal attachment; cons, in vet. & invet. last cytoplasmic region 341 GAG -» AAG 342 ACG -» ATG 345 GTG -» TTG or CTG 345 CTG -» ATC 347 CCG -» GCG 347 CCG -» CGG 347 CCC -> CAC 347 CCC -»• CTG 347 CCG -» TCG 347 CCC -» ACC last cytoplasmic region; conserved in vertebrates last cytoplasmic region, phosphorylated last cytoplasmic region; conserved in vertebrates last cytoplasmic region; conserved in vertebrates last cyto. region; penultimate proline conserved in vertebrates last cyto. region; penultimate proline conserved in vertebrates last cyto. region; penultimate proline conserved in vertebrates last cyto. region; penultimate proline conserved in vertebrates last cyto. region; penultimate proline conserved in vertebrates last cyto. region; penultimate proline conserved in vertebrates Mutation 6 Clinical consequences; clinical phenotype; comment 0 References dominant congenital stationary night blindness; constitutively activates transducin w/o chromophore dominant RP; type D, severe; constitutively activates transducin Dryja et al. 1993 dominant RP; profound early rod loss, central cone involvement minimal dominant RP Sullivan et al. 1993 dominant RP; mild Scott et al. 1993 Thr342Met (T342M) dominant RP; C —* T in CpG Stone et al. 1993 Val345Leu (V345L) dominant RP Vaithinathan et al. 1993 Val345Met (V345M) dominant RP; moderate, variable severity; class I Pro347Ala (P347A) dominant RP Berson et al. 1991c; Bunge et al. 1993; Dryja et al. 1991 Stone et al. 1993 Pro347Arg (P347R) dominant RP; type 1, moderate; multiple families Pro347Gln (P347Q) dominant RP Pro347Leu (P347L) dominant RP; type 1, diffuse, early onset; class I, multiple families including European and Japanese; C —* T in CpG Pro347Ser (P347S) dominant RP; type 1; class I Apfelstedt-Sylla et al. 1992; Berson et al. 1991b; Bunge et al. 1993; Dryja et al. 1990; Fujiki et al. 1992; Hotta et al. 1992; Nakazawa et al. 1991; Orth et al. 1991; Shiono et al. 1992a Bunge et al. 1993; Dryja et al. 1990a Pro347Thr (P347T) dominant RP Rodriguez et al. 1993b Lys296Met (K296M) Leu328Pro (L328P) Glu341Lys (E341K) Hunge et al. 1993; Keen et al. 1991 Rodriguez et al. 1993b Bunge et al. 1993; Gal et al. 1991; Neimeyer et al. 1992 Vaithinathan et al. 1993 (continued) BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 461 Daiger et al.: Inherited retinal degeneration Table 2. (Continued) Codon no. Codon change Codon location or function; sequence conservation" 64 C A G ^ TAG 1st cytoplasmic loop 249 GAG-> TAG remainder of protein; conserved in vertebrates 344 C A C - * TAG last cytoplasmic region through carboxyterminus Mutation 6 Clinical consequences; clinical phenotype; comment 0 References B. Disease-causing premature stop mutations Gln64stop (Q64X) Glu249stop (E249X) Gln344stop (Q344X) dominant RP; mild, some with normal retina recessive RP; remainder missing starting with 6th transmemb. domain; heterozygotes at lower limit of ERG response to dim blue flash; early night blindness dominant RP; mild, some with normal retina; class I; deletes terminal GlnVal-Ala-Pro-Ala Jacobson et al. 1994; Macke et al. 1993 Rosenfeld et al. 1992 Jacobson et al. 1991; Jacobson et al. 1994; Sung et al. 1991a C. Disease-causing insertions, deletions and splice-site mutations 68-71 — 1st cytoplasmic loop A68-71 (496dell2) 255/256 ATC -> — 6th transmembrane AIle255/256 (1057del3) 264 TGC - * — ACys264 (1084del3) dominant RP (312313) — 6th transmembrane; conserved in vertebrates last cytoplasmic region through carboxyterminus 1230 + 1G (312313) last cytoplasmic region through carboxyterminus 1231 - 1G -> A (312313) last cytoplasmic region through carboxyterminus 1231-23del 30insl50 dominant RP; highly variable expression (Macke et al. 1993), some with normal retina or unaffected (Rosenfeld et al. 1992); intron 4 donor splice site G —» T eliminating terminal 36 amino acids dominant RP; type 2, mild; intron 4 acceptor splice site G —* A, effect on protein unknown dominant RP; type 2, regional; 30 bp deletion of 3' intron 4 (23 bp) and 5' expon 5 (7 bp) plus 150 bp insertion dominant RP; type 1-type 2; 17 bp out of frame deletion, new stop at codon 346 dominant RP; mild, late onset, diffuse; 1 bp deletion replaces terminal 9 AA's with 19 new AA's (preserves terminal Pro-Ala) dominant RP; mild, slow progression; deletion of stop codon may lengthen to 386 amino acids 332 GAG -+ G— last cytoplasmic region, phosphorylated A332stop (1289dell7) 340 ACG -» A-G last cytoplasmic region, phosphorylated A340 (1313del C) 340-348 — last cytoplasmic region through carboxyterminus, phosphorylated A340-348 (1312del24) dominant RP; mild; class Ila;, deletes Leu-ArgThr-Pro, does not bind 11-cts retinal dominant RP; type 1, severe, diffuse; class Ila, multiple families Keen et al. 1991; Min et al. 1993 Artlich et al. 1992; Bunge et al. 1993; Inglehearn et al. 1991 Vaithinathan et al. 1993 Jacobson et al. 1994; Macke et al. 1993; Rosenfeld et al. 1992 Rodriguez et al. unpub. Al-Maghtheh et al. 1994a; Kim et al. 1993 Rodriguez et al. 1993b Horn et al. 1992 Restagno et al. 1992 (continued) 462 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 Daiger et al.: Inherited retinal degeneration Table 2. (Continued) Codon no. Codon change 341-343 Codon location or function; sequence conservation" last cytoplasmic region, phosphorylated Clinical consequences; clinical phenotype; comment 1 Mutation* A341-343 (1315del8) dominant RP; mild; 8 basepair deletion replaces terminal 8 AA's with 9 new AA's (preserves terminal Pro-Ala) References Horn et al. 1992; Bunge et al. 1993 D. Silent substitutions, benign variants and polymorphisms — — 5' noncoding region 296A - • G 104 C T C - * ATC VallO4Ile (V104I) intron 1 — 1st intradiscal loop; conserved in vertebrates — 120 GGC -> CCT 3rd transmembrane 146 T T C - » TTT 160 ACC -» ACT 2nd cytoplasmic loop, transducin interaction 4th transmembrane 173 CCC -* GCT 4th transmembrane 186 TCG^-TCA 244 CAG - ^ CAA 248 AAG -» AAA 297 AGC -* AGT 2nd intradiscal loop; conserved in vertebrates 3rd cytoplasmic, transducin interaction; conserved in vertebrates 3rd cytoplasmic, transducin interaction 7th transmembrane — — — — intron 4 3' noncoding region normal variant— polymorphic with 14% frequency normal variant Sung et al. 1991c Weber & May 1989 Glyl20Gly (654C -> T) Polymorphic microsatellite variation in intron 1 normal variant—silent substitution Phel46Phe (732C -> T) normal variant—silent substitution Thrl60Thr (774C -» T) Reig et al. 1994; Sung et al. 1991a; Macke et al. 1993 Alal73Ala (813C -» T) Serl86Ser (852G -> A) normal variant—infrequent silent substitution; ACC -> ACA (T160T) found in a Spanish family normal variant—silent substitution; C -» T in CpG normal variant—silent substitution Gln244Gln (1026G - » A) normal variant—silent substitution Rosenfeld et al. 1992 Lys248Lys (1038G -* A) Ser297Ser (1185C -H> T) normal variant—infrequent silent substitution normal variant—silent substitution; C —> T in CpG Sung et al. 1991a 1231-23G -» A 1338 + 46C -» A normal variant—infrequent normal variant-polymorphic with 13% frequency [CA]n Macke et al. 1993 Bunge et al. 1993; Dryja et al. 1991; Macke et al. 1993 Macke et al. 1993 Bunge et al. 1993; Dryja et al. 1991 Rosenfeld et al. 1992 Rosenfeld et al. 1992; Saga et al. 1993; Macke et al. 1993 Sung et al. 1991a Sung et al. 1991a "Based on Fryxell and Meyerowitz (1991) with 6 vertebrate opsins (human red, green and blue; human, pig ovine, and bovine rhodopsin), 2 Drosophila opsins, and 10 additional, homologous G-coupled protein receptors. ''Nomenclature based on Beaudet and Tsui (1993), numbering nucleotides from the 5' start of the cDNA (including 294 untranslated nucleotides) (GenBank Accession K02281). 'Class refers to rhodopsin cDNA transfection experiments in human embryonic kidney cells by Sung et al. (1991b; 1993). Class I = wild typo characteristics; class Ha = distinctly abnormal and remains in endoplasmic reticulum; class lib = distinctly abnormal but some localizes in plasma membrane. Diseases caused by mutations in peripherin/RDS run the gamut of dominant retinal degeneration, including adRP, adMD, retinitis punctate albescens, fundus flavimaculatus, and various pattern dystrophies (Kajiwara et al. 1993; Keen et al. 1994; Nichols et al. 1993; Wells et al. 1993). Furthermore, one mutation, a 3 bp deletion of a lysine at codon 153 or 154, causes retinitis pigmentosa, pattern macular dystrophy, and Stargardt dystrophy in different members of the same family (Weleber et al. 1993). In the absence of a clear understanding of the functional domains of peripherin/RDS it would be naive to try to explain these findings mechanistically. Since the gene for peripherin/RDS is expressed both in rods and in cones, some domains may be important for rod function, BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 463 Daiger et al.: Inherited retinal degeneration some for cone function, and some for both. ROM1 is not expressed in cones, so one distinguishing domain may be the R0M1 binding site. (Mutations in patients with "digenic" disease may point to sites of interaction between peripherin/RDS and R0M1; McGee et al. 1994.) It is reasonable to speculate that destabilization of the disc membrane would be very damaging to the photoreceptor. Why some mutations are highly variable in their expression is currently unknown, though this suggests that genetic background and environmental factors play important roles. At present, the limited number of known, disease-causing mutations is insufficient to make meaningful correlations. However, the accumulating number of pedigrees with mutations in the peripherin/RDS gene may prove instructive. 3.2.3. Phosphodiesterase p-subunit(PDEB). Mutations in PDEB are associated with arRP and adCSNB (Gal et al. 1994; McLaughlin et al. 1993). Reduced or absent PDEB should retard hydrolysis of cGMP and make photoreceptors less sensitive to light. Thus, dominant CSNB is a logical outcome. Recessive degeneration, though, is less expected. The reported PDEB mutations causing arRP are compound heterozygotes of three possible nonsense mutations, one out-of-frame deletion, and one missense mutation (McLaughlin et al. 1993). These mutations cause typical retinitis pigmentosa with early onset and no detectable rod ERG in adulthood. They are likely to be nullfunction mutations that, among other consequences, would elevate cGMP levels. 3.2.4. Why do these mutations lead to cell death? For many photoreceptor mutations the initiating cause of retinal degeneration can be inferred, such as membrane instability, aberrant activation of phototransduction, or increased cGMP levels. Studies of retinal degeneration in animals suggest that increased cGMP may be toxic to photoreceptors (Lolley et al. 1977; Ulshafer et al. 1980). The connection between increased cGMP and cell death is not known, but increased cellular permeability, direct cytotoxicity, or unbalanced Ca 2 + /Na + flows are possibilities (Hargrave & McDowell 1992b). Why these various conditions lead to degeneration is unclear. One possibility is a simple necrotic process, wherein the cell ceases its metabolic activity, its membranes decompose, and phagocytic activity - either by the retinal pigment epithelium (RPE) or by infiltrating macrophages - removes the cellular debris. An alternative is an enhanced, catabolic interaction between the RPE and the ROS, possibly exacerbated by a feedback loop triggered by accumulating disc debris. However, recent evidence implicates apoptosis, a form of programmed cell death, as the mechanism responsible for photoreceptor cell death in rodent models of retinal degeneration. In apoptosis, genetic switches in the cell initiate a process that turns off normal cellular activity and causes condensation of nuclei, DNA degradation, and cell death. The cardinal features of apoptosis have been observed in the rd mouse, the rds mouse, and a transgenic mouse with the rhodopsin Pro347His mutation (Chang et al. 1993; Lolley et al. 1994). Thus, it seems likely that some forms of retinal degeneration in humans may result from apoptosis. However, these observations do not preclude alternative mechanisms in other forms of degeneration. Further464 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 more, they beg the question of what triggers apoptosis in the first place. 4. Clinical consequences of rhodopsin mutations Table 2 summarizes data on human rhodopsin mutations from many sources (references are in Table 2 unless otherwise noted). The table gives the codon, nucleotide change, and amino acid change for each mutation. In addition, functional properties, protein domain, 3 and evolutionary conservation of affected amino acids are given (Fryxell & Myerowitz 1991; Hargrave & O'Brien 1991; Nathans et al. 1993). Also listed is the type of disease observed, clinical features of the disease, if reported, and comments. Included in the comments is the biochemical class (I, Ha, or lib) based on transfection-expression experiments by Sung and colleagues (Sung & Schneider et al. 1991; 1993). The table is grouped into disease-causing missense mutations; nonsense mutations; deletions, insertions, or splice-site mutations; and benign variants. There are 73 reported disease-causing mutations of the rhodopsin gene, including 63 missense mutations, 3 nonsense mutations, and 7 rearrangements. Of these, 70 cause dominant retinal degeneration, 2 cause adCSNB, and 1 causes recessive retinal degeneration (only homozygotes or compound heterozygotes are affected). The locations of the missense mutations are diagrammed in Figure lb. This organization emphasizes the missense mutations, because simple amino acid substitutions are easiest to interpret in terms of functional consequences. Thus, most of the following discussion is based on the 63 missense mutations, although the other mutations are also considered. Important functional domains in rhodopsin include the intradiscal, transmembrane, and cytoplasmic domains. Amino acids with critical functions include the acetylated amino-terminus, glycosylated asparagines at codons 2 and 15, cystines at codons 110 and 187, which form a disulfide bound (Karnik & Khorana 1990), the glutamate at codon 113, which provides a counterion for the Schiff base linkage to codon 296, lipid-binding cystines at codons 322 and 323, which attach to the cytoplasmic side of the membrane, and phosphorylation sites at codons 334, 336, 338, 340, 342, and 343 (Figure la; Hargrave & O'Brien 1991). Mutations in the rhodopsin gene have also been classified by the severity of the disease produced, if reported. There are many ways to classify retinitis pigmentosa: by type, by progression, by fundus findings, by ERG findings, and so on. Several classification schemes have been proposed (Fishman et al. 1985; Heckenlively 1988; Lyness et al. 1985; Massof & Finkelstein 1981). Unfortunately, different patients have been classified in different ways. Lacking uniform classification criteria (an undertaking we strongly advocate), we have adopted the simplest possible clinical categories. Thus in what follows we equate "mild" adRP with type 2 or R (late onset and slow progression), and "severe" adRP with type 1 or D (early onset and rapid progression). "Moderate" is intermediate. We also consider adCSNB and arRP as inherently mild. We do not incorporate ERG profiles or funduscopic findings though, needless to say, they would be relevant to a more sophisticated analysis. Based on these Daiger et al.: Inherited retinal degeneration cytoplasmic Figure lb. Missense mutations criteria, a total of 40 rhodopsin mutations can be classified, admittedly a small sample. 4.1. General considerations 4.1.1. Population distribution. The majority of mutations in Table 2 are unique, occurring in one patient or one family only. A few mutations occur in multiple, unrelated families. Among the latter are the Thrl7Met, Thr58Arg, and Pro347Leu mutations found in .both Europe and Japan; the Pro23His mutation found in nearly 10% of American Caucasian adRP patients (but not in Europeans; Farrar et al. 1990a); the Thrl78Cys and AIle255/256 mutations found in the British Isles; and the GlylO6Arg, GlulSlLys, Aspl90Asn, His211Arg, and Pro347Arg mutations found in European and American families. Do these multiple pedigrees represent multiple mutational events or descent from a common ancestor (founder effect)? A single chromosomal haplotype is in complete linkage disequilibrium with the Pro23His mutation, providing strong evidence for founder effect (Dryja et al. 1991). In addition, those mutations confined to a limited geographical range, such as the Tyrl58Cys and AIle255/ 256 mutations, are probably the result of founder effect. By contrast, molecular evidence suggests that the Thr58Arg and Pro347Leu mutations arose on multiple occasions. Also, it seems likely that the mutations found in both Europe and Japan arose independently. The main effect of this uncertainty is to imply that Table 2 should include multiple entries for some of the mutations. Unfortunately the multifamily mutations are not: helpful in population screening because, taken in aggregate, they represent only a small percent of all reported rhodopsin mutations. 4.1.2. General distribution and types of mutations. Based on Table 2, the ratio of transitions to transversions is 43:32, which is not significantly different from the expected ratio of 2:1 in mammalian genes (Nei 1987). Although a few C—»T transitions in CpG dinucleotides are observed, the number is not exceptional (Barker et al. 1984). Most of the amino acid substitutions change the charge, hydrophobicity, or size of the wild-type residue, but this also is not exceptional, since most random changes would affect one or more of these properties. However, there are 18 evolutionarily conserved sites in human rhodopsin (Fryxell & Meyerowitz 1991); 29% of the 63 missense mutations occur at conserved sites, whereas only 5% are expected by change alone. Thus, substitutions at conserved sites are significantly more likely to cause retinal degeneration. There are two other striking anomalies in the distribution of mutations - first, in the number of codons with multiple, independent mutations and, second, in the distribution of mutations within protein domains. For the pedigrees with rhodopsin-based retinal degeneration studied to date, 305 of the 348 amino acids in human rhodopsin have zero mutations, 32 have one mutation BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 465 Daiger et al.: Inherited retinal degeneration each, 6 have two, 3 have three, 1 has five (arginine at codon 135), and 1 has six mutations (proline at codon 347). The expected random distribution would be 290 amino acids with zero mutations, 53 with one, 5 with two, 0.30 with three, and fewer than 10~6 with five or more (Chakraborty 1993). Thus, codons 135 and 347 are highly unusual, particularly considering the multiple, independent families with a Pro347Leu mutation. Note that although both codons contain a CpG dimer, this is an insufficient explanation, since many other codons contain the same dimer. At a superficial level the mutations seem uniformly distributed between the intradiscal, transmembrane, and cytoplasmic regions of the protein. About 50% of rhodopsin amino acids are within the membrane and 25% are on each side. The observed distribution of mutations is 19:32:12, which is not significantly different from 1:2:1. However, nearly all the cytoplasmic mutations are in the" eight terminal amino acids; there is a paucity of mutations in the four cytoplasmic loops. The cytoplasmic loops include the binding sites for the a-subunit of transducin (Konig et al. 1989; Min et al. 1993) and arrestin. We speculate that mutations may be as frequent here as elsewhere in the molecule, but that they do not cause adRP. 4.1.3. General distribution of clinical types. There are no striking anomalies in the distribution of mild, moderate, or severe mutations. Six intradiscal mutations are mild and four are severe; two transmembrane mutations are mild (four counting adCSNB) and eight are severe; and two cytoplasmic mutations are mild and two are severe. Thus, protein domain alone does not determine severity of disease. It is interesting that many of the "mild" forms of adRP caused by rhodopsin mutations show a sectorial phenotype in some patients. These include the Thrl7Met, Pro23His, Thr58Arg, GlylO6Arg, Glyl82Ser, and Pro267Leu mutations. Given that clinicians will differ in attributing this phenotype to a particular patient, the evidence suggests that sectorial adRP is not a separate genetic category. 4.1.4. Distribution of biochemical classes. Sung and colleagues (Sung & Schneider et al. 1991; 1993) have used site-directed mutagenesis to distinguish broad classes of rhodopsin mutations. They constructed expression vectors containing the human rhodopsin sequence with specific mutations and used these to transfect human embryonic kidney cells. They then evaluated synthesis, transport, opsin binding capability, and immunologic properties of the mutant rhodopsins. In total, they tested 21 human mutant types. Based on their findings they classified the mutants into three classes, class I with wildtype properties, class Ha with aberrant properties and complete failure to leave the site of synthesis, and class lib with aberrant properties but partial transport to the plasma membrane. Class I mutants clustered within the first transmembrane domain and at the extreme carboxylterminus. Class II mutants clustered within the other transmembrane and extracellular domains (Sung et al. 1993). There are clinical data on 18 of the 20 mutations tested. Nine of the mutations produce mild disease, 7 produce severe disease, and 2 produce moderate disease. Surprisingly, there is no simple correlation between se466 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 verity and class: the ratios of mild-to-severe for class I are 1:2, for class Ila, 6:3, and for class lib, 2:2. 4.2. The connection between phenotype and genotype of specific rhodopsin mutations Within the limits of the data cited above, specific rhodopsin genotypes can be correlated with broad phenotypic categories. Different mutations cause retinal degeneration for distinctly different reasons. However, there is an organizing principal: the clinical phenotype is a consequence of where and when the mutation affects the function of rhodopsin. 4.2.1. Mutations producing no protein or unstable protein will be mild (or recessive) in heterozygotes. This may seem counterintuitive, since it is often assumed that a defective protein will accumulate at the point of synthesis and "poison" the cell. However, the cellular system for degradation and clearance of aberrant proteins is highly efficient (Klausner & Sitia 1990). Furthermore, a 50% diminution in rhodopsin levels may be mildly deleterious at worst. This category includes mutations producing no message or unstable message, mutations producing truncated or otherwise aberrant proteins, or mutations producing proteins that are not transported to or inserted into the disc membrane because of incorrect folding or incorrect post-translational modification. Examples include early stop mutations, such as Gln64stop and Glu249stop, and substantial deletions, such as A68-71 and 1231-23del30insl50 that produce mild disease (or recessive disease in the case of Glu249stop). The two splice-site mutations, 1230 + lG-»Tand 1231-1G-»A, produce very mild disease and are highly variable in expression. This may be because the splice-site mutations are spliced differently in different individuals, that is, they are "leaky" mutations. Finally, the Asnl5Ser mutation would preclude glycosylation of this highly conserved site, a potentially major defect. However, the phenotype of this mutation is also mild, suggesting that this change may simply interfere with targeting or insertion into the disc (Hargrave & O'Brien 1991). 4.2.2. Heterozygous mutations producing proteins that are inserted into but destabilize the membrane will be severe. Since the discs are geometrically precise structures with complex functional requirements and rapid turnover, and since rhodopsin constitutes the bulk of the disc protein, destabilizing mutations are expected to be severe. We speculate that this is why the Leu40Arg, Leu46Arg, Argl35Leu, Argl35Trp, and Met207Arg mutations are severe. The Argl35 mutations are particularly noteworthy, because this is a highly conserved site at the margin of the membrane. Several mutations at this site cause severe adRP. In addition, the Argl35Leu and Argl35Trp mutations are class lib. If results from embryonic kidney cells can be extrapolated to intact photoreceptors, this implies that some protein gets to the disc membrane. Furthermore, this amino acid may have a critical role in rod function beyond simple membrane integrity, since Argl35Leu and Argl35Trp mutants fail to activate transducin (Fahmy & Sakmar 1993; Min et al. 1993). However, as we consider subsequently, this effect alone is not likely to cause severe dominant disease. Daiger et al.: Inherited retinal degeneration The Cysl87Tyr mutation eliminates the disulfide bond with CysllO and produces severe disease, possibly by destabilizing the membrane. The A255/256IIe deletion, another transmembrane mutation, also produces severe disease. On the other hand, some transmembrane mutations, such as Thr58Arg, only produce mild disease, suggesting that mutations of transmembrane regions may not necessarily perturb the membrane in significant ways. 4.2.3. Heterozygous mutations that fail to incorporate 11-c/s retinal or disrupt the Schiff base counterion will constitutively activate transducin and produce severe disease. Artificial mutations that change the lysine at codon 296 or the glutamate at codon 113 result in constitutive activation of transducin (Robinson et al. 1992). No human pedigrees expressing a Glull3 mutation have been identified. However, the Lys296Glu and Lys296Met mutations that produce very severe, dominant degeneration, Opsins with substitutions at the Lys296 position fail to bind 11-cis retinal and are constitutively active. Thus, these mutations are likely to induce high levels of cGMP with toxic consequences (Lolley et al. 1977; Ulshafer et al. 1980). a heterozygote may be diminished sensitivity to dim light or adCSNB (or recessive CSNB), but not significant degeneration. Finally, heterozygous mutations affecting phosphorylation sites may affect recovery from photobleaching but not cell survival. Possible examples of mild mutations that meet these criteria are Thrl7Met, Pro23His, GlylO6Arg, and Glu341Lys. The Thr58Arg mutation is an example of a mild phenotype associated with diminished transducin activation (Min et al. 1993). Such examples may help to explain the paucity of disease pedigrees caused by dominant mutations within the cytoplasmic loops. One intriguing observation is the importance of the penultimate proline, Pro347, in the carboxyl-terminal cytoplasmic domain (Humphries et al. 1993). The two terminal amino acids (Pro-Ala) are highly conserved in mammals. This is the most common site of mutations causing adRP, and the disease phenotype is severe. By contrast, two mutations that delete these amino acids, 1313delC and 1515del8, have a mild phenotype. However, by coincidence, the terminal Pro-Ala residues are preserved in each of these mutations. Unfortunately for the argument, this is not the case for the 1289dell7 and 1312del24 mutations, which also produce mild disease. 4.2.4. Heterozygous mutations that bind 11-c/s retinal normally but also activate transducin in the absence of chromophore will be mild or stationary. The revealing examples in this case are the two mutations that cause adCSNB, Gly90Asp and Ala292Glu. Both mutant opsins bind chromophore are inactive in the dark, and are activated normally by light (Dryja et al. 1993; Rao et al. 1994). However, in contrast to wild-type opsin, these proteins can also activate transducin in the absence of chromophore. Thus, unlike the Lys296 mutations, which are permanently activated, these proteins cause inappropriate activation of transducin when all-trans retinal leaves opsin during the regeneration cycle. This minor abnormality is sufficient to diminish sensitivity to dim light but is not sufficient to damage the photoreceptor. Significantly, the Lys296, Glull3, and Gly90 sites all fall in close proximity within the three-dimensional structure of rhodopsin (Baldwin 1993), suggesting that the glycine participates in chromophore stabilization. 4.2.6. We do not know enough about rhodopsin to guess which mutations might interfere with disc assembly, progression along the ROS, shedding, or phagocytosis; nor can we know their clinical consequences. We do not know what role rhodopsin plays, if any, in disc morphogenesis, axial displacement, shedding, and phagocytosis by the RPE. We also do not know whether defects in these processes might be related to human retinal degeneration. In any case, we know so little about these processes that guessing may not be productive. It is our expectation that new disease-causing mutations will reveal previously unrecognized functional sites in rhodopsin that are involved in disc membrane renewal. 4.2.5. Heterozygous mutations in amino acids outside of the membrane are likely to be mild or recessive unless indirectly involved in membrane stability or chromophore binding. We speculate that the following general principles apply to such mutations. First, mutations of some of the amino-terminal amino acids will affect targeting of rhodopsin to the disc and will have mild consequences (sect. 4.2.1). Second, some of the amino acids in the intradiscal and cytoplasmic loops will be essential for proper protein folding. If mutations in these sites prevent insertion of the protein into the membrane, the consequences will be mild (sect. 4.2.1). Third, heterozygous mutations in the intradiscal loops are not likely to affect phototransduction directly unless they disturb chromophore binding or isomerization (sects. 4.2.3; 4.2.4). Fourth, mutations in the cytoplasmic loops will affect transducin binding or activation (Franke et al. 1988; Konig et al. 1989). The consequences of such mutations in NOTES 1. Since more than one mammalian protein is named peripherin, "peripherin/RDS" is used to identify the photoreceptor protein (Travis 1991). 2. Gene symbols, such as PDEB, are assigned by the Nomenclature Committee of the Genome Data Base (GDB) and are recorded in GDB or the OnLine McKusick Catalog (OMIM) (McKusick 1992; Pearson et al. 1992). GDB and OMIM are accessible by Internet or modem. At the request of the Nomenclature Committee, gene symbols in humans are all upper case letters (lower case italics in nonhumans; personal communication). Symbols for diseases with multiple causes, such as autosomal dominant retinitis pigmentosa, i.e., "adRP," include both upper- and lower-case letters to distinguish them from genes (Beaudet & Tsui 1993). 3. Determining the exact location of transmembrane boundaries is somewhat speculative (Dryja et al. 1991; Fryxell & Meyerowitz 1991; Nathans 1992). We base our boundaries on Nathans (1992). ACKNOWLEDGMENTS Supported by grants from the RP Foundation Fighting Blindness and the George Gund Foundation, and NIH NRSA Grant EY06467 to L. S. S., and a Fellowship from the Schissler Foundation to J.A.R. BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 467 Commentary!Controversies in Neuroscience III Open Peer Commentary Commentary submitted by the qualified professional readership of this journal will be considered for publication in a later issue as Continuing Commentary on this article. Integrative overviews and syntheses are especially encouraged. Calcium/calmodulin-sensitive adenylyl cyclase as an example of a molecular associative integrator Thomas W. Abrams Department of Neuroscience, University of Pennsylvania, Goddard Labs, Philadelphia, PA 19104-6018. tabrams@mail.sas.upenn.edu Abstract: Evidence suggests that the Ca 2+ /calmodulin-sensitive adenylyl cyclase may play a key role in neural plasticity and learning in Aplysia, Drosophila, and mammals. This dually-regulated enzyme has been proposed as a possible site of stimulus convergence during associative learning. This commentary discusses the evidence that is required to demonstrate that a protein in a second messenger cascade actually functions as a molecular site of associative integration. It also addresses the issue of how a dually-regulated protein could contribute to the temporal pairing requirements ofclassical conditioning: that relationship between stimuli display both temporal contiguity and predictability. [XIA ET AL.] It is widely accepted that modulatory changes in neurons, as in other cells, are frequently mediated by second messenger cascades involving covalent modifications of existing proteins via phosphorylation and dephosphorylation (Greengard 1979). Moreover, cellular studies suggest that neuronal changes produced by learning may be mediated by these intracellular signaling cascades (e.g., Kandel & Schwartz 1982; Malenka et al. 1989; Malinow et al. 1989; Silva et al. 1992). An intriguing possibility is that during associative learning, duallyregulated proteins within these intracellular signaling cascades may serve as the sites at which inputs from multiple stimuli or behaviors converge (Abrams & Kandel 1988; Bourne & Nicoll 1993). According to this hypothesis, these proteins serve an associative role: when two stimuli are paired with the appropriate temporal relationship, the protein responds to the pairing by triggering a second messenger cascade that initiates persistent alteration of neuronal properties. As discussed by XIA ET AL., Ca 2+ /CaM-sensitive adenylyl cyclase in Aplysia was the first such example of a molecular site of associative integration shown to contribute to neural plasticity. Based on cellular studies, Abrams and colleagues (Abrams 1985; Abrams & Kandel 1988; Abrams et al. 1984; Kandel et al. 1983) and Occor et al. (1985) proposed that during classical conditioning in Aplysia, the dually-regulated Ca 2 + / CaM-sensitive adenylyl cyclase serves as a site of associative integration that detects the pairing of the cellular signals from the conditioned and unconditioned stimuli. They found that optimal stimulation of cAMP synthesis required temporal pairing of activity and Ca 2 + influx in the presynaptic sensory neuron, which are triggered by the conditioned stimulus (CS), and release of modulatory transmitter, which is triggered by the unconditioned stimulus (US); cAMP, in turn, initiates synaptic facilitation in the sensory neurons of the CS pathway (Ghirardi et al. 1992; Goldsmith & Abrams 1991). The second example of a molecular mechanism of associative integration that has been well characterized in the context of associative neural plasticity is the NMDA-type glutamate receptor in hippocampus (Collingridge 1987; Madison et al. 1991). critical to distinguish between an actual role as an associative integrator, where the protein serves as a molecular site of associative signal convergence, and a role as an essential relay step (or amplification step) in a signal transduction cascade. In many experimental systems, it is difficult to demonstrate that a protein is functioning to integrate paired stimuli, even when in vitro experiments demonstrate the protein can be synergistically activated by two signals. For example, most studies that utilize inhibitors of an enzyme or receptor in a second messenger cascade (Davis et al. 1992) or studies of mutants lacking the relevant protein cannot distinguish between an associative and a relay role. Neither the studies of the rutabaga mutant in Drosophila, which lacks Ca 2+ /CaM-sensitive adenylyl cyclase (Dudai & Zvi 1984; Levin et al. 1992; Livingstone et al. 1984), nor the experiments of XIA ET AL. on a mouse mutant in which the type I adenylyl cyclase gene has been disrupted, demonstrate that the affected enzyme is playing an associative role (as has been suggested in Aplysia); it is possible that stimulation of cAMP synthesis is simply an essential step in learning, or even that a basal level of cAMP-dependent phosphorylation is permissive for neural plasticity. In order to demonstrate an associative function, it is necessary to conduct cellular studies measuring the response of the enzyme or receptor to paired or unpaired signals. Nevertheless, it is exciting that a similar Ca 2+ /CaM-sensitive adenylyl cyclase may play acritical role in learning in three separate phyla, suggesting that molecular mechanisms for learning are highly conserved through evolution. 2. Detecting relationships between stimuli: temporal contiguity. Because a species of adenylyl cyclase is activated synergistically by two stimuli, it could, as XIA ET AL. point out, function as what has been termed a "molecular-coincidence detector" (Bourne & Nicoll 1993; Hille 1992). However, the optimal timing of paired stimuli for proteins that perform associative integration may not be always be simultaneous. This is the case particularly for proteins that mediate the learning of predictive relationships between paired stimuli. In the conditioning of the defensive withdrawal reflex of Aplysia, as in many Pavlovian paradigms, training is more effective if the CS precedes the onset of the US by a short interval than if stimulus onset is simultaneous; moreover, associative learning disappears if the US begins first, even if the US continues through the CS. Using a perfused membrane cyclase assay that permits transient exposures to stimulating ligands, Yovell and Abrams (1992) found that forward pairing of Ca 2 + and serotonin (with Ca 2 + beginning first) gave more powerful cyclase activation than backward pairing (with serotonin before Ca 2 + ); i.e., adenylyl cyclase activation was more effective when the signal representing the CS began before the signal representing the US. Thus, the cyclase displayed a sequence preference analogous to that of the conditioning that the CS precede the US during stimulus pairing. Recently, Abrams and Galun (1993) found that CaMmediated stimulation of Ca 2+ /CaM-sensitive adenylyl cyclase was relatively delayed, and began after a brief transient initial period of direct inhibition of cyclase by Ca 2 + . This delayed Ca 2 + stimulation may contribute to the requirement that the CS (and Ca 2 + influx) begin before the US (and transmitter binding). The NMDA receptor also shows temporal asymmetry in the dependence of its activation on the relative timing between paired signals (Gustafsson et al. 1987). 3. Detecting relationships between stimuli: Contingency and predictability. Learning about relationships during Pavlovian conditioning requires discriminating not only temporal contiguity of paired stimuli, but also contingency between stimuli; the learning of the associative relationship depends upon how reliably the CS predicts the occurrence of the US (Hearst 1988; Rescorla 1988). For example, presentations of the CS in the f. Distinction between a site of associative convergence and a relay step in a linear cascade. In analyzing the contributions of absence of the US or presentations of the US in the absence of the CS degrade associative learning. As yet, we know nothing dually-regulated proteins to associative neural plasticity, it is 468 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 Commentary /Controversies in Neuroscience III about whether molecular integrators are sensitive to contingency. In other words, do unpaired presentations of either the signal from the CS or the signal from the US degrade the activation of the molecular cascade? It may be that an individual enzyme or receptor cannot integrate across separate trials in which stimuli are presented. The fact that a late pulse of Ca 2 + accelerates turn-off of Aplysia neural adenylyl cyclase activated by transmitter (or by transmitter paired with Ca 2 + ) (Abrams et al. submitted) may cause extra unpaired CSs to reduce overall stimulation ofcAMP synthesis; however, because cyclase activation decays in approximately 60 sec, this effect of an extra unpaired CS could occur only within a relatively short time window. It is also possible that other downstream proteins in the intracellular cascade(s) that have longer "memories" than the integrator protein itself (e.g., a protein kinase) feed back, modulating the integrator protein; such interactions could cause activation of the associative integrator protein by paired stimuli to be influenced by unpaired stimuli at earlier times. As yet there have been no serious attempts to evaluate whether the activation of either Ca 2+ /CaM-sensitive adenylyl cyclase or the NMDA receptor by paired stimuli is decreased by a previous unpaired presentation of one of the two stimuli. ACKNOWLEDGMENTS The experimental results cited were obtained with support from a National Institutes of Health grant NS 25788 to the author and from a Dana Foundation grant to the University of Pennsylvania. The determination of rhodopsin structure may require alternative approaches Arlene D. Albert ab and Philip L. Yeagleb Departments of Biochemistry" and Ophthalmology, University at Buffalo School of Medicine (SUNY), Buffalo, NY 14214. bchphll(H)ubvms.cc.buffalo.edu Abstract: The structure of rhodopsin is a subject of intense interest. Solving the structure by traditional methods has proved exceedingly challenging. It may therefore be useful to confront the problem by a combination of alternate techniques. These include FTIR (Fourier transform infrared spectroscopy) and AFM (atomic force microscopy) on the intact protein. Furthermore, additional insights may be gained through structural investigations of discrete rhodopsin domains. [HARCRAVE] Determination of rhodopsin structure is central to understanding the function of rhodopsin in visual transduction. The importance of a structure of rhodopsin is emphasized by its membership in the family of G-protein receptors. Due to the homology among members of this family, a structure of rhodopsin will be helpful in understanding function of other G-protein receptors. As described by HARCRAVE, while great progress has been made in understanding the relationship between structure and function of rhodopsin, major hurdles remain. In particular, no true three-dimensional crystal structure has as yet been reported for rhodopsin. Since crystallizing membrane proteins in forms suitable for structure determination has been successful in only a very few cases so far and since the structure of rhodopsin is so critically needed, it is apparent that alternative methods to structure determination should be explored to obtain as much structural information as possible. Fourier transform infrared (FTIR) spectroscopy has recently provided interesting information on the structure of rhodopsin. This technique is especially appealing in that it does not require crystallization of the protein. Consistent with earlier circular dichroism (CD) studies, FTIR studies indicate extensive a-helix. Furthermore, the data from these studies suggest some P-sheet and p-turns in rhodopsin (Brown et al. 1993, Lamba et al. 1994). Pistorius and deGrip have employed an interesting use of FTIR to gain further structural insight. Proteolytic digestion was used to remove exposed peptides from the cytoplasmic side of rhodopsin, and the FTIR spectral results before and after proteolysis were compared. These experiments provided evidence for P-sheet and a P-turn near the C terminal region (Pistorius & deGrip 1994). It may be useful to consider structural studies of rhodopsin from the standpoint of protein structural domains. The structures of many proteins have been shown to consist of multiple structural domains that are independently stabilized. Membrane proteins also exhibit structural domains. A notable example of this is glycophorin which has both extramembranous and intramembranous domains. Circular dichroism studies have shown that these domains retain their secondary structure after tryptic cleavage (Schulte & Marchesi 1979). Rhodopsin has also been shown to exhibit intramembranous structural domains (Albert & Litman 1978, Litman et al. 1982). After proteolysis the rhodopsin fragments remain associated as a complex. However, after detergent solubilization and exposure to light the fragments disassociate. Although when detergent solubilized rhodopsin is bleached there is loss of ot-helical structure, the protein retains at least 30% of its helical structure. Interestingly, the secondary structure which remains after bleaching intact rhodopsin, is not further lost when the opsin proteolytic fragments are separated. These fragments may serve as a useful starting point for additional structural work. If they were isolated in a fully deuterated detergent they may be amenable to NMR studies. Recently, several studies have reported interesting results using fragments of rhodopsin. For example, a peptide incorporating residues 325-338 of the carboxyl terminal of rhodopsin was reported to interact with transducin (Phillips & Cerione 1994). A peptide incorporating the sequence of the third cytoplasmic loop of rhodopsin inhibited binding between rhodopsin and arrestin (Krupnick et al. 1994). Finally, phosphorylated derivatives of the sequence 337-348 of rhodopsin were found to bind to rhodopsin kinase (Pullen et al. 1993). The sites of palmitoylation of rhodopsin are on the carboxyl terminal of the protein. The role of these sites in receptor function has not been identified. Recent fluorescent labeling of these sites with a fluorescent acyl CoA opens the possibility of exploring the function of these sites, and of gaining additional structural information about rhodopsin through fluorescence techniques (Moench et al. 1994). Atomic force microscopy (ATM) may also soon prove to be a powerful tool for investigating rhodopsin structure. This technique has been shown to be sensitive enough to detect changes in the structure of lysozome upon substrate binding (Radmacher et al. 1994). This technique may then be especially useful in investigating the changes occurring upon metarhodopsin II formation. Mutant forms of rhodopsin have provided extensive information regarding the importance of particular residues. Together with the biophysical structure techniques there is reason to anticipate progress on a molecular structure of rhodopsin in the near future. Finally, it is important to consider the role of the lipid bilayer on the structure of rhodopsin. This is especially important in considering the structure of Metarhodopsin II. A number of studies have shown that the formation of this intermediate is dependent upon an appropriate lipid environment. These include studies in reconstituted systems on the role of chain length and unsaturation (Baldwin & Hubbell 1985a; 1985b) and on the role of cholesterol (Mitchell et al. 1990; Straume et al. 1990). The role of cholesterol is especially relevant in that it naturally occurs in the ROS membranes and is capable of inhibiting the cGMP cascade initiated by rhodopsin bleaching in the ROS plasma membrane (Boesze-Battaglia & Albert 1990). BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 469 Commentary /Controversies in Neuroscience III Mechanisms of photoreceptor degenerations Colin J. Barnstable Department of Ophthalmology and Visual Science, Yale University School of Medicine, New Haven, CT 06520-8061. colin.barnstable@quickmail.yale.edu Abstract: The candidate gene approach has identified many causes of photoreceptor rod cell death in retinitis pigmentosa. Some mutations lead to increased cyclicGMP concentrations in rods. Rod photoreceptors are also particularly susceptible to some mutations in housekeeping genes. Although many more cases of macular degeneration than retinitis pigmentosa occur each year, there is much less known about both genetic and sporadic forms of this disease. [OAICER ET AL. ] The target article by DAIGER ET AL. provides an excellent and up to date review of the cloned genes, and other mapped loci, in which mutations result in photoreceptor degenerations. Although the title refers to retinal degenerations, the text is focused on photoreceptor degenerations and does not discuss other retinal diseases, such as some inherited forms of glaucoma that lead to degeneration of retinal ganglion cells. In addition, a number of neural degenerative diseases that can have a strong genetic component, such as Alzheimer's disease, lead to a significant death of retinal ganglion cells. Table 1 in the target article lists loci defined by either linkage or the candidate gene approach. The latter has succeeded in increasing our understanding of photoreceptor degenerations. In the coining years we hope genes will be assigned to the number of anonymous loci. Several of the loci listed in Table 1 of the target article encode housekeeping or widely distributed proteins, suggesting that photoreceptors show a phenotypic effect of the mutation because of some unique sensitivity. Perhaps the best example of this is dominant gyrate atrophy, which is caused by mutations in the ornithine aminotransferase gene on chromosome 10. This gene is expressed in many tissues, including liver, kidney, and brain as well as retina. Since these other tissues show no obvious effects of the mutation it would appear that the photoreceptors are uniquely sensitive to increased concentrations of ornithine. The idea that photoreceptors are more sensitive to otherwise mild variations in proteins may explain why some opsin mutations lead to degeneration. The light-sensitive outer segments are complex membranous extensions of a cilium. To synthesize the membrane removed each day by phagocytosis requires a high level of protein synthesis. Small structural changes in opsin, the major outer segment protein, may slow the movement of protein through the endoplasmic reticulum (ER) and Golgi by just enough to disrupt outer segment biosynthesis and thus lead to eventual cell death. Although the review suggests that "degradation and clearance of aberrant proteins is highly efficient" (sect. 4.2.1, para. 1) this may apply less to the metabolically hyperactive photoreceptors that to other cell types such as fibroblasts. Support for this idea comes from studies with transgenic mice expressing an additional normal opsin gene (Sung et al. 1994). These mice show rod photoreceptor degeneration whose rate is roughly proportional to the level of expression of the transgene. Perhaps some of the mutations listed in Table 1 do not cause structural changes in the opsin molecule itself but rather mark changes in the level of expression. One striking feature of photoreceptor degenerations is that mutations in many components of the transduction pathway can lead to disease. In section 3.2.4 the authors repeat some of the current ideas about why these mutations might cause cell death. Direct measurements of rd mutant mice, which have an inactive rod photoreceptor cGMP-phosphodiesterase, have shown that just before the major period of rod degeneration the levels of cGMP are over 10-fold higher than in normal animals (Farber & Lolley 1974; Fletcher et al. 1986). Normal levels of cGMP are sufficient te activate only a few percent of the cGMP-gated 470 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 cation channels in the outer segment. A 10-fold increase in cGMP would probably activate all the channels and cause complete cation equilibrium, including large increases in intracellular calcium. In section 4.2.3 the authors describe mutations that cause severe, dominant degeneration and in experimental studies have been shown to be constitutively active. In the intact rod photoreceptor it is not clear how the mechanisms that shut down rhodopsin activity, namely phosphorylation and binding of arrestin, act on the constitutively active molecules. The authors are probably incorrect when they state that these mutations will lead to high levels of cGMP. Activation of rhodopsin normally leads to a reduction in cGMP levels by activation of a cGMPphosphodiesterase. If this occurs in these mutants, then it would seem that both abnormally high and abnormally low levels of cGMP might cause photoreceptor degeneration. One point that is not stressed in this review is that the macular degenerations affect many more people than the various forms of retinitis pigmentosa. A recent estimate is that 1,700,000 people in the United States suffer from age-related macular degeneration as compared to 100,000 people with all forms of retinitis pigmentosa. As with many degenerative conditions that affect the aged, it is difficult to determine what proportion of cases have a genetic basis and what proportion are truly sporadic. Many family studies are confounded by the death of subjects before they show clinical symptoms of macular degeneration. An important question in future studies of macular degeneration will be what proportion is due to defects within the retina and what proportion to defects in the underlying tissues. The macular region is avascular and the retina in this region is dependent upon the underlying choroidal blood supply for oxygen and nutrients. Anything that alters the permeability or function of the intervening retinal pigment epithelial cells, or alters the structure of the extracellular matrix between the choroid and the RPE might lead to macular degeneration. Perhaps the strongest point brought out by DAIGER ET AL.'S target article is that studies of photoreceptor degenerations have advanced beyond those of most other neurological degenerative diseases. We know more mutations that can give rise to the diseases, we have better hypotheses as to the mechanism of action of the mutations, we have good natural and transgenic animal models that seem to exhibit very similar phenotypes to those seen in humans. While there is no simple drug to alleviate the problems associated with retinitis pigmentosa or macular degeneration, experimental use of growth factors, transplantation, and targeted gene replacement are all showing great promise. Once again the eye may serve as a window on the brain and knowledge gained from retinal degenerations may prove invaluable for other conditions such as Alzheimer's, Parkinson's, and Huntington's diseases. Genetic and clinical heterogeneity in tapetal retinal dystrophies A. A. B. Bergen The Netherlands Ophthalmic Research Institute, 1100 AC Amsterdam, The Netherlands, bergen@amc.uva.nl Abstract: Large scale DNA-mutation screening in patients with hereditary retinal diseases greatly enhances our knowledge about retinal function and diseases. Scientists, clinicians, patients, and families involved with retinal disorders may directly benefit from these developments. However, certain aspects of this expanding knowledge, such as the correlation between genotype and phenotype, may be much more complicated than we expect at present. [DAIGER ET AL.] DAIGER and colleagues provide another example of how molecular genotyping of a large number of patients with hereditary tapetal retinal dystrophies has dramatically expanded our knowledge about retinal function and disease. In Commentary /Controversies in Neuroscience III summary, the benefit of large-scale DNA-mutation screening of patients with retinitis pigmentosa (RP) and allied disorders is at least four-fold. Specific knowledge about mutated genes in RP provides biochemists with clues about the function of regions of the proteins corresponding to these genes. Thus, biochemical research strategies can be adapted to address functional problems more specifically. Electrophysiological, pathological, and epidemiological studies benefit directly. For the first time, the large number of clinically different types of hereditary retinal disorders can be unambiguously grouped together on the basis of their primary DNA defect. Thus, the results of functional, genetic, and morphological studies can be directly compared within, and between, these specific and uniform groups of disorders (e.g., Zong-Yi et al. 1994). The molecular characterization of RP patients is also important for future therapies. At least some possible therapies, such as gene therapy, will most likely be mutation-specific. Thus, patients already typed at a molecular level will benefit most from these new developments. Clinicians and patients also benefit directly. For a patient, a more accurate differential diagnosis is possible if a functional DNA mutation is found that can be implicated in his disease. The analysis of the segregation of that functional DNA mutation in pedigrees may also lead to improved genetic counselling. Moreover, as DAICER and colleagues describe in their article, more accurate prognostic information about the severity and course of the disease may become available. However, although the contribution of Daiger et al. is very valuable, the data presented should be interpreted with caution. First of all, the number of patients who can be thoroughly characterized both at the genotypic and phenotypic level, is still limited. Unfortunately, with a few exceptions (Jacobson et al. 1994), molecular geneticists tend to summarize the ophthalmic data in their articles, while ophthalmologists usually minimize the amount of molecular genetic information. Also, it is not clear in all cases whether the phenotypic consequences of (groups of) mutations can be compared directly; for instance: How is a "mild" form of RP defined? Can "severe" forms of RP from different studies be directly compared? Although the results of DNA studies can be compared directly and easily, the importance of standard and internationally accepted clinical protocols must (again) be emphasized here. These protocols are extremely useful in order to establish a correct correlation between genotype and phenotype, and also to obtain comparable clinical results. Although mtrafamilial clinical variability in RP has been well documented, little is known yet about the variability of clinical expression within larger pedigrees in which RP is caused by defined DNA mutations (Weleber et al. 1993). It may well be that this clinical variability is caused by the "molecular genetic background" in which the defective gene functions. In other words, while a certain mutated gene may actually cause RP, several other "modifier genes" may influence the clinical expression of the disease. These "modifier genes" may contribute strongly to the expression of the disease (similar to the situation in "digenic RP" (Kajiwara et al. 1994)) or only weakly, thereby influencing only the severity, expression, or progression of the disease. As an example, we recently described a pedigree from a genetic isolate in The Netherlands, in which at least two different types of RP (both genetically and clinically) are segregating within one pedigree (van Soest et al. 1994). We may have localized a modifier gene to chromosome lq, segregating in only part of the pedigree and causing the phenotype retinitis pigmentosa plus para-arteriolar preservation of the retinal pigment epithelium. At present, it is our hypothesis that the location of the gene, which actually causes RP in the entire pedigree, remains to be elucidated. Clearly, other genes directly implicated in RP may be candi- date modifier genes, although other genes that are more fundamental to the cell-structure, or in DNA- or RNA- processing, or metabolism cannot be excluded. Thus, it can be postulated that other genetic factors besides the actual disease causing gene may influence the clinical picture in an individual RP patient. Finally, the phenotype of the patient may also be influenced by nongenetic factors, such as overall physical condition, exposure to light, or smoking. Such factors clearly complicate the phenotypic assignment of patients given their genotype. In conclusion, in my opinion, much more extensive clinical and molecular data of RP patients, especially in larger families, are needed in order to establish a more definite correlation between phenotype and genotype in inherited retinal degeneration. Molecular insights gained from covalently tethering cGMP to the ligand-binding sites of retinal rod cGMP-gated channels R. Lane Brown and Jeffrey W. Karpen R S. Dow Neurological Sciences Institute, Portland, OR 97209 and Department of Physiology, University of Colorado School of Medicine, Denver, CO 80262. karpenj@essex.hsc.colorado.edu Abstract: A photoaffinity analog of cGMP has been used to biochemically identify a new ligand-binding subunit of the retinal rod cGMP-activated ion channel, as well as amino acids in contact with cGMP in the original subunit. Covalent tethering of this probe to channels in excised menbrane patches has revealed a functional heteogeneity in the ligand-binding sites that may arise from the two biochemically identified subunits. [MOLDAY & HSU] In recent years we have been interested in identifying ligand contact points within retinal rod cGMP-gated channels, and in elucidating the mechanism by which cGMP binding leads to channel opening. For these purposes we developed a new photoaffinity analog of cGMP (8-p-azidophenacylthiocGMP; APT-cGMP) that specifically labels the channel from bovine rods, and irreversibly activates the channel in excised patches from amphibian rods by covalently tethering cGMP to the channel's binding sites. This probe has enabled us to address, from a unique perspective, some of the issues raised by MOLDAY & HSU regarding the subunit composition of the channel and how each subunit contributes to overall channel function. In 1993, we reported that APT-[32P]-cGMP specifically labeled the 63-kDa channel subunit originally purified by Cook et al. (1987), as well as a 240-kDa associated protein in a partiallypurified biochemical preparation (Brown et al. 1993a). We subsequently isolated a CNBr-peptide from the 63-kDa subunit that contained virtually all of the label. Sequence analysis indicated that this peptide is contained within the 110 aminoacid region that shows homology with other cyclic nucleotidebinding proteins. The label was further localized to three amino acids (val 524-ala 526) within a larger hydrophobic stretch of residues (Brown et al. 1995). This may explain the channel's preference for cGMP derivatives containing hydrophobic substituents at the C 8 position (Brown et al. 1993b). A model of the cGMP-binding domain proposed by Kumar and Weber (1992), which places these residues on a P-strand near C 8 , is consistent with our photoaffinity labeling result. The recent cloning of a second subunit provided explanations for several functional discrepancies between native channels and expressed a-subunits (Chen et al. 1993). Like most important discoveries, however, it posed a number of new questions. First of all, the biochemical identity of the new subunit was a mystery because no protein of the predicted molecular weight (71 or 102 kDa, depending on the splice variant) was present in highly-purified channel preparations. Although a 240-kDa pro- BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 471 Commentary /Controversies in Neuroscience III tein was known to copurify with the bovine channel, it seemed an unlikely candidate for the second subunit because of the obvious molecular weight discrepancy. Furthermore, antibodies against this protein were shown to cross-react with spectrin, a cytoskeletal protein from red blood cells. We were prompted to take a closer look at the 240-kDa protein, however, when our photoaffinity probe specifically labeled this protein in biochemical preparations. Sequencing of a labeled 8-kDa CNBr fragment revealed a stretch of 16 amino acids that is identical to part of the subunit cloned by Chen et al. (1993). Furthermore, alignment of the amino acid sequences of the two cloned subunits based on homology indicates that the peptide labeled in the 240-kDa protein corresponds to that labeled in the a subunit (Brown et al. 1995). This result agrees with the conclusions of Molday and his colleagues (Chen et al. 1994) that the cloned P subunit is contained within the 240-kDa protein complex. Together with the molecular genetic data, these results provide compelling evidence that the 240-kDa protein binds cCMP and is a second subunit of the native rod channel. A second major question that arises is how the binding of cGMP to each subunit contributes to channel activation. The ability to covalently tether cGMP to channel binding sites in excised membrane patches (Brown et al. 1993a) affords a unique opportunity to investigate the allosteric mechanism of channel activation. As the fraction of covalently-activated channels increased, the dose-response relation for the remaining closed, but partially-liganded, channels became progressively shallower. This behavior is expected from channels that are activated by the binding of only one or two molecules of cGMP, rather than the full complement of Iigands. In fact, after 80% of the channels in a patch have been covalently activated, binomial statistics predicts that the remaining channels will be missing only thefinalligand. Dose-response relations for this population of channels revealed a previously unknown heterogeneity in the channel's cGMP-binding sites. The relations were shallower than predicted by single-site activation models; however, the relations were fit rather well by a model assuming two populations of binding sites, present in roughly equal proportions, which bind cGMP with widely different apparent affinities (Brown & Karpen 1994). This heterogeneity appears to reflect an intrinsic property of the native channel because, although neither affinity alone was consistent with the original doseresponse relation of the patch, both affinities taken together provided a goodfit.Although the origin of this heterogeneity is unknown, we are testing two intriguing possibilities: it may arise from the two subunits described above, or it may be due to differential modification of binding sites. Gordon et al. (1992) have reported spontaneous shifts in the channel's dose-response relation to lower concentrations following patch excision. These shifts were blocked by protein phosphatase inhibitors and accelerated by protein phosphatases, suggesting that phosphorylation may play a role in modulating the channel's affinity for cGMP. Because covalent tethering of cGMP to the channel's binding sites permits the functional isolation of channels with a fixed number of bound Iigands, it promises to be a valuable tool to dissect the contribution of individual subunits to the overall function and regulation of this type of channel. The structure of rhodopsin and mechanisms of visual adaptation Rosalie K. Crouch3 and D. Wesley Corson ab Departments of "Ophthalmology and of bPathology and Laboratory Medicine, Medical University of South Carolina, Charleston, SC 29425. crouchro@musc.edu and corsondw@musc.edu Abstract: Rapidly advancing studies on rhodopsin have focused on new strategies for crystallization of this integral membrane protein for x-ray 472 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 analysis and on alternative methods for structural determination from nuclear magnetic resonance data. Functional studies of the interactions between the apoprotein and its chromophore have clarified the role of the chromophore in deactivation of opsin and in photoactivation of the pigment. [BOWNDS & ARSHAVSKY; HARCRAVE] HARCRAVE'S target article is a thorough and penetrating analysis of the structure and function of rhodopsin that is current up through 1993. Little new structural information has been added since that time. The author summarizes three methods of obtaining atomic level structural data for proteins: x-ray diffraction of threedimensional crystals, electron diffraction of two dimensional crystals, and nuclear magnetic resonance analysis of rhodopsin in solution. The author dismisses the latter technique due to the size of rhodopsin. However, the latter point is debatable as data are now being obtained on proteins and receptor-ligand complexes in the 35-40kD range (Clore & Gronenborn 1994). Considerable progress has been made on obtaining both twoand three-dimensional crystals suitable for x-ray diffraction and the majority of the paper is a discussion of suggestions for improving this process. The importance of the choice of detergent was emphasized. Since HARCRAVE'S target article was written, Schafmeister et al. (1993) have reported the design of a homogeneous peptide for use as a detergent to solubilize integral membrane proteins. The general plan was to have a peptide that spanned the membrane and which would "pack around the protein in a rigid, wellordered, parallel a-helical arrangement. " This detergent was found to solubilize both rhodopsin and bacteriorhodopsin. The approach offers clear advantages to the more traditional detergents which are likely to cause disorder in the crystals. As HARGRAVE has considered, the inherent photosensitivity of rhodopsin is a major problem for crystallization. A potential solution to this problem is to reconstitute the pigment with a chromophore that cannot isomerize and is therefore not photosensitive. The use of ring locked retinal derivatives would seem to be the most straightforward approach. Unfortunately, as discussed in the target article, activation of these pigment analogues is still observed, even in the absence of full isomerization about a double bond. An alternative approach might be to use fragments of retinal that do not form a Schiff base with the protein but which have been shown to induce a "rhodopsin-like" conformation by physiological studies (Jin et al. 1993). Partial occupation of the binding site by fragments of retinal may serve to maintain the protein in a sufficiently rigid conformation for crystallization without the problems arising from light activation of extended chromophores. A third approach which might be considered is to use peptides from transducin to stabilize the Meta II state of rhodopsin for crystallization in a more soluble complex. Recent studies have shown binding of peptides from transducin to the activated form of rhodopsin (Hamm & Rarick 1994). This approach has recently been used to crystallize the complex of alpha transducin (Noel et al. 1993). Although this latter study did not involve integral membrane proteins, peptides binding to the activated form of rhodopsin may impart sufficient stability to the complex for crystallization to occur. BOWNDS & ARSHAVSKY provide a thorough and perceptive overview of adaptation. They cover recent work on photoreceptor adaptation up through July of 1994. A few comments on background and bleaching adaptation are in order. In considering bleaching adaptation, that is, the total desensitization that occurs after bleach of a substantial amount of pigment, it is important to distinguish among the forms of desensitization produced by (1) pigment depletion, (2) the short-lived effects of background adaptation produced by the bleaching light, (3) the long-lived but finite effects of accumulated bleaching photoproducts (see Lamb 1990), and (4) the effects of accumulated opsin which persist indefinitely in cells in Commentary the absence of 11-cis retinal (Corson et al. 1990; Perlman et al. 1982). BOWNDS & ARSHAVSKY emphasize recently discovered similarities between background adaptation and opsin desensitization (see Cornwall & Fain 1994). However, several significant differences remain to be explained. First, the diffusional space constant for opsin desensitization is considerably smaller than that of background adaptation (Cornwall & Pan 1985; Cornwall ct al. 1983). Second, Fain & Cornwall (1993) have recently reported that the noise associated with opsin desensitization is not similar to the photon shot noise produced by dim backgrounds or the shot noise that occurs transiently after small bleaches. Third, the two mechanisms display markedly different sensitivities to IBMX (Cornwall et al. 1990). Finally, although a sufficiently bright background light can completely suppress the light response, this is not the case with bleaching adaptation (Cornwall et al. 1990). These differences are not easily reconciled with a simple single mechanism that explains both phenomena. Recent experiments on adaptation have probed the physiological lifetime of photoactivated pigment (R*) (see Pepperberg et al. 1994). One measure of the physiological lifetime of R* can be derived from the intensity dependence of the response saturation following bright flashes (Pepperberg et al. 1992). By this measure, the physiological lifetime of R* (~2 seconds at ~20°C in amphibian rods) appears to be independent of both background and bleaching adaptation (Corson et al. 1994b). It can, however, be modified through manipulation of the chromophore as might be expected for this fundamental process (Corson et al. 1994a; 1994b). However, other physiological experiments by Lagnado & Baylor (1994) suggest that R* is subject to control by the intracellular calcium concentration during a narrow window at the onset of a flash. Finally, other recent work is directed to the role of the chromophore in adaptation and pigment cycling. Detachment of the all-trans chromophore from opsin, conversion to retinol and export from the photoreceptors have recently become active topics of investigation (Hofmann et al. 1992; Palczewski et al. (1994) but see Jones et al. 1989). ACKNOWLEDGMENT Supported by N1H grants EY04939 and EY07543 and Research to Prevent Blindness. The key to rhodopsin function lies in the structure of its interface with transducin Edward A. Dratz Department of Chemistry and Biochemistry, Montana State University, Bozeman, MT 59717. uched@msu.oscs.montana.edu or dratz@chemistry.montana.edu Abstract: Light activated rhodopsin functions by catalyzing the exchange of GTP for GDP on numerous copies of transducin. Peptide mapping has shown that at least six regions, three on rhodopsin and three on the transducin alpha subunit, are involved in the active interface between the two proteins. The most informative structural studies of rhodopsin should include focus on the transducin interaction. [HARCRAVE] Achieving a detailed understanding of biological mechanisms requires knowledge of the molecular structures and dynamics of the component molecules and of the functional complexes between components. The first step of signal amplification in visual excitation is accomplished by the interaction of light excited rhodopsin with the CTP binding protein, transducin. Until very recently, concepts of rhodopsin structure have of necessity relied on theoretical models. For example, the model of rhodopsin proposed by Dratz and Hargrave (1983) contained a bundle of seven (bent) transmembrane helices /Controversies in Neuroscience III (7TMH) and a detailed alignment of the amino acid sequence in the membrane. This and related models helped to organize thinking about the structure of rhodopsin and other G-proteincoupled receptor (GPCR) proteins. The general features of the 1983 vintage rhodopsin model survive today. For example, proline residues, that were proposed to bend the transmembrane helices, have been found to be among the most conserved features of the growing super-family of GPCR that now includes perhaps 850 members. However, most features of the model have not been tested with experimental data. 2D crystals thought to be due to rhodopsin, were grown and studied by Corless et al. (1982) and similar crystals shown to be due to rhodopsin by Dratz et al. (1985), were consistent with the membrane surface are of the 7TMH model. The crystals were stained with heavy metals to improve contrast, which restricted resolution to ca. 20 A and structural detail was limited. Recently, Schertler et al. (1993) were able to obtain much better resolution data (ca. 9 A in plane) on unstained, two dimensional crystals of rhodopsin, where some of the long-imagined transmembrane helices could be visualized in the structure for the first time and the approximate packing of the helices was apparent. Transducin can be eluted from the retinal rod disk membrane surface in soluble form and has been much more amenable to crystallization and structural study than rhodopsin. Noel et al. (1993) recently published a high resolution (2.2 A) x-ray crystal structure of the a subunit of transducin. Lambright et al. (1994) were able to further refine the transducin a structure (to 1.8 A) and show structural changes that occur in the crystals between the Ga complex with GDP (the "off" state of the protein) and the Ga complex with a nonhydrolyzable GTP analog (the "on" state of the protein). This elegant work on Ga structure did not, however, reveal Ga interactions with the excited rhodopsin, where Ga has an empty nucleotide binding site (the required intermediate between GDP release and GTP uptake). Secrets of visual excitation and the function of rhodopsin lie in the interaction between rhodopsin and transducing. Peptide mapping has provided information on the interacting interfaces between rhodopsin and transducin. Hamm and coworkers (1988) first reported that peptides which mimicked three regions of Ga (residues 1-23, 311-329, and 340-350) were able to block formation of the complex between light excited rhodopsin (Mil) and transducin. Soon after, Konig and coworkers (1989) reported that peptides mimicking three regions of rhodopsin, thought to be on cytoplasmic loops of the structure, were also able to block the Mf]-transducin complex. Therefore, at least six regions, three on each protein, are involved in the active interface between Mil and Ga. The peptide mapping experiments stimulated Dratz and coworkers (1990; 1991) to apply NMR techniques to study the structures of the interfacial peptides when they are bound to their intact protein partner. Recently, Dratz and coworkers (1993) reported the use of NMR methods to study the three dimensional structures of Ga peptide homolog, N-acetyl 340350 (K341R), when it was bound to rhodopsin in the dark (called the "precoupled" state). In addition, structural changes were found in the bound peptide when rhodopsin was excited by light. The peptide interactions with their intact protein interfaces have rather low affinity but appear to be quite specific. For example, changes of single amino acids from Gly—»Ala or Cys—»Ser or amidination of the C-terminal carboxyl group in Ga 340-350 each greatly reduce the activity of the peptides. The crystal structure of Ga was not useful for comparison with the NMR structure of Ga 340-350 bound to rhodopsin or Mil, because most of the 340-350 region was disordered in the crystals (Noel et al. 1993; Lambright et al. 1994). Furthermore, binding to rhodopsin or Mil would be expected to affect the structure. The NMR approach reported by Dratz et al. (1993) showed feasibility, but the results were preliminary since structure refinement (such as used in the development of x-ray structures) was not yet reported. These NMR methods are BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 473 Commentary /Controversies in Neuroscience HI yielding more detailed information on the structure of the rhodopsin/Mfl-transducin interface as structure refinement and data acquisition proceed on several of the peptides on the contract interface between the two proteins (Busse, Furstenau & Dratz, in preparation). In addition, enhanced NMR methods, using G protein peptides combined with constitutive rhodopsin mutants (Cohen et al. 1993) and rhodopsin mutants which bind but do not activate G proteins upon bleaching (Franke et al. 1990), may be expected to illuminate the molecular basis of the activation mechanism. NMR methods also have the potential for obtaining information on the dynamics of the structure and the structural interfaces. However, it is not yet clear if the resolution of the structures determined by NMR on bound peptides, which mimic the protein-protein interfaces, will be as high as those obtainable with x-ray diffraction on high quality 3D crystals. Therefore, approaches to obtain better ordered 2D and 3D crystals should certainly be vigorously pursued. Current effort to improve the 2D crystals of rhodopsin, described by HARGRAVE in the target article, are providing new information on the structure of rhodopsin. HARGRAVE cites a manuscript in preparation in 1994, by Schertler and Hargrave, where the planer (2D) projection of frog rhodopsin was obtained with 6 A resolution. If tilt series data can be reconstructed on such 2D crystals, a 3D structure could be obtained with a ca. 12A resolution perpendicular to the plane of the membrane. Such a structure would be sufficient to reveal the location of helices rather clearly, but would not resolve even the largest amino acid side chains. Many years of work were required on bacteriorhodopsin to obtain a 3.5 A 2d/ca. 5.5A 3D structure, which was sufficient to resolve the positions of about 21 of the largest amino acid side chains in the transmembrane helices and thus reveal much of the alignment of the amino acid sequence in the structure (Henderson et al. 1990). The electron microscopy method uses a series of 2D images, obtained by tilting the specimen, to mathematically reconstruct the 3D electron density profile of the protein in the 2D array. This process suffers from the so called "missing cone" problem since data can be obtained up to a maximum tilt angle of about 60°. Therefore, with this method the 3D reconstruction of a cylindrical helix yields a long, skinny "morphed" football shape with tapered ends. The most detailed information is obtained by tilt series image reconstruction in the center of the membrane, whereas most of the information on the structure of the interconnecting loops on the surface is lost in the missing cone. The regions of rhodopsin that are implicated in coupling to transduction by peptide mapping (Konig et al. 1989) are thought to be located in or near the cytoplasmic surface loops. Formation of transducin complexes with light-excited 2D crystals of rhodopsin would shift the center of the electron density to near the interface of interest between the two proteins. However, it has not been possible to pack transducin (GotB-y = 80kD) above about 1 transducin per 4 rhodopsins in the native membrane. The surface density of rhodopsin in the native membrane is 2-3 times lower than in 2D crystals and so it may not be possible to bind more than 1 transducin per 8-10 rhodopsins in 2D rhodopsin crystals, and this would presumably provide poorly ordered transducing arrays. Therefore the smaller 40 kDa Got subunit of transducin should be used instead to bind light excited 2D rhodopsin crystals, since it should be possible to obtain a much tighter packing of Got than the whole transducin. Presumably, it would be important to attain a 1:1 complex of Mfl rhodopsin and Got, since Got stabilizes Mil extremely well and uncomplexed Mfl would decay to later intermediates and disorder the structure. If the whole Got subunit cannot be made to pack in a 1:1 complex with Mfl in 2D crystals, perhaps a large fragment of Got could be produced that would do so. It seems that the most desirable approach to provide the structural information of interest is to seek 3D crystals contain- 474 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 ing rhodopsin. By far the most desirable type of 3D crystals would be complexes of light excited rhodopsin and transducin (Mil-Gap"/) or light-excited rhodopsin and the Got subunit (Mil-Got). Hargrave presents arguments that it might be easier to crystallize these complexes than pure rhodopsin, since crystal packing forces are strengthened by increasing the size of the aqueous domains. A major problem impeding progress is the "tradition" in structural biology in the USA that investigators generally cannot obtain grant support for growing crystals of macromolecules. Investigators in Europe can often obtain support for seeking crystals, and so much of the leading work is going on there. Some of the labs in the United States that work with these proteins have worked on the crystallization problem, but they have had to do so on a relatively small scale with little direct support. I urge that attacks on the crystallization and structure determination of rhodopsin/transducin be funded and pursued more vigorously. Progress is most likely to occur if financial support is offered to consortia of investigators who combine a wide range of different expertise. As pointed out by HARGRAVE, many different types of expertise can be identified as being important for this eflFort: protein purification; site specific mutagenesis; protein expression; amphiphile, detergent, and lipid synthesis; protein characterization; crystal growth; crystal evaluation; synchrotron source and crystallography. I take the opportunity in this commentary to urge (and ask readers to urge) the US National Eye Institute to issue a Request for Proposals (RFP) inviting the formation of competing consortia to attack the crystallization and structure determination of rhodopsin and the rhodopsin-G protein complex. If the attack were widened to include other GPCR, than several NIH Institutes could reasonably team to cooperatively fund crystallization/structure determination consortia and perhaps more consortia could be added. It may well also be possible to obtain industrial/pharmaceutical partners for such an effort. Members of each consortium could be located anywhere in the world and communicate efficiently by Internet and video conferencing. A detailed 3D structure of any one of the GPCR complexes with its cognate G protein would provide an enormously increased understanding of this major signal transduction pathway in biology. Such an effort would also be expected to develop new methodology for structural studies of other membrane proteins. Discovery of the detailed molecular structures of membrane receptors would greatly facilitate structure-based design of improved drugs to specifically modulate different signal transduction pathways. The atomic structure of visual rhodopsin: How and when? R. Michael Garavito Department of Biochemistry and Molecular Biology, The University of Chicago, Chicago, IL 60637. garavito@biovax.uchicago.edu Abstract: Strong arguments are presented by Hargrave suggesting that the crystallization of visual rhodopsin for high resolution analysis by X-ray crystallography or electron microscopy is feasible. However, the eflFort needed to achieve this goal will most likely exceed the resources of a single laboratory and a concerted approach to the research is necessary. 1. Introduction: The dilemma, [HARGRAVE] Twenty years has elapsed since Henderson and Unwin (1975)firstpresented their low resolution structure of bacteriorhodopsin from purple membrane. Since 1975, the atomic structures of several integral membrane proteins have been reported (for a recent overview, see Sowadski 1994) that have significantly altered the way we view membrane protein structure. Moreover, this emphasizes Commentary that effective techniques for growing 3-dimensional and 2-dimensional crystals of membrane proteins have been developed (Garavito 1995; Garavito & Picot 1990; Kuhlbrandt 1988; 1992). While this seems to promise a bright future, the handful of successes pale when compared to the exponential growth in the number of structure determinations of soluble proteins (see Bowie et al. 1991). The bottom line is that the effort needed to bring a membrane protein structure project to fruition remains enormous and long-term. The questions posed by my title I consider critical: how should investigators approach this goal and how long might it take? As molecular biology and computer modeling (Baldwin 1993), has improved, some have argued that we will gain more from these techniques, over the same span of time, than from experimental structure determination, HARCRAVE raises this issue with regard to visual rhodopsin and discusses many of the facets of the research problem that highlights the dilemma that most investigators in this field face. 2. The growth and analysis of 3-dlmenslonal crystals of rhodopsin. X-ray crystallography requires well ordered, 3-dimensional crystals for the analysis to succeed. The physical, biochemical, and biological characteristics of rhodopsin satisfy most, though not all, of the criteria (Garavito 1995) I feel are essential before engaging in a membrane crystallization experiment: (1) the protein should be obtainable in milligram quantities; (2) the protein is stable under a reasonably wide range of detergent conditions; (3) native (although perhaps inactivated) protein can be isolated and purified; (4) the purified protein is chemically homogeneous; and (5) the detergent-solubilized protein is physically monodisperse (i.e., it forms a unform population of protein-detergent aggregates). Most preparations of rhodopsin satisfy all but the last two criteria. Not only is rhodopsin heterogeneously glycosylated (Hargrave, sect. 9, para. 1) but may also be contaminated by phosphorylated opsin due to incomplete dark adaptation (Hargrave, sect. 9, para. 3) and nonstoichiometric palmitoylation (Hargrave, sect. 9, para. 3). The possibility of chemical heterogeneity being introduced after rhodopsin is purified, particularly from stray-light induced opsin formation (Hargrave, sects. 17 and 18) and depalmitoylation (Hargrave, sect. 9, para. 3), remain major concerns. Furthermore, the physical state of the rhodopsin-dctergent complex, although well-studied compared to other membrane proteins (Hargrave, sect. 11), is still not sufficiently defined. The idea that monovalent Fab! fragments from monoclonal IgCs might be used to add 50 kDa of "soluble" protein mass to a rhodopsin molecule (Hargrave, sect. 20, para. 2) is an excellent one and has been bandied about for more than 10 years. If specific, high avidity monoclonal IgGs are available, these experiments should be done. Moreover, the idea of crystallizing fusion proteins (Hargrave, sect. 20, para. 4) may also work for smaller (<50 kDa), monomeric membrane proteins although the linker region between the fused proteins may need to be carefully engineered. The lectin-rhodopsin complexes (Hargrave, sect. 20, para. 3) may be less likely to succeed as lectin binding would be expected to be less specific and more heterogeneous given the glycosylation present. For both the Fabrhodopsin and fusion protein "complexes," they will still need to be solubilized in detergent to prevent nonspecific aggregation. However, the increase in the polar surface area should decrease the detergent-dependence on crystallization (Garavito & Picot 1990). If a chemically and physically homogeneous preparation of rhodopsin, preferably light-stable, can be produced, the chances of finding appropriate conditions for the growth of X-ray quality crystals are greatly improved. Poor quality 3-dimensional crystals of rhodopsin have already been obtained (Hargrave, sect. 6), so the potential is there. However, the successful determination of a crystal structure by X-ray diffraction depends /Controversies in Neuroscience HI on crystal quality: they should display well-ordered Bragg diffraction to better than 3.4 A resolution if a moderately accurate atomic model is to be built. The history of the bacteriorhodopsin crystal structure research (Michel 1982; Michel & Oesterhelt 1980; Schertler et al. 1993b) demonstrates the effort and time expended in searching for conditions which yield X-ray quality crystals. However, the crystallization potential of purified, detergent-solubilized rhodopsin warrants further, rigorous investigation. 3. The growth and analysis of 2-dlmenslonal crystals of rhodopsin. The advances made by high resolution electron microscopy (Henderson et al. 1990; Kuhlbrandt et al. 1994; Subramaniam et al. 1993) have clearly demonstrated that this technique has come of age and promises to be quite powerful. The consensus of the field (Robert Glaeser, University of California, personal communication) is that upcoming technical advances and refinements promise even wider application. Kuhlbrandt (1992) presents the straight-forward methodology for obtaining 2-dimensional crystals, although the exact roles of many of the components (e.g., specific lipids) in the crystallization process remain to be determined (Nussberger et al. 1993). However, a distinguishing difference between X-ray and electron microscopic crystallographic techniques is that even poor 2-dimensional crystals can yield useful low resolution structures of membrane proteins (Hargrave, sect. 13, para. 2; Kuhlbrandt 1992), including rhodopsin (Hargrave, sect. 13, para. 3; Schertler et al. 1993a). It may also be more feasible to devise experiments to trap light-induced structural changes in 2-dimensional crystals (Subramaniam et al. 1993). 4. Conclusions: Where should one go? A careful reading of the target article by HARGRAVE reveals two important aspects of the research. First, the potential for the successful determination of the rhodopsin atomic structure is excellent, either by X-ray diffraction or electron microscopy. Second, the work ahead promises to be exhausting and time-consuming. It is no coincidence that most of the successful structure determinations of membrane proteins have required the collaboration of two or more laboratories and many investigators. Thus, in many respects, the real issue is: can sufficient resources be brought to bear on the problem over the long-term? Hargrave shows it is feasible, but how long remains a question to be answered. Does calmodulin play a functional role in phototransduction? Mark P. Gray-Keller and Peter B. Detwiler Department of Physiology and Biophysics, University of Washington School of Medicine, Seattle, WA 98195. mgkeller@u.washington.edu and detwiler@ii.washington.edu Abstract: Molday and Hsu review results from in vitro experiments, which indicate that Ca-bound calmodulin reduces the cGMP sensitivity of the cyclic nucleotide-gated channel of photoreceptor cells, and speculate about the role they might play in the recovery of the light response. We discuss results from in vivo experiments that argue against the participation of Ca-calmodulin in photorecovery. [MOLDAY & HSU] In vitro experiments are discussed by MOLDAY & HSU that indicate the cGMP binding affinity of the cyclic nucleotide-gated channel from retinal rods is reduced by calmodulin in a Ca-dependent manner. The authors speculate about the possibility that Ca-calmodulin regulation of the channel plays a role in shaping the recovery of the light response under physiological conditions. In previous studies on functionally intact rod outer segments we have found, however, that neither the amplitude of circulating dark current nor the sensitivity and kinetics of the electrical light response were affected by either exogenous calmodulin (Gray-Keller et al. 1993) or high BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 475 Commentary /Controversies in Neuroscience III concentrations of calmodulin inhibitors (trifluoperazine, W-7, calmidazolium, and mastoparan). Our brief commentary focuses on possible explanations for the differences between the results of in vitro and in vivo studies of the role of calmodulin in phototransduction. One option is to ignore the inhibitor experiments on the grounds that negative results are notoriously problematic in their interpretation. Other studies, however, using truncated rod outer segments (Lagnado & Baylor 1994) and excised outer segment membrane patches (Gordon et al. 1994) have also failed to demonstrate effects of calmodulin inhibitors. Taken together, these experiments on more intact preparations than those used by Hsu and Molday (1993) suggest that the negative results with calmodulin and its antagonists deserve consideration. The observation that exogenous calmodulin had no effect on the light response in dialyzed rod outer segments may indicate the presence of a saturating amount of endogenous calmodulin. The ineffectiveness of calmodulin inhibitors suggests that, during a light response, internal free Ca (Ca,) does not fall low enough to dissociate Ca-calmodulin from the channel complex. Using Indo-dextran, a fluorescent Ca indicator, we have recently measured Ca( in dialyzed rod outer segments in the dark and during subsaturating flash and background responses (GrayKeller & Detwiler 1994). In darkness steady-state Ca( was 554 ± 25 nM (n = 28) and briefly declined to a minimum of—325 nM shortly (~600 msec) after the peak of a flash response that transiently suppressed the circulating dark current by ~70%. On the basis of results reported by Hsu and Molday (1993) the calmodulin-dependent shift in the cGMP affinity of the channel would not be expected to be triggered by a fall in Ca, from 550 to 325 nM. Hsu and Molday found that calmodulin had its largest effect on the channel when Cas was between 50 to 100 nM. Our measurements suggest that Cas would reach these values only during very bright illumination. In steady saturating light, for example, after permanent closure of all cGMP-gated channels, Ca, dropped to a minimum level of about 50 nM in approximately 20 seconds. Another possible explanation for the negative results with the calmodulin inhibitors is that rods contain an endogenous factor that affects the cyclic nucleotide affinity of the channel-like calmodulin but is not calmodulin. This suggestion is supported by recent experiments on truncated rods (Nakatani et al. 1995) and excised patches (Gordon et al. 1995), which show a large irreversible increase in cGMP-induced current following exposure to low Ca (tens of nanomolar). These results are consistent with an endogenous factor that continually reduces the cGMP binding affinity of the channel and can be removed when Ca levels are in the vicinity of 20 to 60 nM. In this scenario, the endogenous inhibitory factor would only dissociate from the channel in functionally intact cells during bright steady light when Ca, falls to the lower end of its physiological range. In summary, the physiological importance of the interaction between the channel and Ca-calmodulin, or some other unidentified endogenous Ca-binding protein, has not been established. Recent studies suggest that the Ca-dependent regulation of the cGMP binding affinity of the channel may only operate in bright steady light when Cas falls low enough to dissociate Ca from the channel-associated binding protein. describing how structure determines function is only just beginning. The discovery that the affinity of the rod channel for its agonist can be modulated indicates that the relationship between intracellular cGMP and the channel's open probability (current) during the course of the photoresponse may be more complex than previously thought. Structure and physiology of photoreceptor cGMP-gated cation channels The presence of an S4-like region (the presumptive voltagesensor of voltage-gated channels) in these nearly (rod) or completely (cone) voltage-insensitive channels is another mystery. As MOLDAY & HSU point out, the S4 region of the cGMP-gated channels have lost three or more of the positively charged repeats. This might explain the loss of voltage-dependence in the gating process, but on the other hand, parts of the gating process are strongly voltage-dependent even though there is no net voltage-dependence. This can be seen from the voltage- Lawrence W. Haynes Department of Medical Physiology, University of Calgary, Calgary, Alberta, T2N 4N1, Canada, haynes@acs.ucalgary.ca Abstract: The primary sequence of two subunits of the rod and one subunit of the cone cGMP-gated channel have been described, but 476 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 [MOLDAY & HSU] An excellent review of the cloning and expression of the vertebrate rod photoreceptor's cGMP-gated cation channel has been provided by MOLDAY & HSU. It clearly demonstrates the power of molecular biology and just as clearly shows the limitations of this approach. These limitations are apparent both in what the authors say and in what they are forced to omit. Although we have learned a great deal about these channels, what we have yet to learn about their structure and how structure determines function is enormous. If the goal of those of us working on the cGMP-gated channels is to understand how, at the molecular level, these channels function, then cloning the proteins that comprise the channel is a necessary first step, after which investigation of structurefunction relationships using site-directed mutagenesis can be undertaken. These studies, in turn, rely implicitly upon knowing the biophysical properties of the native channel since this is the standard against which the mutations must be judged. Thus, it was clear from the gating properties and diltiazem sensitivity of the expressed homomeric channel as cloned by Kaupp et al. (1989) that, while they had a cGMP-gated channel, it was not the same as the rod channel. Something was missing. This, in turn, led to the discovery of a second subunit by Chen et al. (1993). Several questions arise. First, are their more subunits? MOLDAY & HSU suggest a third subunit that is tightly associated with the P subunit in vivo, but the function of this subunit is not understood. Second, howdoesthePsubunitinteractwiththeasubunit to alter the gating of the channel? Third, what is the stoichiometry for the subunits of the channel? While knowing the primary sequence of the channel is important, knowing the three dimensional structure is crucial to understanding how the protein functions. At present, our knowledge of the structure of these channels allows us to draw only crude cartoons about how we imagine the pieces may fit together. This can be misleading since, as pointed out above, we do not yet know if we have all of the pieces to this puzzle. Nevertheless, the gaps in the picture can be illuminating. For example, simply examining the amino acid sequence of the clone can lead to some mistaken conclusions. Consider the sequence of the putative pore region of the cloned rod and cone a subunits. This sequence is identical in the two types of channel, yet the electrophysiological evidence obtained from native channels suggest striking differences between the rod and cone channels. Under identical physiological conditions, the rod channel shows strong outward rectification while the cone channels show exponential increases in current at both hyper-and depolarized potentials (reviewed in Yau & Baylor 1989). Clearly, the putative pore sequence cannot explain such differences. Either there is a cone p subunit that changes the permeation properties of the channel, or there are other regions of the channel a subunit that also form part of the pore and which differ in sequence between rods and cones. Site-directed mutagenesis studies can help answer this question, but knowing the 3D structure of the channel would help greatly by indicating which regions of the protein to probe. Commentary/Controversies in Neuroscience III dependence of diltiazem "block.' Diltiazem does not, in fact, block the pore of the channel, but is instead a gating-modifier (Haynes 1992). One can speculate that diltiazem immobilizes a segment of the protein that must move during the gating process. One can further speculate that there are at least two such segments moving in opposite directions (or the same direction but having opposite charge) during normal gating. This would give rise to a gating process with no net voltageindependence except when one of the two segments is immobilized by diltiazem. If this scheme is correct, which segments are moving and how far do they travel? Since the P subunit confers both diltiazem sensitivity and modifies the gating of the a subunit, the binding site for diltiazem is probably located within the P subunit. The location of this site and how it affects the conformation of the channel need to be determined. This is only part of a larger question, which is how the interaction of a ligand with its binding site in one portion of the channel protein leads to conformational changes that give rise to the opening of a pore in another, distant, part of the channel. The description of the gating of these channels at a level deeper than rather crude kinetic models is a complete mystery. The observation of Hsu and Molday (1993) that calciumcalmodulin reduced the apparent affinity of the rod cGMP-gated channel for its ligand in vitro is very interesting. Together with the observations of Gordon et al. (1992) that the rod channel may undergo phosphorylation-dephosphorylation reactions, this offers a possible mechanism that can explain the high degree of variability in the apparent affinity of these channels for their ligand as well as a mechanism that may accelerate the recovery of the response to light. It remains to be seen if this process works in vivo, however. In the only experiment dealing with the physiology of calmodulin, Gray-Keller et al. (1993) found dialysis of calmodulin into a rod from a patch pipette had little effect on the shape of the light response. This might be expected if the concentration of calmodulin within the rod outer segment was enough to saturate any calmodulin binding sites. It would be interesting, therefore, to see if the calmodulin inhibitor mastoparen prolongs the light response without affecting the rising phase as Hsu and Molday's hypothesis predicts. The time-course of modulation is, of course, of crucial importance here. It is unfortunate that Hsu and Molday's experiments could not provide this information. If the affinity of calmodulin for calcium is - 6 0 iiM (Fig. 3 of Hsu and Molday 1993), then the mean time that it takes calcium to unbind from calmodulin is ~170 msec (assuming diffusion limited binding and no cooperativity). Clearly, this length of time is not important when considering the long responses elicited by bright flashes or steps, but it may be significant when considering responses to dim flashes. One might expect that cones, with their shorter light responses, would undergo modulation by calmodulin to a greater degree. Contrary to this expectation, I have found that while catfish rod channels are modulated by calmodulin, cone channels are not under identical conditions. This would represent a major departure between rods and cones in terms of the mechanisms used for phototransduction. It may be that the unbinding rate for calcium is too slow for the cone response and so this mechanism is not used. If, as MOLDAY & HSU'S Figure 6 sug- gests, the calmodulin binding site is on the P (or 7) subunit then the cone analog of this subunit must lack the calmodulin binding site. Testing this prediction must await the cloning of the cone channel's other subunits. ACKNOWLEDGMENT The author is an Alberta Heritage Foundation for Medical Research and Medical Research Council of Canada Scholar. This work was supported by the MRC. Long term potentiation and CaM-sensitive adenylyl cyclase: Long-term prospects Warren Heideman Division of Pharmacology, School of Pharmacy, University of Wisconsin, Madison, Wl 53706. heldeman@macc.wisc.ed Abstract: The type I CaM-sensitive adenylyl cyclase is in a position to integrate signals from multiple inputs, consistent with the requirements for mediating long term potentiation (LTP). Biochemical and genetic evidence supports the idea that this enzyme plays an important role in LTP. However, more work is needed before we will be certain of the role that CaM-sensitive adenylyl cyclases play in LTP. [XIA ET AL.] The work of Westcott et al. (1979), published some 15 years ago, describing an adenylyl cyclase that could respond to both GTP and to calmodulin (CaM) pointed to a potential crossroad for two major signal transduction systems. The fact that this enzyme could respond to both G protein regulation and to signals regulating calcium and calmodulin pointed to the possibility that this form of adenylyl cyclase could play a pivotal role in integrating multiple signals coming into neuronal cells. XIA ET AL. have collected data showing that this integration of signals indeed exists. In particular, this integration can take the form of a synergistic burst of cAMP production that is much greater than that produced in response to a single input. This synergism is relevant to long term potentiation (LTP) in that, with LTP, combinations of signals produce responses that are distinctly different from those produced by individual signals alone. Experiments manipulating cAMP, and Ca + 2 in neuronal cells also point to a role for these systems in LTP. Thus, activation of these two pathways correlates with induction of LTP. The evidence from invertebrates is perhaps even more compelling (Abrams et al. 1991a). It is perhaps remarkable that learned responses can be observed in organisms that are enormously divergent. Thus, the molecular mechanisms that underlie this process must have been conserved over hundreds of millions of years. The fact that calmodulin-sensitive adenylyl cyclases are associated with learning in creatures as divergent as nudibranchs and fruit flies suggests more than a coincidence. This is strengthened by the finding that the CaM-sensitive type I adenylyl cyclase is restricted to only certain regions of the mammalian brain. With these arguments in mind, can we say that the CaMsensitive adenylyl cyclases play a role in neuroplasticity? It seems likely that indeed they must play at least some role. More work remains ahead before we can describe that role in exact terms. The disruption experiments should be very revealing, but the tantalizing early results reported by XIA ET AL. point to the possibility that this is a redundant pathway in rodents. The modest effects on CaM-sensitive adenylyl cyclase activity and learning suggest that there may be another CaM-sensitive adenylyl cyclase that is not the type I and perhaps not the type III that can also mediate learning in the absence of the type I enzyme. Putting this aside, the genetic experiments in both mice and Drosophila have some limitations in interpretation. It is possible, for example, that the CaM-sensitive adenylyl cyclase plays a permissive role in the process rather than a regulatory role. In such a model, the CaM-sensitive adenylyl cyclase would be necessary for neuroplasticity by allowing some important process to occur but would perhaps not play a direct role in regulating the response. With either a direct or permissive role, a loss of function mutation in the CaM-sensitive adenylyl cyclase will block LTP. Ultimately, we will need to set up some sort of mammalian model along the lines of the Aplysia system in which we can manipulate the production of cAMP and measure the effects on neuronal activity. Also, because the mechanisms involved in BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 477 Commentary /Controversies in Neuroscience III LTP probably can be better represented by a web of signals rather than as a pathway, we will need a thorough understanding of the messengers produced in response to the external stimuli, as well as the downstream events, including the genes that are transcribed and are set in motion by these signals. Channel structure and divalent cation regulation of phototransduction Richard L. Hurwitz,3 Devesh Srivastava," and Mary Y. Hurwitza "Departments of Pediatrics and Cell Biology, Baylor College of Medicine, Houston, TX 77030 and bCollege of Optometry, University of Houston, Houston, TX 77204. rhurwltz@msmail.his.tch.tmc.edu Abstract: The identification of additional subunits of the cGMP-gated cation channel suggests exciting questions about their regulatory roles and about structure/functional relationships. How do the different subunits interact? How is the complex assembled into the plasma membrane? Divalent cations have been implicated in the regulation of adaptation. One often overlooked cation is magnesium. Could this ion play a role in phototransduction? & ARSHAVSKY; MOLDAY & HSU] An understanding of the photoreceptor response to light requires an insight into the structure/function of the individual members of the phototransduction cascade and an understanding of the interactions between them and the regulatory mechanisms that modulate their response. The importance of both calcium and cGMP in visual signal transduction is recognized in the target articles by [BOWNDS MOLDAY & HSU a n d BOWNDS & ARSHAVSKY. MOLDAY & HSU have detailed a thorough view of exciting recent discoveries concerning the structure and function of the cGMP-gated cation channel which regulates the lightdependent intracellular concentration of cations in the photoreceptor. As in all good investigations, these discoveries point to many more questions that need to be answered in order to understand the relationships of the various channel subunits (oc and P) to each other and to the regulation of phototransduction. Chen et al. (1993) have demonstrated the existence of the putative P subunit of the cGMP-gated channel which, when coexpressed with the a subunit in a mammalian cell line, imparts diltiazem sensitivity to the cation flux. Molday and his colleagues have found that the deduced amino acid sequence of this P subunit is identical to sequenced polypeptides derived from the 240 kDa polypeptide (sect. 2.4, para. 1) that copurifies with the a subunit. The p subunit cDNA, however, codes for a protein that is less than half the size of the 240 kDa polypeptide. Unraveling the mystery of the 240 kDa polypeptide may provide substantial insight into the structure of the channel complex. In our hands, the purified, reconstituted channel was inhibited by 1 cis diltiazem and this inhibition could be eliminated by preincubation with trypsin (Hurwitz & Holcombe 1991). Cook et al. (1987) have also purified the channel, however their preparation did not appear to be 1 cis diltiazem sensitive. Both of these channel preparations contain the 240 kDa polypeptide and therefore would be predicted to be affected by diltiazem. Definition of the requirements necessary for effective interaction between the a and P subunits should provide insight into the function of the channel complex in vivo. Localization of the channel exclusively in the plasma membrane (Cook et al. 1989) raises another fascinating question: How is the channel segregated into the plasma membrane while rhodopsin is directed to both the plasma and disk membranes? The mechanism of this specific localization may provide insight into membrane protein trafficking for other macromolecules as well. The observation that the amino terminus of the channel appears to be missing when the channel is isolated from photoreceptors but not when the channel is isolated from in vivo 478 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 expression systems using oocytes or mammalian cultured cell lines (Molday et al. 1991) may suggest a functional role for this apparent posttranslational modification. Molday and his colleagues provide suggestive evidence that this cleavage may occur in vivo and is not an artifact of the channel purification. Both the mechanism of this proteolysis and any functional differences that may result should provide additional insight into the functional channel complex. BOWNDS & ARSHAVSKY provide a thorough discussion and summary of the diverse work that addresses the topic of photoreceptor adaptation. The authors correctly point out that experiments have been performed using various rod outer segment preparations, stimulus conditions, and, at times, nonphysiological assay conditions (sect. 3, para. 2), which make interpretation of the literature as well as arriving at basic conclusions difficult. The authors associate light-induced changes in intracellular calcium concentrations with the kinetics of the adaptive process, but they acknowledge that calcium may not be the sole regulator of the adaptive process (sect. 9, para. 6). Another cation that may regulate the kinetics of both excitation and recovery is magnesium. Magnesium ions comprise 5% of the current that flows through the cGMP-gated cation channel (Yau & Baylor 1989). The manner by which magnesium exits the photoreceptor is not known. Somlyo and Walz (1985) have used electron probe analysis to demonstrate that light decreased the total concentration of magnesium in both the inner and outer segments of the frog photoreceptor. The free concentration of intracellular magnesium ([Mg2+],) and the effect of light on [Mg2+], are unknown. In other tissues, [Mg2+]j has consistently been ~0.5 mM (Chen et al. 1992; Heinonen & Akerman 1987; Quamme & Rabkin 1990; Raju et al. 1989). What impact would potential changes in [Mg2+], have on mechanisms that contribute to adaptation? Bownds and Arshavsky identified major sites of adaptation (sect. 10, para. 2). These sites affect cGMP hydrolysis and synthesis and the phosphorylation of rhodopsin. These reactions all require Mg2+ as a cofactor. In our laboratory we demonstrated an almost 2-fold decrease in the Vmax of the isolated, trypsin-activated bovine rod phosphodiesterase when [Mg2+] was decreasedfrom500 to 100 \x.M (Srivastava et al. 1994). The Km of the enzyme also decreased as [Mg2+] decreased. We further demonstrated that Mg2+ and cGMP bind to the phosphodiesterase in a random order prior to the hydrolysis of cGMP. The substrate for guanylate cyclase, rhodopsin kinase, and protein kinase C is a Mg-NTP complex. A decrease in [Mg2+]( would result in a decrease in available substrate. A decrease in available substrate would result in a decrease in the rate of cyclic nucleotide production or protein phosphorylation by cyclases of kinases, respectively. Therefore, a light-induced decrease in [Mg2+], may decrease the kinetics of excitation or recovery. With the acceptance of the essential nature of Mg2+, its contribution to the current through the cGMP-gated channel, and its effect on the kinetics of numerous reactions, perhaps this cation deserves more attention. First, future experiments will need to determine whether there are light-dependent changes in the free [Mg2+],. Second, these potential changes need to be related to each step of excitation and recovery. Third, the mechanism and regulation of the extrusion of Mg2+ to the extracellular space needs to be identified. Once these are better understood, perhaps our knowledge of the mechanisms of photoreceptor adaptation will be more complete. Linking genotypes with phenotypes in human retinal degenerations: Implications for future research and treatment Michael W. Kaplan fl. S. Dow Neurological Sciences Institute, Portland, OR 97209-1595. kaplanm@ohsu.edu Commentary /Controversies in Neuroscience HI Abstract: Although undoubtedly it will be incomplete by the time it is published, the target article by DAJCER ET AL. organizes mutations in genes that produce retinal degenerations in humans into categories of clinically relevant phenotypes. Such classifications should help us understand the link between altered photoreceptor cell proteins and subsequent cell death, and they may yield insight into methods for preventing consequent blindness. [DAICER ET AL. ] The application of genetic analysis and molecular biology to determine mutations responsible for retinitis pigmentosa in humans has been both informative and surprising. Although variability in the inheritability, course, and severity of the disease suggest multiple etiologies, the number of underlying genetic defects that have been discovered so far is both daunting and bewildering. Studies of specific pedigrees provide a rapidly expanding genetic taxonomy of the disease that will be important for deciphering how the underlying mutations lead to retinal cell death. The review by DAICER ET AL. organizes the individual mutations into phenotypic families, and provides a useful framework for building general principles of how the consequent disease progresses. Grouping the mutations into functional and clinical phenotypes should also facilitate the development of any strategies for treatment and remediation. Faced with the multiplicity of genetic defects, are there reasonable strategies that can be adopted for developing therapies for inherited retinal degenerations? The answer to this question will depend upon whether information about specific mutations can be translated into an understanding of how the resulting aberrant protein produces cell death and whether therapeutic intervention is possible. Further understanding will be facilitated by appropriate animal models, cell cultures, and cell-free systems. The correspondence of naturally occurring retinal degenerations in mice (rd, rd-3, rds, nr, pcd) (Chang et al. 1993; Mullen et al. 1976; Sidman & Green; 1965; 1970; Van Nie et al. 1978), dogs (Aguirre et al. 1978), cats (Narfstrom 1983), and other species to retinitis pigmentosa in humans is not exact. The genotypes usually differ from human mutations, and there may be species-dependent effects of altering given proteins upon photoreceptor cell development and ultimately upon photoreceptor cell death. The value of natural mutations of photoreceptor-specific murine and canine genes such as the genes for PDEp (rd) (Bowes et al. 1990; Farber et al. 1992; Pittler & Baehr 1991; Suber et al. 1992) and rds/peripherin (rds) (Council et al. 1991; Travis et al. 1989; 1991) has been to show that their abnormal gene products can cause photoreceptor cell death. They have focused the search for mutations in the equivalent human genes. The outcome has been an extensive catalog of mutations in human PDEB and r<is/peripherin and a depressingly varied list of the categories of retinitis pigmentosa that they cause. An approach to understanding the consequences of given mutation identified in humans has been to develop a corresponding transgenic animal model incorporating the same mutation. Transgenic mice potentially can answer many important questions about the physiological consequences of a particular mutation (Huang etal. 1993; Naashetal. 1993; Roof etal. 1994). For example: How does the mutation affect the function of its gene product? Does it affect the transport and localization to appropriate sites within the cell? Does it affect the way that its gene product interacts with other proteins? Why do different mutations of the same gene sometimes have profoundly different consequences in terms of the course and severity of the retinal degeneration? How and why does the mutation cause photoreceptor cells to die? Why does a mutation in a rod-specific gene ultimately cause the death of cone cells? Why do mutations in some proteins of the visual transduction cascade cause autosomal dominant disease (opsin) while mutations in other transduction proteins (PDEB) cause autosomal recessive disease? Given the extent of the ever expanding list of mutations that cause retinal degeneration in humans, it is unlikely that equivalent transgenic animal models will be developed for them all. A more reasonable expectation is that categories of disease phenotypes, such as those presented by Daiger et al., will be used to select for critical analysis those mutations that represent a particular category or functional domain or retinal degeneration. Unsolved issues in S-modulin/recoverin study Satoru Kawamura Department of Physiology, Keio University School of Medicine, Shinanomachi 35, Shinjuku-ku, Tokyo 160, Japan Abstract: S-Modulin is a frog homolog of recoverin. The function and the underlying mechanism of the action of these proteins are now understood in general. However, there remain some unsolved issues including; two distinct effects of S-modulin; Ca 2+ -dependent binding of S-modulin to membranes and a possible target protein; S-modulin-like proteins in other neurons. These issues are considered in this commentary. [HURLEY] S-modulin in frog and recoverin in bovine were reported almost at the same time in 1991. Even though their reported molecular weights and molecular characteristics were very similar, the reported functions were different: S-modulin was shown to inhibit phosphodiesterase activation at high [Ca 2+ ] and recoverin to activate guanylate cyclase at low [Ca 2+ ]. There had been a dispute about the function(s) of these proteins. However, as reviewed in HURLEY'S target article, it now appears that recoverin is a bovine homolog of S-modulin and does not activate the cyclase but instead regulates PDE (phosphodiesterase) activation through inhibition of rhodopsin phosphorylation. In section 2.3 of HURLEY'S target article, studies of S-modulin are briefly reviewed, but there remain some unsolved issues that are not addressed in the target article. I will elaborate on these issues to contribute to future studies of this protein family (for details, see Kawamura 1994). Two distinct effects of S-modulln. S-modulin seems to consist of (at least) two components, each of which exerts a distinct effect. This notion came from the observation that S-modulin purified with rod outer segment (ROS) membranes only affects the fractional activation of the cGMP phosphodiesterase (PDE) (Kawamura 1992; Kawamura & Murakami 1991) while S-modulin purified with a phenyl Sepharose column affects not only the fractional activation of PDE but also the inactivation of PDE (Kawamura 1993). In both purifications, S-modulin was obtained as a single band on an SDS-PAGE gel. From these results, there seem to be two populations of S-modulin whose hydrophobic characteristics are different. Unfortunately, no attempts have been made so far to isolate these components. As reviewed in HURLEY'S target article, one of the S-modulin effects, namely that on PDE inactivation, is attained by suppressing rhodopsin phosphorylation. The mechanism of the other effect, the effect on the fractional activation, is not known. However, the fractional activation also seems to be attained by regulation of phosphorylation by S-modulin, since without ATP or S-modulin, this effect was not seen (Kawamura 1993). It is interesting to note that ATP has two distinct effects on PDE; one was observed on the fractional activation of PDE with a half effect at 50 n-M ATP and the other on the PDE inactivation with a half effect at 2 p,M ATP (Kawamura 1983). Recent electrophysiological work has revealed that PDE fractional activation is influenced by both Ca 2 + and a soluble Ca 2+ -binding protein; this effect is distinct from the regulation of PDE inactivation (Lagnado & Baylor 1994). This observation supports the possibility mentioned above. Ca2+-dependent binding of S-modulln to membranes. S-mod- ulin (Kawamura et al. 1992) and recoverin (Dizhoor et al. 1993; BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 479 Commentary /Controversies in Neuroscience III Zozulya & Stryer 1992) bind to ROS membranes in a Ca 2 + dependent manner. It has been thought that this binding is essential for S-modulin function. However, the [Ca 2+ ] required for the half-effect of the membrane-binding of S-modulin (10 |j.M; Kawamura et al. 1992) is almost 100 times higher than for half-inhibition of rhodopsin phosphorylation (=100 nM; Kawamura 1993). In addition to this, it has recently been found that N-myristoylation of recoverin is essential for the membrane-binding (Dizhoor et al. 1993; Zozulya & Stryer 1992) but not for the inhibition of rhodopsin phosphorylation (Kawamura et al. 1994). These two observations raised the possibility that membrane-binding at \x.M range of [Ca 2 + ], and therefore N-myristoylation, is not essential for the function of S-modulin. S-modulin is washed out from a truncated ROS at 1 u.M Ca 2 + (Kawamura & Murakami 1991). Thus, the binding affinity to the membrane is low in intact cell in agreement with the binding experiment mentioned above. To elucidate the role ofthe membranebinding and thus N-myristoylation, further work is required. Target molecule of S-modulin. Even though S-modulin exerts its effect by inhibiting rhodopsin phosphorylation, the target molecule has not been clearly identified. Rhodopsin kinase and photolyzed rhodopsin are the potential candidates for the target molecule (Kawamura 1993). Actually, it has been claimed that recoverin binds to the rhodopsin kinase in a Ca 2+ -dependent manner (Gorodovikova & Philippov 1993). However, the [Ca 2+ ] used in this study was very high (100 u,M). At this very high calcium concentration, physiologically less important membranebinding of recoverin takes place. Therefore the observed binding may not be physiologically relevant; we wish to see the binding at physiological (<(j,M) [Ca 2 + ]. In the original report of the finding of recoverin (Dizhoor et al. 1991), this protein was isolated as a binding protein of rhodopsin. It remains to be determined which of the molecules, rhodopsin kinase or photolyzed rhodopsin, is the real target. At [Ca 2+ ] over 1 p.M, S-modulin/recoverin has been reported to bind to ROS membranes or phospholipids (Dizhoor et al. 1993; Kawamura et al. 1992; Zozulya & Stryer 1992), which raises the possibility that S-modulin/recoverin first binds to the membrane and then interacts with rhodopsin kinase or photolyzed rhodopsin. However, the membrane-binding is not essential for the inhibition of rhodopsin phosphorylation (Kawamura et al. 1994), and thus this possibility seems to be slight. S-modulln homologs In other tissues. As reviewed in section 3 of HURLEY'S target article, many S-modulin/recoverin homologs have been found in nonretinal tissue. These homologs can be classified into three groups (Nef et al., submitted.) We picked up S-modulin and recoverin from group 1, NCS-1 from group 2, and Vilip 1 and hippocalcin from group 3, and examined the effects of these proteins on rhodopsin phosphorylation. All of the proteins inhibited rhodopsin phosphorylation at 10 u,M Ca 2 + (Nef et al., submitted; Kawamura & Takamatsu, unpublished observation cited in Kawamura 1994). The site (or sites) that interact with rhodopsin or its kinase must be conserved in this protein family. Even though the real functions of these proteins may not be the same as in photoreceptors, they could possibly regulate phosphorylation reactions in a Ca 2 + dependent manner in the host-cells. Crucial steps in photoreceptor adaptation: Regulation of phosphodiesterase and guanylate cyclase activities and Ca 2+ -buffering Karl-Wilhelm Koch Institut fur Biologische Informationsverarbeitung, Forschungszentrum JQIich, Postfach 1913, 0-52425 Julich, Germany. Inboo8ez.am00l.zam.kfa-juelich.de 480 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 Abstract: This commentary discusses the balance of phosphodiesterase and guanylate cyclase activities in vertebrate photoreceptors at moderate light intensities. The rate of cGMP hydrolysis and synthesis seem to equal each other. Ca2+ as regulator of both enzyme activities is also effectively buffered in photoreceptor cells by cytoplasmic buffer components. [BOWNDS & ARSHAVSKY; HURLEY] Since the light-triggered cGMP cascade of visual transduction has been investigated in more detail, a comprehensive description of its amplification and the kinetics of its activation steps has become possible (Pugh & Lamb 1993). Recently, interest has also been focused on other aspects of phototransduction, for example, the inactivation steps that restore the dark state of a photoreceptor cell and the adaptational processes of which a significant part takes place in the photoreceptor itself. A rich complexity of biochemical control functions has emerged in the last 3-4 years following the initial observation from electrophysiological recordings that an important signal of light adaptation is the change in intracellular Ca 2 + (Matthews et al. 1988; Nakatani & Yau 1988). Bownds and Arshavsky in this volume discuss many steps on phototransduction as putative points in the modulation of photoreceptor recovery and light adaptation, emphasizing that the most important site of light adaptation cannot be determined at present (sect. 4). This is certainly true; it will become necessary in the future to work out the quantitative contribution of the different control steps and feedback loops (involving phosphodiesterase, guanylate cyclase, the cGMP-gated channel, and other unidentified proteins). I will comment mainly on two points related to the target articles of BOWNDS & ARSHAVSKY and HURLEY: (1) What can be concluded about the balance between guanylate cyclase (GC) and phosphodiesterase (PDE) activity after light stimulation or "is cyclase activation sufficient to contribute to adaptation during background light?" (BOWNDS & ARSHAVSKY para. 10). (2) What is known about the functional role of intracellular Ca2+-bufFers in photoreceptors? 1. Balance of PDE and GC activities. The discrepancy between the kcat values of activated GC (GC*) at low free Ca 2 + and activated PDE (PDE*) after light stimulation is obvious (kcat of GC = 10 sec" 1 ; kcat of PDE* = 2000 sec" 1 ) and would point to an effective shutdown of PDE* activity under control of the Ca 2+ -feedback. However, considering the cellular situation, the rate of cGMP hydrolysis should be compared with the rate of cGMP synthesis. According to Pugh and Lamb (1993) the rate of cGMP hydrolysis is (equation 11 in their paper): dcG/dt = -PDE* (t) [k^, (Km Nnv Vcyl BP "']cG BP is the buffering power of the cytoplasm for cGMP. Nnv is the Avogadro number and Vcyt is the cytoplasmic volume of a rod. The free substrate concentration is cG. A similar equation should be valid for the synthesis of cGMP by the GC. The rate of transforming GTP to cGMP is: -dGTP/dt = dcG/dt = GC*(t) [keal (Km Nov Vcyl )"']CTP A half-saturating response activates about 180 rhodopsin molecules (Rh*) per sec in bovine photoreceptors (Nakatani et al. 1991). With an activation of 2500 transducin molecules (G*) per Rh* and per sec at 30°C (Kahlert & Hofmann 1991), we have 4.5 x 105 PDE*. With the numbers for a mammalian rod provided by Pugh and Lamb (1993): 2 kcat = 4000 sec-1; Km = 100 |xM; Vcyt cG = 4 (xM 40fl;BP = 2; the rate of cGMP hydrolysis by PDE* is dcG/dt = 720 (J.M sec"1. Assuming that at half-saturating responses the cytoplasmic Ca 2 + drops to a value below 0.1 (iM in one second by operation of the powerful Na:Ca,K-exchanger (Lagnado et al. 1992) all GC molecules become activated by the Ca z+ -feedback. The GC to Commentary/Controversies in Neuroscience III rhodopsin ratio in a bovine photoreceptor was determined to be 1:100 yielding about 1.5 x 106 GC* (Koch 1991). Free GTP is about 1 mM and the Km of GC with MG 2 + as cofactor is 1 mM (Koch et al. 1990). Abstract: Light adaptation is modulated almost exclusively by changes in intracellular Ca 2 + concentration, and other Ca 2+ -independent mechanisms are likely to play only a minor role. Changes in Ca 2+ , may be not only necessary for light adaptation to take place but sufficient to cause it. the rate of cCMP synthesis by GC* is dcG/dt = 600 (xM sec-1. In their target article BOWNDS & ARSHAVSKY say that calcium is unlikely to be the sole regulator ofadaptation in vertebrate photoreceptors. Wewouldargue(withKoutalosetal. 1994) that adaptation can be accounted for by reactions that are modulated by changes in intracellular Ca 2 + concentration (Ca2+j). We believe recent results indicate that changes in Ca 2+ , are not only necessary for light adaptation to take place but are sufficient to cause it. It is well known that adaptation to steady light leads to graded acceleration of the recovery of the response to a bright flash. This acceleration can be abolished if the fall in Ca 2+ j induced by light is prevented by superfusing the outer segment with low Ca 2 + /0 Na + solution, which simultaneously minimizes Ca 2 + influx and efflux (Matthews et al. 1988; Nakatani & Yau 1988). The falling phase of the flash response is unaffected by prior exposure to steady light if the light is presented in low Ca 2 + /0 Na + solution (Fain 1989). Together with other observations, this result indicates that the light-induced fall in Ca 2 + , is necessary for adaptation to take place (Fain & Matthews 1990). When low Ca 2 + /0 Na + solution is used to hold the Ca 2 + in the cell near its dark level and steady light is presented, the circulating current can be completely suppressed by quite a dim background light (Fain et al. 1989). Partial inhibition of the PDE (phosphodiesterase) with IBMX can then be used to restore the dark current to its normal level despite the continued presence of the background. Under these circumstances, the response to a dim flash exhibits dark-adapted kinetics, instead of the more rapid kinetics normally seen in the presence of the background when Ca 2 + , is allowed to fall to the appropriate light-adapted level (Matthews 1992b; 1995). This observation shows that adaptational changes in the kinetics of the dim flash response require a fall in Ca 2 + i ( which cannot be mimicked by light alone. However, it should be noted that this approach is intended to restore the cyclic G M P concentration to around its normal value in darkness, and so cannot directly rule out a feedback role for cyclic GMP. If the photoreceptor is allowed to adapt to steady light in Ringer and then exposed to low Ca 2 + /0 Na + solution, the Ca2"1", can be held near the value appropriate for the steady light intensity. If the background is then extinguished, the response to a subsequently-presented bright flash is still accelerated to the same degree as when the steady light was present, even though in Ringer the same period of darkness would have allowed a substantial degree of recovery of response kinetics (Matthews 1995). Comparable acceleration of bright flash response recovery can be seen when Ca 2 + | is artificially reduced in darkness by removing Ca 2+ O (Matthews 1992a; 1995). These results strongly suggest that lowered Ca 2+ j is sufficient to cause adaptation even in the absence of light itself. These observations place severe constraints on the possible significance of any adaptational mechanism operating other than through changes in Ca 2+ ,. However, Ca 2+ ( can only be reliably controlled by supervision with low Ca 2 + /0 Na + solution for 1015 s (Fain et al. 1989). Thus these results do not directly exclude the possibility that some other calcium-independent mechanism might operate on a longer time scale. However, as the great majority of adaptation takes place within 10-15 s in Ringer, the quantitative significance of any calciumindependent mechanism would seem likely to be rather limited. Apparently, both enzyme activities almost match each other at moderate light intensities. A decrease of free cGMP would only affect PDE* since CC activity is independent ofcGMP(but GC is inhibited by the other product pyrophosphate; see for example Hakki & Sitaramayya 1990; Yang & Wensel 1992). Thus, regulation of PDE* by substrate level seems to be significant. A similar conclusion is drawn by Miller and Korenbrot (1993) based on their observations on cone photoreceptors of striped bass. 2. Recoverin as Ca2*-buffer. Ca 2 + as modulator of phototransduction exerts its effects mainly via Ca 2+ -binding proteins (see recent review, Koch 1994). Another crucial aspect of Ca 2 + and its role in recovery and light adaptation is the Ca 2 + buffering in an intact photoreceptor cell. Ca 2+ -binding proteins are good candidates to serve this function, HURLEY discusses properties of recoverin, a prototype of a new class of neuronal Ca 2+ -binding proteins originally isolated from photoreceptor preparations. There is presently good evidence that recoverin or the amphibian homolog S-modulin prolongs the lifetime of PDE*. I will add a few remarks on a possible function of recoverin as a Ca2+-buffer. Lagnado et al. (1992) have demonstrated the existence of two intracellular Ca2+-buffers in tiger salamander rods. One buffer is present in high amounts and has a low affinity for Ca 2 + , the other buffer is present in smaller amounts, but has a higher affinity. Particularly the high-affinity buffer has important functions: (a) It determines the rate of relaxation in intracellular Ca 2 + after closure of the lightsensitive channel. Thus the onset of Ca 2+ -dependent light adaptation would be controlled by this buffer (Lagnado et al. 1992). (b) It would continuously supply Ca 2 + from its binding sites and therefore be crucial for the operation of the Na:Ca,Kexchanger. (c) It would participate in any new steady state of Ca 2+ -homeostasis at different background light intensities. Lagnado et al. (1992) have derived some molecular properties of the high-affinity buffer from their electrophysiological recordings. The buffer is a diffusible macromolecule with a buffering capacity of 35 |i,M in aequorin-loaded rods or 241 u,M in intact rods. The apparent Michaelis constant Kbllfrwas £ 0.7 u,M. DO the properties of recoverin fulfill these requirements? Recoverin can be isolated in high amounts from rod outer segments at least in a ratio of 1:100 to rhodopsin corresponding to a cellular concentration of 30 u,M (reviewed in Koch 1994). Although KD values for Ca 2+ -binding have not been documented, the threedimensional structure of recombinant unrnyristoylated recoverin gave first insight into its Ca 2+ -binding properties (Flaherty et al. 1993). Recoverin has two canonical EF-hands that probably bind Ca 2 + with KD-values of 0.15 u.M and 2 u.M. Therefore, due to its relative abundance and high-affinity binding of Ca 2 + , recoverin could function as a prominent Ca2+-buffer in photoreceptor cells. For a full description of Ca 2+ -homeostasis it will be important to determine the intracellular concentrations of photoreceptor Ca 2+ -binding proteins and their corresponding KD-values. Reduced cytoplasmic calcium concentration may be both necessary and sufficient for photoreceptor light adaptation H. R. Matthews and G. L. Fain Physiological Laboratory, University of Cambridge, Downing Street, Cambridge CB2 3EG, England and Department of Physiological Science, University of California at Los Angeles, Los Angeles, CA 90025-1527. hrml@phx.cam.ac.uk; gordonf@physcl.lifesci.ucla.edu Gene therapy, regulatory mechanisms, and protein function in vision James F. McGinnis Department of Anatomy and Cell Biology, University of California, Los Angeles, Los Angeles, CA 90024. ianw|fm@mvs.oac.ucla.edu BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 481 Co7nmentari//Controversies in Neuroscience III Abstract: Hereditary retinal degeneration due to mutations in visual genes may be amenable to therapeutic interventions that modulate, either positively or negatively, the amount of protein product. Some of the proteins involved in phototransduction are rapidly moved by a lightdependent mechanism between the inner segment and the outer segment in rod photoreceptor cells, and this phenomenon is important in phototransduction. [DAICER ET AL.; BOWNDS & ABSHAVSKY; HURLEY] T h e target article by DAIGER ET AL. has done an excellent job of documenting, organizing, and correlating specific mutations, especially in rhodopsin, with hereditary blindness. Although it is still not yet possible to predict the severity of the disease based on precise mutations in rhodopsin, the amount of knowledge accumulated enables one to begin to envision therapeutic methods that can be devised to prevent the onset or to stop the progression of the degeneration of photoreceptor cells. In those cases where the accumulation of defective rhodopsin molecules is correlated with blindness, a logical strategy would be to inhibit the production of such molecules. This could be accomplished through the development of "antisense" therapies using either antisense oligonucleotides and/or vectors that express antisense molecules capable of distinguishing between the "good" mRNA and the "bad" mRNA. Similarly, studies that identify and characterize cts-acting and trans-acting regulatory regions may provide alternative approaches to suppress the synthesis of a specific mRNA. In those cases of blindness due to the absence of a specific function, the development of appropriate vectors that can target the expression of the correct gene to the photoreceptor cells offers a real possibility to preserve vision by preventing the death of the cells. A "good" gene product, even at half the concentration found in the normal individual, should be sufficient to stop the progression of the disease. Cell culture techniques for growing and maintaining retina tissues and cells in vitro will provide the opportunity to develop and test the efficacy of such approaches. These in vitro techniques may also enable small numbers of photoreceptor cells to be removed from an individual, be maintained and "modified" in vitro, and then transplanted to the same or different person to provide a source of hormone, nutrient, growth factor, or drug for nearby cells or as functional replacements for defective cells. The compilation and analysis of mutations in photoreceptor cell-specific genes such as presented in this article are necessary for progressing to the next level of research on retinal degeneration; the preservation and/or restoration of vision in patients with retinitis pigmentosa or other hereditary eye diseases. The target article by BOWNDS & ARSHAVSKY does not include a discussion of the contributions and/or effects of the lightdependent movement of some of the proteins involved in phototransduction. We have shown, at the light microscopic level, that the intracellular localization of transducin and arrestin in mouse photoreceptors is transient and dependent on the lighting environment to which the retina is exposed (McGinnis et al. 1992b; Whelan & McGinnis 1988). Similar data have been obtained in rats and frogs by other laboratories (Brann & Cohen 1987; Mangini & Pepperberg 1988; Philip etal. 1987). In the dark when the rods are most active, the inhibitory protein, arrestin, is in the inner segments while the activating protein, transducin, is in the outer segments. When the lights are turned on, or with sunrise, the positions of these proteins rapidly and simultaneously reserve with the majority of arrestin moving to the outer segment and transducin moving to the inner segment. This movement has already begun at the earliest time point tested (30 seconds) and is readily reversed by changing the lighting environment (McGinnis et al. 1992b). The outer segments from lightadapted mice were shown to contain at least three times as much arrestin as the outer segmentsfromdark adapted mice (Whelan & McGinnis 1988), supporting the conclusion that the proteins are actually moving between the inner and outer segments. Temporal and spatial changes in the cellular and subcellular concentrations of photoreceptor cell gene products appear to be important 482 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 features of phototransduction in rod photoreceptor cells. The positional shifts of arrestin and transducin result in dramatic changes in the subcellular concentrations of these photoreceptor specific proteins and, therefore, dramatic changes in the rates of the reactions catalyzed by these proteins. We think that these light-dependent translocations are functionally related to the mechanism of phototransduction and photoreceptor adaptation and that they may be incorporated into any model that seeks to explain the molecular basis of these processes. I would like to add to the information presented in the target article by HURLEY. Using a novel immunological approach, we (McGinnis & Leveille 1985) identified four proteins as being photoreceptor cell-specific. One of these, 23kDa, was subsequently cloned (McGinnis et al. 1992a) from our mouse and bovine retina cDNA expression libraries in lambda gtll and shown to be identical to the amino acid sequence published for the protein cloned and named recoverin (Dizhoor et al. 1991) and the cancer-associated retinopathy protein (Polans et al. 1991; Thirkill et al. 1992) or S-modulin (Kawamura & Murakami 1991). We subsequently mapped the mouse 23kDa (recoverin) gene to chromosome 11 adjacent to the tumor suppressor gene trp53 (McGinnis et al. 1993) and the human recoverin gene to 17pl3.1, adjacent to the human tumor suppressor gene p53 (McGinnis et al. 1994). Recoverin was shown to be developmentally regulated in the mouse and to be present in the retinas of a variety of species (Stepanik et al. 1993). We cloned frog recoverin, showed it to be identical to frog S-modulin and demonstrated, using coupled in vitro transcription and translation of a rhodopsin kinase cDNA, that rhodopsin kinase specifically binds to mouse and frog recoverin (Subbaraya et al. 1994). This supports our suggestion that the function of recoverin is to inhibit the phosphorylation of rhodopsin that would prevent its interaction with arrestin and thereby prolong the light state. Structure of the cGMP-gated channel Daniel D. Oprian Graduate Department of Biochemistry and Volen Center for Complex Systems, Brandeis University, Waltham, MA 02254 Abstract: The subunit structure of the cGMP-gated cation channel of rod photoreceptors is rapidly being defined, and in the process the mode of regulation by Ca 2+ -calmodulin unraveled. Intriguingly, early results suggest that additional subunits of unknown function are associated with the channel and remain to be identified. [MOLDAY & HSU] The target article by MOLDAY & HSU provides a timely review of the recent history of the cGMP-gated channel of rod photoreceptor cells. Thisfieldhas advanced by leaps and bounds since the seminal observation by Fesenko et al. (1985) that the cation conductance of rod cell plasma membranes was controlled by cGMP. The identity of the channel, which formed the foundation of a lively controversy just a few years ago, is now resolved, at least to the extent that the a-subunit cloned by Kaupp et al. (1989) forms the cGMP-gated pore. However, the complete formation of the channel with all of its pharmacological and electrophysiological nuance requires in addition a (5-subunit isolated by Chen et al. (1993) and perhaps other as yet unidentified polypeptides. It is here that Molday and Hsu bring us to the current issues in the structure of the channel. Purification of the channelfrombovine retina by immunoaffinity chromatography using a monoclonal antibody raised to the osubunit results in two proteins identifiable by SDS-PAGE: a63 kDa subunit corresponding to a proteolytically processed form of the a-subunit in which the amino-terminal 92 amino acids have been removed, and a much larger protein migrating with an apparent mobility of240 kDa. The identity of the 240 kDa protein remained a mystery until very recently. Hsu and Molday (1993) showed that the affinity of the channel for cGMP is regulated by Ca2+-calmodulin, Commentary/Controversies in Neuroscience III and that the channel can be purified on a calmodulin affinity column. The channel purified in this manner contains the same two proteins that were purified by immunoaffinity chromatography and blots of SDS gels demonstrated that calmodulin binds to the 240 kDa protein. Very recently, Molday and Yau and coworkers (Chen et al. 1994) have shown that antibodies raised to the P-subunit bind to the 240 kDa protein and that the amino acid sequence of peptides from this protein match closely the sequence of the cloned human P-subunit. Furthermore, electrophysiological measurements of the channel expressed in HEK 293 cells after connection of mRNA for both the a- and p-subunits demonstrates that the P-subunit confers on the channel sensitivity to Ca 2+ -calmodulin. Prior to these studies, the p-subunit was known to be inactive on its own, but when coexpressed with the ot-subunit gives the channel the flickering response and sensitivity to l-cis diltiazem long associated with the intact channel from rod outer segment plasma membranes. The more recent results clearly demonstrate that the P-subunit is an integral component of the 240 kDa protein and that the regulatory properties of Ca2+-calmodulin stem from an interaction with the P-subunit. However, major questions remain. For example, what other polypeptides are components of the 240 kDa protein and what is the subunit composition and structure of this complex. It is clear that there are other polypeptides. To begin, the calculated molecular weight of the cloned p-subunit is 102 kDa not 240 kDa. Also, the P-subunit polypeptide isolated from transfected HEK 293 cells migrates on SDS gels with a mobility of 130-150 kDa, again distinct from 240 kDa. Finally, some of the peptides isolated and sequenced from the 240 kDa complex are clearly different from the cloned P-subunit. It will be very interesting to discover what additional polypeptides are components of the 240 kDa complex and why this assembly survives dissociation on SDS gels. Answers to these questions will provide much needed information for resolution of the channel's structure. Recoverin is the tumor antigen in cancerassociated retinopathy new information is now available about the role of recoverin in the development and progression of the CAR syndrome. Our laboratories recently identified recoverin in the tumor of a CAR patient diagnosed with small cell carcinoma of the lung. We performed a Western blot analysis using biopsy material and the patient's own serum. A prominent band at 23 kDa was reactive along with a few nonspecific bands that also reacted with normal human sera. Sera from other CAR patients gave the identical staining profile. Since human recoverin migrates at 23 kDa, we used a series of anti-peptide antibodies generated against different regions of the recoverin sequence (Polans et al. 1993), and the same 23 kDa protein was stained with each of the antibodies. Our experiments, therefore, demonstrate that recoverin is expressed in the tumors of CAR patients. We have analyzed similar tumors from individuals who do not have the associated retinopathy, and recoverin cannot be detected in those tumors. Therefore, it appears likely that the disease originates as the result of the aberrant expression of recoverin in a subset of tumors. Recoverin is photoreceptor-specific and normally expressed in the eye, which is an "immunologically privileged" site. Under normal conditions, potential antigens like recoverin are unavailable for immunological recognition at the inductive stage of the immune response due to sequestration within photoreceptors. Owing to the expression by the tumor, however, the released recoverin can trigger the autoimmune response, which in the patient results in both autoantibodies and specific T-cells. The immunological response then correlates with the degeneration of the target cells, photoreceptors. If the disease is autoimmune mediated then a potential autoantigen (recoverin) should induce immunological mechanisms that lead to the degeneration of photoreceptors. Two recent studies have demonstrated this finding (Adamus et al. 1994; Gery et al. 1994). After inoculation of Lewis rats with a single dose of recoverin, a specific immunological response occurred. Antibodies to recoverin could be detected seven days after the initial immunization, and the peak of the antibody response occurred around day 28. The response was specific for recoverin; no antibodies could be detected using arrestin (Santigen), IRBP or rhodopsin in specific ELISAs (Fig. 1). Impor- Arthur S. Polans and Grazyna Adamus R S. Dow Neurological Sciences Institute, Legacy-Good Samaritan Hospital and Medical Center, Portland, OR 97209 Abstract: Considerable progress has been made toward understanding the involvement of recoverin in a cancer-associated retinopathy (CAR) that results in blindness. We describe the expression of recoverin in tumors of individuals afflicted with CAR, characterize the hnmunological response towards recoverin in these patients, and demonstrate how the disease can be induced in rodents using recoverin as an immunogen. [HURLEY] Recoverin was identified simultaneously in 1991 by four laboratories. Dizhoor et al. (1991) and Lambrecht and Koch (1991) characterized the protein in terms of its possible involvement in the calcium-dependent regulation of guanylyl cyclase. Kawamura and Murakami (1991) isolated the frog homologue of recoverin, S-modulin, and proposed its regulation of cyclic GMP phosphodiesterase. In studies unrelated to phototransduction, Polaris et al. (1991) identified recoverin as the autoantigen in a parancoplastic disease of the retina known as cancer-associated retinopathy (CAR). In this disorder individuals with cancer at a site other than the nervous system develop visual abnormalities and eventually blindness. Autoantibodies specific for recoverin were found only in the sera of CAR patients, and the presence of such antibodies suggested that this paraneoplastic disease may result from an autoimmune response based on antigenic mimicry. The occurrence of autoantibodies could be the result of selfantigen(s) released from the cancerous tissue followed by a response of the immune system to these antigens. In CAR, the presence of such antibodies correlates with the degeneration of rods and cones, resulting in loss of visual function. Considerable IT) O <* Q O PRE-IMM REC RHO IRBP SAG ANTIGENS Figure 1 (Polans & Adamus). ELISA of rat antiserum to various photoreceptor proteins. Dilutions of antiserum (1:1,000) obtained from a rat after immunization with 50 p.g recoverin were added to plates coated with 50 ng of the following retinal proteins: rhodopsin (RHO), interphotoreceptor retinoid binding protein (IRBP), S-antigen (S-AG), and recoverin (REC). Preimmune serum from the same animal also was tested. BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 483 Commentary/Controversies in Neuroscience III B Figure 2 (Polans & Adamus). Histopathological changes in the retinas of a human CAR patient (A) and Lewis rat 39 days after immunization with 50 |xg recoverin (B). There is limited inflammatory infiltrate detectable in both tissues, while the photoreceptor layers have completely degenerated (asterisks). C, choroid; I, inner nuclea layer; G, ganglion cell layer. Paraffin embedded sections stained with hematoxylin and eosin. tantly, there was no secondary response to other retinal antigens that might become available during photoreceptor degeneration. As measured by a lymphocyte proliferation assay, cellular responses towards recoverin also were detected and reached a peak 17 days after immunization. Furthermore, noT-cells were activated by S-antigen, IRBP or rhodopsin in comparable cell proliferation assays. During the same period as the cellular responses to recoverin, inflammation of the retina also was observed. By four weeks after the single injection of recoverin, photoreceptor cells had degenerated while the remainder of the inner retina and pigment epithelium appeared normal, thus paralleling the end stage of the CAR syndrome in humans (Figs. 2A and 2B). The combined data indicate that photoreceptor degeneration is not the result of an epiphenomenal event involving induction by other retinal antigens. Rather, development of a specific immunological response to recoverin underlies the degenerative event. Furthermore, the autoimmune etiology of CAR is supported by the finding of recoverin in the tumors of CAR patients, the presence of autoantibodies and T-cells to recoverin in those patients, and studies of animal models. It has yet to be determined how the immune response leads to photoreceptor degeneration, and specifically the role of autoantibodies. In previous experiments (Adamus et al. 1994), we demonstrated that the retinopathy can be produced in Lewis rats by passive transfer of T-cells obtained from animals immunized with recoverin, but the pathogenicity of the anti-recoverin antibodies still needs to be studied. In regards to other experiments, it is noteworthy that the gene encoding recoverin maps to human chromosome 17pl3.1, which also is the site of the tumor suppressor gene encoding p53. A low frequency mutation in p53 might explain both the development of the tumor as well as the aberrant expression of recoverin in a subset of tumors. This no doubt is a simplistic model, since the fine mapping between p53 and recoverin has not been performed, nor do we know whether other retinal/neural gene products mapped to the same region of chromosome 17 also are expressed by the CAR patient's tumor. The regulation of recoverin expression in these tumors, nonetheless, will be an important aspect of future research. Adenylyl cyclase, G proteins, and synaptic plasticity Mark M. Rasenick Departments of Physiology, Biophysics and Psychiatry, University of Illinois, College of Medicine, Chicago, IL 60612-7342. raz@uic.edu 484 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 Abstract: It has been suggested that type I adenylyl cyclase may play a unique role in long-term potentiation, due to both unique regulatory properties as well as a specialized distribution within the mammalian brain. This would allow an integration of the signals wrought by increased intracellular calcium with those conveyed into the cellular milieu via increased cAMP. These results are discussed in the context of changes in cellular structure, because of changing interactions between G proteins and cytoskeletal components, which might be expected to accompany chronic synaptic activation. [XIA ET AL.] The target article by XIA ET AL. makes interesting observations that suggest a role for a specific adenylyl cyclase subtype (type I) in the synaptic events subserving long term potentiation. A model is presented that would integrate calcium and cyclic AMP signals and provide for the possible long-term elevation of cAMP, which might be needed to evoke the synaptic remodeling (physical or chemical) that likely accompanies this process. It is suggested that type I adenylyl cyclase might serve as a unique neuromodulator due to its unique distribution and localization to some brain regions often associated with long term potentiation (LTM). It must be pointed out, however, that mRNA hybridization was used to localize this enzyme. Most of the localization of mRNA is confined to the cell body. While there appears to be protein synthesis in the dendrite, it is likely that the amount of mRNA there is sufficiently small that most hybridization techniques would not detect this. If LTM is a process which is largely presynaptic (this author is not attempting to join the debate), then Ca2+ sensitive adenylyl cyclase at the axon terminals would be expected to have the type I enzyme. The cell bodies have a diffuse representation, not all of which appears positive for type I adenylyl cyclase. Further, axons might project some distance from the synapse. Sustained Ca2+ entry into synaptic terminals could also activate type I adenylyl cyclase, but the precise localization of that enzyme awaits the development of selective antibodies of sufficient quality to be used in immunoelectron microscopy. The authors state that receptors and Gs do not activate the type I enzyme in the absence of calmodulin. These unpublished findings appear to contradict recent work (Sutkowski et al. 1994), which demonstrates significant (five-fold) activation of Type I adenylyl cyclase in SF9 membranes with activated Gsa. This does not undermine the authors' intriguing proposition that there exists some mechanism for the receptor-independent stimulation of adenylyl cyclase. We have suggested (Roychowdhury & Rasenick 1994; Yan & Rasenick 1990) that, since the net neurotransmitter (quantitatively) available to stimulate adenylyl cyclase is minuscule next to that involved in the inhibition of that enzyme, there must be some indirect mechanism for activating that enzyme. Commentary/Controversies in Neuroscience III Studies from the groups of either Cooper or Gnegy (Ahlijanian ct al. 1987; Treisman et al. 1983) suggest that Ca 2 + may indeed play a role in such an indirect activation of the enzyme. These studies also suggested that calmodulin activation of adenylyl cyclase (in brain membranes) took two forms, one of which was independent and one of which required GTP. The GTPindependent form likely represented the type I enzyme. The calmodulin-GTP sensitive adenylyl cyclase connection need not have been direct. In fact, we have postulated that the role of calmodulin (in this indirect activation of adenylyl cyclase) is to activate CAM kinase II, subsequently phosphorylating microtubule associated protein 2 (MAP2), which occurs complexed with CAM Kinase, MAP 2 and tubulin on the post-synaptic membrane. MAP2 subsequently reduces its affinity for tubulin, which binds to Gsa or Gial with a Kd of about 130 nM and activates these G proteins due to direct transfer of GTP (Roychowdhury et al. 1993; Wang et al. 1990), and in the case where Gsa is activated, effects a calmodulin-dependent yet indirect activation of adenylyl cyclase. Tubulin is capable of transferring GTP to a recombinant Got under conditions where that Cot is incapable of binding nucleotide from the medium (Roychowdhury & Rasenick 1994). In fact, tubulin is capable of bypassing a "tightly coupled" receptor to activate a G protein (Popova et al. 1994). It is suggested that tubulin, perhaps in response to a neurotransmitter not normally coupled to a G protein, can be engaged to activate Gs or Gil and their attendant intracellular effectors. This could be especially important with respect to the possibility that adenylyl cyclase can be activated (depending on subtype) by f}"y subunits, liberated, perhaps from a Ga coupled to a receptor nor normally associated with the stimulation of adenylyl cyclase (Andrade 1993). Such stimulation cannot occur without the prior activation of Gsa. The mechanism proposed here would allow tubulin, liberated, perhaps from the bonds of MAP2 on the post-synaptic membrane in response to an increase in intracellular Ca 2 + , to activate Gs, Gil, orGq in a receptor-independent fashion. Thus, tubulin may provide a conduit for the interactions among neurotransmitters. Clearly, there is a relationship between the ordered environment of the synaptic membrane and G protein-mediated signal transduction systems. This might be particularly true in the nervous system, where rapid and discrete response is a hallmark of synaptic transmission. Further, as pointed out by the authors, changes in synaptic form may accompany changes in synaptic efficiency. Ca 2+ -calmodulin-mediated processes may well be involved in such shape changes. Further, it is possible that G proteins at the synapse may be involved in synaptic shape change through their modulation of the assembly state of synaptic microtubules (Want & Rasenick 1991). In this way, signals through a variety of second messenger systems may interface to orchestrate the cellular events that create memory. Regulation of adenylyl cyclase in LTP Erik D. Roberson and J. David Sweatt Division of Neuroscience, Baylor College of Medicine, Houston, TX 77030. erikr@llgand.neusc. bcm.tmc.edu and david@ligand.neusc. bcm.tmc.edu Abstract: Our results on hippocampal long-term potentiation are considered in the context of Xia et al.'s hypothesis. Whereas the target article proposes presynaptic PKC involvement in adenylyl cyclase activation by phosphorylation of nenromodulin, we suggest an additional postsynaptic role involving RC3/nenrogranin. Finally, we examine the possibility that the adenylyl cyclase mutant mouse may display normal learning with a selective impairment of memory. [XIA ET AL.] The target article by XIA ET AL. reviews the evidence supporting the hypothesis that a general mechanism of long-term synaptic modulation is activation of adenylyl cyclase by calcium and calmodulin, leading to altered gene expression via activation of PKA. Our research on the molecular mechanisms of long-term potentiation (LTP) in area CA1 of the rat hippocampus strongly supports this model. Xia et al. document the fact that adenylyl cyclase is activated during LTP induction (Chetkovich et al. 1991; Frey et al. 1993). This activation is blocked by preventing calcium flux through NMDA receptors (Chetkovich et al. 1991; Frey et al. 1993) and requires calmodulin (Chetkovich & Sweatt 1993). Although small, the LTPassociated increase in cAMP concentration is sufficient to support activation of PKA (Roberson & Sweatt, unpublished observations). In addition, the fact that the rise in cAMP is transient (Chetkovich & Sweatt 1993) is consistent with the model in which only a brief activation of PKA is required to initiate the cascade leading to changes in gene expression. The target article ably synthesizes the growing body of evidence supporting a role for calmodulin-stimulated adenylyl cyclases in LTP. We would like to open two additional lines of discussion: 1. Is the Interaction with PKCpre- or postsynaptic^ An issue of great debate in LTP is whether the modifications that support potentiation occur pre- or postsynaptically. Addressing the role of PKC in activating adenylyl cyclase in section 7.2, Xia et al. have focused on regulation of calmodulin buffering by neuromodulin, which is localized presynaptically in axon terminals (Goslin et al. 1988). However, in light of data suggesting that some changes in gene expression during LTP occur postsynaptically (Mackler et al. 1992), we feel that the hypothesis should be extended to include a role for the postsynaptic calmodulinbinding protein RC3/neurogranin (Baudier et al. 1989; Watson et al. 1990). RC3 is a 17 kD protein homologous to neuromodulin in its PKC phosphorylation and calmodulin-binding domains that, like neuromodulin, releases calmodulin when phosphorylated by PKC. However, unlike neuromodulin, RC3 is localized postsynaptically and enriched in dendritic spines (Represa et al. 1990; Watson et al. 1992). We have demonstrated an increase in PKC activity and in the post hoc phosphorylation of a 17 kD PKC substrate during LTP (Klann et al. 1991; 1992; 1993). This 17 kD protein is biochemically and immunologically indistinguishable from RC3 and its phosphorylation in situ increases during LTP as demonstrated by back phosphorylation (Chen, Klann & Sweatt, unpublished observations). Thus, RC3 may serve to release calmodulin postsynaptically during LTP and contribute to the activation of adenylyl cyclase and other postsynaptic effectors, such as the retrograde messenger generator. 2. A selective memory knockout? In section 8.4, XIA ET AL. describe the deficits of the adenylyl cyclase knockout mouse in the Morris water maze task. They stress the defect in the transfer test, attributing the mutant's normal latency of escape onto the hidden platform to the low sensitivity of that index. This is a valid interpretation of the data, but there is an alternative explanation. The hidden platform task is by its nature a relatively immediate test of learning and short-term memory, especially when trials are massed; on the other hand, the transfer test is usually administered after days of training on the hidden platform task and thus is a measure of longer-term memory. It is possible that the mutation in adenylyl cyclase selectively interferes with longterm memory, leaving learning and short-term memory, and thus hidden platform task performance, intact. This would be consistent with evidence suggesting that in the hippocampus, cAMP antagonists selectively block the late phase of LTP, having no effect on earlier stages (Frey et al. 1993). Interestingly, transgenic mice with mutations in a-calcium/ calmodulin-dependent kinase II, which lack early stages of LTP, do exhibit deficiencies on the hidden platform task (Silva et al. 1992a; 1992b). Our hypothesis predicts that protein synthesis inhibitors, BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 485 Commentary /Controversies in Neuroscience III that, like cAMP antagonists, selectively block late LTP (Frey et al. 1988), should cause deficits in the transfer test, but not in the hidden platform task. This would be consistent with data from many other systems, where inhibiting protein synthesis spares initial learning, but selectively interferes with retention of the learned behavior (Agranoff et al. 1965; Tully et al. 1994). We are not aware of any published accounts of the effects of protein synthesis inhibitors on hidden platform task performance, but rats injected with these inhibitors do have deficits in the transfer test (Itoh et al. 1992). XIA ET AL. have examined a strong, testable hypothesis for the mechanism of long-lasting modulation of synaptic function, with supporting evidence in systems as diverse as Aplysia, Drosophila, and the mammalian hippocampus. Such a hypothesis provides a solid conceptual framework around which to construct a future experimental edifice. Modulation of the cGMP-gated channel by calcium Mandeep S. Sagoo and Leon Lagnado Medical Research Council, Laboratory of Molecular Biology, Hills Road, Cambridge CB2 2QH, England. Il1@mrc-lmb.cam.ac.uk and ms1@mrclmb.cam.ac.uk. Abstract: Calcium acting through calmodulin has been shown to regulate the affinity of cyclic nucleotide-gated channels expressed in cell lines. But is calmodulin the Ca-sensor that normally regulates these channels? [MOLDAY & HSU] In the first five sections of their target article, MOLDAY & HSU provide a concise overview of the biochemical and functional properties of the cGMP-gated channel in photoreceptors. In the next two sections they concentrate on the evidence that calcium acts through calmodulin to regulate the affinity of cGMP-gated channel for its ligand. Recent work from their laboratories has demonstrated that this modulatory action cannot be measured in the homo-ologomeric channel formed by the human alpha subunit expressed in a cell line, but it can be measured when the alpha subunit is coexpressed with the longer of the two splice forms of the beta subunit. The modulatory action of calmodulin on the rod cGMP-gated channel is therefore mediated through the beta subunit. In contrast, the effect of calmodulin on the cyclic nucleotide-gated channel from rat olfactory receptors is mediated through the alpha subunit. A second important difference between the two types of receptor is that calmodulin has a stronger action on the olfactory channel: calcium-calmodulin increases the KU2 of the expressed olfactory channel over 10-fold, whereas the K 1/2 of the rodchannel is increased 1.5-to-2-fold. When a rod responds to light, calcium levels fall inside the outer segment, and MOLDAY & HSU suggest that this will relieve the effect of calmodulin on the channel, increasing its affinity for cGMP and therefore promoting the re-opening of channels during the process of light adaptation. Unfortunately, we have relatively little information about this proposed action. GrayKeller and Detwiler (1994) have reported that calmodulin or calmodulin inhibitors introduced into a transducing rod through a patch pipette have no effect on phototransduction. We have made similar observations using an alternative preparation, the truncated salamander rod outer segment. This preparation leaves the outer segment relatively intact and many properties of the phototransduction mechanism are maintained (Lagnado & Baylor 1994), but with the advantage that it is possible to manipulate internal levels of calcium and cGMP. We find that calcium over the range 200nM to lOnM regulates the channels affinity for cGMP. At low cGMP concentrations the current is 486 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 increased about 2-fold when Ca is removed. A diffusible mediator is involved, since the ability to modulate the channel's affinity for cGMP by changing calcium gradually washes out. However, there does not seem to be any effect of adding calmodulin or calmodulin inhibitors such as mastoparan. All these observations are unpublished. A similar observation is found in the original study of Kramer and Siegelbaum (1992) on olfactory receptors, where the ability of calcium to modulate the cAMP-dependent current did not seem to involve calmodulin. Although we are still investigating a possible action of calmodulin, the obvious implication of the results we have so far is that some other calcium-binding protein modulates the channel. Recently, three new calcium-binding proteins have been discovered in the rod outer segment: recoverin, GCAP and p24. All of these are "calmodulin-like," in the sense that they are soluble, possess multiple E-F hands, and have molecular weights of 20-24 kD. It does not seem that recoverin binds to the channel but it will be interesting to see if GCAP or p24 can. The idea that there may be a second calcium-sensitive modulator of the channel has also been suggested by Gordon and Zimmerman (1994), who found evidence for an endogenous modulator of the cGMP-gated channel which was lost at low calcium. ACKNOWLEDGMENT Our work is supported by the Medical Research Council and the Human Frontiers in Science Program. Unique lipids and unique properties of retinal proteins Kamon Sanada and Yoshitaka Fukada Department of Pure and Applied Sciences, College of Arts and Sciences, University of Tokyo, Komaba 3-8-1, Meguro-hu, Tokyo 153, Japan Abstract: Amino-terminal heteroacylation has been identified in retinal proteins including recoverin and a subunit of G-protein, transducin. The tissue-specific modification seems to mediate not only a proteinmembrane interaction but also a specific protein-protein interaction. The mechanism generating the heterogeneity and its physiological role are still unclear, but an interesting idea for the latter postulates a fine regulation of the signal transduction pathway by distinct N-acyl groups. [HURLEY] AS reviewed in HURLEY'S target article, an interesting aspect of recoverin has been heightened by the finding of its unique N-terminal modification by myristate or its related fatty acids. Protein N-myristoylation has been identified in a variety of proteins and the modification is required for membrane targeting of the proteins or protein-protein interactions (reviewed in Gordon et al. 1991). In 1992, two research groups found that retinal G-protein transducin a-subunit (Tot; Kokame et al. 1992; Neubert et al. 1992) and recoverin (Dizhoor et al. 1992) were modified by one of four types of fatty acyl groups, 14:0, 14:1 (5-cis), 14:2 (5-cis, 8-cis) and 12:0 (Fig. 1). After that, similar modification has been identified in other retinal proteins such as guanylyl cyclase-activating protein (GCAP; Palczewski et al. 1994) and a catalytic subunit (C-subunit) of cAMP-dependent protein kinase (Johnson et al. 1994). It is interesting to note that only myristate was detected in the C-subunits purified from bovine heart and brain. A growing body of evidence suggests that the heteroacylation is not a protein-specific but a tissue-specific phenomenon. A simple explanation for the tissue-specific heterogeneity is that the mammalian photoreceptor may have a unique composition of each acyl-CoA pool available to the myristoyl-CoA: protein N-myristoyltransferase (NMT), an enzyme probably Commentary /Controversies in Neuroscience III responsible for this modification. Then the relative amounts of the four acyl groups linked to the retinal proteins are expected to be similar. As shown in Figure 1, however, this is not the case. In contrast to the relative abundance of N-myristoylated forms of recoverin (43%) and the C-subunit (57%), only small percentages (5—7%) of Tot and CCAP are N-myristoylated. The significant difference of the relative amounts of four isoforms among proteins (Fig. 1) raises a question about the mechanism yielding the heterogeneity. Mammalian photoreceptors might have isozymes of protein-specific or N-terminal sequence-specific NMT with different accessibility to each acyl-CoA. Alternatively, the selectivity of a common NMT for acyl-CoAs and proteins might be regulated by unknown factor(s) in photoreceptor cells. Characterization of photoreceptor NMT is required for elucidation of the mechanism. Although the vertebrate retina is unusual tissue enriched with unsaturated fatty acids, C14:2(5-cw, 8-cis) has not been found in the fatty acid composition analysis. The uncommon fatty acid, C14:2, seems to be formed specifically for the acylation of retinal proteins. Enrichment of C14:2 in retinal proteins (Fig. 1) raises the more important question of whether the heteroacylation has a specific physiological function. As described by Hurley, the N-terminal fatty acid of recoverin serves as a Ca 2+ -dependent membrane anchor, but this does not fully account for the significance of the heteroacylation. It seems reasonable to speculate that the acyl groups play another role in protein function. Although no direct evidence has been presented for the functional difference caused by distinct iV-acyl groups, a hint was given in the case of Tot by Kokame et al. (1992). They synthesized two kinds of acylated peptide (C14:0and C12:0-peptide) corresponding to the N-terminal part of Ta and observed that the C14:0-peptide was a more potent compet- itor for the Ta-TfJ-y interaction than the C12:0 peptide. Extend ing this, they synthesized two additional peptides, C14:l- and C14:2-peptide and found the following order of magnitude for the inhibitory effect; C14:0 > C14:l > C14:2 = CI2:0(Shono et al., in preparation). These observations suggest that the N-acyl group provides a specific protein-protein interaction and that the affinity between Ta and T|}"y is altered by the heterogeneous acylation. A weaker ot-(i"y interaction, a unique property of retinal G-protein, due to a small percentage (5%) of N-myristoylated Ta could lead to rapid association-dissociation cycles of transducin subunits and contribute to the rapid and highly amplified photon-signal transduction in photoreceptor cells. HURLEY'S observation of the complex formation between recoverin and rhodopsin kinase has opened the way to explore a possible fine regulation of rhodopsin phosphorylation by the heteroacylation of recoverin. In fact, the magnitude of the inhibition of rhodopsin phosphorylation is different among recoverin isoforms with distinct fatty acyl groups (Sanada et al. 1995). Physiological role of the heterogeneous acylation of retinal proteins discussed here is likely to contribute to the unique properties of high speed, high gain, and low noise signal transduction in photoreceptor cells. ACKNOWLEDGMENTS The work is supported in part by Grants-in-Aid from the Japanese Ministry of Education, Science and Culture, and by grants from Toray Science Foundation, SUNBOR, and Mochida Memorial Foundation. K. Sanada is supported by fellowships of the Japan Society for the Promotion of Science for Young Scientists. a N-terminal Structures Ta be Recoverin c GCAP C-subunit 43.0 C14:0 H Gly- 28.1 C14:1 (5-c/s) GlyC14:2 (5-c/s, 8-cis) H O C12:0 28.9 H Figure 1 (Sanada & Fukada). Relative quantities of N-fatty acyl groups linked to bovine retinal proteins "Data from Kokame et al. (1992) '•Data from Sanada et al. (1995) "Data from Sanada et al. (1994) BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 487 Commentary/Controversies in Neuroscience III Na-Ca + K exchanger and Ca 2 + homeostasis in retinal rod outer segments: Inactivation of the Ca 2 + efflux mode and possible involvement of intracellular Ca 2 + stores in Ca 2 + homeostasis Paul P. M. Schnetkamp Department of Medical Biochemistry, University of Calgary, Calgary, Alberta, T2N 4N1, Canada. schnetka@asb.acs.ucalgary.ca Abstract: Inactivation of the Ca 2 + extrusion mode of the retinal rod NaCa + K exchanger is suggested to be the mechanism that prevents lowering of cytosolic free Ca 2 + to < 1 nM when rod cells are saturated for a prolonged time under bright light conditions. Under these conditions, Ca 2 + fluxes across disk membranes can contribute significantly to Ca 2+ homeostasis in rods. [BOWNDS & ARSHAVSKY] The contribution by BOWNDS & AR- SHAVSKY summarizes the biochemical and physiological evidence that calcium acts as a messenger for light adaptation in the outer segments of retinal rods (ROS). As this contribution does not touch much upon Ca 2 + homeostasis in ROS, I would like to review some of our recent findings on this subject. Dynamic Ca 2 + fluxes occur across the plasma membrane (ROS). In darkness, a large sustained Ca 2 + influx via the cGMP-gated channels is balanced by Ca 2 + efflux via Na-Ca + K exchange, while no evidence has been reported to involve a plasma membrane Ca2+-ATPase. The ROS Na-Ca exchanger differs from the ones in other tissues: it couples Ca 2 + extrusion to both inward Na + and outward K + gradients and operates with a maximal capacity to change total intracellular Ca 2 + in bovine ROS by > 1 mM/s (for recent reviews on Na-Ca + K exchange, see Lagnado & McNaughton 1990; Schnetkamp 1989). The high-capacity Ca 2 + efflux mode was typically observed in ROS containing unphysiologically high Ca 2 + loads. Ca 2 + influx and Ca 2 + efflux mediated by the Na-Ca + K exchanger could cause large (|JUM) and rapid (seconds) changes in cytosolic free Ca 2 + when operating at only 1% of its maximal capacity (Schnetkamp et al. 1991). The cytosolic Ca 2 + buffer capacity was found to be 200 p,M total Ca 2+ /jiM free Ca 2 + (Schnetkamp & Szerencsei 1993). NaCa(+K) exchangers were generally believed to operate in a continuous and reversible fashion with little regulation of the flux rate. In view of this and in view of the exchanger's high capacity to regulate cytosolic free Ca 2 + , intracellular Ca 2 + sequestration and release has received little attention and is thought to be of little importance for regulation of cytosolic Ca 2 + in ROS. ROS may have a 4Na:(lCa + IK) exchanger (as opposed to 3Na:lCa exchange in other tissues) as it is unlikely to reverse under any physiological condition (e.g., depolarized rod cell with elevated cytosolic Na + ) and carry Ca 2 + into the cell. However, during bright daylight, rod cells are saturated and Ca 2 + influx is suspended for hours, during which normal cytosolic Na + is expected to be restored and the 4Na-lCa + IK exchanger is expected to lower cytosolic Ca 2 + to 0.1-0.2 nM. The exchanger is kinetically quite capable of handling low free Ca 2 + concentrations as judged from Ca 2 + influx data (via the reverse mode of the exchanger): Ca 2 + influx could raise intracellular Ca 2 + to 1 u,M at an external free Ca 2+ concentration of only 20 nM, when Na + -loaded and Ca 2+ -depleted ROS were incubated in a KC1 medium with 7.5 mM NaCl (to mimic intracellular milieu) (Fig. 7 in Schnetkamp et al. 1991). Thus, the problem is: how does the exchanger avoid lowering cytosolic Ca 2 + to undesirable low values? Na-Ca exchangers have an intracellular regulatory site for Ca 2 + (e.g., the cardiac Na-Ca exchanger, Matsuokaetal. 1993): Ca 2 + influx via reverse Na-Ca exchange requires intracellular Ca 2 + . However, a similar regulation by intracellular Ca 2+ was not observed in ROS (Schnetkamp 1989). A different type of regulation was suggested by the observation that the Ca 2 + efflux mode of the Na-Ca + K exchanger failed to lower cytosolic free 488 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 Ca 2 + below 50-100 nM when bovine ROS were exposed to external Na + in the presence of EDTA. Instead, addition of Na + caused a rapid efflux phase that was halted rather abruptly at relatively high cytosolic free Ca 2 + levels. Inactivation of the Ca 2 + efflux mode of Na-Ca + K exchange after a brief period of high-velocity activity was shown to be the underlying molecular mechanism (Schnetkamp & Szerencsei 1993). Cytosolic free Ca 2 + levels of 50-100 nM could be maintained for prolonged periods of time (1 hr) when the efflux mode of the Na-Ca + K exchanger was reduced to < 0.03% of its maximal capacity. Under these conditions, Ca 2 + sequestration and release from ROS disks could contribute significantly to Ca 2 + homeostasis in ROS. By combining total Ca 2 + flux measurements (with 45Ca) and free Ca 2 + measurements (with fluo-3) we found that a significant portion of Ca 2 + that enters ROS became sequestered within disks. Ca 2 + fluxes across ROS disk membranes were about 100-fold slower when compared with Ca 2 + fluxes mediated by the high velocity mode of Na-Ca + K exchange. Nevertheless, Ca 2 + fluxes across disk membranes became significant when the exchanger was inactivated and contributed equally to regulation of cytosolic free Ca 2 + . To summarize, inactivation of the Ca 2 + efflux mode of the NaCa + K exchanger is a useful attribute to prevent lowering of cytosolic free Ca 2 + to undesirably low values of < 1 nM when Ca 2 + influx via the cGMP-gated channels is interrupted for a prolonged period of time in bright illumination. The significance and molecular mechanism of Ca 2 + fluxes across disk membranes remains to be explored fully, but observations suggests that the Ca 2 + filling state of disks dictates the Ca 2 + concentration at which the exchanger inactivates (Schnetkamp & Szerencsei 1993). Nuclear magnetic resonance studies on the structure and function of rhodopsin Steven O. Smith Department of Molecular Biophysics and Biochemistry, Yale University, Box 208114, New Haven, CT 06520. smlth@lokl.csb.yale.edu Abstract: Magic angle spinning (MAS) NMR methods provide a means of obtaining high resolution structural data on rhodopsin and its photointermediates. Current work has focused on the structure of the retinal chromophore and its interactions with surrounding protein charges. The recent development of MAS NMR methods for measuring internuclear distances with a resolution of ~0.2 will complement diffraction methods for addressing key mechanistic questions. [HARGRAVE] The major conclusion drawn by HARCKAVE on the future directions of structure/function studies on rhodopsin is that diffraction techniques, particularly electron diffraction of two-dimensional crystals, are the sole methods for obtaining a high-resolution structure of rhodopsin and its photointermediates. Hargrave rules out high-resolution solution NMR methods for structural studies because rhodopsin is a high molecular weight membrane protein. Although solution NMR methods are in fact poorly suited for membrane proteins, I would like to suggest that magic angle spinning NMR methods are promising for addressing some relevant structural questions concerning the reaction mechanism of rhodopsin. The key structural questions focus on (1) the structure and protonation state of the retinal chromophore during the rhodopsin photoreaction, (2) the relative position of Glull3 and the retinal Schiffbase, and (3) the structure of the retinal binding site. The key mechanistic questions involve (1) how the protein raises the pKa of the retinal Schiffbase to effectively shut down protein activity in the dark, (2) how absorbed light energy is stored in bathorhodopsin and channeled into the protein, and (3) how structural changes between the retinal and C l u l l 3 trigger Schiffbase deprotonation and G-protein binding. Commentary/Controversies in Neuroscience III Magic angle spinning (MAS) is a solid-state NMR method for obtaining high-resolution spectra of macromolecules that are rigid or have slow rotational correction times (Smith & Peersen 1992). MAS NMR has been used increasingly in the past six years for structural studies on membrane proteins, in particular on rhodopsin and bacteriorhodopsin (Creuzet et al. 1991; Han & Smith 1995; Han et al. 1993; Harbison et al. 1984; Lakshmi et al. 1993; Smith et al. 1991; Thompson et al. 1992). The focus in these studies has been on the structure and protein environment of the retinal chromophore (i.e., the conformation of the retinal and how the retinal interacts with surrounding protein charges). Analysis of the NMR chemical shifts has been used to establish the relative positions of Glull3 and the retinal chromophore (Han et al. 1993). MAS NMR methods have also been used to determine the configuration and protonation state of the retinal-lysine C = N bond, and the conformation of the retinal C 6 —C 7 bond. These measurements directly address how the protein modulates the visible absorption of the retinal and the pKa of the Schiff base. In the case of metarhodopsin II, NMR measurements resolved a controversy as to whether the chromophore was attached as an unprotonated Schiff base or as a tetrahedral carbinolamine (Smith et al. 1992). Based on NMR constraints, we have recently been able to position the retinal chromophore in a structural model of rhodopsin developed by J. Baldwin (Baldwin 1993; Han & Smith 1995). We are now in a position to address questions involving the reaction mechanism of the protein by characterizing how the structure of the retinal and its interactions within the protein change in each reaction intermediate. As mentioned by HARGRAVE, structural studies on reaction intermediates are possible using low temperature trapping methods. One advantage of NMR over absorption or diffraction methods, is that the contributions of different intermediates trapped in a low temperature experiment can be distinguished on the basis of different NMR chemical shifts. Using low temperature methods, we have characterized the structural changes occurring in the formation of a bathorhodopsin where rapid photoisomerization of the retinal generates a conformationally distorted ll-trans chromophore (Han & Smith 1995; Smith et al. 1991). Steric interactions between the retinal and the protein are responsible for most, if not all, of the 33 kcal/mole of energy stored in this intermediate. Finally, the future appears bright for MAS NMR and high resolution distance measurements in membrane proteins. Over the past five years, a number of methods have been developed for measuring 13 C.... I3 C and I 3 C ... 1 5 N distances out to ~ 7 A with a resolution on the order of 0.1 - 0.2 A (Bennett et al. 1992; Gullion & Schaefer 1989; Raleigh et al. 1988; Tycko & Smith 1993). With the ability to incorporate specific isotopic labels into selected sites in membrane proteins, there is the prospect that high resolution distance measurements can be made to address the local secondary structure and tertiary interactions in both rhodopsin and its photointermediates. These measurements will greatly extend in the structural and mechanistic studies outlined above. A novel protein family of neuronal modulators Ken Takamatsu Department of Physiology, Toho University School of Medicine, 5-21-16, Ohmori-nishi, Ohta-ku, Tokyo 143, Japan. physiken@med. toho-u.ac.jp Abstract: A number of proteins homologous to recoverin have been identified in the brains of the several vertebrate species. The brainderived members originally contain four EF-haml domains, but NH2terminal domain is aberrant. Many of these proteins inhibited lightinduced rhodopsin phosphorylation at high [Ca2+], suggesting that the brain-derived members may act as a Ca2+-sensitive modulator of receptor phosphorylation, as recoverin does. [HURLEY] Recoverin was initially discovered in bovine photoreceptors as an activating factor of guanylyl cyclase (GC) to mediate the Ca 2+ -dependent stimulation of cGMP resynthesis during the recovery of the electrical light response. Recent studies indicate that recoverin does not regulate GC. Instead, it may regulate the rate of deactivation of cGMP phosphodiesterase (PDE) following photoexcitation. Rhodopsin phosphorylation and in consequence cGMP hydrolysis in bovine rod outer segments (ROS) are Ca 2+ -dependent in the presence of ATP. The level of rhodopsin phosphorylation decreases and the lifetime of active PDE increases when free [Ca 2+ ] is raised from 1 nM to about 1 |J.M; in both cases the half maximal effect was observed at 140-170 nM of free [Ca 2 + ]. The Ca 2 + effects observed are mediated by recoverin which inhibits rhodopsin kinase at a high [Ca 2 + ]. Although the protein structure of this protein family was first reported in visinin (Yamagata et al. 1990) and the actual role of recoverin is that of S-modulin, the term of recoverin seems to be suitable for this protein family. During a search for endogenous protein kinase C (PKC) inhibitors, the protein 21-kDa CaBP has been isolated from bovine brain and characterized as a member of the calmodulin superfamily (Walsh et al. 1984). 21-kDa CaBP was reported to have isoforms of 21 kDa and 23 kDa depending on the degree of glycosylation. Since 21-kDa CaBP did not inhibit PKC activity, no further analysis has been done. After the discovery of recoverin, a number of proteins homologous to recoverin were identified in the brain of the several vertebrate species (Hidaka & Okazaki 1993; Kajimoto et al. 1993; Kobayashi et al. 1992; Nef et al., in press; Takamatsu & Uyemura 1992), indicating that multiple isoproteins exist in the brain and form the brain-derived recoverin family. These proteins have been given several different names, such as neurocalcin a.P.'Y.S, hippocalcin and hippocalcinIike protein 1, visinin-like protein (VILIP) 1,2,3, and neural calcium sensor (NCS) 1. These, including recoverin and Smodulin, are classified into three subfamilies according to their sequence homology (Nef et al., in press). All of these proteins originally contained four EF-hand domains. One of the four EFhand domains does not seem to satisfy the structural criteria of a high affinity calcium-binding site; therefore, these proteins may bind less than three moles of Ca 2 + per mole of protein. The NH 2 terminal sequence of these proteins meet the criteria for NH 2 terminal myristoylation: a glycine residue at the NH 2 -terminus, and a serine or threonine residue at position 5. Many of these proteins are known to have a blocked NH 2 -terminus, indicating that these proteins might be NH2-myristoylated. Hippocalcin was confirmed to incorporate myristic acid into its NH 2 -terminus (Kobayashi et al. 1993). NH 2 -terminal myristoylation of hippocalcin was essential for its Ca 2+ -dependent membrane-binding properties as recoverin does. The maximal membrane-binding of hippocalcin was observed at about 30 n,M [Ca 2 + ], and halfmaximal binding at about 5 JJLM [Ca 2 + ]. Contrarily, NH 2 myristoylation was not essential for the inhibition of rhodopsin phosphorylation by these proteins (Kawamura 1994). The physiological roles of NH2-myristoylation are yet to be determined. The physiological function of the brain-derived members of the recoverin family from all three subfamilies were examined using ROS membrane preparation; S-modulin and recoverin from subfamily I, NCS from subfamily II, VILIP and hippocalcin from subfamily III (see Kawamura 1994 for a review). All the proteins tested in this study inhibited rhodopsin phosphorylation at high (10 |xM) [Ca 2 + ], suggesting that all members of the recoverin family act in a similar way as does S-modulin in ROS. At least five G protein-coupled receptor kinases (GRKs) exist in the brain (Inglese et al. 1993). Phosphorylation of receptor molecules by GRKs may be involved in agonistdependent desensitization, however, the mechanism behind the modulation of phosphorylation in the neurons shows diversity and is totally different from the mechanism in the vertebrate photoreceptors. The effect of the brain-derived members on receptor phosphorylation by GRKs are yet to be determined. BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 489 Commentary/Controversies in Neuroscience III It has been known that Ca 2 + affected both the rate of rise (PDE) activity) and the time course of the recovery (PDE deactivation) of a photoresponse in the vertebrate photoreceptors. It is noteworthy that S-modulin purified using ROS membranes affected the PDE activity but not the PDE deactivation (Kawamura & Murakami 1991); however, S-modulin purified using phenylsepharose affected both the PDE activity and deactivation (Kawamura 1993). Recently, Lagnado and Baylor (1994) observed the Ca 2 + effect on the PDE activity using internally dialyzed ROS preparation and found that the Ca 2 + effect seems to be mediated by a soluble Ca 2+ -binding protein. Although the isoforms of S-modulin have not been characterized, thus two Ca 2 + effects seem to be mediated by two distinct forms of S-modulin, one isoform is responsible for the PDE deactivation and the other isoform for the PDE activity. Two distinct Ca 2 + effects on PDE activation are observed only in the presence of both ATP and S-modulin. The mechanism of the Ca 2 + effect on the PDE activity is not known, but is possibly regulated by rhodopsin phosphorylation. Frequenin, a member of the recoverin family identified in Drosophila, modulates neurotransmitter release at the neuromuscular junction (Pongs et al. 1993) and also modulates K + -channel in muscle cells (Poulain et al. 1994) in a Ca 2 + dependent manner. The molecular mechanism of the frequenin actions is not known. The brain-derived members are distributed heterogeneously in the brain. Many of these proteins are distributed in neurons. In some cases, the same neuron expresses at least two different members of this family (Saitoh et al. 1993; 1994). It is not clear whether each protein plays a different role or shares the same role in its host cells, however, the common action on rhodopsin phosphorylation lead us to postulate that all the members of this protein family regulate the signal transduction systems by modulating phosphorylation reaction in their host cells. Glutamate accumulation in the photoreceptor-presumed final common path of photoreceptor cell death Makoto Tamai Department of Ophthalmology, Tohoku University School of Medicine, 1-1 Seiryomachi, Aoba-ku Sendai, 980-77 Japan Abstract: Genetic abnormalities of three factors related to the photoreceptor mechanism have been reported in both animal models and humans. Apoptotic mechanism has also been suggested as a final common pathway of photoreceptor cell death. Our findings of increased level of glutamate in photoreceptor cells in rds mice suggest that amino acid might mediate between these two pathological mechanisms. [DAICER ET AL.] An excellent summary of genetic abnormalities of rhodopsin, peripherin/RDS, and phosphodiesterase b-subunit and phenotypic variations in human is provided by DAICER ET AL. They also describe almost all possible mechanisms of photoreceptor cell death caused by these mutations and proteins. I agree with their descriptions about molecular abnormality and their explanation of the variety of phenotypes, but the abnormality of proteins related to signal transduction does not seem to be enough to explain the biological mechanisms underlying retinitis pigmentosa and the heterogeneity of phenotypes of this disease entity. From our observations, other mechanisms may exist between genetic mutations and trigger of cell death. We have reported increased level of glutamate-like immunoreactivity in the interior of segments of photoreceptors at 2, 3, and 9 weeks postnatally of rds/rds mice comparing to the control BALB/c mice (ARVO, abstract #501,1994). It is very interesting 490 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 that the accumulation was observed to start earlier than the beginning of photoreceptor degeneration. The accumulation of this amino acid has also been reported in the chick retina by Ulshafer et al. (1990). They reported that the uptake of glutamate in photoreceptors occurred normally but that the release of this neurotransmitter did not because glutamate was accumulated and hence the retinal degeneration took place. At present, we have no data about why glutamate is accumulated in rds/rds photoreceptors. We also cannot explain why the accumulation of this amino acid was related to the apoptosis of photoreceptors. (Chang et al. 1993). Kure et al. (1991) have reported that the chromosomal DNA of cultured neurons was degraded into nucleosomal-sized DNA fragments by the addition of glutamate prior to the glutamate-induced neuronal death. Oxidative stress induced by light exposure or other mechanisms might also result in apoptosis-type cell death by the depletion of glutathione (Ratan et al. 1994). Ratan et al. have shown that glutamate could induce glutathione depletion in immature embryonic cortical neurons. Intracellular mechanisms of accumulation and the mode of action of glutamate in photoreceptors need to be studied. In conclusion, as pointed out by DAICER ET AL., retinal dystrophy has genetic and allelic as well as clinical heterogeneity. Such heterogeneity may be affected by environmental or systemic conditions, which might affect retinal circulation and might also be affected by daily lighting conditions or nutrition via the level of vitamin A or trace elements such as M g + + or Z n + + (Wu et al. 1993). The genetic kaleidoscope of vision Douglas Wahlsten Department of Psychology, University of Alberta, Edmonton, Alberta, Canada T6G 2E9. wahtsten@psych.ualberta.ca Abstract: Site-specific phenotypic effects of the 73 known alleles in the rhodopsin gene that cause retinal degeneration are difficult to interpret because most alleles are documented in only one case or one family, which means variation in effects could actually arise from interactions with other loci. However, sample sizes necessary to detect epistatic interaction may place an answer to this question beyond our grasp. [DAICER ET AL.] The comprehensive review of DAICER ET AL. documents genetic heterogeneity, allelic heterogeneity, and clinical heterogeneity in a diverse collection of mutations causing degeneration of the human retina. Listed here are 42 mapped loci, one of which (rhodopsin) has at least 73 alleles causing a wide array of phenotypes in terms of the quality, severity, and consistency of effects. While proposing a classification scheme to bring some order to a bewildering complexity in a single gene, the authors point out that many vital aspects of the gene's role in development remain unexplored. A great deal has been accomplished in the biochemical genetics of retinal degeneration, but continuing study suggests a limitless diversity and provokes questions about the goals of this research and the likelihood of achieving them. 1. The chemically complete retina. Understanding a system of 42 genes is challenging enough, yet there must be thousands of genes expressed in the retina, many of which are fixed in the population and do not produce disease. It makes good sense to concentrate on genes associated with disorders of vision if one hopes to someday prevent blindness. At the same time, the connectedness of the metabolic system in the retina dictates that the effects of a mutation at one locus will not be well understood until its associated protein can be placed in a biochemical context that includes the protein products of fixed loci. Thus, it would be helpful to know how many genes are expressed in the retina and how many are unique to the retina. Is it important for science to identify and characterize all of these, and will this Commentary /Controversies in Neuroscience III knowledge benefit those who suffer from retinal dystrophy? The target article has set forth much that is already known but has given little more than a hint at the goals of this enterprise. In particular, will the knowledge of the DNA sequence of various alleles causing some form of retinal degeneration be used primarily to screen for genetic defects and possibly abort embryos destined to suffer reduced vision in later life, or is there hope for ameliorating these hereditary conditions when the specific allele is known? A simple catalog of alleles and phenotypes is sufficient for eugenic purposes, whereas practical intervention to prevent retinal degeneration will necessitate a much more comprehensive understanding of the relations between rhodopsin and its environment. 2. Site-specific phenotyplc effects? The answers to these questions will depend critically on the general properties of biochemical gene action. In particular, should we expect that changing a specific amino acid in a protein should result in a specific array of phenotypic effects at the level of retinal function in anyone possessing that allele? If yes, then there is no need to understand much about associated proteins derived from fixed loci or polymorphic loci not strongly tied to disease. The familiar catalog of loci related to specific disease syndromes can be expanded into a catalog of alleles or groupings of alleles, each with a characteristic syndrome. This seems to be the answer provided by the authors when they state "the clinical phenotype is a consequence of where and when the mutation affects the function of rhodopsin" (sect. 4.2, para. 1). There is no doubt that an amino acid substitution in a critical protein can alter neural function. Rigorously controlled studies of oisogenic mice differing at a single amino acid in only one protein find that the mutants deviate phenotypically from their normal siblings, at least when group averages are compared. However, it does not follow that the average phenotypic difference is attributable solely to the amino acid difference. Coisogenic littermates share thousands of other genes and a multifaceted laboratory environment in common, and the genetic difference occurs in a context that may be critically important for the phenotypic outcome. For example, epistatic interaction with the genetic background can markedly alter the consequences of mutations such as diabetes and reeler in mice (Caviness et al. 1972; Coleman 1981) and interaction with the rearing environment can sometimes be quite dramatic (e.g., Lee & Bressler 1981). The appropriate research designs for assessing epistasis and gene-environment interaction always require individuals with a specific genotype at a designated locus to be combined with different genetic backgrounds or reared in different circumstances. To achieve this, the investigator must have access to a fairly large number of individuals carrying the identical mutation. In this regard, the yellow flag for caution begins to wave when the authors inform us in section 4.1.1 that "The majority of mutations in Table 2 are unique, occurring in one patient or one family only." Is it not possible or even likely that certain mutations with apparently variable effects are working in concert with other retinal genes that, while not causing disease on their own account, are interactants that strongly influence the consequences of a rare rhodopsin allele? And might not some of the apparently consistent effects of other mutations simply reflect the lack of genetic variation in interacting loci within certain families? The great phenotypic diversity attributed in this article to allelic heterogeneity may perhaps result from epistatic interaction or even gene-environment interaction, as DAICEK ET AL. acknowledge briefly in section 3.2.2. Of course, the argument can work both ways. When a mutation in a gene results in substantially different disease in different people, epistasis might be invoked to explain this, but the effect could just as well arise from allelic diversity. In this respect, reliable knowledge about the DNA sequence in the rhodopsin gene should help to discriminate between these two kinds of causes. We begin with the observations that there are in fact many alleles of the rhodopsin gene and there are also many phenotypes. But are the two variations closely associated and, if so, why? 3. Interaction, the night blindness of statistical analysis. It is instructive to explore the kinds of data that will be needed to address two important questions about allelic heterogeneity. Both of these involve the question of sample size required to render a statistical procedure sufficiently sensitive to more subtle genetic effects. One question is simply whether two alleles do indeed differ significantly in the frequency of associated phenotypes. Suppose most cases can be dichotomized into early and later onset of retinal degeneration, and let the probability of the early form be p, and p 2 f° r two alleles. The null hypothesis is that p, = p 2 = p, whereas the alternative hypothesis is that p , = p + c and p2 = P — c, where c is the deviation from an average of the two alleles. If one plans to test the significance of a difference; in sample proportions of the early onset form using a Z test with a Type I error probability of a, two-tailed, and wants the test to have Type II error probability P, equivalent to statistical power of 1 — P, then the appropriate sample size is given by a formula based on a normal approximation to the binomial distribution. N = • Suppose p = .5. To achieve power of 90% when a = .05, the necessary sample sizes per group are indicated in Table 1. It could be quite difficult to find enough humans with each of the rare alleles. In this regard, it appears the authors are on the right track by seeking commonalities in terms of domain of action of various alleles, where samples could reasonably be pooled across similar alleles. In order to test the classification scheme more rigorously, it would help not only to work with adequate samples but also to obtain continuous measures of the phenotypes associated with retinal degeneration, such as age at onset of symptoms, rate of degeneration, and percent of the retina impaired. These dependent variables could then be assessed with multivariate statistics to determine the significance and strength of relationships with region occupied by a mutation (as in sects. 4.2.2 to 4.2.5). This would allow for exceptions, such as the 1289dell7 and 1312del24 mutations, to weaken an interesting relationship without negating it altogether. The other important question about phenotypic consequences of allelic heterogeneity concerns the possible role of epistasis. Two alleles might have quite different effects on one genetic background but appear to differ less radically on another background. Consider the example in Table 2 where hypothetical group means are given in arbitrary units. The allelic difference is twice as large on background B compared to A. A simple formula that provides a normal approximation to the noncentral t distribution is convenient for estimating the necessary sample sizes (Wahlsten 1991). Suppose the standard deviation within a group is o" = 7.5 units and a planned contrast is used to test the difference between the effects of the two alleles without regard to genetic background. To achieve power of 80% when a two- Table 1 (Wahlsten). Sample size to yield power of 90% c Pi P2 N .05 .10 .15 .2 .55 .6 .65 .7 .45 .4 .35 .3 128 32 14 8 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 491 Commentary /Controversies in Neuroscience III Table 2 (Wahlsten). Model of epistatic interaction Background A Background B Allele 1 Allele 2 10 11 15 21 tailed test with a = .05 is used, the investigators must study about 10 patients in each of the four groups. However, to test the hypothesis of interaction between allele and genetic background with the same level of power, the sample size must be 73 patients per group! As a general rule, the necessary sample size is inversely proportional to the square of the size of the effect. Thus, the sample sizes appropriate for studying many kinds of interactions in a serious way are substantially larger than those that investigators commonly employ when looking at the average effects of an allelic difference [see Wahlsten "Insensitivity of the Analysis of Variance to Heredity-Environment Interaction" BBS 13(1) 1990]. The tyranny of numbers appears to render unanswerable the question of epistatic interaction when there are so few families with rare alleles. If we cannot address this question in a serious way, I believe investigators should inform readers that they cannot argue strongly for the reductionistic thesis that each allele or molecular domain codes for a specific clinical syndrome. Given the statistical difficulties of human genetic research, perhaps it might be helpful to assess allelic interactions in animal models of retinal degeneration using transgenic methods to create animals with different rare alleles. Once accomplished, large numbers of retinal degenerate mice could be produced. More answers about cGMP-gated channels pose more questions Theodore G. Wensel and Joseph K. Angleson Verna and Mans McLean Department of Biochemistry, Baylor College of Medicine, Houston, TX 77030. twensel@bcm.tmc.edu Abstract: Our understanding of the molecular properties and cellular role of cGMP-gated channels in outer segments of vertebrate photoreceptors has come from over a decade of studies which have continuously altered and refined ideas about these channels. Further examination of this current view may lead to future surprises and further refine the understanding of cGMP-gated channels. [MOLDAY & HSU] In the decade or so since the first reports of cGMP-gated ion flux in rod outer segments (Carretta & Cavaggioni 1983, Fesenkoetal. 1985), the standard dogma concerning the channels responsible for it has gone through numerous revisions. A comparison of earlier notions with current views summarized by MOLDAY & HSU highlights both the frequent modification of simplified models that have been necessary, and the clues to new breakthroughs that have been revealed, however dimly at the time, by pesky details that failed to fit neatly into these models. Close examination of remaining loose ends of the otherwise tightly wrapped package presented in the target article may provide hints of future surprises about the structure and function of cGMP-gated channels. Examples of the seeds of new discoveries in the imperfect understanding of old ones abound in this story. For example, early reports of cGMP-gated .cation channels in ROS membranes (Caretta & Cavaggioni 1983; Cavaggioni & Sorbi 1981) met with some skepticism, because several features were hard to fit into ideas then-extant of how visual transduction should work. Channels located in disk membranes (Caretta et al. 1979; 492 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 Koch & Kaupp 1985) and releasing Ca 2 + from intradiskal pools in response to increased cGMP were hard to reconcile with light-induced hydrolysis of cGMP (Bitensky et al. 1981) accompanied by Ca 2 + release from outer segments (Gold & Korenbrot 1980; Yoshikami et al. 1980) generally assumed to be due to increased cytoplasmic [Ca 2 + ]. These puzzling features actually reflected important information about the channels and about the organization of photoreceptor membranes. They arose from the facts (appreciated only later) that the plasma membrane channel is highly permeable to Ca 2 + , a property that figures prominently in shaping the light response, and that disk membranes are so strongly attached to plasma membranes that standard preparations of "disks" contained many plasmamembrane-derived vesicles (Molday & Molday 1987). Following the identification of a 63 kDa protein with the channel by protein purification (Cook et al. 1987) and functional expression of cDNA (Kaupp et al. 1989), and the ruling out of rhodopsin or proteins of similar size as likely candidates (Cook et al. 1989; Molday et al. 1990), a simple picture emerged. The simplest model was that the 63 kDa channel formed a multimeric channel with a central pore, similar to the heteropentameric complex formed by other ligand-gated channels, with cGMP-binding sites on each subunit. The channel was exclusively (or nearly so) located in plasma membranes (Cook et al. 1989). Nonetheless, nagging questions remained: Why was the electrophoretic mobility of the purified channel protein so different from that of the heterologously expressed protein? Why was the purified channel almost always accompanied by a 240 kDa spectrin-like protein? Why did the detailed properties (such as sensitivity to 1-cw-diltiazem) of the purified or expressed channel differ from those of the channel in photoreceptor membranes? As detailed in the target article, answers (or at least partial answers) to these questions not only cleared up some discrepancies, but provided important new information about the channels; for example, photoreceptor-specific proteolytic processing and the existence of an additional channel subunit. Additional discrepancies have persisted, and the unfolding of new explanations for them has continued to deepen our understanding of how these channels work. Here then, are some remaining questions whose answers may provide new insight into channel structure and function; the reader may well be able to think of many more: What is the native channel subunit stoichiometry and how it is governed? Is there only one such stoichiometry, or might there be functional channels with different ratios of a to B subunits? What is responsible for the apparent heterogeneity of cGMP binding sites revealed in excised patches from outer segments (Brown & Karpen 1994)? Is there a function for alternative splicing of the B subunit? Might there be additional regulatory subunits (e.g., Ca 2+ -binding proteins) lost in standard purification schemes? Why are some preparations of the channel apparently devoid of the 240 kDa band corresponding to the B subunit, whereas other preparations following the same procedure yield apparently different ratios of a subunit to the 240 kDa band (Cook et al. 1987)? What is the function of the peptide in the 240 kDa band that is not the B subunit? What is the functional role of proteolytic cleavage of the a subunit amino-terminal peptide? Where and when does this processing occur? Why is the purified channel sensitive to 1-ctsdiltiazem when purified on a cGMP affinity column (Hurwitz & Holcombe 1991) but not when ion exchange and/or dye affinity chromatography is used, even though all these preparations presumably include the B subunit? MOLDAY & HSU cite evidence indicating that proteolytic processing is not the deciding factor, but could it be that cGMP binding enhances subunit-subunit interactions or stabilizes a diltiazem-sensitive domain? How is the channel localized so precisely? Does subunit composition or processing affect the targeting of channels to outer versus inner segments? How do subunit compositions of Commentary /Controversies in Neuroscience III rod and cone channels compare, and what is the molecular nature of cCMP-gated channels recently reported in cone inner segments and synaptic processes (Rieke & Schwartz 1994).? Is there an in vivo role for modulation by calcium binding proteins or phosphorylation? Is the rather small effect of calmodulin on cGMP sensitivity the predominant effect, and if so, why do calmodulin inhibitors appear to have little effect on rod responses (Cray-Keller & Detwiler 1994)? Are there additional calcium binding proteins that can modulate channel function, as has been suggested (Gordon & Zimmerman 1994). What, if any, role does phosphorylation play? Why do phosphatase inhibitors show significant effects in excised patch experiments (Gordon et al. 1992), but not in dialyzed rods (Gray-Keller & Detwiler 1994)? Answers to these questions may well result in more surprises which may lead to yet more questions about cGMP-gated channels. Cyclic nucleotides as regulators of lightadaptation in photoreceptors Barry M. Willardson, Tatsuro Yoshida, and Mark W. Bitensky Biophysics Group, Physics Division, Los Alamos National Laboratory, University of California, Los Alamos, NM 87545. tatsuro@beta.lanl.gov Abstract: Cyclic nucleotides can regulate the sensitivity of retinal rods to light through phosducin. The phosphorylation state of phosducin determines the amount of G, available for activation by Rho*. Phosducin phosphorylation is regulated by cyclic nucleotides through their activation of cAMP-dependent protein kinase. The regulation of phosphodiusteru.se activity by the noncatalytic cGMP binding sites as well as Ca 2+ /calmodulin dependent regulation of cGMP binding to the cation channel are also discussed. [BOWNDS & ARSHAVSKY; HURLEY; MOLDAY & HSu] T h e target articles by HURLEY, BOWNDS & ARSHAVSKY, and MOLDAY & HSU address various aspects of light adaptation in photoreceptors. They reflect a time-honored tradition in vision research to understand the electrophysiological responses that underlie rod signal transduction in terms of the interactions of their highly specialized gene products. The research summarized in these articles continues to elucidate a light adaptation scheme that is orchestrated by the second messengers, cGMP and Ca 2 + , and appears to be regulated at practically every locus of the light activation cascade. f. Phosducin. A fundamental mechanism of signal down regulation mentioned only in passing by BOWNDS & ARSHAVSKY is the regulation of transducin activation by phosducin. Phosducin can compete with G,a for binding to C,P"y (Lee et al. 1992; Yoshida et al 1994). The ability of phosducin to compete with C,a for binding sites on Gfiy depends directly on the phosphorylation state of phosducin (Yoshida et al. 1994). When phosducin was phosphorylated with PKA, its ability to compete with G,a for a binding site on G,P"y was dramatically reduced. This is an important point because, in rods, phosducin is found in its phosphorylated state in the dark and becomes dephosphorylated upon illumination (Lee etal. 1984). Moreover, the endogenous kinase appears to be PKA (Lee et al. 1990), and rod PKA may be activated by p-M concentrations of cGMP. From these data, a mechanism can be postulated in which phosducin regulates the concentration of G,a(3-y in a light dependent manner: in unilluminatcd rods, cyclic nucleotide levels are elevated, PKA is active, and phosducin remains phosphorylated. Under these conditions, phospho-phosducin would not compete with G,a for G,p7 binding, allowing G t a and Gfiy to be maximally associated in their activatable heterotrimeric state. After illumination, cyclic nucleotide levels fall, PKA is inactivated, and phosducin becomes dephosphorylated by unopposed rod phosphatases. Dephospho-phosducin avidly binds to G,P"y and prevents G t a from associating with G,P7. The rate of activation of G,a by Rho* is linearly dependent upon free [C,p-y]. Thus when dephospho-phosducin sequesters Gfiy, only a fraction of the G t a normally activated during the lifetime of a Rho* will be activated in the light-adapted, phosducin-activated state. In this manner, phosducin sequestration of Gfiy could contribute to the smaller response amplitude and shorter response times that are widely observed in light-adapted photoreceptors. Following a light response, cyclic nucleotides eventually return to dark levels and PKA reactivates. Phosducin is subsequently phosphorylated even while complexed to Gfiy (Willardson et al., unpublished observations). Under these conditions, G t a can displace phospho-phosducin from GtP"y, restoring the rod's full compliment of G t aP7, and the system is returned to its high sensitivity state. A similar mechanism of phosducin regulation is also observed in other signaling pathways. Bauer and others (Bauer etal. 1992) have shown that phosducin isolated from bovine brain also inhibits p-adrenergic receptor activation of adenylyl cyclase in A431 cells, and phosphorylation of phosducin with PKA reverses this inhibition. Moreover, they demonstrated that phosducin inhibits the GTPase activity of G s , G,, and Go in a phosphorylation dependent manner. Phosducin may also regulate direct activation of effector enzymes by GP"y subunits. Stimulation of P-adrenergic receptor kinase activity by Gp-y is inhibited by phosducin in a phosphorylation dependent manner (Hekman et al. 1994). All of the findings mentioned above suggest that phosducin plays an important role in downregulating signaling in retinal rods as well as in other G-protein mediated signaling pathways. 2. Noncatalytic cGMP binding sites on PDE. Since their discovery over a decade ago, the noncatalytic binding sites on PDE have been an enigma, BOWNDS & ARSHAVSKY have now de- scribed an attractive mechanism for limiting phosphodiesterase (PDE) activity in amphibian rods by an actual sensing of [cGMP] by the noncatalytic cGMP binding sites on PDE. In assessing the function of the noncatalytic cGMP binding sites on PDE regulation, the rate of release of cGMP from the noncatalytic sites is of critical importance. Depending on this rate, the noncatalytic sites may participate in light adaptation, as suggested by Bownds & Arshavsky, and/or in turning off PDE activity within a single light activation cycle. Recently, Cote and others (Cote et al. 1994) demonstrated that the activation of PDE by G,orGTP"yS accelerates the dissociation of cGMP from the non-catalytic sites. One component of the dissociation process was rapid, with a kll2 of 25 sec, whereas the other component was as slow as that from inactivated PDE with a kl/2 of 4.4 min. The faster rate is comparable with light-adaptation of rods under background illumination, as suggested by Bownds & Arshavsky. Although this rate is too slow for the noncatalytic sites to play an active role in PDE turn-off within a single light response, there are two reasons to believe that the dissociation rate of cGMP from the noncatalytic sites during activation would be more rapid in vivo. First, in preparing disrupted rod outer segments, a significant portion of G, is lost. Moreover, when G,ct is activated by light and GTP-yS, much of the G,ct GTP-yS is released to the supernatant. As a result, only a subset of PDE may be activated. In fact, the fraction of PDE with the slower k1/2 may simply be PDE that was not activated by G^GTP'yS. Second, the method of exchanging prebound [ 3 H]cGMP with lmM cold cGMP to obtain the dissociation rate assures that both noncatalytic cGMP binding sits remain occupied; that is, when [3H]cGMP dissociates from one noncatalytic site, the site is immediately rebound by a cold cGMP. Thus, the two kU2 values are valid only if the two sites are independent. However, if they have cooperative interactions, the ko!r for the second site could be masked by this method. In the physiological situation, both noncatalytic sites can release cGMP as its concentration drops during PDE activation. Once the first site is emptied, release of BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 493 Commentary /Controversies in Neuroscience III cGMP from the second site could be accelerated. Thus, the data of Bownds & Arshavsky do not yet rule out the possibility that the in vivo release of cGMP from the noncatalytic sites on PDE is sufficiently rapid to participate in the turn-off of PDE within a single light response, in the ~ 5 second time scale. BOWNDS & ARSHAVSKY point out that bovine PDE binds cGMP much more tightly than does amphibian PDE, suggesting that the same mechanism of cGMP release from noncatalytic sites may not regulate the Gtor GTPase activity and the lifetime of active PDE in mammals. Indeed, the mechanism of acceleration of G,a GTPase by PDE in bovine rods appears to be very different. The minimal subunit ensemble required for G t a GTPase acceleration is PDEaP, and no G,a-GTPase acceleration was observed in the presence of PDE-y (Antonny et al. 1993; Pages et al. 1993). Moreover, GTPase acceleration was not effected by adding cGMP, presumably because cGMP remains tightly bound to bovine PDE. It appears that the cGMP regulation of light adaptation in bovine rods may not be mediated by the PDE noncatalytic sites but is mediated through cyclic nucleotide effects on phosducin phosphorylation and perhaps other yet to be discovered mechanisms. 3. Ca2 * /calmodulin regulation of the cGMP-gated channel. In the section by MOLDAY & HSU, the most compelling issue (aside from the ultimate definition of subunit composition and topology) is the significance of calcium regulation. At first blush, the significance might appear questionable since the divergence of the plus and minus calcium/calmodulin curves is most apparent at concentrations above 10 u.M. Although there is a fragile consensus that the average free concentrations of cGMP may never exceed 5 u-M; in fact, the actual concentrations of cGMP in the vicinity of the channels are not yet precisely known. It is noteworthy that the effects of calcium and calmodulin on the affinity of cGMP in rod cation channels has persisted in a highly conserved manner over eons of evolutionary time. In view of the ruthlessness of mutational drift in protein domains without function, it appears most prudent to assume that the calcium/calmodulin regulation of the cGMP-gated rod channel has physiological meaning until proven otherwise. Is the lifetime of light-stimulated cGMP phosphodiesterase regulated by recoverin through its regulation of rhodopsin phosphorylation? Akio Yamazaki Kresge Eye Institute, Department of Ophthalmology and Pharmacology, Wayne State University School of Medicine, Detroit, Ml 48201 Abstract: In the current model of visual transduction, the lifetime of active cGMP phosphodiesterase depends upon the period of its interaction with GTP-bound transducin. If recoverin regulates the lifetime of light-activated cGMP phosphodiesterase through inhibition of rhodopsin phosphorylation, rhodopsin should directly interact with cGMP phosphodiesterase and/or GTP-bound transducin complexed with cGMP phosphodiesterase. Is this true? [HURLEY] In vertebrate rod photoreceptors, illuminated rhodopsin stimulates GTP/GDP exchange on Ta (a subunit of 494 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 transducin) followed by dissociation of G T P T a from T(J-y (p-y subunit of transducin). GTP T a activates PDE (cGMP phosphodiesterase) through release or displacement of Fy (inhibitory subunit of PDE) from P a P (catalytic subunit of PDE), resulting in a fall in cytoplasmic cGMP concentration, closure of cGMPregulated channels, and hyperpolarization of photoreceptor plasma membranes. Therefore, illumination of rhodopsin is a trigger for the PDE activation through stimulation of GTP/GDP exchange on T a ; however, illuminated rhodopsin is not directly involved for the PDE activation (Fung et al. 1981). As reviewed concisely in HURLEY'S article, Kawamura and coworkers (Kawamura 1993; Kawamura & Murakami 1991; Kawamura et al. 1993) proposed that recoverin regulates the lifetime of active PDE through its regulation of rhodopsin phosphorylation. They claimed that recoverin (S-modulin) lengthens the lifetime of active PDE at high Ca 2 + concentration (half-maximum effects at 200-400 nM), and that recoverin (Smodulin) inhibits (partially) rhodopsin phosphorylation (half maximum effect at 100 nM Ca 2 + ). Their claim appears to be supported by recent data showing that recoverin forms a complex with rhodopsin kinase (Chen & Hurley 1994; Subbaraya et al. 1994). However, their proposal comprises several questions: (1) how the inhibitory effect of recoverin on rhodopsin phosphorylation is connected to the longer lifetime of active PDE, and (2) how much the partial inhibition of rhodopsin phosphorylation can regulate the PDE activity under low bleaching of rhodopsin. In the current model of visual transduction, inhibition of rhodopsin phosphorylation lengthens the lifetime of active rhodopsin (meta II), resulting in the increase of number of G T P T a . The increase in the number of G T P T a is related to amplification of light signal through increase in the total number of active PDE. In the process of PDE activation, GTP Ta interacts with P 7 to release the inhibitory effect of P 7 from P a P and the molecular ratio between G T P T a and Fy in the interaction is 1:1 (Fung & Griswold-Prenner 1989; Yamazaki et al. 1990). The lifetime of active PDE ( P a 3 or P a P 7 ), therefore, is dependent upon the period in which G T P T a interacts with P , but not upon the number of GTP T a if the concentration of GTP T a is enough for its interaction with PDE subunits. Therefore, without solid data, it is difficult for me to accept that the lifetime of active rhodopsin is related to the lifetime of active PDE. It should be emphasized that the lifetime of active rhodopsin is not related to the lifetime of GTP TQ since GTP T a is believed to be released from active rhodopsin after GTP/GDP exchange on T a (Kiihn 1980). Moreover, the regulation of lifetime of GTP T a by GTP hydrolysis may not be important for the turnoff of active PDE (Erickson et al. 1991; Tsuboi et al. 1994). If recoverin is functional for prolongation of the lifetime of active PDE, recoverin should regulate the interaction between G T P T a and P r In future, we may find a new regulatory mechanism in which recoverin regulates the lifetime of active PDE through its regulation of the interaction between P^ and G T P T a . In such a case, change in Ca concentration may also be crucial for the PDE regulation. I strongly feel that the role of recoverin in visual signal transduction is still unclear. ACKNOWLEDGMENT The author is supported by Jules and Doris Stein Professorship from Research to Prevent Blindness. Response/Controversies in Neuroscience III Authors' Responses Distribution of commentators responded to in Authors' Responses HARCRAVE: BOWNDS & ARSHAVSKY: HURLEY: Future directions for rhodopsin structure and function studies How many light adaptation mechanisms are there? Recoverin, a calcium-binding protein in photoreceptors Albert & Yeagle Crouch & Corson Dratz Garavito Smith Crouch & Corson Hurwitz, Srivastava & Hurwitz Koch Matthews & Fain McCinnis Schnetkamp Willardson, Yoshida & Bitensky Kawamura Koch McGinnis Folans & Adamus Sanada & Fukada Takamatsu Yamazaki XIA & STORM: MOLDAY & HSU: DAICER, SULLIVAN & RODRIGUEZ: Evidence that the type I adenylyl cyclase may be important for neuroplasticity Further insight into the structural and regulatory properties of the cGMP-gated channel Genetic and functional complexity of inherited retinal degeneration Abrams Heideman Rasenick Roberson & Sweatt Brown & Karpen Gray-Keller & Detwiler Haynes Hurwitz, Srivastava & Hurwitz Sagoo & Lagnado Oprian Wensel & Angelson Barnstable Bergen Kaplan McGinnis Tamai Wahlsten Authors' Responses Future directions for rhodopsin structure and function studies Paul A. Hargrave Departments of Ophthalmology, Biochemistry and Molecular Biology, University of Florida, Gainesville, FL 32610. hargrave@eyel.eye.ufl.edu Abstract: NMR (nuclear magnetic resonance) may be useful for determining the structure of retinal and its environment in rhodopsin, but not for determining the complete protein structure. Aggregation and low yield of fragments of rhodopsin may make them difficult to study by NMR. A long-term multidisciplinary attack on rhodopsin structure is required. A variety of structural methods, in combination, is leading to a progressively clearer picture of the threedimensional structure of visual rhodopsin. The most informative to date has been cryoelectron microscopy of two-dimensional crystals of frog rhodopsin (Schertler & Hargrave 1995). This has resulted in a projection map of frog rhodopsin to 6A resolution. More recently, using tilted images with the same crystal form, a projection map has been obtained that resolves each of rhodopsin's seven helices (Unger et al. 1995). This represents the first physical evidence for the reality of the seven transmembrane helices of rhodopsin that have been reasonably proposed to exist for over a decade. In the target article I essentially dismissed nuclear magnetic resonance (NMR) as a currently applicable approach to providing us with complete structural infor- mation for rhodopsin. Smith points out, quite correctly, that solid state magic angle spinning NMR is an important technique for providing selected structural information about rhodopsin, such as the structure and protein environment of the retinal chromophore. It is this kind of information that is most sought in order to understand functional properties of the visual pigments. NMR methods may well provide us with high resolution structural information for the retinal environment in a more timely fashion than will other methods. Crouch & Corson suggest that multi-dimensional het^ eronuclear NMR may in fact have the potential for investigating the structure of proteins as large as the 40 kDa rhodopsin. Although the potential may be there, the reality at present is that only proteins in the 15-25 kDa range can be examined in solution (Clore & Gronenborn 1994). Although 40 kDa is a foreseeable frontier, only an exceptional protein may meet the necessary criteria. The protein would have to be highly soluble and not aggregate at concentrations of 40 mg/ml, would have to be stable to temperatures of over 30° for several weeks, must have little conformational heterogeneity, and must be available in quantity uniformly labelled with 15N and l3C (Clore & Gronenborn 1994). Whether all of these criteria will be met by rhodopsin, and whether advances in NMR methodologies will allow the structures of detergentsolubilized 40 kDa proteins to be determined, should be known during the coming decade. Albert & Yeagle suggest that large proteolytic fragments of rhodopsin that are within the size range of current NMR capabilities to examine might be suitable objects of study to learn about the structure of rhodopsin. This would certainly be the case if the isolated fragments BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 495 Response/Controversies in Neuroscience III retained the structure that they have in the seven-helix bundle of rhodopsin, and if they could be prepared in quantity in concentrated solution in nonaggregated form. Since the first two-fragment complex of thermolysindigested rhodopsin was prepared and separated on an analytical scale (Pober & Stryer 1975), attempts have been made to use similar methods to prepare significant amounts of nondenatured fragments for spectral analysis. Similar attempts in our laboratory led to aggregated fragments in yields of less than 10% (Hargrave & Fong 1977), and we are not aware of other reports in the literature. However, rhodopsin fragments have been coexpressed in E. coli recently (Ridge et al. 1995). It may eventually be possible to find conditions in which some fragments of rhodopsin may be expressed, purified, and concentrated in "native" monomeric form, suitable for structural examination by NMR. Finally, both Dratz and Garavito stress that the difficulty of the task of crystallization of rhodopsin makes it a commitment that requires long-term support. Additionally, it may require the resources and expertise of several research groups (i.e., one lab that knows how to purify and handle the protein, one lab that makes antibodies against it or synthesizes specialized retinal derivatives, and one lab that specializes in X-ray analysis). Progress will probably be incremental over a period of years, as it has been for bacteriorhodopsin. The results will be important not only to those of us who are intrinsically interested in rhodopsin itself, but will be closely followed by the large community who study G-proteinlinked receptors. How many light adaptation mechanisms are there? M. Deric Bownds and Vadim Y. Arshavsky1 Laboratory of Molecular Biology, University of Wisconsin, Madison Wl 53706. mdbownds@facstaff.wisc.edu and vadlm@macc.wisc.edu. Abstract: The generally positive response to our target article indicates that most of the commentators accept our contention that light adaptation consists of multiple and possibly redundant mechanisms. The commentaries fall into three general categories. Thefirstdeals with putative mechanisms that we chose not to emphasize. The second is a more extended discussion of the role of calcium in adaptation. Finally, additional aspects of cGMP involvement in adaptation are considered. We discuss each of these points in turn. The commentators made a number of interesting points, and we are gratified that in general the main points in our target article seem to have passed muster. We are grateful to Drs. R. Cote, A. Dizhoor, N. Mangini, and D. Pepperberg for discussing with us different points raised by the commentators. Our Response can be organized under the headings shown below: R1. Mechanisms that we did not emphasize. Crouch & Corson give a more thorough and useful description of bleaching adaptation than we provided in the target article. We suggested similarities between bleaching and background adaptation but did not want to imply that the issue was resolved. We favor the recent suggestion of Leibrock et al. (1994) that "a bleach induces two closely related phenomena in rods: a process indistinguishable 496 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 from real light, and another process which desensitizes and accelerates the response and causes steady current suppression but which does not cause photon-like events. These results can be explained ifbleached forms of rhodopsin lead both the generation of R* at a low rate and to the direct activation of G-protein with low probability." This model does not appear to be in conflict with most of the observations cited by Crouch & Corson, which, as they point out, are not easy to explain by a simple single mechanism. Both bleaching and background illumination might derive from a discrete collection of mechanisms. The relative contribution of each of these mechanisms during bleaching and background adaptation may be different. Hurwitz et al. appropriately point out that not only calcium but also possibly magnesium concentration inside the photoreceptor cell changes upon illumination. The issue is whether magnesium levels ever actually decrease sufficiently to limit the amount of this ion available as an enzyme cofactor. It would be most interesting to obtain direct measurements of light-induced changes in free magnesium concentration. If approximately 5% of the inward dark current is carried by magnesium, this would correspond to an inward flux on the order of 1-2 u,M/sec compared with dark-free levels of several hundred u,M. A cessation of the inward flux, with no adjustment of magnesium efflux, would be expected to cause a relatively slow change in intracellular magnesium, on a time scale of minutes rather than seconds. These kinetics are substantially slower than the decrease in intracellular calcium known to regulate rapid aspects of light adaptation. Free calcium concentration in the dark is only ~550 nM (GrayKeller & Detwiler 1994) and the dark influx blocked by channel closing is ~3-fold greater than the influx of magnesium (Yau & Baylor 1989). This would appear to restrict magnesium's role to slower aspects of light adaptation. McGinnis mentions light-induced migration of arrestin from inner to outer segments and a reverse movement of transducin. Thus there might be an increase in arrestin concentration in the light adapted outer segment, leading to more rapid quenching of excited rhodopsin. We think it more likely that the concentration of arrestin decreases on illumination and that the movement of arrestin noted by McGinnis reflects an increase in the amount of bound rather than free arrestin in the outer segment. Mangini et al. (1994) have convincingly shown that arrestin's movement can be explained without an active mechanism, but rather could be the simple consequence of the appearance of arrestin binding sites in the outer segment as rhodopsin is bleached and phosphorylated. The movement is abolished if hydroxylamine is used to remove the arrestin binding site. R2. Is Ca sufficient to account for adaptation? Matthews & Fain seem to be alone among the reviewers in their determination to exclude internal messengers other than calcium from playing a role in light adaptation. By pushing such a restrictive agenda, they set themselves a difficult task. While the emphasis on "necessary" and "sufficient" reflects a proper biophysical reductionist stance, we would suggest that biological cells care very little about what is sufficient and often equip themselves with multiple and/or redundant mechanisms. We should emphasize that we completely agree that calcium changes are a necessary and possibly sufficient explanation for Response/Controversies in Neuroscience III background adaptation at dim light intensities. However, Matthews & Fain do not contest the references in section 9, last paragraph, listing dis-correlations between light adaptation and calcium changes. One more argument can be offered for the existence of calcium-independent mechanisms, based on recent direct measurements of the decrease in cytoplasmic calcium that follows saturating illumination (Gray-Keller & Detwiler 1994; McCarthy et al. 1994). If a continuous saturating light is applied to frog or gecko rod photoreceptors the calcium decrease is substantially complete after 10 seconds. However, saturating lights having durations longer than this cause increasing shortening of recovery kinetics for periods up to 50 seconds (cf. Cervetto et al. 1984; Coles & Yamane 1975). This longer period corresponds more closely to the dissociation time of cGMP from the noncatalytic binding sites of PDE, a dissociation that we suggested (sect. 7, para. 7) might be responsible for accelerating response kinetics. It is hard for us to imagine that a mechanism with such detailed correspondence to response kinetics, which operates over the physiological range of cGMP concentration changes, is a fortuitous artifact. Our idea is that cGMP feedback on the lifetime of PDE might be particularly important at bright light levels, in receptors that can operate over a broad range of background light intensity changes. Direct measurements of bound cGMP levels and conductance at high background illumination would be required to prove the point. Both Schnetkamp and Koch make further points that are relevant to calcium homeostasis. Schnetkamp interprets some of his data to suggest that inactivation of the calcium efflux mechanism occurs when the level of free calcium is reduced to ~50 nM. This agrees with recent data from Gray-Keller and Detwiler (1994), who have shown that free calcium in gecko photoreceptors does not fall below ~50 nM even at light levels saturating the photoresponse. Koch's calculations yield the same basic conclusion as ours (see sect. 10, para. 4) namely, that at moderate levels of illumination guanylate cyclase activation may underlie light adaptation. We find ourselves in disagreement, however, with details of Koch's numbers. In section 8 we discuss reasons for setting the dark cGMP flux at approximately 4 (xM/sec, representing a turnover of the entire free cGMP pool every second. All of the available electrophysiological (Cornwall & Fain 1994) and biochemical data (Dawis et al. 1988; Koch & Stryer 1988) argue that saturating light causes no more than a 10-fold increase in cyclase activity. Thus, a cGMP flux level of —40 jjiM/sec represents a reasonable estimate of the amount of PDE activity that could be countered by cyclase. Koch's estimate of guanylate cyclase activity as 600 n-M/sec at intermediate illumination mystifies us. We think that he has erred by overestimating the amount of cyclase in the cell. R3. Further comments on the possible involvement of cGMP in adaptation. Willardson et al. make some very thoughtful comments. They propose that regulation of transducin activation by phosducin may be a mechanism of signal down regulation. In detail, the light-dependent binding of transducin (3,7 to the unphosphorylated form of phosducin is meant to remove it from circulation. This decreases the rate of activation of transducin's a subunit because that activation requires the participation of the (3,7 subunits. The suggestion is that this would both cause a smaller response amplitude and shorter response times. While we agree that a smaller response amplitude may result from such a mechanism, response times are not a function of the amount of available transducin, but rather of the lifetimes of the activated components. We see two major problems with depletion argument: (1) Fung (1983) has shown in a reconstituted system that the maximum rate of transduction a subunit activation requires only about 10% of the total (3,7 subunit pool. Current estimates of phosducin levels in the cell suggest the presence of less phosducin than transducin. Therefore, thereshouldalways be an excess of (3,7 to support recycling of the transducin a subunit. What we need to know is how much (3,7 is required in a real cell. (2) The recent report of Kutuzov and Pfister (1994) that the GDP-loaded a-subunit of transducin can activate PDE raises another suggestion. The removal of (3,7 might actually increase background PDE activity. How does this fit into the down-regulation model? A second major point of Willardson et al. is that the kinetics of cGMP dissociation from PDE noncatalytic sites may be influenced by the technique used to measure them. Namely, the displacement of radioactive bound cGMP by nonradioactive cGMP may affect cGMP dissociation from the second site if dissociation from this site is dependent upon occupancy of the first. Recently we have developed procedures for measuring cGMP dissociation without cGMP re-binding, and find that the kinetics are essentially the same. We agree with the argument that the biphasic behavior described by Cote et al. (1994) may be a consequence of incomplete activation of the PDE in the preparation used. Our more recent measurements indicate that under conditions where the whole PDE pool is activated the dissociation is better described by a single exponential process that is at least four times more rapid than dissociation from inactive PDE. A last point raised by Willardson et al. considers the difference between amphibian and mammalian mechanisms regulating transducin GTPase activity. We believe that frog and bovine systems differ in the affinity of PDE noncatalytic sites for cGMP, but not in the fundamental mechanism of transducin GTPase acceleration. The 7 subunit of PDE catalyzes the acceleration in both cases (Angleson & Wensel 1994; Arshavsky et al. 1994). However, in both cases it requires an additional membraneassociated factor for its manifestation. This is why Antonny et al. (1993) did not note the acceleration in a reconstituted system lacking membranes. Arshavsky et al. (1994) provide a detailed critique that outlines why Pages et al. (1993) as well as Yamazaki et al. (1993) failed to observe the acceleration. NOTE 1. Arshavsky's present address is Howe Laboratories, Harvard Medical School and the Massachusetts Eye and Ear Infirmary, 243 Charles St., Boston, MA 02114. Recoverin, a calcium-binding protein in photoreceptors James B. Hurley Department of Biochemistry and Howard Hughes Medical Institute, SL-1S University of Washington, Seattle, WA 98195. jbhhh@11.washlngton.edu BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 497 Response/Controversies in Neuroscience III Abstract: Recoverin is a Ca2+-binding protein found primarily in vertebrate photoreceptors. The proposed physiological function of recoverin is based on the finding that recoverin inhibits light-stimulated phosphorylation of rhodopsin. Recoverin interacts with rod outer segment membranes in a Ca2+-dependent manner. This interaction requires N-terminal acylation of recoverin. Four types of fatty acids have been detected on the Nterminus of recoverin, but the functional significance of this heterogeneous acylation is not yet clear. The commentaries call attention to several important points I did not stress in my target article on recoverin. The alternative perspectives in the commentaries provide a balanced view of our current understanding of recoverin and its homologues. Since the functions of recoverin in photoreceptors are not yet completely understood, it is very important that the reader consider these different perspectives because only further experimentation and testing of alternative hypotheses can provide a more complete understanding of this protein and its roles in photoreceptor cells. Kawamura, who was the first to correctly identify an important function of recoverin, reviewed the evidence that recoverin/S-modulin regulates stimulation of phosphodiesterase and the rate of rhodopsin phosphorylation. A very important point raised by Kawamura that has not yet been thoroughly investigated is that recoverin appears to have two separate activities, one on the extent of phosphodiesterase activation, the other on the lifetime of activated phosphodiesterase. Kawamura's framework for thinking about these separate activities should inspire clearer strategies for investigations of the mechanisms of recoverin function. The Ca 2+ -dependence of recoverin's effects on phosphorylation and on membrane binding remain controversial since several laboratories have recently found values for the effect of recoverin on phosphodiesterase activity and rhodopsin phosphorylation that are in the micromolar range (e.g., Ames et al. 1995). Takamatsu's commentary focuses attention on a family of recoverin-related proteins and the possibility that they play important roles in signal transduction in other types of neurons besides photoreceptors. It is clear now that recoverin homologues are capable of regulating rhodopsin kinase, but it is not yet clear whether other receptor kinases are regulated by recoverin or its homologues. For example, my laboratory has found that while recoverin inhibits rhodopsin kinase, it fails to have any effect on the activity of the beta-adrenergic receptor kinase. Further studies are required to determine whether other receptor kinases are regulated by recoverin homologues. Alternatively, recoverin homologues may regulate other types of effectors in the same way that G-proteins regulate a diversity of effectors. The role of heterogeneous acylation is addressed in the commentary by Sanada & Fukada. Fukada and his colleagues were co-discoverers of heterogeneous acylation of transducin and it is clear that they have thought seriously about the function of this unusual protein modification. The differences in proportions of different types of fatty acids on different retinal proteins is interesting, but there is an important variable, cell type, that should be kept in mind. For example, the C subunit of cAMP-dependent protein kinase contains a higher proportion of myristoyl residue than T-alpha or GCAP, but it is important to 498 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 remember that this protein is also present in nonphotoreceptor cells in the retina. T-alpha and GCAP are confined to photoreceptors. (Recoverin, however, is primarily expressed in photoreceptors so the cell-type expression does not explain the high proportion of myristoyl residues in recoverin.) It would also be interesting to compare the distribution of N-terminal fatty acids in rods versus cones within the same retina, but this may not be technically feasible. Concerning differences in the ability of nonacylated recoverin and myristoylated recoverin to inhibit rhodopsin kinase, recent results from my laboratory show that both inhibit rhodopsin kinase, but myristoylated recoverin inhibits at ~ 5 times lower concentration than nonacylated recoverin. The myristoyl residue enhances rhodopsin kinase inhibition, but it is not required. Koch raises the possibility that recoverin may function as a Ca2+-buffer within the photoreceptor cell. The concentration of recoverin appears to be in the appropriate range for such an activity although the unresolved controversy concerning the affinity of recoverin for Ca 2 + makes this interesting hypothesis difficult to evaluate. I certainly agree with Yamazaki's suggestions that we should not mistakenly confine investigations of recoverin function to regulation of rhodopsin phosphorylation. However, my laboratory has done one type of experiment that suggests regulation of rhodopsin phosphorylation is important. When hydroxylamine is used, instead of ATP, to inactivate rhodopsin following a light flash, the rate of phosphodiesterase deactivation in the presence of recoverin is independent of Ca 2 + . Nevertheless, this result certainly does not preclude other regulatory interactions of recoverin and the study of such interactions is very important. Finally, Polans & Adamus describe another important process associated with recoverin, its occasional expression in tumors. When recoverin is expressed in tumors it may cause blindness that can precede the detection of the tumor. The diagnostic implication of this observation is very important in itself. The finding, discussed by both Polans & Adamus and McGinness, that recoverin maps near to a tumor suppressor gene, suggests a possible mechanism for expression of recoverin in certain tumors. It will be very important to investigate the molecular mechanisms responsible for oncogenesis in tumors expressing recoverin. Evidence that the type I adenylyl cyclase may be important for neuroplasticity: Mutant mice deficient in the gene for type I adenylyl cyclase show altered behavior and LTP Zhengui Xia and Daniel R. Storm Department of Pharmacology, University of Washington, Seattle, WA 98195. dstorm@u.washlngton.edu Abstract: The regulatory properties of the neurospecific, type I adenylyl cyclase and its distribution within brain have suggested that this enzyme may be important for neuroplasticity. To address this issue, the murine, Ca2+-stimulated adenylyl cyclase (type I), was inactivated by targeted mutagenesis. Ca2+stimulated adenylyl cyclase activity was reduced 40% to 60% in the hippocampus, neocortex, and cerebellum. Long term potentiation in the CA1 region of the hippocampus from mutants Response/Controversies in Neuroscience III was perturbed relative to controls. Both the initial slope and maxim um extent of changes in synaptic response were reduced. Although mutant mice learned to find a hidden platform normally in the Morris water task, they did not display a preference for the region where the platform had been when it was removed. The behavioral phenotype of these mice is very similar to that exhibited by mice which have been surgically lesioned in the hippocampus. These results indicate that disruption of the gene for the type I adenylyl cyclase produces changes in spatial memory and indicate that the cAMP signal transduction pathway may play an important role for synaptic plasticity. R1. Rasenlck. Detailed information concerning the subcellular localization of the type I adenylyl cyclase and other adenylyl cyclases in neurons is not available because appropriate antibodies have not been isolated. However, we have examined the intracellular sorting of the type I adenylyl cyclase in primary cultured neurons (Choi et al. 1992b). We have transfected primary hippocampal and cerebellar neurons with constructs encoding epitope tagged type I adenylyl cyclase, stained with antibodies against the epitope, and discovered that the enzyme is transported to neurite extremities and is not limited to cell bodies. Whether the enzyme is postsynaptic, presynaptic, or both has not been determined. We were also surprised to discover that the type I adenylyl cyclase is not stimulated by Gs coupled receptors in vivo since the purified enzyme is stimulated by addition of purified Gs-alpha (Tang etal. 1991). Treatment of recombinant I-AC with combinations of Ca 2+ /CaM and activated Gs-alpha in vitro produced additive stimulations that were not significantly synergistic (Tang et al 1991). We examined the sensitivity of the type I adenylyl cyclase expressed in HEK-293 cells to beta-adrenergic agonists or glucagon when intracellular Ca 2+ was elevated by Ca 2+ ionophore or carbachol (Wayman et al. 1994). Although previous studies have shown that this enzyme can be directly stimulated by activated Gs in vitro, we demonstrated that it is not stimulated by Gs coupled receptors in vivo. We have also shown that the type I adenylyl cyclase is not stimulated by beta-adrenergic agonists in cultured neurons. However, the enzyme was stimulated by Gs coupled receptors in vivo when it was activated by intracellular Ca 2+ . For example, the Ca 2+ ionophore A23187 stimulated the enzyme 3 ± 0.5 fold (n = 9), isoproterenol alone did not stimulate the enzyme, but the combination of the two stimulated type I adenylyl cyclase 13 ± 2 fold (n = 9) in vivo. Similarly, 500 nM glucagon alone did not stimulate the enzyme but the combination of A23187 and glucagon activated the enzyme 90 ± 8 fold (n = 4). Synergistic stimulation of type I adenylyl cyclase activity was also obtained with combinations of carbachol and isoproterenol or glucagon. This phenomenon was not observed with a mutant enzyme that is insensitive to Ca 2+ and CaM suggesting that conformational changes caused by binding of CaM to the type I adenylyl cyclase enhance binding or coupling to activated Gs. These data illustrated that this adenylyl cyclase can couple Ca 2+ and neurotransmitter signals to generate optimal cAMP levels, a property of the enzyme that may be important for its role in learning and memory in mammals. The studies by Tang et al. (1991) clearly established that the type I adenylyl cyclase has a domain for interaction with Gs but it is important to note that relatively high concentrations of recombinant Gs were used to demonstrate Gs activation of type I adenylyl cyclase in vitro. Furthermore, these studies generally used Gs-alpha complexed to nonhydrolizable analogous of GTP. Our data emphasizes the importance of determining the regulatory properties of each adenylyl cyclase in vivo. An analysis of the Ca 2+ sensitivity of adenylyl cyclase activity in brains membranes is complicated by the existence of several different Ca 2+ sensitive adenylyl cyclases in the same regions of brain. For example, the type I and VIII adenylyl cyclases are both stimulated by Ca 2+ in vivo, but the type I enzyme shows greater Ca 2+ sensitivity and the type VIII enzyme is not synergistically stimulated by Ca 2+ and Gs coupled receptors in vivo. R2. Roberson & Sweatt. As emphasized in the target article, we believe that data from Sweatt's lab showing that LTP and NMDA activation both stimulate cAMP levels in the hippocampus are key observations supporting the role of cAMP in synaptic plasticity. Is the interaction with PKC pre-synaptic or postsynaptic? It has not been established whether type I adenylyl cyclase is pre-synaptic or postsynaptic. Neuromodulin is enriched in axons but we do not feel the protein is exclusively presynaptic. We agree that neurogranin may also play a role in controlling the levels of free CaM postsynaptically. This idea is strengthened by unpublished data from Sweatt's laboratory showing that neurogranin is phosphorylated in response to LTP. The type 1 adenylyl cyclase mutant mice. Our data with the type I deficient mice provided the first evidence that a specific adenylyl cyclase contributes to full development of LTP and may be important for spatial memory in vertebrates (Wu et al. 1995). Our data did not distinguish between effects on long-term and short-term learning. It is notable that the mutant mice did express normal L-LTP but showed a dampened response in the early stages of CA1 LTP. Since the mutant mice still expressed significant levels of Ca 2+ stimulated adenylyl cyclase activity in the hippocampus that may be attributable to type VIII adenylyl cyclase, mice deficient in Type VIII adenylyl cyclase and double mutants will be required to access the role of Ca 2+ stimulated adenylyl cyclase activity for synaptic plasticity. R3. Abrams. As emphasized by Abrams and the target article, the Aplasia system has been a very important tool for demonstrating that adenylyl cyclase activity may play an important role in associative learning. It is also important to note that there are important regulatory differences between the type I adenylyl cyclase and the Aplasia enzyme. Furthermore, the Aplasia adenylyl cyclases have not been cloned and their relationship to the mammalian adenylyl cyclases awaits further purification and cloning studies. R4. Heideman. The type I adenylyl cyclase mutant mice show a 50% reduction in Ca 2+ sensitive adenylyl cyclase activity in the cerebellum and hippocampus. The residual adenylyl cyclase activity in the mutant mice is also less sensitive to Ca 2+ than the wild type mice. It seems likely that the type VIII adenylyl cyclase contributes to the residual Ca 2+ sensitive adenylyl cyclase activity in mutant BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 499 Response/Controversies in Neuroscience III mice. The type I and VIII adenylyl cyclase are, however, not redundant activities and their regulatory properties are quite distinct. Mice deficient in type I adenylyl cyclase show a major deficiency in spatial memory compared to the wild type mice, indicating that this specific adenylyl cyclase may be important for some forms of memory. Further insight into the structural and regulatory properties of the cGMP-gated channel Robert S. Molday and Yi-Te Hsu Department of Biochemistry and Molecular Biology, University of British Columbia, Vancouver, B.C., Canada V6T 1Z3. molday@unixq.ubc.ca Abstract: Recent studies from several different laboratories have provided further insight into structure-function relationships of cyclic nucleotide-gated channel and in particular the cCMPgated channel of rod photoreceptors. Site-directed mutagenesis and rod-olfactory chimeria constructs have defined important amino acids and peptide segments of the channel that are important in ion blockage, ligand specificity, and gating properties. Molecular cloning studies have indicated that cyclic nucleotide-gated channels consist of two subunits that are required to reproduce the properties of the native channels. Biochemical analysis of the cGMP-gated channel of rodcells have indicated that the 240 kDa protein that co-purifies with the 63 kDa channel subunit contains both the previously cloned second subunit of the channel and a glutamic acid-rich protein. The regulatory properties of the cGMP-gated channel from rod cells has also been studied in more detail. Studies indicate that the beta subunit of the cGMP-gated channel of rod cells contains the binding site for calmodulin. Interaction of calmodulin with the channel alters the apparent affinity of the channel for cGMP in all in vitro systems that have been studied. The significance of these recent studies are discussed in relation to the commentaries on the target article. R1. Structural features of the rod cGMP-gated channel. The commentaries of Oprian, Haynes, Hurwitz et al. and Wenzel & Angleson correctly point to the fact that although significant progress has been made on analysis of the structural features and molecular properties of the cGMP-gated channel of photoreceptor cells, many unresolved issues remain. Indeed, since the target article was written, a number of important new studies have been carried out in several different laboratories that deal with some of these unresolved issues. Site-directed mutants have been used to identify the glutamic acid residue within the putative pore segment as the divalent cation binding site (Eismann et al. 1994; Root & MacKinnon 1993) and rod-olfactory chimeras have been constructed that begin to define regions of the channel that are important in ligand specificity and gating mechanisms (Goulding et al. 1994). These studies help to confirm the general structural features of the cyclic nucleotide-gated channels as outlined in the target article and in a recent review article by Eismann et al. (1993). As indicated in the commentary of Haynes, this type of approach, although extremely valuable, will not in itself provide a refined structure of the channel for a rigorous correlation with functional properties. Ultimately, a high resolution threedimensional structure of the channel, such as can be obtained by X-ray crystallographic methods, will be re500 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 quired. However, in the absence of such structural analysis, molecular biology techniques coupled with biochemical and biophysical studies should continue to provide important new insight into segments of the channel that mediate functional and regulatory properties of these channels. Recent cloning studies have also revealed the presence of cyclic nucleotide-gated cation channels in other tissues. Cyclic nucleotide-gated channels related to the photoreceptor and olfactory channels have been reported to be expressed in testis, kidney, and heart (Biel et al. 1994; Weyand et al. 1994). A cyclic nucleotide-gated channel has also been cloned from Drosophila (Baumann et al. 1994). The physiological role and properties of these channels, however, remain to be investigated in detail. Recent biochemical studies have also led to more insight into the subunit structure of the native rod photoreceptor cGMP-gated channel and have addressed some of the questions raised in the commentaries. As outlined in the target article, biochemical studies have indicated that the purified cGMP-gated channel from rod outer segments contains a 240 kDa polypeptide, as well as the 63 kDa a-subunit. These channel preparations, however, do not contain polypeptides of ~70 kDa or 102 kDa, the predicted size of the P-subunit (Chen et al. 1993). The 240 kDa protein has now been isolated and subjected to peptide sequence analysis. These studies and immunochemical analysis indicate that the 240 kDa polypeptide contains both the P-subunit (or subunit 2) of the channel (Chen et al. 1994) and the bovine glutamic acid rich protein (GARP) as first cloned by Sugimoto et al. (1991). In their commentary, Brown & Karpen have indicated that they have also obtained a peptide sequence corresponding to P-subunit of the channel in photoaffinity labeling of the 240 kDa polypeptide. The primary structure of cGMP binding domains for the a-subunit and P-subunit are similar, but not identical. Sequence differences may account for the heterogeneity in cGMP binding affinity observed in their reported studies. The existence of a glutamic acid rich protein as a component of the 240 kDa protein has been determined both by peptide sequence analysis and by expression cloning using two monoclonal antibodies that react with the 240 kDa protein (Ming et al. 1994). What is not clear from these studies, however, is how this glutamic acidrich protein component is linked to the P-subunit to form the 240 kDa polypeptide. One suggestion is that they are synthesized as individual proteins and that these proteins are subsequently covalently linked together via a posttranslational reaction. This possibility is based on previously reported studies indicating that the glutamic acid rich protein (GARP) is expressed as a protein of apparent molecular mass of 65 kDa from a single 2.4 kB mRNA transcript (Sugimoto et al. 1991) and that the human P-subunit is expressed from one or two transcripts as a protein of 70 or 102 kDa (Chen et al. 1993). This has led to the suggestion that the GARP protein may be a third subunit of the channel (Ming et al. 1994). Alternatively, the 240 kDa protein, may be translated from a single mRNA containing coding regions for both the P-subunit and the glutamic acid rich protein. If this is the case, then some of the earlier studies on the GARP protein and the human P-subunit need to be reinvestigated. Direct experimental information on how the glutamic acid rich Response/Controversies in Neuroscience III protein and the p-subunit component combine to form the 240 kDa protein remains an important area of investigation. On the basis of these recent studies, the earlier view that the photoreceptor cGMP-gated channels, as well as the olfactory channels, is composed of identical subunits has now given way to the conclusion that these cyclic nucleotide-gated channels consist of at least two principal subunits (a and P subunit), both of which are needed to reproduce the electrophysiological properties of the native channels (Bradley et al. 1994; Chen 1993; 1994; Liman & Buck 1994). The role of the GARP protein found in the bovine rod 240 kDa channel polypeptide is not yet understood. Clearly, a lot still remains to be learned about the structure-function relationships of the photoreceptor of cGMP-gated channels and their targeting to outer segment plasma membranes as pointed out in the commentaries. R2. Calmodulin regulation of the cGMP-gated channel. Since the initial observation that calmodulin binds to and modulates the affinity of the rod channel for cGMP in rod outer segment membranes, several laboratories have confirmed this observation in different channel preparations including patches from frog rod outer segments (Gordon et al. 1994), purified and reconstituted channel preparations (Hsu & Molday 1994) and mammalian cells coexpressing both the a and (J-subunits of the channel (Chen et al. 1994). Thus, the effect of calmodulin on channel activity, although relatively small, is consistently observed, and at least in ROS membrane vesicle preparations is inhibited by mastoparan (Hsu & Molday 1994). Calmodulin has also been observed to modulate the activity of the olfactory channel (Chen & Yau 1994). In this case the calmodulin mediated effect is much larger and the binding site has been localized to a region close to the N-terminus of the a-subunit (Liu et al. 1994). The calmodulin binding site on the rod cGMP-gated channel has been localized to the P-subunit (Chen et al. 1994) within the 240 kDa protein. The precise binding site has not yet been reported, but examination of the sequence of the P-subunit of the rod channel indicates that an amphipathic helix predicted to serve as a calmodulin binding motif is present on the C-terminal side of the cGMPbinding domain. Preliminary studies in our laboratories have confirmed that fusion peptides in this region do bind calmodulin in a calcium-dependent manner (unpublished results). The physiological role of calmodulin modulation of the channel in phototransduction, however, remains to be determined. As outlined in the target article, we have proposed that calmodulin modulation of the affinity of the channel for cGMP may play a role in the recovery of the photoreceptor cell to its dark state following bleaching. As indicated in the commentary of Gray-Keller & Detwiler, such an effect would only come into play if the physiological concentration of calcium drops to a value in the range of the Kd for the interaction of Ca-calmodulin with the channel. Their measurements indicate that prolonged bleaching conditions are required to reduce Ca levels to this range. This would suggest that calmodulin is bound to the channel in the dark and remains bound during brief flashes of light under conditions in which only a minimal drop in calcium is realized. The suggestion of Gray-Keller and Detwiler that calmodulin modulation of the channel may occur under prolonged bleaching conditions is plausible and warrants further investigation. Studies reported in the commentary of Sagoo & Lagnado and also by Koutalos et al. (1994) using truncated rod outer segments show an effect of Ca on the cGMP induced current which is consistent with a diffusable mediator. Their studies and those of Gordon et al. (1994) lead to the possibility that other calcium binding proteins may also modulate the activity of the channel at least in amphibian rod photoreceptors. It would be of interest to isolate and characterize this calcium-binding protein or mediator for comparison with the action of calmodulin on the channel. We are currently looking into the possibility that other calcium-binding proteins may also regulate the channel in mammalian rod outer segment preparations. Another possibility, however, that needs to be considered is that calmodulin may regulate the channel in mammalian photoreceptors, but another calcium binding protein that is not inhibited by mastoparan regulates the channel in amphibian photoreceptors. This may explain the observations of Gray-Keller & Detwiler, Sagoo & Lagnado, and Gordon et al. (1994). As in other systems, it is likely that regulation of the cGMP-gated channel in photoreceptors is not a simple mechanism, but may involve multiple regulators and modes of regulation that have yet to be discovered. A comprehensive study of the regulatory properties of the channel at both the biochemical and physiological level needs to be carried out to resolve these controversial issues and to extend our understanding of the role of the cGMP-gated channel in phototransduction and adaptation. Genetic and functional complexity of inherited retinal degeneration Stephen P. Daiger, Lori S. Sullivan, and Joseph A. Rodriguez Human Genetics Center, School of Public Health, The University of Texas Health Science Center, Houston, TX 77030. daiger@gsbs21.uth.tmc.edu Abstract: Recent findings emphasize the complexity, both genetic and functional, of the manifold genes and mutations causing inherited retinal degeneration in humans. Knowledge of the genetic bases of these diseases can contribute to design of rational therapy, as well as elucidating the function of each gene product in normal visual processes. Our target article has two central themes. The first is that many different photoreceptor genes (or genes expressed in other retinal cells), and many different mutations within these genes, can cause inherited retinal degeneration in humans. Recent developments continue to support this view. As examples, a locus for type 3 Usher syndrome (USH3) has been mapped to chromosome 3q21-q24 (Sankila et al. 1995), one form of recessive retinitis pigmentosa (RP14) has been mapped to 6p (Knowles et al. 1994), and an X-linked dominant form of cone-rod dystrophy has been mapped to Xp21.1-p21.3 (McGuire et al. 1995). In addition, a gene causing dominant Sorsby fundus dystrophy (SFD; chromosome 22ql3qter) has been cloned (Weber et al. 1994), as has a gene causing recessive type IB Usher syndrome (USH1B; BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 501 Re/erences/Controversies in Neuroscience III chromosome Ilql3.5) (Well et al. 1995). The two cloned genes are tissue inhibitor of metalloproteinases-3 (TIMP3) and myosin VIIA, respectively. Mutations in the myosin VIIA gene also produce the mouse recessive shaker-1 (shl) phenotype, with hearing loss and vestibular dysfunction but without retinal degeneration (Gibson et al. 1995). The second theme of our target article is that diseasecausing mutations in photoreceptor genes, such as rhodopsin can provide valuable insights into the functional roles of particular amino acid sites and protein domains. Thus the clinical consequences of specific mutations contribute to understanding normal biochemistry and physiology and, in turn, biochemistry and physiology contribute to understanding - and predicting - clinical consequences. Our specific conclusion, regarding rhodopsin and other proteins with complex functional roles, is that simplistic categorization of mutations, say, into "transmembrane," or "cytoplasmic" or "intradiscal," is not sufficient to explain the rich variety of clinical consequences. Rather, what is required is a detailed, site-bysite analysis of function. In the target article we propose a tentative classification of functional sites in human rhodopsin, based on the current, limited number of known, disease-causing mutations. The commentaries provide a valuable additional prospective on these themes. Bergen suggests that the correlation between genotype and phenotype may be more complicated than we expect at present. Moreover, he notes that there is a need for standardized protocols for reporting clinical findings in patients and families. We emphatically agree with both points. The simplistic classification system used in our analysis is intended as a starting point for more sophisticated analyses. We were limited in our analysis by the very sketchy clinical details provided in most mutation reports (with a few outstanding exceptions). Thus standard reporting protocols are not only desirable, but also essential for further progress in understanding genotype-phenotype correlations. Kaplan discusses the value of animal models but warns that human physiology may differ significantly. Animal models have been of most value in cloning disease genes later implicated in human retinal degeneration such as peripherin/RDS, phosphodiesterase (i and myosin VIIA. However, the utility of transgenic animals in elucidating the physiologic consequences of specific human mutations remains to be seen. Tamai describes an increase in photoreceptor glutamate levels in mice homozygous for the rds insertion mutation and suggests that increased glutamate might trigger apoptosis. Clearly this will be a useful model for recessive retinal degeneration. McCinnis stresses the great promise of molecular treatments, such as gene therapy or antisense therapy, in treating inherited retinal diseases. These treatments, of course, require knowledge of the specific, underlying genetic defect in each patient. This emphasizes the urgency of continued gene mapping, cloning and mutation screening. Barnstable notes that several forms of retinal degeneration are caused by mutations in genes expressed in many tissues in addition to the retina. He also expresses concern that several of these diseases were not discussed in our target article. There are over 200 distinct forms of 502 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 inherited retinal degeneration known in humans (McKusick 1994); the currently reported forms are likely to represent only a small subset of all possible disease-causing genes. We intentionally limited the discussion to photoreceptorspecific genes because these "simple" cases may best illuminate the various mechanisms that lead to cellular degeneration. In fact, the "housekeeping" genes implicated in retinal degeneration, such as ornithine aminotransferase in gyrate atrophy and geranylgeranyl transferase A in choroideremia, affect the entire retinal structure, not just photoreceptors. In spite of these minor caveats, though, we strongly agree with the points made by Dr. Barnstable: one common element in several of these diseases is the effect on cGMP levels, the retina appears to be especially prone to degeneration as a result of metabolic perturbations, and inherited retinal degeneration serves as an excellent model for other neural degenerative conditions such as Alzheimer disease. Finally, Wahlsten raises an intriguing philosophic question: Is it possible that different mutations produce different clinical phenotypes simply because of individual variation in genetic background and/or environment? This concern arises for two reasons. First, even within families with the same mutation, there is substantial clinical variation; and second, many of the reported mutations are found in only one or very few individuals; hence, the range of expression is unknown. It is clear that genetic and environmental factors must play a role in modulating the clinical consequences of specific mutations. Wahlsten correctly demonstrates the technical difficulties in identifying such factors. (Fortunately, some mutations such as the RhoPro23His mutation, are common enough to make statistically sound studies possible.) We disagree, though, on the relative impact of the primary mutation versus secondary factors. Even though there is variation within families, this is much less than the variation often observed between individuals with different mutations. In fact, for many specific mutations, the sample size is large enough to argue against major consequences of modifying factors. In addition, we intentionally chose a broad classification scheme, that is, "mild" versus "severe," to*emphasize the extremes. However, it may be that subtle differences in clinical phenotype are strongly influenced by genetic background or environment. We hope this is so, since identifying these factors may suggest useful therapies. We look forward to rapid progress in identifying genes and mutations causing inherited retinal degeneration in humans. We expect that thisfloodof new knowledge will eventually lead to effective therapies and to important biological insights. References Letters a and r appearing before authors' initials refer to target article and response respectively. Abeliovich, A., Chen, C , Coda, Y., Silva, A. J., Stevens, C. F. & Tonegawa, S. (1993a) Modified hippocampal long term potentiation in PKC-gamma mutant mice. Cell 75:1253-62. [aZX] Abeliovich, A., Paylor, R., Chen, C , Kim. J. J., Wehner, J. M. & Tonegawa, S. (1993b) PKC gamma mutant mice exhibit mild deficits in spatial and contextual learning. Cell 75:1263-71. [aZX] Abrams, T. W. (1985) Activity-dependent presynaptic facilitation: An References/Controversies in Neuroscience HI associative mechanism in Aplysia. Cellular and Molecular Neurobiology 5:123-45. (TYVA] Abrams, T. \V., Bernier, L., Hawkins, H. D. & Kandel, E. R. (1984) Possible roles of Ca2+ and cAMP in activity-dependent facilitation, a mechanism for associative learning in Aplysia. Society for Neuroscience Abstracts 10:269. [TWA] Abrams, T. W. & Galnn, J. (1993) Temporal aspects of activation of a Ca2+/CaM-sensitive adenylyl cyclase: Possible contribution to detection of CS-US pairing during conditioning. Society for Neuroscience Abstracts 19:1065. |TWA] Abrams, T. W., Jarrard, H. E. & Yovell, Y. (1994) Calcium accelerates adenylyl cyclase dcactivation contributing to CS-US sequence requirement during conditioning in Aplysia. Submitted. [TWA] Abrams, T. W. & Kandel, E. R. (1988) Is contiguity detection in classical conditioning a system or a cellular property? Learning in Aplysia suggests a possible molecular site. Trends in Neuroscience 11:128-35. [TWA] Abrams, T. W , Karl, K. A. & Kandel, E. R. (1991a) Biochemical studies of stimulus convergence during classical conditioning in Aplysia: Dual regulation of adcnylate cyclase by Ca2+/calmodulin and transmitter. Journal of Neuroscience 11:2655-65. [aZX, WH] Adamus, C , Ortega, H., Witkowska, D. & Polans, A. (1994) Recovorin: A potent uvcitogen for the induction of photoreceptor degeneration in Lewis rats. Experimental Eye Research 59:447-56. [ASP] Adamus, C , Zam, Z. S., Arendt, A., Palczewski, K., McDowell, J. H. & Margrave, P. A. (1991) Anti-rhodopsin monoclonal antibodies of defined specificity: Characterization and application. Vision Research SLITSI. [aPAH] Agranoff, B. W., Davis, R. E. & Brink, J. J. (1965) Memory fixation in the goldfish. Proceedings of the National Academy of Sciences USA 54:78893. [EDR] Aguirre, C , Farber, D. Lolley, R., Fletcher, R. T. & Chader, C. (1978) Rodcone dysplasia in Irish setters: A defect in cyclic CMP metabolism in visual cells. Science 201:1133-34. [MWK] Ahlijanian, M. K., Halford, M. K. & Cooper, D. M. F. (1987) Csfi+I cahnodulin distinguishes between guanyl-5'-yl-imidodiphosphate- and opiate-mediated inhibition of rat striata) adenylate cyclase. Journal of Ncurochemistry 49:1308-15. [MMR] Ahmad, I., Korbmacher, C , Segal, A. S., Cheung, P., Boulpaep, E. L. & Barnstable, C. J. (1992) Mouse cortical collecting duct cells show nonselective cation channel activity and express a gene related to the cGMP-gated rod photorecoptor channel. Proceedings of the National Academy of Sciences USA 89:10262-66. [aRSM] Ahmad, I., Redmond, L. J. or Barnstable, C. J. (1990) Developmental and tissue-specific expression of rod photoreceptor cGM P-gated ion channel gene. Biochemical and Biophysical Research Communications 173:46370. [aRSM] Akita. H., Tunis, S. P., Adams, M., Balogh-Nair, V. & Nakanishi, K. (1980) Nonbleachable rhodopsins retaining the full natural chromophore. Journal of the American Chemical Society 102:6370-72. [aPAH] Al-Maghtheh, A., Kim, R. Y., Inglehearn, C. & Bhattacharya, S. S. (1994a) A 150 lip insert in the rhodopsin gene of an autosomal dominant retinitis pigmentosa family. Human Molecular Cenetics 3:205-6. [aSPD] Al-Maghtheh, M., Ingleheam, C. F., Keen, T. J., Evans, K., Moore, A. T., Jay, M., Bird, A. C. & Bhattacharya, S. S. (1994) Identification of a sixth locus for autosomal dominant retinitis pigmentosa on chromosome 19. Human Molecular Cenetics 3:351-54. [aSPD] Al-Maghtheh, M., Ingleheam, C , Lunt, P., Jay, M., Bird, A. & Bhattacharya, S. (1994) Two new rhodopsin transversion mutations (C40R; M216K) in families with autosomal dominant retinitis pigmentosa. Human Mutation 3: 409-10. [aSPD] Albert, A. D. & Litman, B. J. (1978) Independent structural domains in the membrane protein bovine rhodopsin. Biochemistry 17:3893— 3900. [ADA] Alexander, K. A., Cimler, B. M., Meier, K. E. & Storm, D. R. (1987) Regulation of cahnodulin binding to P-57. Journal of Biological Chemistry 262:6108-13. [aZX] Allen, J. P. or Feher, G. (1991). Crystallization of reaction centers from Rhodobacter sphaeroides. In: Crystallization of membrane proteins, ed. H. Michel. CRC Press. [aPAH] Altcnhnfcn, \V., Ludwig, J., Eismann, E., Kraus, W , Bonigk, W. & Kaupp, U. B. (1991) Control of ligand specificity in cyclic nucleotide-gated channels from rod photoreceptors and olfactory epithelium. Proceedings of the National Academy of Sciences USA 88:9868-72. [aRSM] Ames, J. B., Porumbo, T., Tanaka, T., Ikura, M. & Stryer, L. (1995) Aminoturminal myristoylation induces cooperative calcium binding to recoverin. Journal of Biological Chemistry 270:4526-33. [rJBH] Amos, L. A., Henderson, R. & Unwin, P. N. T. (1982) Three-dimensional structure determination by electron microscopy of 2-D crystals. Progress in Biophysics and Molecular Biology 39:183-231. [aPAH] Anant, J. S. & Fung, B. K.-K. (1992) In vivo farnesylation of rat rhodopsin kinase. Biochemical and Biophysical Research Communications 183:46873. [aMDB] Anant, J. S., Ong, O. C , Xie, H., Clarke, S., O'Brien, P. J. & Fung, B. K.-K. (1992) In vivo differential prenylation of retinal cyclic GMP phosphodiesterase catalytic subunits. Journal of Biological Chemistry 267:687-90. [aMDB] Andersson, L. & Porath, J. (1986) Isolation of phosphoproteins by immobilized metal (Fe3+) affinity chromatography. Analytical Biochemistry 154:25054. [aPAH] Andrade, R. (1993) Neuron 10:83-88. [MMR] Andreasen, T. J., Luetje, C. W., Heideman, W. & Storm, D. R. (1983) Purification of a novel calmodulin binding protein from bovine cerebral cortex membranes. Biochemistry 22:4615-18. [aZX] Andrdasson, S., Ehinger, B., Abrahamson, M. & Fex, G. (1992) A sixgeneration family with autosomal dominant retinitis pigmentosa and a rhodopsin gene mutation (arginine-135-leucine). Ophthalmic Pediatrics and Cenetics i3:145-53. [aSPD] Angleson, J. K. & Wensel, T. G. (1993) A GTPase-accelerating factor for transducin, distinct from its effector cGMP phosphodiesterase, in rod outer segment membranes. Neuron 11:939-49. [aMDB] (1994) Enhancement of rod outer segment GTPase accelerating protein activity by the inhibitory subunit of cGMP phosphodiesterase. Journal of Biological Chemistry 269:16290-96. [arMDB] Antonny, B., Otto-Bruc, A., Chabre, M. & Minh Vuong, T. (1993) CTP hydrolysis by purified ot-subunit of transducin and its complex with the cyclic GMP phosphodiesterase inhibitor. Biochemistry 32:864653. [arMDB, BMW] Apel, E. D., Byford, M. F., Au, D., Walsh, K. A. & Storm, D. R. (1990) Identification of the protein kinase C phosphorylation site in neuromodulin. Biochemistry 29:2330-35. [aZX] Apfelstedt-Sylla, E., Kunisch, M., Horn, M., Ruther, K., Gal, A. & Zrenner, E. (1992) Diffuse loss of rod function in autosomal dominant retinitis pigmentosa with pro-347-leu mutation of rhodopsin. German Journal of Ophthalmology 1:319-27. [aSPD] Applebury, M. A. (1991) Molecular determinants of visual pigment function. Current Opinion in Neurobiology 1:263-69. [aPAH] Applebury, M. L. & Hargrave, P. A. (1986) Molecular biology of the visual pigments. Vision Research 26:1881-95. [aPAH] Arikawa, K., Molday, L. L., Molday, R. S. & Williams, D. (1992) Localization of periphcrin/rds in the disk membranes of cone and rod photoreceptors: Relationship to disk membrane morphogenesis and retinal degeneration. Journal of Cell Biology 116:659-67. [aSPD] Arnoux, B., Ducruix, A., Reiss-Husson, F., Lutz, M., Norris, J., Schiffer, M. & Chang, C. G. (1989) Structure of spheroidene in the photosynthetic reaction center from Y Rhodobacter sphaeroides. FEBS Utters 258:4750. [aPAH] Arshavsky, V. Y., Antoch, M. P., Dizhoor, A. M., Rakhilin, S. V. & Philippov, P. P. (1987) Interaction of visual transduction enzymes in a detergent solution. In: Retinal proteins, ed. Y. A. Ovchinnikov. Utrecht: VNU Science Press. [aMDB] Arshavsky, V. Y., Antoch, M. P., Lukjanov, K. A. & Philippov, P. P. (1989) Transducin GTPase provides for rapid quenching of the cGMP cascade in rod outer segments. FEBS Letters 250:353-56. [aMDB] Arshavsky, V. Y. & Bownds, M. D. (1992) Regulation of deactivation of photoreceptor G protein by its target enzyme and cGMP. Nature 357:416-17. [aMDB] Arshavsky, V. Y., Dizhoor, A. M., Shestakova, I. K. & Philipiwv, P. P. (1985) The effect of rhodopsin phosphorylation on the light-dependent activation of phosphodiesterase from bovine rod outer segments. FEBS Letters 181:264-66. [aMDB] Arshavsky, V. Y., Dumke, C. L. & Bownds, M. D. (1992) Non-catalytic cGMP binding sites of amphibian rod cCMP phosphodiesterase control interaction with its inhibitory gamma-subunits—a putative regulatory mechanism of the rod photoresponse. Journal of Biological Chemistry 267:24501-7. [aMDB] (1995) Manuscript in preparation. [aMDB] Arshavsky, V. Y., Dumke, C. L., Zhu, Y., Artemyev, N. O., Skiba, N. P., Hamm, H. E. & Bownds, M. D. (1994) Regulation of transducin GTPase activity in bovine rod outer segments. Journal of Biological Chemistry 269:19882-87. [arMDB] Arshavsky, V. Y., Gray-Keller, M. P. & Bownds, M. D. (1991) cGMP suppresses GTPase activity of a portion of transducin equimolar to phosphodiesterase in frog rod outer segments. Light-induced cGMP decreases as a putative feedback mechanism of the photoresponse. [aM DB1 Journal of Biological Chemistry 266:18530-37. BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 503 Re/erences/Controversies in Neuroscience III Artlich, A., Horn, M., Lorenz, B., Bhattacharya, S. & Gal, A. (1992) Recurrent 3-bp deletion at codon 255/256 of the rhodopsin gene in a German pedigree with autosomal dominant retinitis pigmentosa. American Journal of Human Cenetics 50:876-78. [aSPD] Aton, B. R., Litman, B. J. & Jackson, M. L. (1984) Isolation and Identification of the phosphorylated species of rhodopsin. Biochemistry 23:173741. [aPAH) Bacskai, B. J., Hochner, B., Mahaut-Smith, M., Adams, S. R., Kaang, B. K., Kandel, E. R. & Tsien, R. Y. (1993) Spatially resolved dynamics of cAMP and protein kinase A subunits in Aplysia sensory neurons. Science 260:222-26. [aZX] Baehr, W., Gorczyca, W., Subbaraya, I., Ohguro, H., Johnson, R. S., Walsh, K. A., Gray-Keller, M. P., Detwiler, P. B., Crabb, J. & Palczewski, K. (1994) Isolation, cloning, and functional characterization of bovine retina guanylate cyclase activating protein (GCAP). Investigative Ophthalmology and Visual Science 35:1486. [aMDB] Baehr, W., Morita, E. A., Swanson, R. J. & Applebury, M. L. (1982) Characterization of bovine rod outer segment G-protein. Journal of Biological Chemistry 257:6452-60. [aMDB] Bakalyar, H. A. & Reed, R. R. (1990) Identification of a specialized adenylyl cyclase that may mediate odorant detection. Science 250:1403-6. [aZX] Baldwin, J. M. (1993) The probable arrangement of the helices in G proteincoupled receptors. EM BO Journal 12:1693-1703. [aSPD, SOS] Baldwin. P. A. & Hubbell, W. L. (1985a) Effects of lipicl environment on the light-induced conformational changes of rhodopsin: 1. Absence of metarhodopsin II production in dimyristoylphosphatidylcholine recombinant membranes. Biochemistry 24:2624-32. [ADA] (1985b) Effects of lipid environment on the light-induced conformational changes of rhodopsin: 2. Roles of lipid chain length, unsaturation, and phase state. Biochemistry 24:2633-39. [ADA] Barker, D., Schafer, M. & White, R. (1984) Restriction sites containing CpG show a higher frequency of polymorphism in human DNA. Cell 6:13138. [aSPD] Bascom, R. A., Manara, S., Collins, L., Molday, R. S., Kalnins, V. I. & Mclnnes, R. R. (1992) Cloning of the cDNA for a novel photoreceptor membrane protein (rom-1) identifies a disk rim protein family implicated in human retinopathies. Neuron 8:1171-84. [aSPD] Baudier, J., Bronner, C , Kligman, D. & Cole, R. D. (1989) Protein kinase C substrates from bovine brain: Purification and characterization of neuromodulin, a neuron-specific calmodulin binding protein. Journal of Biological Chemistry 264:1824-28. [EDR] Bauer, P. H., Miiller, S., Puzicha, M., Pippig, S., Obermaier, B., Helmreich, E. J. M. & Lohse, M. J. (1992) Phosducin is a protein kinase A-regulated G-protein regulator. Nature 358:73-76. [BMW] Baiimann, A., Frings, S., Godde, M., Seifert, R. & Kaupp, U. B. (1994) Primary structure and functional expression of a Drosphila cyclic nucleotide-gated channel present in eyes and antennae. EMBO Journal 13:5040-50. [rRSM] Baylor, D. A. (1987) Photoreceptor signals and vision. Investigative Ophthalmology and Visual Science 28:34-49. [aPAH] Baylor, D. A. fit Hodgkin, A. L. (1974) Changes in time scale and sensitivity in turtle photoreceptors. Journal of Physiology (London) 242:729— 58. (aMDB] Baylor, D. A., Lamb, T. D. & Yau, K.-W. (1979a) The membrane current of single rod outer semgments. Journal of Physiology (London) 288:589611. [aMDB] (1979b) Responses of retinal rods to single photons. Journal of Physiology (London) 288:613-34. [aMDB] Baylor, D. A., Matthews, G. & Yau, K.-W. (1980) Two components of electrical dark noise in toad retina! rod outer segments. Journal of Physiology (London) 309:591-621. [aMDB] Beaudet, A. L. & Tsui, L.-C. (1993) A suggested nomenclature for designating mutations. Human Mutation 2:245-48. [aSPD] Bell, C , Converse, C. A., Collins, M. F., Esakowitz, L., Kelly, K. F. & Haites, N. E. (1992) Autosomal dominant retinitis pigmentosa (ADRP): A rhodopsin mutation in a Scottish family. Journal of Medical Cenetics 29:667-68. [aSPD] Bennett, A. E., Ok, J. H., Griffin, R. G. & Vega, S. (1992) Chemical shift correlation specetroscopy in rotating solids: Radio frequency-driven dipolar reeoupling and logitudinal exchange. Journal of Chemical Physics 96:8624-27. |SOS] Bennett, N. & Clere, A. (1989) Activation of cGMP phosphodiestera.se in retinal rods: Mechanism of interaction with the GTP-binding protein (transducin). Biochemistry 28:7418-24. [aMDB] Bennett, N. & Sitaramayya, A. (1988) Inactivation of photoexcited rhodopsin in retinal rods: The roles of rhodopsin kinase and 48-KDa protein (arrcstin). Biochemistry 27:1710-15. [aMDB] Benowitz, L. I. & Routtenberg, A. (1987) A membrane phosphoprotein 504 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 associated with neural development axonal regeneration, phospholipid metabolism, and synaptic plasticity. Trends in Neuroscience 10:52732. [aZX] Berson, E. L. (1993) Retinitis pigmentosa—The Friedenwald Lecture. Investigative Ophthamology and Visual Science 34:1659-76. [aSPD, aPAH] Berson, E. L., Rosner, B., Sandberg, M. A. & Dryja, T. P. (1991) Ocular findings in patients with autosomal dominant retinitis pigmentosa and a rhodopsin gene defect (Pro-23-His). Archives of Ophthalmology 109:92101. [aSPD] Berson, E. L., Rosner, B., Sandberg, M. A., Weigel-DiFranco, C. & Dryja, T. P. (1991a) Ocular findings in patients with autosomal dominant retinitis pigmentosa and rhodopsin, Proline-347-Leucine. American Journal of Ophthalmology 111:614-23. [aSPD] Berson, E. L., Sandberg, M. A. & Dryja, T. P. (1991b) Autosomal dominant retinitis pigmentosa with rhodopsin, valine-345-methionine. Transcripts of the American Ophthalmological Society 89:117-30. |aSPD] Bhattacharya, S. S., Ridge, K. D., Knox, B. E. & Khorana, H. G. (1992) Light-stable rhodopsin: 1. A rhodopsin analog reconstituted with a nonisomerizable ll-cis retinal derivative. Journal of Biological Chemistry 267:6763-69. [aPAH] Bhattacharya, S. S., Wright, A. F., Clayton, J. F., Price, W. U., Phillips, C. I., McKeown, C. M. E., Jay, M., Bird, A. C , Pearson, P. L., Southern, E. M. & Evans, H. J. (1984) Close genetic linkage between X-linked retinitis pigmentosa and a restriction fragment length polymorphism identified by recombinant DNA probe Ll.28. Nature 309:253-55. [aSPD] Biel, M., Altenhofen, W., Hullin, R., Ludwig, J., Freichel, M., Flocherzi, V., Dascal, N., Kaupp, U. B. & Hofmann, F. (1993) Primary structure and functional expression of a cyclic nucleotide-gated channel from rabbit aorta. FEBS Letters 329:134-38. [aRSM] Biel, M., Zong, X., Distler, M., Bosse, R., Klugbauer, N., Murakami, M., Flockerzi, V. & Hoffmann, F. (1994) Another member of the cyclic nucleotide-gated channel family expressed in testis, kidney and heart. Proceedings of the National Academy of Sciences USA 91:35059. [rRSM] Biernbaum, M. S., Binder, B. M. & Bownds, M. D. (1991) Dim background light and Cerenkov radiation from 32P block reversal of rhodopsin phosphorylation in intact frog retinal rods. Visual Neuroscience 7:499503. [aMDB] Biernbaum, M. S. & Bownds, M. D. (1985) Light-induced changes in GTP and ATP in frog rod photoreeeptors: Comparison with recovery of dark current and light sensitivity during dark adaptation. Journal ofCeneral Physiology 85:107-21. [aMDB] Binder, B. M., Biernbaum, M. S. 6t Bownds, M. D. (1990) Light activation of one rhodopsin molecule causes the phosphorylation of hundreds of others. A reaction observed in electropermeabilized frog rod outer segments exposed to dim illumination. Journal of Biological Chemistry 265:15333-40. [aMDB] Binder, B. M., Brewer, E. & Bownds, M. D. (1989) Stimulation of protein phosphorylations in frog rod outer segments by protein kinase activators: Suppression of light-induced changes in membrane current and cGMP by protein kinase C activators. Journal of Biological Chemistry 264:885764. [aMDBJ Bitensky, M. W , Wheeler, M. A., Yamazaki, A., Rasenick, M. M. & Stein, P. J. (1981) Cyclic nucleotide matabolism in vertebrate photoreceptors: A remarkable analogy and an unraveling enigma. Current Topics in Membrane Transportation 15:237-71. [TGW] Blanton, S. H., Heckenlively, J. R., Cottingham, A. W , Friedman, J., Sadler, L. A., Wagner, M., Friedman, L. H. & Daiger, S. P. (1991) Linkage mapping of autosomal dominant retinitis pigmentosa (adRP) to the pericentric region of human chromosome 8. Gcnomics 11:857-69. [aSPD] Blatt, C , Eversole-Cire, P., Colin, V. H., Zollman, S., et al. (1988) Chromosomal localization of genes encoding guanine nuclcotide-binding protein subunits in mouse and human. Proceedings of the National Academy of Sciences USA 85:7642-46. [aSPD] Bodoia, R. D. & Detwiler, P. B. (1985) Patch-clamp recordings of the lightsensitive dark noise in retinal rods from the lizard and frog. Journal of Physiology 367:183-216. [aRSM] Boekema, E. J., Van Heel, M. G. & Van Bniggen, E. F. J. (1986) Preparation of two-dimensional crystals of complex 1 and image analysis. Methods in Enzymology 126:344-53. [aPAH] Boesze-Battaglia, K. & Albert A. (1990) Cholesterol modulation of photoreceptor function in bovine rod outer segments. Journal of Biological Chemistry 265:20727-30. [ADA] Bonigk, W., Altenhofen, W , Muller, F., Dose, A., Illing, M., Molday, R. S. & Kaupp, U. B. (1993) Rod and cone photoreeeptor cells express distinct genes for cyclic GMP-gated channels. Neuron 10: 865-77. [aRSM] References /Controversies in Neuroscience HI Bornancin, F., Pfister, C. & Chabre, M. (1989) The transitory complex between photocxcited rhodopsin and transducin: Reciprocal interaction between the retinal site in rhodopsin and the nucleotide site in transducin. European Journal of Biochemistry 184:68798. laPAH] Bourne, H. R. & Nicoll, R. (1993a) Molecular machines integrate coincident synaptic signals. Cell 72:65-75. [aZX] (1993b) Molecular machines integrate coincident synaptic signals. Neuron 10(Suppl.):65-75. [TWA] Bourne, H. R., Sanders, D. A. & McCormick, F. (1990) The GTPase superfamily: A conserved switch for diverse cell functions. Nature 348:125-32. [aMDB] (1991) Tlie CTPase superfamily: Conserved structure and molecular mechanism. Nature 349:117-27. [aMDB] Bowes, C , Li, T., Dancigcr, M., Baxter, L. C , Applebury, M. L & Farber, D. B. (1990) Kctinal degeneration in the rd mouse is caused by a defect in the 6 subunit of rod cCMP-phosphodiesterase. Nature 347:67780. [aSPD, MWK| Bowes, C , Li, T., Frankel, W. N., Danciger, M., Coffin, J. M., Applebury, M. L. & Farber, D. B. (1993) Localization of a retroviral element within the rd gene coding for the beta subunit of cGMP phosphodiesterase. Proceedings of the National Academy of Sciences USA 90:295559. [aSPD) Bowie, J. U., Liithy, R. & Eisenberg, D. (1991) A method to identify protein sequences that fold into a known three-dimensional structure. Science 253:164-70. [RMC] Bownds, D., Dawes, J., Miller, J. & Stahlman, M. (1972) Phosphorylation of frog photoreceptor membranes induced by light. Nature New Biology 237:125-27. [aMDB] Bownds, M. D., Arshavsky, V. Y., Calvert, P. D. & Dumke, C. I. (1992) Rod outer segment concentration and structure influence the apparent kinetics of cGMP phosphodiesterase. Investigative Ophthalmology and Visual Science 33:873. [aMDB] Bownds, M. D. & Thomson, D. (1988) Control of the sensitivity and kinetics of the vertebrate photoresponse during light adaptation—calculations from simple molecular models. In: Molecular physiology of retinal proteins, ed. T. Hara. Osaka: Japan Scientific Society Press. [aMDB] Bradley, J., Li, J., Davidson, N., Lester, H. A. & Zinn, K. (1994) Heteromeric olfactory cyclic nucleotide-gated channels: A subunit that confers increased sensitivity to cAMP. Proceedings of the National Academy of Sciences USA 91:8890-94. [rRSM] Brann, M. R. & Cohen, L. V. (1987) Diurnal expression of tranducin mRNA and translocation of transdiicin in rods of rat retina. Science 235:585— 88. [JFM] Breer, H. & Boekhoft", 1. (1992) Second messenger signalling in olfaction. Current Opinion in Neurobiology 2:439-43. [aMDB] Bressler, N. M., Bressler, S. B. & Fine, S. L. (1988) Age-related macular degeneration. Survey of Ophthalmology 32:375-413. [aSPD] Brown, N. C , Fowles, C , Sharma, R. & Akhtar, M. (1993) Quantitative characterization of the structure of rhodopsin in disc membrane by means of Fourier transform infrared spectroscopy. Journal of Biological Chemistry 268:2403-9. [ADA] Brown, R. L., Bert, R. J., Evans, F. E. & Karpen, J. W. (1993) Activation of retinal rod cGMP-gated channels: What makes for an effective 8-snbstitutecl derivative of cCMP? Biochemistry 32:10089-95. [RLB] Brown. R. L., Cerber, W. V. & Karpen, J. W. (1993) Specific labeling and permanent activation of the retinal rod cGMP-activated channel by the photaffinity analog 8-p-azidophcnacylthio-cGMP. Proceedings of the National Academy of Sciences USA 90:5369-73. [aRSM, RLB] Brown, R. L., Crainling, R., Bert, R. J. & Karpen, J. VV. (in press) Cyclic GMP contact points within the 63-kDa subunit and a 240-kDa associated protein of retinal rod cGMP-activated channels. Biochemistry. [RLB] Brown, R. L. & Karpen, J. W. (1994) Dissecting the activation mechanism of retinal rod cGMP-gated channels using covalently tethered ligands. Biophysics Journal 66:A350. [RLB, TCW] Buczylko, J., Gutmann, C. & Palezewski, K. (1991) Regulation of rhodopsin kinase by autophosphorylatiou. Proceedings of the National Academy of Sciences USA 88:2568-72. [aMDB] Bunge, S., Wedemann, H., David, D., Terwilliger, D. J., Van Den Bom, L. I., Aulehla-Scholz, C , Sammons, C , Horn, M., Ott, J., Schwinger, E., Schinzel, A., Denton, M. J. & Gal, A. (1993) Molecular analysis and genetic mapping of the retinitis pigmentosa gene in families with uutosomal dominant retinitis pigmentosa. Ccnomics 17:230-33. [aSPD] Burks, C , Cinkosky, M. J., Fischer, W. M., Gilna, P., Hayden, J. E.-D., Keen, G. M., Kelly, M., Kristoflerson, D. & Lawrence, J. (1992) GeiiBaiik. Nucleic Acids Research 20:2065-69. [aSPD] Byrne, J. II. (1987) Presynaptic facilitation as a mechanism for behavioral sensitization in Aplysia. Physiology Review 67:329-439. [aZX] Cali, J. F., Zwaagstra, J. C , Mons, N., Cooper, D. M. F. & Krupinski, J. (1994) Type VIII adenylyl cyclase. Journal of Biological Chemistry 269:12190-95. [aZX] Campbell, J. W., Duee, E., Hodgson, G., Mercer, W. D., Stammers, D. K., Wendell, P. L., Muirhead, H. & Watson, H. C. (1971) X-ray diffraction studies on enzymes in the glycolytic pathway. Cold Spring Harbor Symposium on Quantitative Biology 36:165-70. [aPAH] Caretta, A. & Cavaggioni, A. (1983) Fast ionic flux activated by cyclic GMP in the membrane of cattle rod outer segments. European Journal of Biochemistry 132:1-8. [TGW] Caretta, A., Cavaggioni, A., Grimaldi, R. & Sorbi, R. (1988) Regulation of cyclic GMP binding to retinal rod membranes by calcium. European Journal of Biochemistry 177:139-46. [aRSM] Caretta, A., Cavaggioni, A. & Sorbi, R. T. (1979) Cyclic GMP and the permeability of the disks of the frog photoreceptors. Journal of Physiology 295:171-78. [TGW] Castellucci, V. F., Blumenfeld, H., Goelet, P. & Kandel, E. R. (1989) Inhibitor of protein synthesis blocks long-term behavioral sensitization in the isolated gill-withdrawal reflex of Aplysia. Journal of Neurobiology 20:1-9. [aZX] Castellucci, V. F. & Kandel, E. R.(1976) Presynaptic facilitation as a mechanism for behavioral sensitization in Aplysia. Science 194:117678. [aZX] Cavaggioni, A. & Sorbi, R. T. (1981) Cyclic GMP releases calcium from disc membranes of vertebrate photoreceptors. Proceedings of the National Academy of Science USA 78:3964-68. [TCW] Caviness, V. S., Jr., So, D. K. & Sidman, R. L. (1972) The hybrid reeler mouse. Journal of Heredity 63:341-46. [DW] Cervetto, L., Torre, V, Pasino, E., Marroni, P. & Capovilla, M. (1984) Recovery from light-desensitization in toad rods. In: Photoreceptors, ed. A Borcellino & L. Cervetto. Plenum. [arMDB] Chabre, M. (1985) Molecular mechanism of visual phototransduction in retinal rod cells. Annual Review of Biophysics and Biophysical Chemistry 14:331-60. [aPAH] Chabre, M. & Deterre, P. (1989) Molecular mechanism of visual transduction. European Journal of Biochemistry 179:255-66. [aMDB, aPAH, aRSM] Chakraborty, R. (1993) Generalized occupancy problem and its applications in population genetics. In: Genetics of cellular, individual, family, and population variability, ed. C. F. Sing & C. L. Hanis. Oxford University Press. [aSPD] Chang, B., Heckenlively, J. R., Hawes, N. L. & Roderick, T. H. (1993) New mouse primary retinal degeneration (rd-3). Cenomics 16:45-49. [MWK] Chang, G.-Q., Hao, Y. & Wong, F. (1993) Apoptosis: Final common pathway of photoreceptor death in rd, rds, and rhodopsin mutant mice. Neuron 11:595-605. [aSPD, MT] Charbonneau, H., Prusti, R. K., LeTrong, Sonnenburg, W. K., Mullancy, P. J., Walsh, K. A. & Bcavo, J. A. (1990) Identification of a noncatalytic cGMP-binding domain conserved in both the cGMP-stimulated and photoreceptor cyclic nucleotide phosphodiesterase. Proceedings of the National Academy of Sciences USA 87:288-92. (aMDB] Chen, C.-K. & Hurley, J. B. (1994) Calcium-dependent recoverin/rhodopsin kinase interaction. Investigative Ophthalmology and Visual Science 35:1485. [aMDB, aJBH, AY] Chen, S., Fry, C. H., Illner, H., Kickenweiz, E., McGuigan, J. A. S., Powell, T. & Twist, V. W. (1992) Measurement of intracellular free magnesium in isolated guinea-pig papillary muscles and ventricular myocytes. Journal of Physiology (London) 452:194P. [RLH] Chen, T.-Y, Ming, M., Molday, L. L., Hsu, T.-T, Yau, K.-W. & Molday, R. S. (1994) subunit 2 (or B) of retinal rod cGMP-gated cation channel is a component of the 240-kDa channel-associated protein and mediates Ca2+-ealmodulin modulation. Proceedings of the National Academy of Sciences USA 91:11757-61. [rRSM, RLB, DDO] Chen, T.-Y, Peng, Y.-W, Dhallan, R. S., Ahamed, B., Reed, R. R. & Yau, K.-W. (1993) A new subunit of the cyclic nucleotide-gated cation channel in retinal rods. Nature 362:764-67. [arRSM, RLB, LWH, RLH, DDO] Chen, T.-Y. & Yau, K.-W. (1994) Direct modulation by Ca-calmodulin of cyclic nucleotide-activated channel of rat olfactory receptor neurons. Nature 368:545-48. [rRSM] Chetkovich, D. M., Gray, R., Johnston, D. & Sweatt, J. D. (1991) N-methylD-aspartate receptor activation increases cAMP levels and voltage gated Ca2+ channel activity in area CA1 of hippocampus. Proceedings of the National Academy of Sciences USA 88:6467-71. [aZX, EDR] Cheikovich, D. M. & Sweatt, J. D. (1993) NMDA receptor activation increases cyclic AMP in area CA1 of the hippocampus via calcium/calmodulin stimulation of adenylyl cyclase. Journal of Ncurochemistry 61:1933-42. [aZX, EDR] Cheung, W. Y. & Storm D. R. (1982) Calmodulin regulation of cAMP metabolism. Handbook of Experimental Pharmacology 58:301-17. BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 [aZX] 505 References/Controversies in Neuroscience III Choi, E. J., Wong, S. T , Dittman, A. H. & Storm, D. R. (1993). Phorbol ester stimulation of the type I & type III adenylyl cyclases in whole cells. Biochemistry 32:1891-94. [aZX] Choi, E. J., Wong, S. T., Hinds, T. J. & Storm, D. R. (1992a) Calcium and muscarinic agonist stimulation of type 1 adenylyl cyclase in whole cells. Journal of Biological Chemistry 267:12440-42. [aZX] Choi, E. J., Xia, Z. & Storm, D. R. (1992b) Stimulation of the type III olfactory adenylyl cyclase by calcium and calmodulin. Biochemistry 31:6492-98. [arZX] Cimler, B. M., Andreasen, T. J., Andreasen, K. 1. & Storm, D. R. (1985) P-57 is a neural specific calmodulin-binding protein. Journal of Biological Chemistry 260:10784-88. [aZX] Clack, J. W. & Pepperberg, D. R. (1982) Desensitization of skate photoreceptor by bleaching and background light. Journal of General Physiology 80:863-83. [aMDB] Clack, J. W. & Stein, P. J. (1988) Opsin exhibits cCMP-activated singlechannel activity. Proceedings of the National Academy of Sciences USA 85:9806-10. [aRSM] Clore, G. M. & Gronenborn A. M. (1994a) Structures of larger proteins, protein-ligand and protein-DNA complexes by multidimensional heteronuclear NMR: Young investigator award lecture. Protein Science 3:372-90. [RKC] (1994b) Structures of larger proteins, protein-ligand and protein-DNA complexes by multidimensional heteronuclear NMR. Protein Science 3:372-90. [rPAH] Coleman, D. L. (1981) Inherited obesity-diabetes syndromes in the mouse. In: Mammalian genetics and cancer. Liss. [DW] Coles, J. A. & Yiimane, S. (1975) Effects of adapting lights on the time course of the receptor potential of the anuran retinal rod. Journal of Physiology (London) 247:189-207. [arMDB] Collingridge, G. (1987) Synaptic plasticity: The role of NMDA receptors in learning and memory. Nature 330:604-5. [TWA] Cone, R. A. (1972) Rotational diffusion of rhodopsin in the visual receptor membrane. Nature New Biology 236:39-43. [aPAH] Connell, G., Bascom, R., Molday, L., Reid, D., Mclnnes, R. R. & Molday, R. S. (1991) Photoreceptor peripherin is the normal product of the gene responsible for retinal degeneration in the rds mouse. Proceedings of the National Academy of Sciences USA 38:723-36. [aSPD, MWK] Cook, N. J., Hanke, W. & Kaupp, U. B. (1987) Identification, purification and functional reconstitution of the cyclic CMP-dependent channel from rod photoreceptors. Proceedings of the National Academy of Sciences USA 84:585-89. [aRSM, RLB, RLH, TGW] Cook, N. J., Molday, L. L., Reid, D., Kaupp, U. B. & Molday, R. S. (1989) The cGMP- gated channel of bovine rod phororeceptors is localized exclusively in the plasma membrane. Journal of Biological Chemistry 264:6996-99. [aRSM, RLH, TGW] Corless, J. M., McCaslin, D. R. & Scott, B. L. (1982) Two-dimensional rhodopsin crystals from disk membranes of frog retinal rod outer segments. Proceedings of the National Academy of Sciences USA 79:1116-20. [aPAH EAD] Cornwall, M. C. & Fain, G. L. (1992) Bleaching of rhodopsin in isolated rods causes a sustained activation of PDE and cyclase which is reversed by pigment regeneration. Investigative Ophthalmology and Visual Science 33:1103. [aMDB] (1994) Bleached pigment activates transduction in isolated rods of the salamander retina. Journal of Physiology (London) 480:261279. [rMDB] Cornwall, M. C. & Fain, G. L. (1994) Bleached pigment activates transduction in isolated rods of the salamander retina. Journal of Physiology 480:261-79. [RKC] Cornwall, M. C. & Pan, W.-Z. (1985). Spatial resolution of bleaching and background adaptation in rod photoreceptors of Ambystoma tigrinum. In: The visual system, ed. A. Fein & J. Levine. Alan R. Liss. [RKC] Cornwall, M. C , Fein, A. & MaeNichol, E. F. (1990) Cellular mechanisms that underlie bleaching and background adaptation. Journal of General Physiology 96:345-72. [RKC] (1983) Spatial localization of bleaching adaptation in isolated vertebrate rod photoreceptors. Proceedings of the National Academy of Sciences USA 80:2785-88. [RKC] Corson, D. W., Cornwall, M. C , MaeNichol, E. F., Jin, J., Johnson, R., Derguini, F., Crouch, R. K. & Nakanishi, K. (1990) Sensitization of bleached rod photoreceptors by 11-cis-locked analogues of retinal. Proceedings of the National Academy of Sciences USA 87:682327. [RKC] Corson, D. W., Cornwall, M. C , MaeNichol, E. F., Tsang, S., Derguini, F., Crouch, R. K. & Nakanishi, K. (1994a) Relief of opsin desensitization and prolonged excitation of rod photoreceptors by 9-desmethyl retinal. Proceedings of the National Academy of Sciences USA 91:6952-58. [RKC] 506 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 Corson, D. W., Cornwall, M. C. & Pepperberg, D. R. (1994b) Evidence for the prolonged photoactivated lifetime of an analogue visual pigment containing 11-cis 9-desmethylretinal. Visual Neuroscience 11:91— 98. [RKC] Cote, R. H. & Brunnock, M. A. (1993) Intracellular cGMP concentration in rod photoreceptors is regulated by binding to high and moderate affinity cGMP binding sites. Journal of Biological Chemistry 268:1719098. [aMDB] Cote, R. H., Bownds, M. D. & Arshavsky, V. Y. (1994) cGMP binding sites on photoreceptor phosphodiesterase: Role in feedback regulation of visual transduction. Proceedings of the National Academy of Sciences USA 91:4845-49. [arMDB, BMW] Creuzet, F., McDermott, A., Gebhard, R., van der Hoef, K., Spijker-Assink, M., Herzfeld, J., Lugtenburg, J., Levitt, M. H. & Criffin, R. G. (1991) Determination of membrane protein structure by rotational resonance NMR: Bacteriorhodopsin. Science 251:783-86. [SOS] Davis, S., Butcher, S. P. & Morris, R. G. (1992) The NMDA receptor antagonist D-2-amino-5-phosphonopentanoate (D-AP5) impairs spatial learning and LTP in vivo at intracerebral concentrations comparable to those that block LTP in vitro. Journal of Neuroscience 12:21-34. [aZX, TWA] Dawis, S. M. (1991) A molecular basis for Weber's law. Visual Neuroscience 7:285-320. [aMDB] Dawis, S. M., Graeff, R. M., Heyman, R. A.. Walseth, T. F. & Goldberg, N. D. (1988) Regulation of cyclic GMP metabolism in toad photoreceptors. Journal of Biological Chemistry 263:8771-85. [arMDB] de Crip, W. J. (1982a) Purification of bovine rhodopsin over concanavalin A-sepharose. Methods in Enzymology 81:197-207. [aPAH] (1982b) Thermal stability of rhodopsin and opsin in some novel detergents. Methods in Enzymology 81:256-65. [aPAH] de Grip, W. J., Olive, J. & Bovee-Geurts, P. H. M. (1983) Reversible modulation of rhodopsin photolysis in pure phosphatidylserine membranes. Biochimica et Biophysica Ada 734:168-79. [aPAH] de Grip, W. J., van Oostrum, J., Bovee-Geurts, P. H. M., van der Steen, R., van Amsterdam, L. J. P., Groesbeek, M. & Lugtenburg, J. (1990) 10,20methanorhodopsins: (7E.9E, 13E)- 10,20,methanorhodopsin and (7E.9Z, 13Z)-10,20-methanorhodopsin: 11-cis-locked rhodopsin analog pigments with unusual thermal and photo-stability. European Journal of Biochemistry 191:211-20. [aPAH] de Grip, W. J., van Oostrum, J. & De Caluw, C. L. J. (1992) Studies towards the crystallization of the rod visual pigment rhodopsin. Journal of Crystal Growth 122:375-84. [aPAH] Deese, A. J., Dratz, E. A., Dahlquist, F. W. & Paddy, M. R. (1981) Interaction of rhodopsin with two unsaturated phosphatidylcholines: A deuterium nuclear magnetic resonance study. Biochemistry 20:642027. [aPAH] Deisenhofer, J. & Michel, H. (1989) The photosynthetic reaction center from the purple bacterium Rhodopseudomonas viridis. Science 245:1463— 73. [aPAH] Demin, V. V., Yurkova, E. V., Kuzin, A. P., Barnakov, A. N. & Abdulaev, N. G. (1987). Crystallization of bovine retinal rhodopsin. In: Retinal proteins. VNU Science Press. [aPAH] Dencher, N. A., Dresselhaus, D., Zaccai, G. & Boldt, G. (1989) Structural changes in bacteriorhodopsin during proton translocation revealed by neutron diffraction. Proceedings of the National Academy of Sciences USA 86:7876-79. [aPAH] Derguini, F. & Nakanishi, K. (1986) Synthetic rhodopsin analogs. Photobiochemistnj i> Photobiophysics 13:259-83. [aPAH] Deterre, P., Bigay, J., Forguet, F., Robert, M. & Chabre, M. (1988) cGMP of phosphodiesterase of retinal rods is regulated by two inhibitory subunits. Proceedings of the National Academy of Sciences USA 85:242428. [aMDB] Detwiler, P. B. & Gray-Keller, M. P. (1992) Some unresolved issues in the physiology and biochemistry of phototransduction. Current Opinion in Neurobiology 2:433-38. [aMDB] Dhallan, R. S., Macke, J. P., Eddy, R. L., Shows, T. B., Reed, R. R., Yau, K. W. & Nathans, J. (1992) Human rod photoreceptor cGMP-gated channel: Amino acid sequence, gene structure, and functional expression. Journal of Neuroscience 12:3248-56. (aRSM] Dhallan, R. S., Yau, K. W , Schrader, K. A. & Reed, R. R. (1990) Primary structure and functional expression of a cyclic nucleotide-activated channel from olfactory neurons. Nature 347:184-87. [aRSM] Dittman, A. H., Weber, J. P., Hinds, T. R., Choi, E. J., Migeon, J. C , Nathanson, N. M. & Storm, D. R. (1994) A novel mechanism for coupling of m4 muscarinic acetylcholine receptors to calmodulin-sensitive adenylyl cyclases: Cross-over from G protein coupled inhibition to stimulation. Biochemistry 33:943-51. [aZX] Dizhoor, A. M., Chen, C.-K., Olshevskaya, E., Sinelnikova, V. V., Philipov, Re/erences/Controversies in Neuroscience III P. & Hurley, J. B. (1993) Role of the acylated amino terminus of recoverin in Ca+2-dcpendcnt membrane iutcraction. Science 259:829-32. (aJBH, SK) Dizhnor, A. M., Ericsson, L. H., Johnson, R. S., Kumar, S., Olshevakaya, E., Zozulya, S., Neubert, T. A., Stryer, L., Hurley, J. B. & Walsh, K. A. (1992) The NH2 terminus of retinal recoverin is acylated by a small family of fatty acids. Journal of Biological Chemistry 267:16033-36. [aJBH, KS] Dizhoor, A. M., Lowe, D. C , Olshevskaya, E. V., Laura, R. P. & Hurley, J. B. (1994) The human photorcceptor membrane guanylyl cyclase, RetGC, is present in outer segments and is regulated by calcium and a soluble activator. Neuron 12:1345-52. [aMDB] Dizhoor, A. M., Ray, S., Kumar, S., Niemi, C , Spencer, M., Brolley, D., Walsh, K. A., Philipov, P. P., Hurley, J. B. & Stryer, L. (1991) Recoverin: A calcium sensitive activator of retinal rod guanylate cyclase. Science 251:915-18. [aMDB, aJBH, SK, JFM, ASP] Dorset, D. L., Engel, A., Honor, M., Massalski, A. & Rosenbusch, J. P. (1983) Two-dimensional crystal packing of matrix porin. Journal of Molecular Biology 165:701-10. [aPAH] Downer, N. W. (1985) Cross-linking of dark-adapted frog photoreceptor disk membranes: Evidence for monomeric rhodopsin. Biophysical Journal 47:285-93. [aPAH] Dratz, E. A., Ftirstcnau, J., Hainni, H. E. & Hargrave, P. A. ( 1990) Probing the molecular mechanism of visual excitation using bioactive peptide sequences. Investigative Ophthalmology and Visual Science 31:79. [EAD] Dratz, E. A., Furstenau, J., Lambert, C. C , Thireault, D. L., Rarick, H., Schepers, T , Pakhlevaniants, S. & Hamm, H. E. (1993) NMR structure of a receptor-bound G-protein peptide. Nature 363:276-81. [EAD] Dratz, E. A., Hamm, H. E., Zwolinski, R., Furstenau, J. & Lambert, C. (1991) The 3D structure of the active interface between rhodopsin and G-protein determined using 2D NMR. Biophysical Journal 59:I89a. [EAD] Dratz, E. A. & Hargrave, P. A. (1983) The structure of rhodopsin and the rod outer segment disk membrane. Trends in Biochemical Sciences 8:12831. [aPAH, EAD) Dratz, E. A., Lewis, J. W., Schaechter, L. E., Parker, K. R. & Kliger, D. S. (1987) Retinal rod GTPase turnover rate increases with concentration: A key to the control of visual excitation? Biochemical and Biophysical Research Communications 146:379-86. [aMDB] Dratz, E. A., VaiiBrccincn, J. F. L., Kamps, K. M. P., Keegstra, W. & Vanliruggon, E. F. J. (1985) Two-dimensional crystallization of bovine rhodopsin. Biochimica ct Biophysica Ada 832:337-42. [aPAH, EAD] Dryer, S. E. & Henderson, D. (1991) A cyclic GMP-activated channel in dissociated cells of the chick pineal gland. Nature 353:756-58. [aRSM] Dryja, T. P. (1992) Doyne lecture: Rhodopsin and autosomal dominant retinitis pigmentosa. Eye 6:1-10. [aPAH, aSPD] Dryja, T. P., Berson, E. L., Rao, V. & Oprian, D. D. (1993) Heterozygous missense mutation in the rhodopsin gene as the cause of congenital stationary night blindness. Nature Ccnetics 4:280-83. [aSPD] Dryja, T. P., Hahn, L. B., McCee, T. L., Cowley, G. S. & Berson, E. L. (1991) Mutation spectrum of the rhodopsin gene in patients with autosomal dominant retinitis pigmentosa. Proceedings of the National Academy of Sciences USA 88:9370-74. [aSPDl Dryja, T. P., McGee, T. I., Hahn, L. B., Cowley, G. S., Olsson, J. E., Keichel, E., Sandbcrg, M. A. & Berson, E. L. (1990) Mutations within the rhodopsin gene in patients with autosomal dominant retinitis pigmentosa. New England Journal of Medicine 323:1302-7. [aSPD] Dryja, T. P., McCee, T. I., Rcichol, E., Hahn, L. B., Cowley, G. S., Yandell, D. W., Sandbcrg, M. A. & Berson, E. L. (1990a) A point mutation of the rhodopsin gone in one form of retinitis pigmentosa. Nature 343(6256):364-66. [aSPD] Duclai, V. (1988) Neurogenetic dissection of learning and short-term memory in Drosophila. Annual Review of Neuroscience 11:537-63. [aZX] Dudai, Y. & Zvi, S. (1984) Adenylate cyclase in the Drosophila memory mutant rutabaga displays an altered Ca2+ sensitivity. Neuroscience Utters 47:19-24. [aZX, TWA] (1985) Multiple defects in the activity of adenylyl cyclase from the Drosophila memory mutant rutabaga. Journal of Neurochcmistry 45:35564. [aZX] Dmnke, C. L., Arshavsky, V. Y., Calvert, P. D., Pugh, E. N. & Bownds, M. D. (1994) Rod outer segment structure influences the apparent kinetic parameters of cGMP phosphodiesterase. Journal of General Physiology 103:1071-98. [aMDB] Durell, S. R. & Guy, R. (1992) Atomic scale structure and fijnctional models of voltage-gated potassium channels. Biophysics Journal 62:238-50. [aRSM] Eisman, E., Brinigk, W. 6t Kaupp, U. B. (1993) Structural features of cyclic nucleotide-gated channels. In: Cellular physiology and biochemistry, ed. F. Lang. S. Karger AC. [rRSM[ Eismann, E., Miiller, F., Heinemann, S. H. & Kaupp, V. B. (1994) A single negative charge within the pore region of a cCMP-gated channel controls rectification, Ca2+ blockage and ionic selectivity. Proceedings of the National Academy of Sciences USA 91:1109-13. [rRSM] Erickson, M. A., Robinson, P. & Lisman, J. (1992) Deactivation of visual transduction without guanosine triphosphate hydrolysis by G protein. Science 257:1255-58. [aMDB, AY] Fahmy, K. & Sakmar, T. P. (1993) Regulation of rhodopsin-transducin interaction by a highly conserved carboxylic acid group. Biochemistry 32:7229-36. [aSPD] Fain, G. L. (1976) Sensitivity of toad rods: Dependence on wavelength and background illumination. Journal of Physiology (London) 261:71101. [aMDB] Fain, G. L. & Cornwall, M. C. (1993) Light and dark adaptation in vertebrate photoreceptors. In: Contrast Sensitivity from receptors to clinic, ed. R. Shapley & D.-K. Lam. MIT Press. [RKC] Fain, G. L., Lamb, T. D., Matthews, H. R. & Murphy, R. L. W. (1989) Cytoplasmic calcium concentration as the messenger for light adaptation in salamander rods. Journal of Physiology 416:215-43. [HRM] Fain, G. L. & Lisman, J. E. (1993) Photoreceptor degeneration in vitamin A deprivation and retinitis pigmentosa: The equivalent light hypothesis. Experimental Eye Research 57:335-40. [aMDB] Fain, G. L. & Matthews, H. R. (1990) Calcium and the mechanism of light adaptation in vertebrate photoreceptors. Trends in Neuroscience 13:37884. [HRM] Farber, D. B., Seager, Danciger, J. & Aguirre, G. (1992) The B subunit of cyclic GMP phosphodiesterase mRNA is deficient in canine rod-cone dysplasia 1. Neuron 9:349-56. [MWK] Famsworth, C. & Dratz, E. A. (1976) Oxidative damage of retinal rod outer segment membranes and the role of vitamin E. Biochimica et Biophysica Ada 443:556-70. [aPAH] Farrar, G. J., Findlay, J. B. C , Kumar-Singh, R., Kenna, P., Humphries, M. M., Sharpe, E. & Humphries, P. (1992) Autosomal dominant retinitis pigmentosa: A novel mutation in the rhodopsin gene in the original 3q linked family. Human Molecular Cenetics 1:769-71. [aSPD] Farrar, G. J., Jordan, S. A., Kenna, P., Humphries, N. M., Kumar-Singh, R., McWilliams, P., Allamand, V., Sharp, E. & Humphries, P. (1991) Autosomal dominant retinitis pigmentosa: Localization of a disease gene (RP6) to the short arm of chromosome 6. Cenomics 11:870-4. [aSPD] Farrar, G. J., Kenna, P., Redmond, R., McWilliams, P., Bradley, D. G., Humphries, M. M., Sharp E. M., Inglehearn, C. F., Bashir, R., Jay, M., Waytty, A., Ludwig, M., Schinzel, A., Samanns, C , Gal, A., Bhattacharya, S. & Humphries, P. (1990) Autosomal dominant retinitis pigmentosa: Absence of the rhodopsin proline—histidine substitution (codon 23) in pedigrees from Europe. American Journal of Human Cenetics 47:941-45. [aSPD] Farrar, G. J., Kenna, P., Redmond, R., Shiels, D., McWilliams, P., Humphries, M. M., Sharp, E. M., Jordan, S., Kumar-Singh, R. & Humphries, P. (1991a) Autosomal dominant retinitis pigmentosa: A mutation in codon 178 of the rhodopsin gene in two families of Celtic origin. Cenomics 11:1170-71. [aSPD] Farrar, G. J., McWilliams, P., Bradley, D. G., Kenna, P., Lawler, M., Sharp, E. M., Humphries, N. M., Eiberg, H., Conneally, P. M., Trofatter, J. A. & Humphries, P. (1990) Autosomal dominant retinitis pigmentosa: Linkage to rhodopsin and evidence for genetic heterogeneity. Genomics 8:35-40. [aSPD] / Feany, M. B. (1990) Rescue of the learning defect in dunce, a Drosophila learning mutant, by an allele of rutabaga, a second learning mutant. Proceedings of the National Academy of Sciences USA 87:279599. [aZX] Federman, A. D., Conklin, B. R., Schrader, K. A., Reed, R. R., Bourne, H. R. (1992) Hormonal stimulation of adenylyl cyclase through G,-protein beta gamma subunits. Nature 356:159-61. Feher, G., Atten, J. P., Okamura, M. Y. & Rees, D. C. (1989) Structure and function of bacterial photosynthetic reaction centers. Nature 339:11116. [aPAH] Feinstein, P. G., Schrader, A., Bakalyar, H. A., Tang, W. J., Krupinski, J., Oilman A. G. & Reed R. R. (1991) Molecular cloning and characterization of a calcium calmodulin insensitive adenylyl cyclase (typell) from rat brain. Proceedings of the National Academy of Sciences USA 88:1017377. |aZX] Ferrell, R. E., Hittner, H. M. & Antoszyk, J. H. (1983) Linkage of atypical vitelliform macular dystrophy (GPT1) locus. American Journal of Human Genetics 35:78-84. [aSPD] Fesenko, E. E., Kolensnikov, S. S. & Lyubarsky, A. L. (1985) Induction by cyclic GMP of cationic conductance in plasma membrane of retinal rod outer segments. Nature 313:310-13. [aRSM, DDO, TGW] Findlay, J. B. C. (1986). The biosynthetic, functional, and evolutionary BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 507 References /Controversies in Neuroscience III implications of the structure of rhodopsin. In: The molecular mechanism of photoreception. Springor-Verlag. [aPAH] Firestein, S. (1991) A noseful of odor receptors. Trends in Neuroscience 14:270-72. [aMDB] Fishman, G. A . (1990) Inherited macular dystrophies: A clinical overview. Australia and New Zealand Journal of Ophthalmology 18:12328. [aSPD] (1978) Retinitis pigmentosa: Genetic percentages. Archives of Ophthalmology 96:822-26. [aSPD] Fishman, G. A., Alexander, K. R. & Anderson, R. J. (1985) Autosomal dominant retinitis pigmentosa: A method of classification. Archives of Ophthalmology 103:366-74. [aSPD] Fishman, C. A., Stone, E. M., Cilbert, L. D., Kenna, P. & Sheffield, V. C. (1991) Ocular findings associated with a rhodopsin gene codon 58 transversion mutation in autosomal dominant retinitis pigmentosa. Archives of Ophthalmology 109:1387-93. [aSPD] Fishman, G. A., Stone, E. M., Gilbert, L. D. & Sheffield, V. C. (1992) Ocular findings associated with a rhodopsin gene codon 106 mutation glycine-to-arginine change in autosomal dominant retinitis pigmentosa. Archives of Ophthalmology 110:646-53. [aSPD] Fishman, G. A., Stone, E. M., Sheffield, V. C., Gilbert, L. D. & Kimura, A. E. (1992a) Ocular findings associated with rhodopsin gene eodon 17 and codon 182 transition mutation in dominant retinitis pigmentosa. Archives of Ophthalmology 110:54-62. [aSPD] Fishman, G. A., Vandenburgh, K., Stone, E M . , Gilbert, L. D., Alexander, K. R. & Sheffield, V. C. (1992b) Ocular findings associated with rhodopsin gene codon 267 and codon 190 mutations in dominant retinitis pigmentosa. Archives of Ophthalmology 110:1582-88. [aSPD] Flaherty, K. M., Zozulya, S., Stryer L. & McKay D. H. (1993) Threedimensional structure of recoverin, a calcium sensor in vision. Cell 75:709-16 [aMDB, aJBH, KWK] Flieslcr, S. J. & Anderson, R. E. (1983) Chemistry and metabolism of lipids in the vertebrate retina. Progress in Lipid Research 22:79-131. [aMDB] Fong, S.-L., Tsin, A. T. C , Bridges, C. D. B. & Lion, G. I. (1982) Detergents for extraction of visual pigments: Types, solubilization, and stability. Methods in Enzymology 81:133-40. [aPAH] Forti, S., Menini, A., Rispoli, C. & Torre, V. (1989) Kinetics of phototransduction in retinal rods of the newt Triturus cristatus. Journal of Physiology (London) 419:265-95. [aMDB] Fowles, C , Charnia, R. & Akhtar, M. (1988) Mechanistic studies on the phosphorylation of photoexcited rhodopsin. FEBS Letters 238:5660. [aMDB] Franke, R. R., Konig, B., Sakmar, T. P., Khorana, H. G. & Hofmann, K. P. (1990) Rhodopsin mutants that bind but fail to activate transducin. Science 250:123-25. [aPAH, EAD] Franke, R. R., Sakmar, T. P., Graham, R. M. & Khorana, H. G. (1992) Structure and function in rhodopsin: Studies of the interaction between the rhodopsin cytoplasmic domain and transducin. Journal of Biological Chemistry 267:14767-74. [aPAH] Franke, R. R., Sakmar, T. P., Oprian, D. D. & Khorana, H. G. (1988) A single amino acid substitution in rhodopsin (Lysine 248 Leucine) prevents activation of transducin. Journal of Biological Chemistry 263:211922. [aSPD] Frey, U., Huang, Y. Y. & Kandel, E. R. (1993) Effects of eAMP simulate a late stage of LTP in hippocampal CA1 neurons. Science 260:166164. |aZX, EDR] Frey, U., Krug, M., Reynumn, K. G. & Matthies, H. (1988) Anisomyein, an inhibitor of protein synthesis, blocks late phases of LTP phenomena in the hippocampal CA1 region in vitro. Brain Research 45:257-65. [EDR, aZX] Frey, U., Matthies, H., Reymann, K. G. & Matthies, H. (1991). The effect of dopaminergic Dl receptor blockade during totalization on the expression of long-term potentiation in the rat CA1 region in vitro. Neuroscience Letters 29:111-14. [aZX] Frost, W. N., Castellucei, V. F., Hawkins, R. D. & Kandel, E. R. (1985) Monosynaptic connections from the sensory neurons of the gill- and siphon-withdrawal reflex in Aplysia participate in the storage of long-term memory for sensitization. Proceedings of the National Academy of Sciences USA 82:8266-69. [aZX] Fryxell, K. J. & Meyerowitz, E. M. (1991) The evolution of rhodopsins and ncurotransmitter receptors. Journal of Molecular Evolution 33:36768. [aSPD] Fujiki, K., Hotta, Y., Hayakawa, M., Sakuma, H., Shiono, T , Noro, M., Shakuma, T., Tamai, M., Hikiji, K., Kawaguchi, R. & Hoshi, A. (1992) Point mutations of rhodopsin gene found among Japanese families with antosomal dominant retinitis pigmentosa (ADRP). Japanese Journal of Human Genetics 37:125-32. [aSPD] Fukuda, M. N., Papermaster, D. S. & Hargrave, P. A. (1979) Rhodopsin 508 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 carbohydrate: Structure of small oligosaecharides attached at two sites near the NH2 terminus. Journal of Biological Chemistry 254:82017. [aPAH] Fukada, Y., Schichida, Y, Yoshizawa, T., Ito, M., Kodama, A. & Tsukida, K. (1984) Studies on structure and function of rhodopsin by use of cyclopentatrienylidene 11-cis-locked-rhodopsin. Biochemistry 23:582632. [aPAH] Fukada, Y., Takao, T , Ohguor, H., Yoshizawa, T., Akino, T. & Shimonishi, Y. (1990) Farnesylated gamma subunit of photoreceptor G protein indispensable for GTP-binding. Nature 346:658-60. [aMDB] Fung, B. K.-K. (1993) Characterization of transducin from bovine retinal rod outer segments: 1. Separation and reconstitution of the subunits. Journal of Biological Chemistry 258:10495-502. [rM DB] Fung, B. K.-K. & Griswold-Prenner, I. (1989) G-protein-effector coupling: Binding of rod phosphodiesterase inhibitory subunit to transducin. Biochemistry 28:3133-37. [AY] Fung, B. B.-K., Hurley, J. B. or Stryer, L. (1981) Flow of information in the light-triggered cyclic nucleotide cascade of vision. Proceedings of the National Academy of Sciences USA 78:152-56. [aM DB, AY] Fung, B. K.-K., Young, ]. H., Yamane, H. K. & Griswold-Prenner, I. (1990) Subunit stoichiometry of retinal rod cGMP phosphodiesterase. Biochemistry 29:2657-64. [aMDB] Furman, R. E. & Tanaka, J. C. (1990) Monovalent selectivity of the cyclic guanosine monophosphatu-activated ion channel. Journal of General Physiology 96:57-82. [aRSM] Gal, A., Artlich, A., Ludwig, M., Niemeyer, C., Olek, K., Schwinger, E. or Schinzel, A. (1991) Pro-347-Arg mutation of the rhodopsin gene in autosomal dominant retinitis pigmentosa. Cenomics 11:468-70. [aSPD] Gal, A., Orth, U., Baehr, W., Schwinger, E., & Rosenberg, T. (1994) Heterozygous missense mutation in the rod cGMP phosphodiesterase G-subunit gene in autosomal dominant stationary night blindness. Nature Genetics 7, 551:64-67. [aSPD] Gannon, A. M., Rodriguez, J. A., Humphries, P., Birch, D. G., Heckenlively, J. R. & Daiger, S. P. (1993) Mutations in peripherin/RDS in patients with retinitis pigmentosa: A 12 base-pair deletion in exon 2. American Journal of Human Genetics 53:1160. [aSPD] Gao, B. or Gilman, A. (1991) Cloning and expression of a widely distributed (type IV) adenylyl cyclase. Proceedings of the National Academy of Sciences USA 88:10178-82. [aZXl Garavito, R. M. (1991). Crystallizing membrane proteins: Experiments on different systems. In: Crystallization of membrane proteins, ed. H. Michel. CRC Press. [aPAH] (in press) Strategies for crystallizing membrane proteins. Journal of Bioenergetics and Biomembrancs. [RMG] Garavito, R. M. & Picot, D. (1990) The art of membrance protein crystallization. Methods: A Companion to methods in Enzymology 1:5769. [RMG] Cery, I., Chanaud III, N. P. & Anglade, E. (1994) Recoverin is highly uveitogenic in Lewis rats. Investigative Ophthalmology and Visual Science 35:3342-45. [ASP] Ghirardi, M., Braha, O., Hochner, B., Montarolo, P. G., Kandel, E. R. & Dale, N. (1992) Roles of PKA and PKC in facilitation of evoked and spontaneous transmitter release at depressed and nondepressed synapses in Aplysia sensor)' neurons. Neuron 9:479-89. [aZX, TWA] Gibson, F., Walsh, J., Mburu, P., Varela, A., Brown, K. A., Antonio, M., Belsel, K. W., Steel, K. P. & Brown, S. D. M. (1995) A type VII myosin encoded by the mouse deafiiess gene shaker-1. Nature 374:6264. [rSPD] Gillespie, P. G. & Beavo, J. A. (1988) Characterization of a bovine cone photoreceptor phosphodiesterase purified by cyclic GMP-sepharose chromatography. Journal of Biological Chemistry 263:8133-41. [aM DB] (1989) cGMP is tightly bound to bovine retinal rod phosphodiesterase. Proceedings of the National Acadenuj of Sciences USA 86:4311— 15. [aMDB] Glatt, C. E. or Snyder, S. H. (1993) Cloning and expression of an adenylyl cyclase localized to the corpus striatum. Nature 363:679-80. [aZX] Gnegy M., Muirhead, N., Roberts-Lewis, J. M. & Treisman, G. (1984) Calmodulin stimulates adenylyl cyclase activity and increases dopamine activation in bovine retina. Journal of Neuroscience 4:2712-17. [aZX] Gnegy, M. & Treisman, G. (1981) Effect of calmodulin on dopamine-sensitive adenylate cyclase activity in rat striatal membranes. Molecular Pharmacology 19:256-63. [aZX] Gold, G. H. & Korenbrot, J. J. (1980) Light-induced calcium release by intact retinal rods. Proceedings of the National Academy of Sciences USA 77:5557-61. [TGW] Goldberg, M. F. (1994) Molecular heterogeneity in retinitis pigmentosa. Ophthalmic Genetics 15(2):47-50 [AB] Goldsmith, B. A. & Abrams, T. W. (1991) Reversal of synoptic depression by References/Controversies in Neuroscience HI serotonin ut Aplysia sensory neuron synapses involves activation of adenylate cyclase. Proceedings of the National Academy of Sciences USA 88:9021-25. [TWA] Gorczyca, W. A., Cray-Keller, M. P., Detwiler, P. B. & Palczewski, K. (1994) Purification and physiological evaluation of a guanylate eyclase activating protein from retinal rods. Proceedings of the National Academy of Sciences USA 91:4014-18. [aMDB] Gordon, J. 1., Duronio, R. J., Rudnick, D. A., Adams, S. P. & Gokel, G. W. (1991) Protein iV-myristoylation. Journal of Biological Chemistry 266:8647-50. [KS] Gordon, S. E., Brautigan, D. L. & Zimmerman, A. L. (1992) Protein phosphatases modulate the apparent agonist affinity of the ligandregulated ion channel of retinal rods. Neuron 8:739-48. [aMDB, aRSM, RLB. LWH, TGW] Gordon, S. E-, Downing-Park, J. & Zimmerman, A. L. (in press) Modulation of the cGMP-gated ion channel in frog rods by calmodulin and an endogenous inhibitory factor. Journal of Physiology. [MPG-K, rRSM] Gordon, S. E. & Zimmerman, A. L. (1994) Modulation of the rod cCMPgated ion channel by calmodulin and an endogenous factor distinct from calmodulin. Biophysics Journal 66-.A355. [MSS, TCW] Gorodovikova, E. N. & Philippov, P. P. (1993) The presence of a calciumsensitive p26-containing complex in bovine retinal rod cells. FEBS Letters 335:277-79. ISK] Goslin, K., Schreyer, D. J., Skene, J. H. P. & Banker, G. (1988) Development of neuronal polarity: GAP-43 distinguishes axonal from dendritic growth cones. Nature 336:672-74. |EDR] Colliding, E. H., Ngai, J., Kramer, R. H., Colicos, S., Axel, R., Siegelbaum, S. A. & Chess, A. (1992) Molecular cloning and single channel properties of the cyclic nuclco tide-gated channel from catfish olfactory' neurons. Neuron 8:45-58. [aRSM] Colliding, E. H., Tibbs, G. R., Liu, D. & Siegelbaum, S. A. (1993) Role of the 115 domain in determining pore diameter and ion permeation through cyclic nuclcotide-gated channels. Nature 364:61-64. [aRSM] Colliding, E. H., Tibbs, C. R. & Siegelbaum, S. A. (1994) Molecular mechanism of cyclic-nucleotide-gated channel activation. Nature 372:36974. [rRSM] Grant, S. G. N., O'Dell, T. J., Karl, K. A., Stein, P. L., Soriano, P. & Kandel, E. R. (1992) Impaired long-term potentiation, spatial learning and hippocampal development in fyn mutant mice. Science 258:190392. [aZX] Gray-Keller, M. P., Bicnibaiim, M. S. & Bownds, M. D. (1990) Transducin activation in electropermeabilized frog rod outer segments is highly amplified, and a portion equivalent to phosphodiesterase remains membrane-bound. Journal of Biological Chemistry 265:1532332. |aMDB] Gray-Keller, M. P. & Detwiler, P. B. (1994a) Intracellular calcium measurements in isolated rod photoreceptors. Investigative Ophthalmology and Visual Science 35:1486. [aMDB] (1994b) Tlie calcium feedback signal in the phototransduction cascade of vertebrate rods. Neuron 13(4):849-61. [MPC-K, MSS, TCW, rMDB] Gray-Keller, M. P., Pblans, A. S. Palczewski, K. 6c Derwiler, P. B. (1993) The effect of recoverin-like calcium-binding proteins on the photoresponse of retinal rods. Neuron 10:523-31. [aMDB, aJBH, MPC-K, LWH] Grecksch, G. & Matthies, H. (1980) Two sensitive periods for the amnesic effect of anisomycin. Pharmacology and Biochemistry of Behavior 12:66365. [aZX] Greenberg, J., Goliath, R., Brighton, P., & Raniesar, R. (1994) A new locus for autosomal dominant retinitis pitmentosa on the short arm of chromosome 17. Human Molecular Cenetics 6:915-18. [aSPD] Greenberg, S. M., Castellueci, V. F., Bayley, H. or Schwartz, J. H. (1987) A molecular mechanism for long-term sensiti'zation in Aplysia. Neuron 329:62-65. [aZXl Greengard, P. (1979) Cyclic nucleotides, phosphorylated proteins, and the nervous system. Federation Proceedings 38:2208-17. [TWA] Groskoph, S. L., Hammett, D. J. & Cote, R. H. (1992) Nucleotide energy metabolism in electropormeabilizcd rod photoreceptors. Investigative Ophthalmology and Visual Science 33:1008. [aM DB] Cullion, T & Schaefer, J. (1989) Detection of weak heteronuclear dipolar coupling by rotational echo double resonance nuclear magnetic resonance. In: Advances in magnetic resonance, vol. 13., ed. W. S. Warren. Academic Press. [SOS] Custafsson, B., Wigstrom, H., Abraham, W. C. & Huang, Y. Y. (1987) Longterm potentiation in the hippocampus using depolarizing current pulses as the conditioning stimulus to single volley synaptic potentials. Journal of Neuroscience 7:774-80. [TWA] Hagiwara, M., Brindle, P., Harootunian, A., Armstrong, R., Rivier, J., Vale, J., Tsien, R. & Montminy, M. R. (1993) Coupling of hormonal stimulation and transcription via the cAMP responsive factor CREB is rate limited by nuclear entry of protein kinase A. Molecular and Cellular Biology 13:4852-59. [aZX] Hakki, S. & Sitaramayya, A. (1990) Guanylate cyclase from bovine rod outer segments: Solubilization, partial purification, and regulation by inorganic pyrophosphate. Biochemistry 29:1088-94. [KWK] Hall, M. D., Hoon, M. A., Ryba, N. J. P., Pottinger, J. D. D., Keen, J. N., Sabil, H. R. & Findlay, J. B. C. (1991) Molecular cloning and primary structure of squid (Loligo forbesi) rhodopsin, a phospholipase C-directed C-protein-linked receptor. Biochemistry Journal 274:35-40. [aPAH] Halloran, S. L. (1985) A multidisciplinary analysis of retinitis pigmentosa: Genetic, cpidemiological and clinical parameters. Doctoral thesis, Virginia Commonwealth University. [aSPD] Hamm, H. (1990) Regulation by light of cyclic nucleotide-dependent protein kinases and their substrates in frog rod outer segments. Journal of General Physiology 95:545-67. [aMDB] Hamm, H. E. & Bownds, M. D. (1986) Protein complement of rod outer segments of frog retina. Biochemistry 25:4512-23. [aMDB] Hamm, H. E., Deretic, D., Arendt, A., Hargrave, P. A., Koenig, B. & Hofmann, K. P. (1988) Site of C protein binding to rhodopsin mapped with synthetic peptides from the alpha subunit. Science 241:83235. [EAD] Hamm, H. E. & Rarick, H. M. (1994) Use of synthetic peptides for mapping G protein interaction sites with receptors. Methods in Enzymology 237:423-36. [RKC1 Han, M., DeDecker, B. & Smith, S. O. (1993) Localization of the retinal protonated schiff base counterion in rhodopsin. Biophysics Journal 65:899-926. [SOS] Han, M. or Smith, S. O. (in press) NMR constraints on the location of the retinal chromophore in rhodopsin and bathorhodopsin. Biochemistry. [SOS] Harbison, G. S., Smith, S. O., Pardoen, J. A, Winkel, D. C , Lugtenburg, J., Herzfeld, J., Mathies, R. & Griffln, R. G. (1984) Dark-adapted Bacteriorhodopsin contains 13-cis, 15-si/n and a\\-trans, 15-anti retinal schiff bases. Proceedings of the National Academy of Sciences USA 81:1706-9. [SOS] Hardie, R. C. & Minke, B. (1992) The trp gene is essential for a lightactivated Ca2+ channel in Drosophila photoreceptors. Neuron 8:64351. [aRSM] Hargrave, P. A. & Fong, S.-L. (1977) The amino-and carboxyl-terminal sequence of bovine rhodopsin. Journal of Supramolecular Structure 6:559-70. [rPAH] Hargrave, P. A., Hamm, H. E. & Hofmann, K. P. (1993) Interaction of rhodopsin with the G-protein, transducin. BioEssays 15:43-50. [aPAH] Hargrave, P. A. & McDowell, J. H. (1992a) Rhodopsin and phototransduction: A model system for G protein-linked receptors. FASEB Journal 6:232331. [aMDB, aSPD, aPAH] (1992b) Rhodopsin and phototransduction. International Review of Cytology 137B:49-97. [aSPD] (1993) Rhodopsin and phototransduction. International Review of Cytology 137B:49-97. |aPAH] Hargrave, P. A., McDowell, J. H., Feldmann, R. J., Atkinson, P. H., Rao, J. K. M. & Argos, P. (1984) Rhodopsin's protein and carbohydrate structure: Selected aspects. Vision Research 24:1487-99. [aPAH] Hargrave, P. A. & O'Brien, P. J. (1991) Speculations of the molecular basis of retinal degeneration in retinitis pigmentosa. In: Retinal degeneration, ed. J. G. Hollyfield, R. E. Anderson & M. M. LaVail. CRC Press. [aSPD] Hayakawa, M., Hotta, Y, Imai, Y., Fujiki, K., Nakamura, A., Yanashima, K. & Kanai, A. (1993) Clinical features of autosomal dominant retinitis pigmentosa with rhodopsin gene codon 17 mutation and retinal neovascularization. American Journal of Ophthalmology 115:16873. [aSPD] Hayashi, F. (1994) Light-dependent in vivo phosphorylation of an inhibitory subunit of cGMP-phosphodiesterase in frog rod photoreceptor outer segments. FEBS Letters 338:203-6. [aMDB] Hayashi, F., Lin, G. Y., Matsumoto, H. & Yamazaki, A. (1991) Phosphatidylinositol-stimulated phosphorylation of an inhibitory subunit of cGMP phosphodiesterase in vertebrate rod photoreceptors. Proceedings of the National Academy of Sciences USA 88:433337. [aMDB] Haynes, L. W. (1992) Block of the cyclic CMP-gated channel of vertebrate rod and cone photoreeeptors by L-cis-diltiazem. Journal of General Physiology 100:783-801. [LWH] Haynes, L. W. & Yau, K.-W. (1985) Cyclic CMP-sensitive conductance in outer segment membrane of catfish cones. Nature 317:61-64. [aRSM] Hearst, E. (1988) Fundamentals of learning and conditioning. In: Steven's handbook of experimental psychology, ed. R. C. Atkinson, R. J. Herrnstein, G. Lindzey & R. D. Luce. Wiley Interscience. [TWA] Hebert, H., Xian, Y., Hacksell, I. & Mardh, S. (1992) Two-dimensional BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 509 Re/erences/Controversies in Neuroscience III crystals of membrane-bound gastric H,K-ATPase. FEBS Letters 299:15962. [aPAH] Heck, M. & Hofmann, K. P. (1993) G-protein-effector coupling: A real-time light-scattering assay for transducin-phosphodiesterase interaction. Biochemistry 32:8220-27. [aMDB] Heckenlively, J. R. (1988) Rctinitis pigmentosa. Lippincott. [aSPD] Heckenlively, J. R., Rodriguez, J. A. & Daiger, S. P. (1990) Autosomal dominant sectoral retinitis pigmentosa: Two families with transversion mutation in codon 23 of rhodopsin. Archives of Ophthalmology 109:8491. [aSPD] Heginbotham, L., Abramson, T. & MacKinnon, R. (1992) A functional connection between the pores of distantly-related ion channels as revealed by mutant K+ channels. Science 258:1152-55. [aRSM] Hcinonen, E. & Akerman, K. E. (1987) Intracellular free magnesium in synaptosomes measured with entrapped eriochrome blue. Biochimica et Biophysica Ada 898:331-37. [RLH] Hekman, M., Bauer, P. H., Sohlemann, P. & Lohse, M. J. (1994) Phosducin inhibits receptor phosphorylation by the B-adrenergic receptor kinase in a PKA-regulated manner. FEBS Letters 343:120-24. |BMW] Henderson, R., Baldwin, J. M., Ceska, T. A., Zemlin, F., Beckmann, E. & Downing, K. H. (1990) Model for the structure of baeteriorhodopsin based on high-resolution electron cryomicroscopy. Journal of Molecular Biology 213:899-929. [aPAH, EAD. RMC] Henderson, R. & Unwin, P. N. T. (1975) Three-dimensional model of purple membrane obtained by electron microscopy. Nature (London) 275:2832. [RMG] Hcrmolin, J., Karell, M. A., Hamm, H. E. & Bownds, M. D. (1982) Calcium and cyclic GMP regulation of light-sensitive protein phosphorylation in frog photoreceptor membranes. Journal of General Physiology 79:63355. [aMDB] Hidaka, H. & Okazaki, K. (1993) Neurocalcin family: A novel calcium-binding protein abundant in bovine central nervous system. Neuroscience Research 16:73-77. [KT] Hille, B. (1992) Ionic channels of excitable membranes, 2d ed. Sinauer. [TWA] Hodgkin, A. L., McNaughton, P. A. & Nunn, B. J. (1985) The ionic selectivity and calcium dependence of the light-sensitive pathway in toad rods. Journal of Physiology 358:447-68. [aRSM] Hodgkin, A. L. & Nunn, B. J. (1988) Control of light-sensitive current in salamander rods. Journal of Physiology (London) 403:439-71. [aMDB] Hofmann, K. P. (1986) Photoproducts of rhodopsin in the disc membrane. Photochemistry 6 Photobiophysics 13:309-27. [aPAH] Hofmann, K. P., Pulvermuller, A., Buczylko, J., Van Hooser, P. & Palczewski, K. (1992) The role of arrestin and retinoids in the regeneration pathway of rhodopsin. Journal of Biological Chemistry 267:15701-6. [aMDB, RKC] Hong, K. & Hubbell, W. L. (1973) Lipid requirements for rhodopsin regenerability. Biochemistry 12:4517-23. [aPAH] Hopkins, W. F. & Johnston, D. (1988) Noradrenergic enhancement of longterm potentiation at mossy fiber synapses in the hippocampus. Journal of Neurophysiology 59:667-87. [aZX] Horn, M., Humphries, P., Kunisch, M., Marchese, C , Apfelstedt-Sylla, E., Fugi, L., Zrenner, E., Kenna, P., Gal, A. & Farrar, J. (1992) Deletions in exon 5 of the human rhodopsin gene causing a shift in reading frame and autosomal dominant retinitis pigmentosa. Human Genetics 90:25557. [aSPD] Hotta, Y., Shiono, T , Hakawa, M., Hashimoto, T , Kanai, A., Nakajima, A., Noro, M., Sakuma, T. & Fujiki, K. (1992) Molecular biological study of the rhodopsin gene in Japanese patients with autosomal dominant retinitis pigmentosa. Nippon Canka Gakkai Zasshi 96:237-42. [aSPD] Hovmuller, S., Slaughter, M., Berriman, J., Karlsson, B., Weiss, H. & Leonard, K. (1983) Structural studies of cytochrome reductase: Improved membrane crystals of the enzyme complex and crystallization of a subcomplex. Journal of Molecular Biology 165:401-6. [aPAH] Hsu, Y.-T. & Molday, R. S. (1993) Modulation of the cGMP-gated channel of rod photoreceptor cells by calmodulin. Nature 361:76-79. [aMDB, aJBH, aZX, MPG-K, LWH, DDO] (1994) Interaction of calmodulin with the cyclic GMP-gated channel of rod photoreceptor cells. Journal of Biological Chemistry 269:2976570. |rRSM] Hu, S., Franklin, P. J., Wang, J., Ruiz Silva, B. E., Derguini, F. & Nakanishi, K. (1994) Unbleachable rhodopsin with an 11-cis-locked eightmembered ring retinal: The visual transduction process. Biochemistry 33:408-16. (aPAH] Huang, P. C , Gaitan, A. E., Hao, Y., Petters, R. M. & Wong, F. (1993) Cellular interactions implicated in the mechanism of photoreceptor degeneration in transgenic mice expressing a mutant rhodopsin gene. Proceedings of the National Academy of Sciences USA 90:8484-88. [M WK] 510 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 Humphries, P., Farrar, G. J. & Kenna, P. (1993) Autosomal dominant retinitis pigmentosa: Molecular, genetic and clinical aspects. Progress in Retinal Research 12:231-45. [aSPD] Humphries, P., Kenna, P. & Farrar, G. J. (1992) On the molecular genetics of retinitis pigmentosa. Science 256:804-8. [aSPD] Hundal, S. P., Difrancesco, D., Mangoni, M., Grammar, W. J. & Conley, E. C. (1993) An iso form of the cGMP-gated retinal photoreceptor channel gene expressed in the sinoa trial node (pacemaker) region of rabbit heart. Biochemical Society Transcripts 21:119S. [aRSM] Hurley, J. B. (1992) Signal transduction enzymes of vertebrate photoreceptors. Journal of Bioenergetics and Biomembranes 24:219-26. [aMDB] Hurley, J. B., Dizhoor, A. M., Ray, S. & Stryer, L. (1993) Recovering role: Conclusion withdrawn. Science 260:740. [aMDB, aJBH] Hurley, J. B. & Stryer, L. (1982) Purification and characterization of the gamma regulatory subunit of the cGMP phosphodiesterase from retinal rod outer segments. Journal of Biological Chemistry 257:1109499. [aMDB] Hurwitz, R. & Holcombe, V. (1991) Affinity purification of the photoreceptor cGMP- gated cation channel. Journal of Biological Chemistry 266:797577. [aRSM, RLH, TGW] Illing, M., Colville, D. A., Williams, A. J. & Molday, R. S. (1994) Sequencing, cloning and characterization of the cyclic nucleotide-gated channel complex of rod outer segments. Investigative Ophthalmology and Visual Science [Abstract 1022] 35:1474. [rRSM] Impey, S., Wayman, G., Wu, Z. & Storm, D. R. (in press) The type I adenylyl cyclase functions as a coincidence detector for eontrpl of CREmediated transcription by Ca 2 + and isoproterenol. Molecular and Cellular Biology. Inglehearn, C. F., Bashir, R., Lester, D. H., Jay, M., Bird, A. C. & Bhattacharya, S. S. (1991) A 3-bp deletion in the rhodopsin gene in a family with autosomal dominant retinitis pigmentosa. American Journal of Human Genetics 48:26-30. [aSPD] Inglehearn, C. F., Carter, S. A., Keen, T. J., Lindsey, J., Stephenson, A. M., Bashir, R., Al-Maghtheh, M., Moore, A. T., Jay, M., Bird, A. C. & Bhattacharya, S. S. (1993) A new locus for autosomal dominant retinitis pigmentosa (adRP) on chromosome 7p. Nature Genetics 4:51— 53. [aSPD] Inglehearn, C. F., Keen, T. J., Bashir, R., Jay, M., Fitzke, F., Bird, A. C , Crombie, A. & Bhattacharya, S. S. (1992) A complete screen for mutations of the rhodopsin gene in a panel of patients with autosomal dominant retinitis pigmentosa. Human Molecular Genetics 1:41— 45. [aSPD] Inglese, J., Freedman, N. J., Koch, W. J. & Lefkowitz, R. J. (1993) Structure and mechanism of the G protein-coupled receptor kinases. Journal of Biological Chemistry 268:23735-38. [KT] Ishikawa, Y., Katsushika, S., Chen, L., Halnon, N. J., Kawabe, J. & Homey, C. J. (1992) Isolation and characterization of a novel cardiac adenylyl cyclase cDNA. Journal of Biological Chemistry 267:13553-57. [aZX] Ito, M., Kodama, A., Tsukida, K., Fukada, Y, Shichida, Y. & Yoshizawa, T. (1982) A novel rhodopsin analogue possessing the cyclopentatrienylidene structure as the 11-cis-locked and the full planar chromophore. Chemical 6 Pharmaceutical Bulletin 30:1913-16. [aPAH] Itoh, S., Takshima, A. & Macda, Y. (1992) Protective effect of cerulein on memory impairment induced by protein synthesis inhibitors in rats. Peptides 13:1007-12. [EDR] Jacobowitz, O., Chen, J., Premont, R. T. & Iyengar, R. (1993) Stimulation of specific types of Gs-stimulated adenylyl cyclases by phorbol ester treatment. Journal of Biological Chemistry 268:3829-32. [aZX] Jacobson, F. A. S. & Baseom, G. I. (1993) Phosducin as a candidate gene for retinitis pigmentosa and Usher syndrome type H. Investigative Ophthalmology and Visual Science 34:1458. [aSPD] Jacobson, S. G., Kemp, C. M., Cideciyan, A. V., Macke, J. P., Sung, C.-H. & Nathans, J. (1994) Phenotypes of stop codon and splice site rhodopsin mutations causing retinitis pigmentosa. Investigative Ophthalmology and Visual Science 35:2521-34. [aSPD, AB] Jacobson, S. G., Kemp, C. M., Sung, C.-H. & Nathans, J. (1991) Retinal function and rhodopsin levels in autosomal dominant retinitis pigmentosa with rhodopsin mutations. American Journal of Ophthalmology 112:256— 71. [aSPD] Jan, L. Y. & Jan, Y. N. (1989) Voltage-sensitive ion channels. Cell 56:1325. [aRSM] (1990) A superfamily of ion channels. Nature 345:672. [aRSM] Jansen, J. J. M., Mulder, W. R., De Caluw, G. L. J., Vlak, J. M. & DeGrip, W. J. (1991) In vitro expression of bovine opsin using recombinant baculovirus: The role of glutamic acid (134) in opsin biosynthesis and glycosylation. Biochimica et Biophysica Acta 1089:68-76. [aPAH] Jin, J., Crouch, R. K., Corson, D. W , Katz, B. M., MacNichol, E. F. & Cornwall, M. C. (1993) Noncovalent occupancy of the retinal-binding References/Controversies in Neuroscience III pocket of opsin diminishes bleaching adaptation of retinal cones. Neuron 11:513-22. [aMDB, RKC] Johnson, R. S., Ohguro, H., Palczewski, K., Hurley, J. B., Walsh, K. A. & Neubert, T. A. (1994) Heterogeneous N-acylation is a tissue- and speciesspecific posttranshitional modification. Journal of Biological Chemistry 269:21067-71. IKS] Jones, G. J., Crouch, R. K., VViggcrt, B., Cornwall, M. C. & Chader C. J. (1989) Rctinoid requirements for recovery of sensitivity after visual pigment bleaching in isolated photoreccptors. Proceedings of the National Academy of Sciences USA 86:9606-10. [RKC] Jordan, S. A., Farrar, C. J., Kenna, P., Humphries, N. M., Shells, D. M., Kumar-Singh, R., Sharp, E. M., Ayuso, C , Benitez, J. & Humphries, P. (L993) Localization of an autosomal dominant retinitis pigmentosa gene to chromosome 7q. Nature Cenctics 4:54—58. [aSPDl Jordan, S. A., Farrar, G. J., Kumar-Singh, R., Kenna, P., Humphries, M. M., Allamand, V., Sharp, E. M. & Humphries, P. (1992) Autosomal dominant retinitis pigmentosa (adRP; RP6): Cosegregation of RP6 and the peripherin-RDS locus in a late-onset family of Irish origin. American Journal of Human Genetics 50:634-39. [aSPD) Kanng, B. K., Kandel, E. R. & Grant, S. G. (1993) Activation ofcAMPresponsive genes by stimuli that produce long-term facilitation in Aplysia sensory neurons. Neuron 10:427-35. [aZX] Kahlert, M. & Hofmann, K.-P. (1991) Reaction rate and collisions! efficiency of the rhodopsin-transducin system in intact retinal rods. Biophysical Journal 59:375-86. [KWK] Kahlert, M., Pepperberg, D. R. & Hofmann, K. P. (1990) Effect of bleached rhodopsin on signal amplification in rod visual receptors. Nature 345:53739. [aMDB] Kajimoto, Y., Shirai, Y., Mukai, H., Kuno, T. & Tanaka, C. (1993) Molecular cloning of two additional members of the neural visinin-like Ca2+binding protein gene family. Journal of Ncurochemistry 61:109196. |KT] Kajiwara, K., Berson, E. L. & Dryja T. P. (1994) Digenic retinitis pigmentosa due to mutations at the unlinked pheripherin/RDS and ROM-1 loci. Science 264:1604-8. [aSPD, AB] Kajiwara, K., Huhn, L. B., Mukai, S., Travis, G. H., Berson, E. L. & Dryja, T. P. (1991) Mutations in the human retinal degeneration slow gene in autosomal dominant retinitis pigmentosa. Nature 354:480-3. [aSPD] Kajiwara, K., Sandberg, M. A., Berson, E. L. & Dryja, T. P. (1993) A null mutation in the human pcriphcrin/RDS gene in a family with autosomal dominant retinitis punctata albescens. Nature Cenetics 3:20812. [aSPD] Kandel. E. R.. Abrams, T , Bcmiur, L., Caruw, T. J., Hawkins, R. D. & Schwartz, J. H. (1983) Classical conditioning and sensitization share aspects of the same molecular cascade in Aplysia. Cold Spring Harbor Symposium on Quantitative Biology 2:821-30. [TWA] Kandel, E. R. & Schwartz, J. H. (1982) Molecular biology of learning: Modulation of transmitter release. Science 218:433—43. [aZX, TWA] Kaplan, J.,'Cerbor, S., Bonneau, D., Rozet, J., Briard, M., Dufler, J., Munnich, A. & Frezal, J. (1991) Probable location of Usher type I gene on chromosome 14q by linkage with D14S13 (MLJ14 probe). Cytogenetics and Cell Genetics 58:1988. [aSPD] Karnik, S. S. & Khorana, H. C. (1990) Assembly of functional rhodopsin requires a disulfide bond between cysteine residues 110 and 187. Journal of Biological Chemistry 265:17520-24. [aSPD] Kauer, J. A., Malenka, R. C. & Nicoll, R. A. (1988) NMDA application potentiates synaptic transmission in the hippocampus. Nature 334:25052. [aZX] Kaupp, U. B. (1991) The cyclic nucleotide-gated channels of vertebrate photoreccptors and olfactory epithelium. Trends in Neuroscience 14:15057. [aRSM, DDO] Kaupp, U. B. & Koch, K.-W. (1992) Role of cCMP and Ca2+ in vertebrate photoreceptor excitation and adaptation. Annual Review of Physiology 54:153-75. [aMDB, aRSM] Kaupp, U. B., Niidoine, T , Tanabe, T., Terada, S., Bonigk, W., Stuhmer, W., Cook, N. J., Kangawa, K., Matsuo, H., Hirose, T , Miyata, T. & Numa, S. (1989) Primary structure and functional expression from complementary DNA of the rod photoreceptor cyclic GMP-gated channel. Nature 342:762-66. [aRSM, LWH, TGW] Kawaniura, S. (1983) Involvement of ATP in activation and inactivation sequence of phosphodicsterase in frog rod outer segments. Biochimica Biophysica Ada 732:276-81. [SK] (1992) Light-sensitivity modulating protein in frog rods.Photochemistry and Photohiology. 56:1173-80. [SK] (1993) Rhodopsin phosphorylation as a mechanism of cyclic GMP phosphodiesterase regulation by S-modulin. Nature 362:85557. (aMDB, aJBH, aRSM, SK, KT, AY] (1994) Photoreceptor light-adaptation mediated by S-modulin, a member of a possible regulatory protein family of protein phosphorylation in signal transduction. Neuroscience Research 20:293-98. [SK, KT] Kawamura, S. & Bownds, M. D. (1981) Light adaptation of the cyclic GMP phosphodiesterase of frog photoreceptor membranes mediated by ATP and calcium ions. Journal of General Physiology 77:571-91. [aMDB] Kawamura, S., Cox, J. A. & Nef, P. (1994) Inhibition of rhodopsin phosphorylation by non-myristoylated recombinated recoverin. Biochemical and Biophysical Research Communications 203:121— 27. [SK] Kawamura, S., Hisatomi, O., Kayada, S., Tokunaga, F. & Kuo, C. H. (1993) Recoverin has S-modulin activity in frog rods. Journal of Biological Chemistry 268:14579-82. [aJBH, KT, AY] Kawamura, S. & Murakami, M. (1991) Calcium-dependent regulation of cyclic GMP phosphodiesterase by a protein from frog retinal rods. Nature 349:420-23. [aMDB, aJBH, JFM, SK, ASP, KT, AY] Kawamura, S., Takamatsu, K. & Kitamura, K. (1992) Purification and characterization of S-modulin, a calcium-dependent regulator on cGMP phosphodiesterase in frog rod photoreceptors. Biochemical and Biophysical Research Communications 186:411-17. [aMDB, SK] Keen, T. J., Inglehearn, C. F., Lester, D. H., Bashir, R., Jay, M., Bird, A. C., Jay, B'. & Bhattacharya, S. S. (1994) Autosomal dominant retinitis pigmentosa: Four new mutations in rhodopsin, one of them in the retinal attachment site. Genomics 11:199-205. [aSPD] Khan, S. M. A., Bolen, W , Hargrave, P. A., Santoro, M. M. & McDowell, J. H. (1991) Differential scanning calorimetry of bovine rhodopsin in rod outer segment disk membranes. European Journal of Biochemisty 200:53-59. [aPAH] Khorana, H. G. (1992) Rhodopsin, photoreceptor of the rod cell: An emerging pattern for structure and function. Journal of Biological Chemistry 267:14. [aMDB, aPAH] Kim, R. Y, Al-Maghtheh, M., Fitzke, F. W., Arden, G. B., Jay, M., Bhattacharya, S. S. & Bird, A. C. (1993) Dominant retinitis pigmentosa associated with two rhodopsin gene mutations. Leu-40-Arg and an insertion disrupting the 5'-splice junction of exon 5. Archives of Ophthalmology 111:1518-24. [aSPD] Kimberling, W. J., Moller, C. G., Davenport, S., Priluck, A., Beighton, P. H., Greenberg, J., Reardon, W., Weston, M. D., Kenyon, J. B., Grunkemeyer, J. A., Dahl, S. P., Overbeck, L. D., Blackwood, D. J., Brower, A. M., Hoover, D. M., Rowland, P. & Smith, R. J. H. (1992) Linkage of Usher syndrome type I gene (USH1B) to the long arm of chromosome 11. Cenomics 14:988-94. [aSPD] Kimberling, W. J., Weston, M. D., Moller, C , Davenport, S. L. H., Shugart, Y. Y, Priluck, I. A., Martini, A., Milani, M. & Smith, R. J. (1990) Localization of Usher syndrome type II to chromosome lq. Genomics 7:245-49. [aSPD] Kjelsberg, M. A., Cotecchia, S., Ostrowski, J., Caron, M. G. & Lelkowitz, R. J. (1992) Constitutive activation of the alpha-lB-adrenergic receptor by all amino acid substitutions at a single site: Evidence for a region which constrains receptor activation. Journal of Biological Chemistry 267:143033. [aPAH] Klann, E., Chen, S.-J. & Sweatt, J. D. (1991) Persistent protein kinase activation in the maintenance phase of long-term potentiation. Journal of Biological Chemistry 266:24253-56. [EDR] (1992) Increased phosphorylation of a 17-kDa protein kinase C substrate in long-term potentiation. Journal of Neurochemistry 58:1576-79. [EDR] (1993) Mechanism of protein kinase C activation during the induction and maintenance of long-term potentiation probe using a selective peptide substrate. Proceedings of the National Academy of Sciences USA 90:833741. [EDR] Klausner, R. D. & Sitia, R. (1990) Protein degradation in the endoplasmic reticulum. Cell 62:611-14. [aSPD] Knowles, J. A., Shugart, Y, Banerjee, P., Gilliam, T. C , Lewis, C. A., Jacobson, S. G. & Ott, J. (1994) Identification of a locus, distinct from RDS-peripherin, for autosomal recessive retinitis pigmeutosa on chromosome 6p. Human Molecular Genetics 3:1401-3. [rSPD] Kobayashi, M., Takamatsu, K., Saitoh, S. & Noguchi, T. (1993) Myristoylation of hippocalcin is linked to its calcium-dependent membrane association properties. Journal of Biological Chemistry 268:18898-904. [KT] Kobayashi, M., Takamatsu, K., Saitoh, S., Miura, M. & Noguchi, T. (1992) Molecular cloning of hippocalcin, a novel calcium-binding protein of the recoverin family exclusively expressed in hippocampus. Biochemical and Biophysical Research Communications 189:511-17. [aJBH, KT] Koch, K.-W. (1991) Purification and identification of photoreeeptor guanylate cyclase. Journal of Biological Chemistry 266:8634-37. [KWK] (1992) Biochemical mechanism of light adaptation in vertebrate photoreceptors. Trends in Biochemical Science 17:307-11. [aMDB] BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 511 Re/erences/Controversies in Neuroscience III (1994) Calcium as modulator of phototransduction in vertebrate photoreceptor cells. Reviews of Physiology, Biochemistry and Pharmacology 125:149-92. [KWK] Koch, K.-W., Eckstein, F. fit Stryer, L. (1990) Stereochemical course of the reaction catalyzed by guanylate cyclase from bovine retinal rod outer segments. Journal of Biological Chemistry 265:9659-63. [KWK] Koch, K.-W. fit Kaupp, U. B. (1985) Cyclic GMP directly regulates a cationic conductance in membranes of bovine rods by a cooperative mechanism. Journal of Biological Chemistry 260:6788-6800. [aBSM, TGW] Koch, K.-W. & Stryer, L. (1988) Highly cooperative feedback control of retinal rod guanylate cyclase by calcium ions. Nature 334:64-66. [arMDB, aJBH, aRSM] Kohn, H. (1980) Light-and GTP-regulated interaction of CTPase and other proteins with bovine photoreceptor membranes. Nature 283:58789. [aPAHl Kohn, H. & McDowell, J. H. (1977) Isoelectric focusing of phosphorylated cattle rhodopsin. Biophysics of Structure and Mechanism 3:199203. [aPAH] Kohnken, R. E., Chafouleas, J. G., Eadie, D. M., Means, A. R. fit McConnell, D. G. (1981) Calmodulin in bovine rod outer segments. Journal of Biological Chemistry 256:12517-22. [aRSM] Kokame, K., Fukada, Y., Yoshizawa, T., Takao, T. & Shimonishi, Y. (1992) Lipid modification at the N terminus of photoreceptor G-protein a-subunit. Nature 359:749-52. [KSJ Konig, B., Arendt, A., McDowell, J. H., Kahlert, M., Hargrave, P. A. & Hofmann, K. P. (1989) Three cytoplasmic loops of rhodopsin interact with transducin. Proceedings of the National Academy of Sciences USA 86:6878-82. [aSPD, aPAH, EAD] Korenbrot, J. I., Brown, D. T. fit Cone, R. A. (1973) Membrane characteristics and osmotic behavior of isolated rod outer segments. Journal of Cell Biology 56:389-98. [aMDB] Koutalos, Y., Nakatani, L. & Yau, K.-W. (1994) Modulation of the cGMP-gated channel of retinal rods by Ca2+. Investigative Ophthalmology and Visual Science [Abstract 1023] 35:1474 . [HRM, rRSM] Koutalos, Y. fit Yau, K.-W. (1993) A rich complexity emerges in phototransduction. Current Opinion in Neurobiology 3:513-19. [aMDB] Kramer, R. H. & Siegelbaum, S. A. (1992) Intracellular Ca2+ regulates the sensitivity of cyclic nucleotide-gated channels in olfactory receptor neurons. Neuron 9:897-906. [aRSM, MSS] Kranich, H., Bartkowski, S., Denton, M. J., Krey, S., Dickinson, P., Duvigneau, C. fit Gal, A. (1993) Autosotnal dominant "sector" retinitis pigmentosa due to a point mutation predicting an Asn-15-Ser substitution of rhodopsin. Human Molecular Genetics 2:813-14. [aSPD] Krebs, E. G. & Beavo, J. A (1979) Phosphorylation-dephosphorylation of enzymes. Annual Review of Biochemistry 48:923-59. [aZX] Kremer, H., Pinckers, A., van den Helm, B., Deutman, A. F., Ropers, H. H. fit Mariman, E. C. M. (1994) Localization of the gene for dominant cystoid macular dystrophy on chromosome 7p. Human Molecular Ccnetics 3:299-302. [aSPD] Krug, M., Lossner, B. & Ott, T. (1984) Anisomycin blocks the late phase of long-term potentiation in the dentate gyrus of freely moving rats. Brain Research Bulletin 13:39-42. [aZX] Krupinski, J., Coussen, F., Bakalyar, H. A., Tang, W. J., Feinstein, P. G., Orth, K., Slaughter, C , Reed, R. R. fit Oilman, A. G. (1989) Adenylyl cyclase amino acid sequence: Possible channel- or transporter-like structure. Science 244:1558-64. [aZX] Krupnick, J. G., Gurevich, V. V., Schepers, T., Hamm, H. E. & Benovic, J. L. (1994) Arrestin-rhodopsin interaction: Multi-site binding delineated by peptide inhibition. Journal of Biological Chemistry 269:322632. [ADAl Kiihlbrandt, W. (1984) Three-dimensional structure of the light-harvesting chlorophyl a/b-protein complex. Nature 307:478-80. [aPAH] (1988) Three-dimensional crystallization of membrane proteins. Quarterly Reviews of Biophysics 21:429-77. [aPAH, RMG] (1992) Two-dimensional crystallization of membrane proteins. Quarterly Reviews of Biophysics 25:1-49. [aPAH, RMG] Kiihlbrandt, W., Wang, D. N. fit Fujiyoshi, Y. (1994) Atomic model of plant light-harvesting complex by electron crystallography. Nature 367:614— 21. [aPAH, RMG] Kiihn, H. (1980) Light- and CTP-regulated interaction of GTPase and other proteins with bovine photoreceptor membranes. Nature 283:58789. [AY] Kumar, V. D. fit Weber, 1. T. (1992) Molecular model of the cyclic GMPbintling domain of the cyclic GMP-gated ion channel. Biochemistry 31:4643-49. [RLB] Kumar-Singh, R., Farrar, G. J., Mansergh, F., Kenna, P., Bhattacharya, S., Gal, S. & Humphries, P. (1993) Exclusion of the involvement of all known retinitis pigmentosa loci in the disease present in a family of Irish 512 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 origin provides evidence for a sixth autosomal dominant locus (RP8). Human Molecular Genetics 2:875-78. [aSPD] Kure, S., Tominago, T., Yoshimoto, T., Tada, K. fit Narisawa, K. (1991) Glutamate triggers internucleosomal DNA cleavage in neuronal cells. Biochemical and Biophysical Research Communications 179:3945. [MT] Kutuzov, M. & Pfister, C. (1994) Activation of the retinal cGMP-specific phosphodiesterase by the GDP-loaded a-subunit of transducin. European Journal of Biochemistry 220:963-71. [rMDB] Lagnado, L. & Baylor, D. (1992) Signal flow in visual transduction. Neuron 8:995-1002. [aMDB, aJBH] (1994) Calcium controls light-triggered formation of catalytically active rhodopsin. Nature 367:273-77. [aMDB, RKC, MPG-K, SK, MSS, KT] Lagnado, L., Cervetto, L. & McNaughton, P. (1992) Calcium homeostasis in the outer segment of retinal rods from the tiger salamander. Journal of Physiology (London) 455:111-42. [aMDB, KWK] Lagnado, L. & McNaughton, P. A. (1990) Electrogenic properties of the Na:Ca exchange. Journal of Membrane Biology 113:177-91. [PPMS] Lakshmi, K. V., Auger, M., Raap, J., Lugtenburg, J., Griffin, R. C., fit Herzfeld, J. (1993) Solid-state 13C NMR rotational resonance distance measurement in the M photointermediate of bacteriorhodopsin: Determination of the configuration about the Schiffbase. Biophysics Journal 64:A212. [SOS] Lamb, T. D. (1984) Effects of temperature changes on toad rod photocurrents. Journal of Physiology (London) 346:557-78. [aM DB] (1990) Dark adaptation: A re-examination. In: Night vision: Basic, clinical and applied aspects, ed. R. F. Hess, L. T. Sharp fie K. Nordby. Cambridge University Press. [RKC] Lamb, T. D., Matthews, H. R. fit Torre, V. (1986) Incorporation of calcium buffers into salamander retinal rods: A rejection of the calcium hypothesis of phototransduction. Journal of Physiology (London) 372:31549. [aMDB] Lamb, T. D., McNaughton, P. A. & Yau, K.-W. (1981) Spatial spread of activation and background desensitization in toad rod outer segments. Journal of Physiology (London) 319:463-96. [aMDB] Lamb, T. D. & Pugh, E. N., Jr. (1992) A quantitative account of the activation steps involved in phototransduction in amphibian photoreceptors. Journal of Physiology (London) 449:719-58. [aMDB] Lamba, O. P., Borchman, D. fit O'Brien, P. J. (1994) Fourier transform infrared study of the rod outer segment disk and plasma membranes of vertebrate retina. Biochemistry 33:1704-12. [ADA] Lambrecht, H. G. fie Koch, K.-W. (1991) A 26 kd calcium binding protein from bovine rod outer segments as modulator of photoreceptor guanylate cyclase. EMBO Journal 10:793-98. [aMDB, aJBH, ASP] (1992) Recoverin, a novel calcium-binding protein from vertebrate photoreceptors. Biochimica et Biophysica Acta 1160:63-66. [aJBH] Lambright, D. G., Noel, J. P., Hamm, H. E. fit Sigler, P. B. (1994) Structural determinants for activation of the alpha subunit of a heterotrimeric G protein. Nature 369:621-28. [aMDB, EAD] Lee, R. H., Brown, B. M. & Lolley, R. N. (1984) Light-induced dephosphorylation of a 33K protein in rod outer segments of rat retina. Biochemistry 23:1972-77. [BMW] (1990) Protein kinase A phosphorylates retinal phosducin on serine 73 in situ. Journal of Biological Chemistry 265:15860-66. [BMW] Lee, R. H., Fowler, A., McCinnis, J. F., Lolley, R. N. fir Craft, C. M. (1990a) Amino acid and cDNA sequence of bovine phosducin, a soluble phosphoprotein from photoreceptor cells. Journal of Biological Chemistry 265:15867-73. [aMDB] Lee, R. H., Lieberman, B. S. & Lolley, R. N. (1987) A novel complex from bovine visual cells of a 33,000-dalton phosphoprotein with beta and gamma transducin: Purification and subunit structure. Biochemistry 28:3983-90. [aMDB] (1990b) Retinal accumulation of the phosducin/Tgamma and transducin complexes in developing normal mice and in mice and dogs with inherited retinal degeneration. Experimental Eye Research 51:32533. [aMDB] Lee, R. H., Ting, T. D., Lieberman, B. S., Tobias, D. E., Lolley, R. N. fit Ho, Y.-K. (1992) Regulation of retinal cGMP cascade by phosducin in bovine rod photoreceptor cells. Journal of Biological Chemistry 267:25104-12. [BMW] Lee, R. H., Whelan, J. P., Lolley, R. N. & McCinnis, J. F. (1988) The photoreceptor-specific 33 kDa phosphoprotein of mammalian retina: Generation of monospecific antibodies and localization by immunocytochemistry. Experimental Ei/c Research 46:829-40. ]aMDB] Lee, S. M. fit Bressler, R. (1981) Prevention of diabetic nephropathy by diet control in the dbldb mouse. Diabetes 30:106-11. [DW] Leibrock, C. S., Reuter, T. & Lamb, T. D. (1994) Dark adaptation of toad rod References /Controversies in Neuroscience III photorcceptors following small bleaches. Vision Research 34:27872800. [rMDB] Lenz, S. E., Henschel. Y., Zopf, D., Voss, B. & Gundelfinger, E. D. (1992) VILIP, a cognate protein of the retinal calcium binding proteins visinin and rccoverin, is expressed in the developing chicken brain. Brain Research 15:133-40. [aJBH] Levin, L. R., Han, P. L., Hwang, P. M., Feinstein, P. C , Davis, R. L. & Reed, R. R. (1992) The Drosophila learning and memory gene rtttabaga encodes a Ca2+/calmodulin-responsive adenylyl cyclase. Cell 68:47989. [aZX, TWA] Levine, M. A., Modi, W. S. & O'Brien, S. J. (1990) Chromosomal localization of the genes encoding two forms of the G protein B polypeptide, Bl and B3, in man. Gcnomics 8:380-86. [aSPD] Lewis, R. A., Otterud, B., Stauffer, D., Lalouel, J.-M. & Leppert, M. (1990) Mapping recessive ophthalmic diseases: Linkage of the locus for Usher syndrome type III to a DNA marker on chromosome lq. Gcnomics 7:25056. [aSPD] Li, H., Volpp, K. & Applebury, M. L. (1990) Bovine cone photoreceptor cGMP phosphodiesterase structure deduced from a cDNA clone. Proceedings of the National Academy of Sciences USA 87:293— 97. [aMDB] Liang, C.-J., Yamashita, K., Muellenberg, C. G., Shichi, H. & Kobata, A. (1979) Structure of the carbohydrate moieties of bovine rhodopsin. Journal of Biological Chemistry 254:6414-18. [aPAH] Liebman, P. A. & Entine, G. (1968) Visual pigments of frog and tadpole. Vision Research 8:761-75. [aMDB] Liebman, P. A., Parker, K. R. & Dratz, E. A. (1987) The molecular mechanism of visual excitation and its relation to the structure and composition of the rod outer segment. Annual Review of Physiology 49:765-91. [aMDB, aPAH] Liebman, P. A. & Pugh, E. N. (1979) The control of phosphodiesterase in rod disk membranes: Kinetics, possible mechanisms and significance for vision. Vision Research 19:375-80. [aMDB] (1980) ATP mediates rapid reversal of cyclic CMP phosphodiesterase activation in visual receptor membranes. Nature 287:734—37. [aMDB] Light, D. B., Corbin, J. D. & Stanton, B. A. (1990) Dual ion-channel regulation by cyclic GMP and cyclic GMP-dependent protein kinase. Nature 344:336-39. [aRSM] Liman, E. R. & Buck, L. B. (1994) A second subunit of the olfactory cyclic nucleotide-gated channel confers high sensitivity to cAMP. Neuron 13:611-21. (rRSM] Linden, D. J. & Rcjuttcnbcrg, A. (1989) The role of protein kinase C in longterm potentiation: A testable model. Brain Research Reviews 14:27996. [aZX] Litman, B. J. (1982) Purification of rhodopsin by concanavalin A affinity chromatography. Methods in Enzymology 81:150-53. [aPAH] Litman, B. J., Aton, B. & Hartley, J. B. (1982) Functional domains of rhodopsin. Vision Research 22:1439-42. [ADA] Liu, M., Chen, T.-Y., Ahamed, B., Li, J. & Yau, K.-W. (1994) Calciumcalmodulin modulation of the olfactory cyclic nucleotide-gated cation channel. Science 266:1348-54. [rRSM] Liu, Y. L. & Storm, D. R. (1989) Dephosphorylation of neiiromodulin by calcineurin. Journal of Biological Chemistry 264:12800-4. [aZX] (1990) Regulation of free calmodulin levels in neurons by neuromodulin: Relationship to neuron growth and regeneration. Trends in Pharmacological Sciences 11:107-11. [aZX] Livingston, M. S. (1985) Genetic dissection of Drosophila adenylyl cyclase. Proceedings of the National Academy of Sciences USA 82:5992— 96. [aZX] Livingston, M. S., Sziber, P. P. & Quinn, W. G. (1984). Loss of catcium/calmodulin responsiveness in adenylyl cyclase of rutabaga, a Drosophila learning mutant. Cell 37:205-15. [aZX, TWA] Lolley, R. N., Farber, D. B., Rayborn, M. E. & Hollyfield, J. G. (1977) Cyclic GMP accumulation causes degeneration of photoreceptor cells: Simulation of an inherited disease. Science 196:664-66. [aSPD] Lolley, R. N., Rong, H. & Craft, C. M. (1994) Linkage of photoreceptor degeneration by apoptosis with inherited defect in phototransduction. Investigative Ophthalmology ami Visual Science 35:358-62. [aSPD] Ludwig, J., Marglit, T , Eismann, E., Lancet, D. & Kaupp, U. B. (1990) Primary structure of a cAMP-gated channel from bovine olfactory epithelium. FEBS Letters 270:24-29. [aRSM] Lyness, A. L., Ernst, W., Quintan, M. P., Clover, G. M., Arden, G. B., Carter, R. M., Bird, A. C..& Parker, J. A. (1985) A clinical, psychophysic.il and electroretinographic survey of patients with autosomal dominant retinitis pigmentosa. British Journal of Ophthalmology 69:32639. [aSPD] Macke, J. P., Davenport, C. M., Jacobson, S. G., Hennessey, J. C , Conzales- Feraandez, F., Conway, B. P., Heckenlively, J., Palmer, R., Maumenee, I. H., Sieving, P., Gouras, P., Good, W. & Nathans, J. (1993) Identification of novel rhodopsin mutations responsible for retinitis pigmentosa: Implications for the structure and function of rhodopsin. American Journal of Human Genetics 53:80-89. [aSPD] Mackler, S. A., Brooks, B. P. & Eberwine, J. H. (1992) Stimulus-induced coordinate changes in mRNA abundance in single postsynaptic hippocampal CA1 neurons. Neuron 9:539-48. [EDR] MacNicol, M. & Schulman, H. (1992) Cross-talk between protein kinase C and multifunctional Ca2+/calmodulin-dependent protein kinase. Journal of Biological Chemistry 267:12197-201. [aZX] Madison, D. V, Malenka, R. C. & Nicoll, R. A. (1991) Mechanisms underlying long-term potentiation of synaptic transmission. Annual Review of Neuroscience 14:379-97. [TWA] Malenka, R. C , Kauer, J. A., Perkel, D. J., Mauk, M. D., Kelly, P. T , Nicoll, R. A. & Waxham, M. N. (1989) An essential role for postsynaptic calmodulin and protein kinase activity in long-term potentiation. Nature 340:554-57. [TWA] Malenka, R. C , Kauer, J. A., Zucker, R. S. & Nicoll, R. A. (1988) Postsynaptic calcium is sufficient for potentiation of hippocampal synaptic transmission. Science 242:81-84. [aZX] Malinow, R., Schulman, H. & Tsien, R. W. (1989) Inhibition of postsynaptic PKC or CaMKII blocks induction but not expression of LTP. Science 245:862-66. [TWA] Mangels, L. A. & Gnegy, M. E. (1990) Muscarinic receptor mediated transolocation of calmodulin in SK-N-SH human neuroblastoma cells. Molecular Pharmacology 37:820-26. [aZX] Mangini, N. J., Gamer, G. L., Okajima, T.-L. L., Donoso, L. A. & Pepperberg, D. R. (1994) Effect of hydroxylamine on the subcellular distribution of arrestin (S-antigen) in rod photoreceptors. Visual Neuroscience 11:561-68. [rMDB] Mangini, N. J. & Pepperberg, D. R. (1988) Immunolocalization of 48K in rod photoreceptors: Light and ATP increase OS labeling. Investigative Ophthamology and Visual Science 29:1221-34. |JFM] Mariuzza, R. A., Phillips, S. E. V. & Poljak, R. J. (1987) The structural basis of antigen-antibody recognition. Annual Review of Biophysics and Biophysical Chemistry 16:139-59. [aPAH] Massof, R. W. & Finkelstein, D. (1981) Two forms of autosomal dominant primary retinitis pigmentosa. Documenta Ophthalmologica 51:289— 346. [aSPD] Matesic, D. & Liebman, P. A. (1987) cGMP-dependent cation channel of retinal rod outer segments. Nature 326:600-3. [uRSMI Matsuoka, S., Nicoll, D. A., Reilly, R. F., Hilgemann, D. W. & Philipson, K. D. (1993) Initial localization of regulatory regions of the cardiac sarcolemmal Na+-Ca2+ exchanger. Proceedings of the National Academy of Sciences USA 90:3870-74. [PPMS] Matthews, G. (1986) Spread of the light response along the rod outer segment: An estimate from patch-clamp recordings. Vision Research 26:535-41. [aMDB] Matthews, H. R. (1992a) Effects of lowered cytoplasmic calcium concentration and light on the responses of rod photoreceptors isolated from the retina of the tiger salamander. Journal of Physiology 446:109P. [HHM] (1992b) Inability of light to accelerate the dim flash response without lowered calcium concentration in rod photoreceptors isolated from the retina of the tiger salamander. Journal of Physiology 44(>:559P. [HRM] (in press) Effects of lowered cytoplasmic calcium concentration and light on the responses of salamander rod photoreceptors. Journal of Physiology. [HRM] Matthews, H. R., Murphy, R. L. W , Fain, G. L. & Lamb, T. D. (1988) Photoreceptor light adaptation is mediated by cytoplasmic calcium concentration. Nature 334:67-69. [aMDB, aRSM, KWK, HRM] Matthies, H., Frey, U., Reymann, K., Krug, M., Jork, R. or Schroeder, H. (1990) Different mechanisms and multiple stages of LTP. Advances in Experimental Medical Biology 268:359-68. [aZX] May, D., Ross, E., Gilman, A. & Smigel, M. (1985). Reconstitution of catecholamine-stimulatcd adenylate cyclase activity using three purified proteins. Journal of Biological Chemistry 260:15829-33. [aZX] McCarthy, S. T , Younger, J. P. & Owen, W. G. (1993) Disc structure, buffering and kinetics of cytosolic free calcium in bullfrog rod outer segments. Investigative Ophthalmology and Visual Science 34:1327. [aMDB] (1994) Free calcium concentrations in bulfrog rods determined in the presence of multiple forms of Fura-2. Biophysics Journal 67:207689. [rMDB] McDowell, J. H. (1993) Preparing rod outer segment membranes, regenerating, and determining rhodopsin concentration. In: Photoreceptor cells, ed. P. A. Hargrave. Academic Press. [aPAH] BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 513 References/Controversies in Neuroscience III McDowell, J. H., Khan, S. M. A., Bolen, D. W., Santoro, M. M. & Hargrave, P. A. (1992) Structural stability of rhodopsin and opsin studied by differential scanning calorimetry. In: Signal transduction in photorecejttor cells, cd. P. A. Hargrave, K. P. Hofmann & U. B. Kaupp. Springer-Verlag. [aPAH] McDowell, J. H., Nawrocki, J. P. & Hargrave, P. A. (1993) Phosphorylation sites in bovine rhodopsin. Biochemistry 32:4968-74. [aMDB] McFall-Ngai, M. J. & Horwitz, J. (1990) A comparative study of the thermal stability of the vertebrate eye lens: Antarctic ice fish to the desert iguana. Experimental Eye Research 50:703-9. [aPAH] McGee, T. L., Berson, E. L. & Dryja, T. P. (1992) Search for point mutations in the interstitial retinoid binding protein gene in patients with hereditary retinal degenerations. Investigative Ophthalmology and Visual Science 33:1396. [aSPD] McCee, T. L., Lin, D., Berson, E. L. & Dryja, T. P. (1994) Defects in the rod cGMP-gated gene in patients with retinits pigmentosa. Investigative Ophthalmology and Visual Science 35:1716. [aSPD] McCinnis, J. F., Austin, B. J., Klisak, I., Heinzmann, C , Kojis, J., Sparkes, R. S., Bateman, B. J. & Lerious, V. (1995) Chromosomal assignment of the human gene for the cancer associated retinopathy protein (recoverin). Journal of Neuroscience Research 40:165-168. [JFM] McCinnis, J. F., Lerious, V., Pazik, J. & Elliott, R. W. (1993) Chromosomal assignment of the recoverin gene and cancer-associated retinopathy. Mammalian. Genome 4:43-45. [JFM] McCinnis, J. F. & Leveille, P. J. (1985) Soluble proteins associated with photoreceptor cell death in the rd mouse. Current Ei/c Research 4:112735. [JFM] McCinnis, J. F., Stepanik, P. L., Baehr, W., Subbaraya, I. & Lerious, V. (1992a) Cloning and sequencing of the 23 kDa mouse photoreceptor cellspecific protein. FEBS Letters 302:172-76. [JFM] McCinnis, J. F., Whelan, J. P. & Donoso, L. A. (1992b) Transient, cyclic changes in mouse visual cell gene products during the light-dark cycle. Journal of Neuroscience Research 31:584-90. [JFM] McGuire, R. E., Sullivan, L. S., Blanton, S. H., Church, M., Heekenlively, J. R. & Daiger, S. P. (in press) X-linked dominant cone-rod dystrophy: Linkage mapping of a new locus (CORD3) to Xp22.13—p22.11. American Journal of Hitman Genetics. [rSPD] McKusick, V. A. (1994) Mendclian inheritance in man, 11th ed. The Johns Hopkins University Press. [arSPD] McLaughlin, M. E., Sandberg, M. A., Berson, E. L. & Dryja, T. P. (1993) Recessive mutations in the gene encoding the (3-subunit of rod phosphodiesterase in patients with retinitis pigmentosa. Nature Genetics 4:130-34. [aSPD] McNaughton, P. A. (1990) Light response of vertebrate photoreceptors. Physiological Revietos 70:847-84. [aMDB] McVVilliams, P., Farrar, G. J., Kenna, P., Bradley, D. G., Humphries, M. M., Sharp, E. M., McConnell, D. J., Lawler, M., Shiels, D., Ryan, C., Stephens, K., Daiger, S. P. & Humphries, P. (1989) Autosomal dominant retinitis pigmentosa (ADRP): Localization of an ADRP gene to the long arm of chromosome 3. Genomics 5:619-22. [aSPD] Michel, H. (1982) Characterization and crystal packing of three-dimension bacteriorhodopsin crystals. EMBO Journal 1:1267-71. [RMG] (1991) General and practical aspects of membrane protein crystallization. In: Crystallization of membrane proteins, ed. H. Michel. CRC Press. [aPAH] Michel, H. & Oesterhelt, D. (1980) Three-dimensional srystals of membrane proteins: Bacteriorhodopsin. Proceedings of the National Academy of Sciences USA 77:1283-85. [RMG] Milam, A. H., Daeey, D. M. & Dizhoor, A. M. (1992) Recoverin immunoreactivity in mammalian cone bipolar cells. Visual Neuroscience 10:1-12. [aJBH] Miljanich, G. P., Brown, M. F., Mabry-Gaud, S., Dratz, E. A. & Sturtevant, J. M. (1985) Thermotropic behavior of retinal rod membranes and dispersions of extracted phospholipids. Journal of Membrane Biology 85:79-86. [aPAH] Miller, C. (1989) Genetic manipulation of ion channels: A new approach to structure and mechanism. Neuron 2:1195-1205. [aRSM] (1991) Annus mirabilis of potassium channels. Science 252:109296. [aRSM] Miller, J. L., Fox, D. A. & Litman, B. J. (1986) Amplification of phosphodiesterase activation is greatly reduced by rhodopsin phosphorylation. Journal of Biological Chemistry 25:4983-88. [aMDB] Miller, J. L. & Korenbrot, J. I. (1993) In retinal cones, membrane depolarization in darkness activates the cGMP-dependent conductance. Journal of General Physiology 101:933-61. [KWK] Miller, K. R. & Jacobs, J. S. (1983) Two-dimensional crystals formed from photosynthetic reaction centers. Journal of Cell Biology 97:126670. [aPAH] 514 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 Min, K. C , Zvyaga, T. A., Cypess, A. M. & Sakmar, T. P. (1993) Characterization of mutant rhodopsins responsible for autosomal dominant retinitis pigmentosa: Mutations on the cytoplasmic surface affect transducin activation. Journal of Biological Chemistry 268:9400-4. [aSPD] Minocherhomjee, A. M., Selfe, S., Flowers, N. W. & Storm, D. R. (1987) Interaction between the catalytic subunit of the calmodulin sensitive adenylate cyclase with [125I] and [125I] calcodulin. Biochemistry 26:444448. [aZX] Mitchell, D. C , Straume, M. & Litman, B. J. (1992) Role of sn-1saturated,sn-2-polyunsaaturated phospholipids in control of membrane receptor conformational equilibrium: Effects of cholesterol and acyl chain unsaturation on the metarhodopsin I—Metarhodopsin II equilibrium. Biochemistry 31:662-70. [aPAH] Mitchell, D. C , Straume, M., Miller, J. L. & Litman, B. J. (1990) Modulation of metarhodopsin formation by cholesterol-induced ordering of bilayer lipids. Biochemistry 29:9143. [ADA] Mitchell, P. J. & Tjian, R. (1989) Transcriptional regulation in mammalian cells by sequence specific DNA binding proteins. Science 245:37178. [aZX] Moench, S. J., Terry, C. E. & Dewey, T. G. (1994) Fluorescence labeling of the palmitoylation sites of rhodopsin. Biochemistry 33:5783-89. [ADA] Molday, L. L., Cook, N. J., Kaupp, U. B. & Molday, R. S. (1990) The cGMPgated cation channel of bovine rod photoreceptor cells is associated with a 240 kDa protein exhibiting immunochemical cross-reactivity with spectrin. Journal of Biological Chemistry 265:18690-95. [aRSM, TGW] Molday, R. S. (1989) Monoclonal antibodies to rhodopsin and other proteins of rod outer segments. Progress in Retinal Research 8:173-209. [aPAH] Molday, R. S. & Molday, L. L. (1987) Differences in the protein composition of bovine retinal rod outer segment disk and plasma membranes isolated by a ricin-gold-dextran density perturbation method. Journal of Cell Biology 105:2589-2601. [TGW] Molday, R. S., Molday, L. L., Dose, A., Clark-Lewis, I., Illing, M., Cook, N. J., Eismann, E. & Kaupp, U. B. (1991) The cGMP-gated channel of the rod photoreceptor cell: Characterization and orientation of the amino terminus. Journal of Biological Chemistry 266:21917-22. [aRSM, RLH] Molday, R. S., Reid, D. M., Connell, G. & Molday, L. L. (1992) Molecular properties of the cGMP-gated channel of rod photoreceptor cells as probed with monoclonal antibodies. In: Signal transduction in photoreceptor cells, ed. P. A. Hargrave, K. P. Hofmann & U. B. Kaupp. Springer-Verlag. [aRSM] Mollner, S. & Pfeuffer, T. (1988) Two different adenylyl cyclases in brain distinguished by monoclonal antibodies. European Journal of Biochemistry 171:265-71. [aZX] Mollner, S., Simmoteit, R., Palm, D. & Pfeuffer, T. (1991) Monoclonal antibodies against various forms of the adenylyl cyclase catalytic subunit and associated proteins. European Journal of Biochemistry 195:28186. [aZX] Montarolo, P. G., Coelet, P., Castellucci, V. F., Morgan, J., Kandel, E. R. & Schacher, S. (1986) A critical period for macromolecular synthesis in long-term heterosynaptic facilitation in Aplysia. Science 234:124954. [aZX] Moore, A. T., Fitzke, F. W., Kemp, C. M., Arden, C. B., Keen, T. J., Ingleheam, C. F., Bhattacharya, S. S. & Bird, A. C. (1992) Abnormal dark adaptation kinetics in autosomal dominant sector retinitis pigmentosa due to rhodopsin mutation. British Journal of Ophthalmology 76:465— 69. [aSPD] Morris, R. G. (1990) Toward a representational hypothesis of the role of hippocampal synaptic plasticity in spatial and other forms of learning. Cold Spring Harbor Symposium on Quantative Biology 50:16173. [aZX] Morris, R. G., Anderson, E., Lynch, C. S. & Baudry, M. (1986) Selective impairment of learning and blockade of long-term potentiation by an N-inethyl-D-aspartate receptor antagonist, AP5. Nature 319:77476. [aZX] Morris, R. C , Garrud, P., Rawlins, J. N. & O'Keefe, J. (1982) Place navigation impaired in rats with hippocampal lesions. Nature 297:68183. |aZX] Mullen, R. J., Eicher, E. M. & Sidman, R. L. (1976) Purkinje cell degeneration, a new neurological mutation in the mouse. Proceedings of the National Academy of Sciences USA 73:208-12. [MWK] Musarella, M. A. (1992) Gene mapping of ocular diseases. Survey of Ophthalmology 36:285-312. [aSPD] Musarella, M. A., Anson-Cartwright, L., Leal, S. M., Gilbert, L. D., Worton, R. G., Fishman, G. A. & Ott, J. (1990) Multipoint linkage analysis and heterogeneity testing in 20 X-linked retinitis pigmentosa families. Genomics 8:286-96. [aSPD] Naash, M. I., Hollyfield, J. G., Al-Ubaidi, M. R. & Baehr, W. (1993) Simulation of human autosomal dominant retinitis pigmentosa in Re/erences/Controversies in Neuroscience HI transgcnic mice expressing a mutated murine opsin gene. Proceedings of tint national Academy of Sciences USA 90:5499-5503. [MWK] Nairn, A. C , Hcmmings, H. C. & Greengard, P. (1985) Protein kinases in the brain. Animal Review of Biochemistry 54:931-76. [aZX] Nakamura, T. & Gold, G. H. (1987) A cyclic nucleotide-gated conductance in olfactory receptor cilia. Nature 325:342-44. [aRSMj Nukuno, A., Terasawa, M., Watanabe, M., Usuda, N., Morita, T. & Hidaka, H. (1992) Ncurocalcin, a novel calcium binding protein with three EFhand domains, expressed in retinal amacrine cells and ganglion cells. Biochcm.Biophys.Res.Commun. 186:1207-11. [aJBH] Nakatani, K., Koutalos, Y. & Yau, K.-W. (1995) Ca+2 modultion of the cGMPdated channel of bullfrog retinal rod photoreceptors. Journal of Physiology 484:1:69-76. [MPC-K] Nakatani, K., Tanuira, T. & Yau, K.-W. (1991) Light adaptation in retinal rods of the rabbit and two other nonprimate mammals. Journal of General Physiology 97:413-35. (aMDB, KWK] Nakatani, K. & Yau, K.-W. (1988a) Calcium and light adaptation in retinal rods and cones. Nature 334:69-71. [aMDB, KWK, HRM] (1988b) Cuanosine 3'-5'-cyclic monophosphate-activatcd conductance studied in a truncated rod outer segment of the toad. Journal of Physiology (London) 395:731-53. [aMDB] Nakazawa, M., Kikawn-Araki, E., Shiono, T. & Tamai, M. (1991) Analysis of rhodopsin gene in patients with retinitis pigmentosa using allcle-specific polymerase chain reaction. Japanese Journal'of Ophthalmology 35:38693. [aSPD] Nfirfstrom, K. (1983) Hereditary progressive retinal atrophy in the Abyssinian cat. The Journal of Heredity 74:273-76. [MWK] Nathans, J. (1990) Determinants of visual pigment absorbance: Role of charged amino acids in the putative transmembrane segment. Biochemistry 29:937-42. [aPAHJ (1992) Rhodopsin: Structure, function, and genetics. Biochemistry 31:492331. [aMDB, aSPD, aPAH] Nathans, J., Maumcncc, I. H., Zrenner, E., Sadowski, B., Sharpe, L. T., Lewis, R. A., Hanson, E., Rosenberg, T , et al. (1993) Genetic heterogeneity among blue-cone monochromats. American Journal of Human Cenetics 53:987-1000. [aSPD] Nathans, J., Mcrbs, S. L., Sung, C.-H., Weitz, C. J. & Wang, Y. (1992a) Molecular genetics of human visual pigments. Annual Review of Cenetics 26:403-24. [aSPD] Nathans, J., Weitz, C. J., Agarwal, N., Nir, I. & Papermaster, D. S. (1989) Production of bovine rhodopsiu by mammalian cell lines expressing cluned cDNA: Spectrophotometry and siibccllular localization. Vision Research 29:907-14. [aPAH] Natsukari, N., Hanai, H., Matsunaga, T. & Fujita, M. (1990) Synergistic activation of brain adenylate cyclase by calmodulin, and either GTP or catccholnmines including dopamine. Brain Research 534:170-76. [aZX] Nawy, S. & Jahr, C. E. (1990) Suppression by glutamate of cGMP-activated conductance in retinal bipolar cells. Nature 346:269-71. [aRSM] Nef, S., Do Castro, E., Martone, M. E., Edelman, V. M., Ellisman, M. H., Fiiimclli, H., Comte, M., Cox, J. A., Lenz, S. E., Kawamura, S. & Nef, P. (in press) Relgulation of receptor phosphorylation by a novel family of neuronal calcium sensors (NCS) and distribution of NCS-1 in the avian and rodent nervous systen. Neuron. [SK] Nef, S., Fiumclli, H., De Castro, E., Raes, M. B. & Nef, P. (in press) Identification of a neuronal calcium sensor (NCS-1) possibly involved in the regulation of receptor phosphorylation. Journal of Receptor Research IKT] Nei, M. (1987) Molecular evolutionary genetics. Columbia University Press. [aSPD] Nelson, R. B. & Routtenberg, A. (1985) Characterization of protein Fl: A kinase C substrate directly related to neural plasticity. Experimental Neurology 89:213-24. [aZX] Nestler, E. J. & Greengard, P. (1983) Protein phosphorylation in the brain. Nature 305:583-88. [aZX] Neubert, T. A., Johnson, R. S., Hurley, J. B. & Walsh, K. A. (1992) The rod transducin a subunit amino terminus is hcterogeneously fatty acylated. Journal of Biological Chemistry 267:18274-77. [aMDB, aJBH, KS] Newton, A. C. & Williams, D. S. (1991) Involvement of protein kinase C in the phosphorylation of rhodopsin. Journal of Biological Chemistry 266:17725-28. [aMDB] (1993) Rhodopsin is the major in situ substrate of protein kinase C in rod outer segments of photoreceptors. Journal of Biological Chemistry 268:18181-86. [aMDB] Nicol, G. D. & Bownds, M. D. (1989) Calcium regulates some, but not all, aspects of light adaptation in rod photoreceptors. Journal ofCeneral Physiology 9-1:233-59. [aMDB] Nicol, G. D., Kaupp, U. B. & Bownds, M. D. (1987) Transduction persists in rod photoreceptors after depletion of intracellular calcium. Journal of General Physiology 89:297-319. [aMDB] Nicoll, R. A., Kauer, J. A. & Malenka, R. C. (1988) The current excitement in long-term potentiation. Neuron 1:97—103. [aZX] Nichols, B. E., Sheffield, V. C , Vandenburgh, K., Drack, A. V., Kimura, A. E. & Stone, E. M. (1993) Butterfly-shaped pigment dystrophy of the fovea caused by a point mutation in codon 167 of the RDS gene. Nature Cenetics 3:202-6. [aSPD] Niemeyer, G., Trub, P., Schinzel, A. & Gal, A. (1992) Clinical and ERG data in a family with autosomal dominant RP and Pro-347-Arg mutation in the rhodopsin gene. Documenta Ophthalmologica 79:303-11. [aSPD] Nigg, E. A., Hilz, H., Eppenberger, H. M. & Dutly, F. (1985) Rapid and reversible translocation of the catalytic subunit of cAMP-dependent protein kinase type II from the Golgi complex to the nucleus. EMBO Journal 4:2801-6. [aZX] Noel, J. P., Hamm, H. E. & Sigler, P. B. (1993) The 2.2 A crystal structure of transducin- complexcd with GTPgaminaS. Nature 366:654-63. [aMDB, RKC, EAD] Noro, Y., lshiguro, S.-I. & Tamal, M. (1994) Accumulation of glutamate-like reactivity in the rds/rds mouse retina. Investigative Ophthaltnology and Visual Science 35:1362. [MT] Nussberger, S., Dorr, K. Wang, D. N. & Kiihlbrandt, W. (1993) Lipid-protein interactions in crystals of plant light-harvesting complex. Journal of Molecular Biology 234:347-56. [RMG] O'Brien, P. J., St. Jules, R. S., Reddy, T. S., Bazan, N. G. & Zatz, M. (1987) Acylation of disc membrane rhodopsin may be nonenzymatic. Journal of Biological Chemistry 262:5210-15. [aPAH] O'Brien, P. J. & Zatz, M. (1984) Acylation of bovine rhodopsin by [3H]palmitic acid. Journal of Biological Chemistry 259:5054-57. [aPAH] Ocorr, K. A., Walters, E. T. & Byrne, J. H. (1985) Associative conditioning analog selectively increases cAMP levels of tail sensory neurons in Aplysia. Proceedings of the National Academy of Sciences USA 82:254852. [TWA] Ohguro, H., Fukada, Y., Takao, T., Shimonishi, Y., Yoshizawa, T. & Akino, T. (1991) Carboxyl methylation and famesylation of transducin gammasubunit synergistically enhance its coupling with metarhodopsin II. EMBO Journal 10:3669-74. [aMDB] Ohguro, H., Palczewski, K., Ericsson, L. H., Walsh, K. A. & Johnson, R. S. (1993) Sequential phosphorylation of rhodopsin at multiple sites. Biochemistry 32:5718-24. [aMDB] Okazaki, K., Watanabe, M., Ando, Y., Hagiwara, M., Terasawa, M. & Hidaka, H. (1992) Full sequence of neurocalcin, a novel calcium-binding protein abundant in central nervous system. Biochemical and Biophysical Research Communications 185:147-53. [aJBH] Oprian, D. D. (1992) The ligand-binding domain of rhodopsin and other G protein-linked receptors. Journal of Bioenergetics ir Biomembranes 24:211-17. [aPAH] Oprian, D. D., Molday, R. S., Kaufman, R. J. & Khorana, H. G. (1987) Expression of a synthetic bovine rhodopsin gene in monkey kidney cells. Proceedings of the National Academy of Sciences USA 84:887478. [aPAH] Orth, U., Samanns, C , Gusseck, H., Niemeyer, G., Ludwig, M., Meitinger, T , Schinzel, A., Schwinger, E. & Gal, A. (1991) Autosomal dominant hereditary retinopathia pigmentosa with genetic heterogeneity. Fortschritte der Opththalmologie. 88:455-59. [aSPD] Ott, J., Bhattacharya, S. S., Chen, J. D., Denton, M. J., Donald, J., DuBay, C , Farrar, G. J., Fishman, G. A. et al. (1990) Localizing multiple X chromosome-linked retinitis pigmentosa loci using multilocus homogeneity tests. Proceedings of the National Academy of Sciences USA 87:701-4. [aSPD] Ovchinnikov, Y. A., Abdulaev, N. G. & Bogachuk, A. S. (1988) Two adjacent cysteine residues in the C-terminal cytoplasmic fragment of bovine rhodopsin are palmitylated. FEBS Letters 230:1-5. [aPAH] Ovchinnikov, Y. A., Abdulaev, N. G., Zolotarev, A. S., Artamonov, I. D., Bespalov, I. A., Dergachev, A. E. & Tsuda, M. (1988) Octopus rhodopsin: Amino acid sequence deduced from cDNA. FEBS Letters 232:6972. [aPAH] Pages, F., Deterre, P. & Pfister, C. (1992) Enhanced GTPase activity of transducin when bound to cGMP phosphodiesterase in bovine retinal rods. Journal of Biological Chemistry 267:22018-21. [aMDB] (1993) Enhancement by phosphodiesterase subunits of the rate of GTP hydrolysis by transducin in bovine retinal rods. Essential role of the phosphodiesterase catalytic core. Journal of Biological Chetnistry 268:26358-64. [arMDB, BMW] Palczewski, K., Buczylko, J., Kaplan, M. W , Polans, A. S. & Crabb, J. W. (1991) Mechanism of rhodopsin kinase activation. Journal of Biological Chemistry 266:12949-55. [aMDB] BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 515 References!Controversies in Neuroscience III Palczewski, K., Jager, S., Buczylko, J., Crouch, R. K., Bredberg, D. L., Hofmann, K. P., Asson-Batres, M. A. & Saari, J. C. (1994) Rod outer segment retinol dehydrogenase: Substrate specificity and role in phototransduction. Biochemistry 38:13741-50. [RKC] Palczewski, K., McDowell, J. H. & Hargrave, P.A. (1988a) Purification and characterization of rhodopsin kinase. Journal of Biological Chemistry 263:14067-73. [aMDB] (1988b) Rhodopsin kinase: Substrate specificity and factors that influence activity. Biochemistry 27:2306-12. [aMDB] Palczewski, K., Rispoli, G. & Detwiler, P. B. (1992) The influence of arrestin (48K protein) and rhodopsin kinase on visual transduction. Neuron 8:117— 26. [aMDB] Palczewski, K., Subbaraya, I., Gorczyca, W. A., Helekar, B. S., Ruiz, C. C , Ohguro, H., Huang, J., Zhao, X., Crabb, J. W , Johnson, R. S., Walsh, K. A., Cray-Keller, M. P., Detwiler, P. B. & Baehr, W. (1994) Molecular cloning and characterization of retinal photoreceptor guanylyl cyclaseactivating protein. Neuron 13:395-404. [KS] Papac, D. I., Oatis, J. E., Jr., Crouch, R. K. & Knapp, D. R. (1993) Mass spectrometric identification of phosphorylation sites in bleached bovine rhodopsin. Biochemistry 32:5930-34. [aMDB] Papac, D. I., Thornburg, K. R., Bullesbach, E. E., Crouch, R. K. & Knapp, D. R. (1992) Palmitylation of a G-protein coupled receptor: Direct analysis by tandem mass spectrometry. journal of Biological Chemistry 267:16889-94. [aPAH] Papermaster, D. S. & Dreyer, W. J. (1974) Rhodopsin content in the outer segment membranes of bovine and frog retinal rods. Biochemistry 13:2438-44. [aPAH] Pearson, P. L., Matheson, N. W., Flescher, D. C , Robbins, F. J. (1992) The CDB[TM] human genome data base anno 1992. Nucleic Acids Research 20:2201-6. [aSPD] Peng, Y.-W, Robishaw, J. D., Levine, M. A. & Yau, K.-W. (1992) Retinal rods and cones have distinct G protein B and 7 subunits. Proceedings of the National Academy of Sciences USA 89:10882-86. [aSPD] Pepperberg, D. R., Cornwall, M. C , Kahlert, M., Hofmann, K. P., Jin, J., Jones, G. J. & Ripps, H. (1992) Light-dependent delay in the falling phase of the retinal rod photoresponse. Visual Neuroscience 8:918. [aMDB, RKC] Pepperberg, D. R., Jin, J. & Jones, G. J. (1994) Modulation of transduction gain in light adaptation of retinal rods. Visual Neuroscience 11:53— 62. [aMDB, RKC] Perez-Sala, D., Tan, E. W., Canada, F. J. & Rando, R. R. (1991) Methylation and demethylation reactions of guanine nucleotide-binding proteins of retinal rod outer segments. Proceedings of the National Academy of Sciences USA 88:3043-46. [aMDB] Perlman, J. I., Nodes, B. R. & Pepperberg, D. R. (1982) Utilization of retinoids in the bullfrog retina. Journal of General Physiology 80:885913. [RKC] Pfcnninger, K. H., Ellis, L., Johnson, M. P., Friedman, L. B. & Somlo, S. (1983) Nerve growth cones isolated from fetal rat brain: Subcellular fractionation and characterization. Cell 35:573-84. [aZX] Pfister, C , Bennett, N., Bruckert, F., Catty, P., Clerc, A., Pages, F. & Deterre, P. (1993) Interactions of a G-protein with its effector: Transducin and cGMP phosphodiesterase in retinal rods. Cellular Signalling 5:23551. [aMDB] Philip, N. J., Chang, W. & Long, K. (1987) Light-stimulated movement in rod photoreceptor cells of the rat retina. FEBS Letters 225:127-31. [JFM] Phillips, W. J. &r Cerione, R. A. (1994) A C-terminal peptide of bovine rhodopsin binds to the transducin alpha-subunit and facilitates its activation. Biochemical Journal 299(Pt 2):351-57. [ADA] Piascik, M. T , Piascik, M. F., Hitzemann, R. J. & Potter, J. D. (1981) Ca2+ -dependent regulation of rat caudate nucleus adenylate cyclase and effects on the response to dopamine. Mol Pharmacol. 20, 2: 319-25. [aZX] Picones, A. & Korenbrot, J. I. (1992) Permeation and interaction of monovalent cations with the cGMP-gated channel of cone photoreceptors. Journal of Ceneral Physiology 100:647-73. [aRSM] Pistorius, A. M. & deGrip.W. J. (1994) Rhodopsin's secondary structure revisited: Assignment of structural elements. Biochemical and Biophysical Research Communications 3:1040-45. [ADA] Pittler, S. J. or Baehr, W. (1991) Identification of a nonsense mutation in the rod photoreceptor cGMP phosphodiesterase B-subunit gene of the rd mouse. Proceedings of the National Academy of Sciences USA 88:832226. [aSPD, MWK] Pittler, S. J., Keeler, C. E., Sidman, R. L. & Baehr, W. (1993) PCR analysis of DNA from 70-year-old sections of rodless retina demonstrates identity with the mouse rd defect. Proceedings of the National Academy of Sciences USA 90:9616-19. [aSPD] Pittler, S. J., Lee, A. K., Altherr, M. R., Howard, T. A., Seldin, M. F., 516 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 Hurwitz, R. L., Wasmuth, J. J. & Baehr, W. (1992) Primary structure and chromosomal localization of human and mouse rod photoreceptor cGM Pgated cation channel. Journal of Biological Chemistry 267:625762. [aRSM] Plantner, J. J., Le, M.-L. & Kean, E. L. (1991) Enzymatic deglycosylation of bovine rhodopsin. Experimental Eye Research 53:269-74. [aPAH] Pober, J. S. & Stryer, L. (1975) Light dissociates enzymatically-cleaved rhodopsin into two different fragments. Journal of Molecular Biology 95:477-81. [rPAH] Polans, A. S, Buczylko, J., Crabb, J. & Palczewski, K. (1991) A photoreceptor calcium binding protein is recognized by autoantibodies obtained from patients with cancer-associated retinopathy. Journal of Cell Biology 112:981-89. [aJBH, JFM, ASP] Polans, A. S., Burton, M. D., Haley, T. L., Crabb, J. W. & Palczewski, K. (1993) Recoverin, but not visinin, is an autoantigen in the human retina identified with a cancer-associated retinopathy. Investigative Ophthalmology and Visual Science 34:81-90. [aJBH, ASP] Polans, A. S., Hermolin, J. & Bownds, M. D. (1979) Light-induced phosphorylation of two proteins in frog rod outer segments. Journal of Ceneral Physiology 74:595-613. [aMDB] Pongs, O., Lindemeier, J., Zhu, X. R., Theil, T., Engelkamp, D., KrahJentgens, I., Lambrecht, H. G., Koch, K. W., Schwemer, J., Rivosecchi, R., Mallart, A., Galceran, J., Canal, I., Barbas, J. A. & Ferrus, A. (1993) Frequenin: A novel calcium-binding protein that modulates synaptic efficacy in the Drosophila nervous system. Neuron 11:15— 28. [KT] Popova, J. S., Johnson, G. L. & Rasenick, M. M. (1994) Journal of Biological Chemistry 269:21748-54. [MMR] Poulain, C , Ferrus, A. & Mallart, A. (1994) Modulation of type A K+ current in Drosophila larval muscle by internal Ca2+; effects of the overexpression of frequenin. European Journal of Physiology 427:71— 79. [KT] Probst, W. C , Snyder, L. A., Schuster, D. I., Brosius, J. & Sealfon, S. C. (1992) Sequence alignment of the G-protein coupled receptor superfamily. DNA and Cell Biology 11:1-20. [aPAH] Pugh, E. N., Jr. & Lamb, T. D. (1990) Cyclic GMP and calcium: The internal messengers of excitation and adaptation in vertebrate photoreceptors. Vision Research 30:1923-48. [aMDB] (1993) Amplification and kinetics of the activation steps in phototransduction. Biochimica et Biophysica Ada 1141:111-49. [aMDB, KWK] Pullen, N., Brown, N. C., Sharma, R. P. or Akhtar, M. (1993) Cooperativity during multiple phosphorylations chatlyzed by rhodopsin kinase: Supporting evidence using synthetic phosphopeptides. Biochemistry 32:3958-64. [ADA] Pulvermuller, A., Palczewski, K. & Hofmann, K. P. (1993) Interaction between photoaetivated rhodopsin and its kinase: Stability and kinetics of complex formation. Biochemistry 32:14082-88. [aMDB] Quamme, G. A. & Rabkin, S. W. (1990) Cytosolic free magnesium in cardiac myocytes: Identification of a Mg2+ influx pathway. Biochemical and Biophysical Research Communications 167:1406-12. [RLH] Radmacher, M., Fritz, M., Hansma, H. C. & Hansma, P. K. (1994) Direct observation of enzyme activity with the atomic force microscope. Science 265:1577-79. [ADA] Raju, B., Murphy, E., Levy, L. A., Hall, R. D. or London, R. E. (1989) A fluorescent indicator for measuring cytosolic free magnesium. American Journal of Physiology 256:C540-48. [RLH] Raleigh, D. P., Levitt, M. H. & Griffin, R. G. (1988) Rotational resonance in solid state NMR. Chemistry b Physics Letters 146:71-76. [SOS] Rao.V. R., Cohen, G. B. & Oprian, D. D. (1994) Rhodopsin mutation G90D and a molecular mechanism for congenital night blindness. Nature 367:639-42. [aSPD] Ratan, R. R., Murphy, T. H. or Baraban, J. M. (1994) Oxidative stress induces apoptosis in embryonic cortical neurons. Journal of Neurochemistry 62:376-79. [MT] Ratto, G. M., Payne, R., Owen, W. G. or Tsien, R. Y. (1988) The concentration of cytosolic free calcium in vertebrate rod outer segments measured with Fura-2. Journal of Neuroscience 8:3240-46. [aMDB] Ray, S., Zozulya, S., Niemi, G. A. et al. (1992) Cloning, expression and crystallization of recoverin, a calcium sensor in vision. Proceedings of the National Academy of Sciences USA 89:5705-9. [aJBH] Reig, C , Llecha, N., Antich, J., Gaen, E., Tcjada, I., Molina, M., Revents, J. or Carballa, M. (1994) A missense mutation (211HisArg) and a silent (160Thr) mutation within the rhodopsin gene in a Spanish autosomal dominant retinitis pigmentosa family. Human Molecular Cenelics 3:19596. [aSPD] Represa, A., Deloulme, J. C , Sensenbrenner, M., Ben-Ari, Y. & Baudier, J. (1990) Neurogranin: Immunocytochemica] localization of a brain- Re/erences/Controversies in Neuroscience HI specific protein kinase C substrate. Journal of Neuroscience 10:378292. [EDR] Rescorla, R. A. (1988) Behavioral studies of Pavlovian conditioning. Annual Review of Neuroscience 11:329-52. [TWA] Rcstagno, C , Maghtheh, M., Ghattacharya, S., Ferrone, M., Garnersoe, S., Samuclly, R. & Carbonara, A. (1992) A large deletion at the 3' end of the rhodopsin gene in an Italian family with a diffuse form of autosomal dominant retinitis pigmentosa. Human Molecular Genetics 2:207— 8. [aSPD] Richards, J. E., Kuo, C. Y., Boehnke, M. & Sieving, P. A. (1991) Rhodopsin Thr58Arg mutation in a family with autosomal dominant retinitis pigmentosa. Ophthalmology 98:1797-1805. [aSPD] Ridge, K. D., Bhattacharya, S., Nakayama, T. A. & Khorana, H. G. (1992) Light-stable rhodopsin: 2. An opsin mutant (Trp-265 —* Phe) and a retinal analog with a nonisomerizablc 11-Cis configuration form a photostable chromophore. Journal of Biological Chemistry 267:6770-75. [aPAH] Ridge, K. D., Lee, S. S. J. & Yao, L. L. (in press) In vivo assembly of rhodopsin from expressed polypeptide fragments. Proceedings of the National Academy of Sciences USA. [rPAH) Ricko, F. & Schwartz, E. A. (1994) A cCMP-gated current can control exocytosis at cone synapses. Neuron 13:863-73. [TGW] Ringens, P. J., Fang, M., Shinohara, T , Bridges, C. D., Lerea, C. L., Berson, E. L. & Dryja, T. P. (1990) Analysis of genes coding for S-antigen, interstitial retinol binding protein, and the alpha-subunit of cone transducin in patients with rctinitis pigmentosa. Investigative Ophthalmology and Vistial Science 31:1421-26. [aSPD] Rispoli, C. & Detwiler, P. B. (1989) Light adaptation in Gecko rods may involve changes in both the initial and terminal stage of the transduction cascade. Biophysical Journal 55:380. [aMDB] (1991) Interactions between calcium and cGMP in dialyzed detached retinal rod outer segments. In: Signal transduction in photoreceptor cells, ed. P. A. Hargrave & K. P. Hofmann. Springer-Verlag. [aMDB] Rispoli, G., Sather, W. A. & Detwiler, P. B. (1993) Visual transduction in dialysed detached rod outer segments from lizard retina. Journal of Physiology (London) 465:513-37. [aMDB] Robinson, P. R., Cohen, G. B., Zhukovsky, E. A. & Oprian, D. D. (1992) Constitutively active mutants of rhodopsin. Neuron 9:719-25. [aMDB, aSPD, aPAH] Robinson, P. R., Kawamura, S., Abramson, B. & Bownds, M. D. (1980) Control of the cyclic GMP phosphodiesterase of frog photoreceptor membranes. Journal of General Physiology 76:631-45. [aMDB] Rodriquez, J. A.. Herrera, C. A., Birch, D. C. & Daiger, S. P. (1993) A leucine to argininc amino acid substitution at codon 46 of rhodopsin is responsible for a severe form of autosomal dominant retinitis pigmentosa. Human Mutation 2:205-13. [aSPD] Rodriguez, J. A., Herrera, C. A., Birch, D. G., Heckenlively, J. R. & Daiger, S. P. (1993a) Rhodopsin mutations in patients with retinitis pigmentosa. American Journal of Human Cenetics 53:1224. [aSPD] Ronnett, G. V. & Snyder, S. H. (1992) Molecular messengers of olfoction. Trends in Neuroscience 15:508-13. [aMDB] Roof, D. J., Adamian, M. & Hayes, A. (1994) Rhodopsin accumulation at abnormal sites in retinas of mice with a human P23H rhodopsin transgene. Investigative Ophthalmology and Visual Science 35:404962. [MWKJ Root, M. J. & MacKinnon, R. (1993) Identification of an external divalent cation-binding site in the pore of a cGMP-octivated channel. Neuron 11:459-66. [arRSM] Rosenberg, C.B., Minocherhomjee, A. & Storm, D. R. (1987b) Reconstitution of calmodulin-sensitive adenylate cyclase. Methods in Enzymology 139:776-79. [aZX] Rosenberg, G. B. & Storm, D. R. (1987a) Immunological distinction between calmodulin-sensitive and calmodulin-insensitive adenylyl cyclases. Journal of Biological Chctnistry 262:7623-28. [oZX] Rosenfeld, P. J., Cowley, G. S., McGee, T. L., Sandberg, M. A., Berson, E. L. & Dryja, T. P. (1992) A null mutation in the rhodopsin gene causes rod photoreceptor dysfunction and autosomal recessive retinitis pigmentosa. Nature Genetics 1:209-13. [aSPD] Ross, E. M. & Oilman, A. G. (1980) Biochemical properties of hormonesensitive adenylyl cyclase. Annual Review of Biochemistry 49:53364. [aZX] Roth, N. S., Campbell, P. T , Caron, M. G., Lefkowitz, R. J. & Lohse, M. J. (1991) Comparative rates of desensitization of B-adrenergic receptors by the B-adrenergic receptor kinase and the cyclic AM P-dependent protein kinase. Proceedings of the National Academy of Sciences USA 88:62014. [aMDB] Routtenberg, A. (1995) Phosphoprotein regulation of memory formation: enhancement and control of synaptic plasticity by protein kinase C and F|. Annual of the Neto York Academy of Sciences 444:203-11. Roychowdhury, S. & Rasenick, M. M. (1994) Tubulin-G protein associations stabilizes GTP binding and activates GTPase. Biochemistry 33:98005. [MMR] Roychowdhury, S., Wang, N. & Rasenick, M. M. (1993) Biochemistry 32:4955-61. [MMR] Ryba, N. J. P. & Marsh, D. (1992) Protein rotational diffusion and lipid/protein interactions in recombinants of bovine rhodopsin with saturated diacylphosphatidylcholines of different chain lengths studied by conventional and saturation-transfer electron spin resonance. Biochemistry 31:7511-18. [aPAH] Saga, M., Mashima, Y., Kawashima, S., Akeo, K., Oguchi, Y., Kudoh, J. & Shimizu, N. (1993) Rhodopsin gene analysis for Japanese autosomal dominant retinitis pigmentosa patients. Investigative Ophthalmology and Visual Science 34:1459. [aSPD] Saitoh, S., Takamatsu, K., Kobayashi, M. & Noguchi, T. (1993) Distribution of hippocalcin mRNA and immunoreactivity in rat brain. Neuroscience Letters 157:107-10. [KT] (1994) Immunohistochemical localization of neural visinin-like Ca2+-binding protein 2 in adult rat brain. Neuroscience Letters 171:155-58. [KT] Sakmar, T. P., Franke, R. R. & Khorana, H. G. (1989) Glutamic acid-113 serves as the retinylidine schiffbase counterion in bovine rhodopsin. Proceedings of the National Academy of Sciences USA 86:830913. [aPAH] Salter, R. S., Krinks, M. H., Klee, C. B. & Neer, E. J. (1981) Calmodulin activates the isolated catalytic unit of brain adenylate cyclase. Journal of Biological Chemistry 256:9830-33. [aZX] Sanada, K., Kokame, K., Yoshizawa, T , Takao, T. & Fukada, Y. (1995) Role of heterogeneneous N-terminal acylation of recoverin in rhodopsin phosphorylation. Journal of Biological Chemistry 267:18274-18277 Sankila, E.-M., Pakarinen, L.. Kaariainen, H., Aittomaki, K , Korjalainen, S., Sistonen, P. & de la Chapelle, A. (1995) Assignment of an Usher syndrome type III (USH3) gene to chromosome 3q. Human Molecular Genetics 4:93-98. [rSPD] Sano, M. (1985) Calmodulin-dependent adenylate cyclase in rat retina and the response to dopamine. Brain Research 345:337-40. [aZX] Sather, W. A. & Detwiler, P. B. (1987) Intracellular biochemical manipulation of phototransduction in detached rod outer segments. Proceedings of the National Academy of Sciences USA 84:9290-94. [aM DB] Schacher, S., Castellucci, V. F. & Kandel, E. R. (1988) cAMP evokes longterm facilitation in Aplysia sensory neurons that requires new protein synthesis. Science 240:1667-69. [aZX] Schafrneister, C. E., Miercke, L. J. W. & Stroud, R. M. (1993) Structure at 2.5 A of a designed peptide that maintains solubility of membrane proteins. Science 262:734-39. [RKC] Schertler, G. F. X. (1992) Overproduction of membrane proteins. Current Opinion in Structural Biology 2:53444. [aPAH] Schertler, G. F. X., Bartunik, H. D., Michel, H. & Oesterhelt, D. (1993) Orthorhombic crystal form of bacteriorhodopsin nucleated on benzamidine diffracting to 3.6 A resolution. Journal of Molecular Biology 234:156-64. [RMG] Schertler, G. F. X. & Hargrave, P. A. (in press) Projection structures of frog rhodopsin in two crystal forms. Proceedings of the National Academy of Sciences USA. [raPAH] Schertler, G. F. X., Villa, C. & Henderson, R. (1993) Projection structure of rhodopsin. Nature 362:770-72. [aPAH, EAD, RMG] Schnetkamp, P. P. M. (1989) Na-Ca or Na-Ca-K exchange in the outer segments of vertebrate rod photoreceptors. Progress in Biophysics and Molecular Biology 54:1-29. [PPMS] (1990) Cation selectivity of and cation binding to the cGMP-dependent channel in bovine rod outer segment membranes. Journal of General Physiology 96:517-34. [aRSM] Schnetkamp, P. P. M., Li, X-.B., Basu, D. K. & Szerencsei, R. T. (1991) Regulation of free cytosolic Ca2+ concentration in the outer segments of bovine retinal rods by Na-Ca-K exchange measured with Fluo-3. Journal of Biological Chemistry 266:22975-82. [PPMS] Schnetkamp, P. P. M. & Szerencsei, R. T. (1993) Intracellular Ca2+ sequestration and release in intact bovine retinal rod outer segments. Journal of Biological Chemistry 268:12449-57. [PPMS] Scholz, K. P. & Byrne, J. H. (1988) Intracellular injection of cAMP induces a long-term reduction of neuronal K+ currents. Science 240:166466. [aZX] Schulte, T. H. & Marchesi.V. T. (1979) Conformation of the human erythrocyte glycophorin A and its constituent peptides. Biochemistry 18:275-80. [ADA] Scott, K., Sieving, P. A., Bingham, E., Bhagat, V. J., Sullivan, ]., Alpern, M. 6r Richards, J. E. (1993) Rhodopsin mutations associated with autosomal dominant retinitis pigmentosa. American Journal of Human Genetics 53:147. [aSPD] BEHAVIORAL AND BRAIN SCIENCES (1995) 18.3 517 Re/erences/Controversies in Neuroscience III Shapley, R. & Enroth-Cugell, C. (1984) Visual adaptation and retinal gain controls. Progress in Retinal Research 3:263-46. [aMDB] Sheffield, V. C , Fishman, G. A., Beck, J. S., Kimura, A. E. & Stone, E. M. (1991) Identification of novel rhodopsin mutations associated with retinitis pigmentosa by GC-damped denaturing gradient gel electrophoresis. American Journal of Human Genetics 49:699-706. [aSPD] Shichi, H., Lewis, M. S., Irreverre, F. & Stone, A. L. (1969) Biochemistry of visual pigments: 1. Purification and properties of bovine rhodopsin. Journal of Biological Chemistry 244:529-36. [aPAH] Shinozawa, T., Sokabe, M., Terada, S., Matsusaka, H. & Yoshizawa, T. (1987) Detection of cyclic GMP binding protein and ion channel activity in frog rod outer segments. Journal of Biochemistry 102:281-20. [aRSM] Shiono, T., Hotta, Y., Noro, M., Sakuma, T., Tamai, M., Hayakawa, M., Hashimoto, T., Fujiki, K., Kanai, A. & Nakajima, A. (1992) Clinical features of Japanese family with autosomal dominant retinitis pigmentosa caused by a single point mutation in codon 347 of rhodopsin gene. Japanese Journal of Ophthalmology 36:69-75. [aSPD] Shnyrov, V. L. & Berman, A. L. (1988) Calorimetric study of thermal denaturation of vertebrate visual pigments. Biomedica el Biochimica Acta 47:355-62. [aPAH] Shuster, T. A. & Farber, D. B. (1984) Phosphorylation in sealed rod outer segments: Effects of cyclic nucleotides. Biochemistry 23:51521. [aMDB] Shuster, T. A., Nagy, A. K., Conly, D. C. & Farber, D. B. (1992) Direct zinc binding to purified rhodopsin and disc membranes. Biochemical Journal 282:123-28. [aPAH] Sidman, R. L. & Green, M. D. (1965) Retinal degeneration in the mouse: Location of the rd locus in linkage group XVII. Journal of Heredity 56:23-29. [MWK] (1970) Nervous, a new mutant mouse with cerebellar disease. In: Les mutants pathologiques chez ianimal, ed. M. Sabourdy. Centre National de la Recherche Scientifique. [MWK] Sieving, P. A., Richards, J. E., Bingham, E. L. & Naarendorp, F. (1992) Dominant congenital complete nyctalopia and Gly90Asp rhodopsin mutation. Investigative Ophthalmology and Visual Science 33:1397. [aSPD] Silva, A. J., Paylor, R., Wehner, J. M. & Tanegawa, S. (1992) Impaired spatial learning in a-calcium-calmodulin kinase II mutant mice. Science 257:206— 211. [aZX, TWA, EDR] Silva, A. J., Stevens, C. F., Tonegawa, S. & Wang, Y. (1992) Deficient hippocampal long-term potentiation in alpha-calmodulin kinase II mutant mice. Science 257:201-6. [aZX, EDR] Sitaramayya, A. (1986) Rhodopsin kinase prepared from bovine rod disk membranes quenches light activation of cGMP phosphodiesterase in a reconstituted system. Biochemistry 25:5460-68. [aMDB] Sitaramayya, A. & Liebman, P. A. (1983a) Mechanism of ATP quench of phosphodiesterase activation in rod disc membranes. Journal of Biological Chemistry 258:1205-9. [aMDB] (1983b) Phosphorylation of rhodopsin and quenching of cyclic GMP phosphodiesterase activation by ATP at weak bleaches. Journal of Biological Chemistry 258:12106-9. [aMDB] Skene, J. H. P. (1989) Axonal growth-associated proteins. Annual Review of Neuroscience 12:127-56. [aZX] Skene, J. H. P., Jacobson, R. D., Snipes, G. J., McGuire, C. B., Norden, J. J. & Freeman, J. A. (1986) A protein induced during nerve growth (GAP-43) is a major component of growth-cone membranes. Science 233:783-86. [aZX] Skene, J. H. P. & Willard, M. J. (1981a) Changes in axonally transported proteins during axon regeneration in toad retinal ganglion cells. Journal of Cell Biology 89:86-95. [aZX] Slack, J. R. & Pockett, S. (1991) Cyclic AMP induces long-term increase in synaptic efficacy in CA1 region of rat hippocampus. Neuroscience Letters 130:69-72. [aZX] Small, K. W., Weber, J. L., Roses, A., Lennon, F., Vance, J. M. & PericakVance, N. U. (1992) North Carolina macular dystrophy is assigned to chromosome 6. Cenomics 13:681-85. [aSPD] Smigel, M. D. (1986) Purification of brain adenylyl cyclase using forskolin affinity chromatography and WGA-Sepharose. Journal of Biological Chemistry 261:1976-82. [aZX] Smith, R. J. H., Lee, E. C , Kimberling, W. J., Daiger, S. P., Pelias, M. Z., Keats, B. J. B., Jay, M. L., Bird, A., Reardon, W., Guest, M., Agyagri, R. & Hejtmancik, F. (1992) Localization of two genes for Usher syndrome type I to chromosome 11. Cenomics 14:995-1002. [aSPD] Smith, S. J. & Augustine, G. J. (1988) Calcium ions, active zones and synaptic transmitter release. Trends in Neuroscience 11:458-64. [aZX] Smith, S. O., Courtin, J., de Groot, H., Gebhard, R. & Lugtenburg, J. (1991) 13 C magic angle spinning NMR studies of bathorhodopsin, the primary photoproduct of rhodopsin. Biochemistry 30:7409-15. [SOS] 518 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 Smith, S. O., de Groot, H., Gebhard, R. & Lugtenburg, J. (1992) Magic angle spinning NMR studies on the metarhodopsin II intermediate of bovine rhodopsin: Evidence for an unprotonated Scruff base. Photochemistry b Photobiology 56:1035-39. [SOS] Smith, S. O. & Peersen, O. B. (1992) Solid-state NMR approaches for studying membrane protein structure. Annual Review of Biophysical and Biomolecular Structure 21:25-47. [SOS] Sneyd, J. & Tranchina, D. (1989) Phototransduction in cones: An inverse problem in enzyme kinetics. Bulletin of Mathematical Biology 51:74984. [aMDB] Somlyo, A. P. & Walz, B. (1985) Elemental distribution in rana pipiens retinal rods: Quantitative electron probe analysis. Journal of Physiology (London) 358:183-95. [RLH] Sowadski, J. M. (1994) Crystallization of membrane proteins. Current Opinion in Structural Biology 4:761-64. [RMG] Squire, L. R. & Davis, H. P. (1981) The pharmacology of memory: A neurobiological perspective. Annual Review of Pharmacology and Toxicology 21:323-56. [aZX] Srivastava, D., Fox, D. A. & Hurwitz, R. L. (in press) The role of magnesium in the hydrolysis of cGMP by the bovine retina] rod cGMP phosphodiesterase. Biochemical Journal. [RLH] Srivastava, D., Hurwitz, R. L. & Fox, D. A. (submitted) Effects of magnesium on cyclic GMP hydrolysis by the bovine retinal rod cyclic GMP phosphodiesterase. [RLH] Stanton, P. K. & Sarvey, J. M. (1985a) The effect of high-frequency electrical stimulation and norepinephrine on cyclic AMP levels in normal versus norepinephrine-depleted rat hippocampal slices. Brain Research 358:34348. [aZX] (1985b) Depletion of norepinephrine, but not serotonin, reduces long-term potentiation in the dentate gyrus of rat hippocampal slices. Journal of Neuroscience 5:2169-76. [aZX] Steinberg, R. H., Fisher, S. K. & Anderson, D. H. (1980) Disc morphogenesis in vertebrate photoreceptors. Journal of Comparative Neurology 190:501-18. [aMDB] Stepanik, P. L., Lerious, V. & McGinnis, J. F. (1993) Developmental appearance, species and tissue specificity of mouse 23-kDa, a retinal calcium-binding protein (recoverin). Experimental Eye Research 57:18997. [JFM] Stern, J. H., Kaupp, U. B. & MacLeish, P. R. (1986) Control of the lightregulated current in rod photoreceptors by cyclic GMP, calcium and L-cis-diltiazem. Proceedings of the National Academy of Sciences USA 83:1167. [aRSM] Stone, E. M., Kimura, A. E., Nichols, B. E., Khadivi, P., Fishman, G. A. & Sheffield, V. C. (1991) Regional distribution of retinal degeneration in patients with proline to histidine mutation in codon-23 of the rhodopsin gene. Ophthalmology 98:1806-13. [aSPD] Stone, E. M., Nichols, B. E., Streb, L. M., Kimura, A. E. & Sheffield, V. C. (1992a) Genetic linkage of vitelliform macular degeneration (Best's disease) to chromosome Ilql3. Nature Cenetics 1:246-50. [aSPD] Stone, E. M., Vandenburgh, K., Kimura, A. E., Lam, B. L., Fishman, G. A., Heckenlively, J. R., Castillo, T. A. & Sheffield, V. C. (1993) Novel mutations in the peripherin (RDS) and rhodopsin genes associated with autosomal dominant retinitis pigmentosa (ADRP). Investigative Ophthalmology and Visual Science 34:1149. [aSPD] Straume, M., Mitchell, D., Miller, J. & B. J. Litman. (1990) Interconversions of metarhodopsins I and II: A branched photointermediate decay model. Biochemistry 29:9135-42. [ADA] Stryer, L. (1986) Cyclic GMP cascade of vision. Annual Review of Neuroscience 9:87-119. [aSPD, aRSM] (1991) Visual excitation and recovery. Journal of Biological Chemistry 266:10711-14. [aMDB, aJBH, aRSM] Subbaraya, I., Qin, N., McGinness, J. F. & Baehr, W. (1994) Cloning and expression of frog recoverin and its interaction with rhodopsin kinase. Investigative Ophthalmology and Visual Science 35:1485. [aJBH, JFM, AY] Suber, M. L., Pittler, S. J., Qin, N., Wright, G. C , Holcombe, V., Lee, R H., Craft, C M . , Lolley, R. N., Baehr, W. & Hurwitz, R. L. (1993) Irish setter dogs affected with rod/cone dysplasia contain a nonsense mutation in the rod cGMP phosphodiesterase p-subunit gene. Proceedings of the National Academy of Sciences USA 90:396872. [aSPD, MWK] Subramaniam, S., Gerstein, M., Osterhelt, D. & Henderson, R. (1993) Electron diffraction analysis of structural changes in the photocycle of bacteriorhodopsin. EMBO Journal 12:1-8. [aPAH, RMG] Sugimoto, Y., Yatsunami, K., Tsujimoto, M., Khorana, H. G. 6r Ichikawa, A. (1991) The amino acid sequence of a glutamic acid-rich protein from bovine retina as deduced from the cDNA sequence. Proceedings of the National Academy of Sciences USA 88:3116-19. [rRSM] References I'Controversies in Neuroscience III Suh, K. H. & Hamin, H. (1988) Components I and II form a complex with the beta gamma subunils of C protein in frog rod outer segments. Investigative Ophthalmology and Visual Science 28:218. [aMDB] Sullivan, J. M., Mnkris, C. S., Dickinson, P., Mulhall, L. E. M., Sc, D. A., Forrest, S., Cotton, R. C. H. & Loughnan, M. S. (1993) A new codon 15 rhodopsin gene mutation in autosomal dominant retinitis pigmentosa is associated with suctorial disease. Archives of Ophthalmology 111:151217. [aSPD] Sullivan, J. M., Scott, K. M., Falls, B. E., Richards, J. E. & Sieving, P. A. (1993) A novel rhodopsin mutation at the retinal binding site (LYS-296MET) in ADRP. Investigative Ophthalmology and Visual Science 34:1149. [aSPDl Sung, C.-H., Davenport, C. M., Hennessy, J. C , Maumenee, I. H., Jacobson, S. C , Heckenlively, J. R., Nowakowski, R., Fishman, C , Counis, P. & Nathans, J. (1991) Rhodopsin mutations in autosomal dominant retinitis pigmentosa. Proceedings of the National Academy of Sciences USA 88:6481-85. [aSPD] Sung, C.-H., Davenport, C. M. & Nathans, J. (1993) Rhodopsin mutations responsible for autosomal dominant retinitis pigmentosa. Clustering of functional classes along the polypeptide chain. Journal of Biological Chemistry 268:26645-49. [aSPD] Sung, C.-H., Schneider, B. C , Agarwal, N., Papermaster, D. S. & Nathans, J. (1991) Functional heterogeneity of mutant rhodopsins responsible for autosomal dominant retinitis pigmentosa. Proceedings of the National Academy of Sciences USA 88:8840-44. [aSPD] Sutkowski, E. M., Tang, W., Broome, C. W , Robbins, J. D. & Seamon, K. B. (1994) Biochemistry 33:12852-859. [MMR] Takamatsu, K. & Uyemura, K. (1992) Identification of recoverin-like immiinorcactivity in mouse brain. Brain Research 571:350-53. [KT] Tamnra, T., Nakatani, K. & Van, K.-W. (1991) Calcium feedback and sensitivity regulation in primate rods. Journal of General Physiology 98:95-130. [aMDB] Tang, W. J. & Cilman, A. C. (1991) Type-specific regulation of adenylyl cyclase by G protein beta gamma subunits. Science 254:1500-03. [TWA] Tang, W. J., Krupinski, J. & Cilman, A. C. (1991) Expression and characterization of calmodulin-activated (type I) adenylyl cyclase. Journal of Biological Chemistry 266:8595-8603. [arZX, TWA, MMR] Taussig, R. T., Quarmby, L. R. & Cilman, A. C. (1993) Regulation of purified type I & type III adenylyl cyclases by C protein beta/gamma subunits. Journal of Biological Chemistry 268:9-12. [aZX] Thirkill, C. E., Tait, R. C , Tyler, N. K., Roth, A. M. & Keltner, J. L. (1992) The cancer-associated retinopathy antigen is a recoverin-like protein. Investigative Ophthalmology and Visual Science 33:2768-72. [aJBH, JFM] Tiberi, M. & Caron, M. C. (in press) Journal of Biolgical Chemistry. [MMR] Ting, T. D. & Ho, Y.-K. (1991) Molecular mechanism of GTP hydrolysis by bovine transducin: Pre-steady-state kinetic analyses. Biochemistry 30:8996-9007. [aMDB] Toscano, W. A., Jr., Westcott, K. R., LaPorte, D. C. & Storm, D. R. (1979) Evidence for a dissociable protein subunit required for calmodulin stimulation of brain adenylate cyclase. Proceedings of the National Academy of Sciences USA 76:P5582-86. [aZX] Tola, M. R., Xia, Z., Storm, D. R. & Schimerlik, M. I. (1990). Reconstitution of muscarinic receptor mediated inhibition of adenylyl cyclase. Molecular Pharmacology 37:950-57. [aZX] Toyoshima, C. & Unwin, N. (1990) Three-dimensional structure of the acetylcholine receptor by cryoelectron microscopy and helical image reconstruction. Journal of Cell Biology 111:2623-35. [aPAH] Tranchina, D., Sneyd, J. At Cadenas, I. D. (1991) Light adaptation in turtle cones: Testing and analysis of a model for phototransduction. Biophysical Journal 60:217-37. [aMDB] Travis, C. H . (1991) Name dropping. Nature 349:24. [aSPD] Travis, G. H, Brennan, M. B., Danielson, P. E., Kozak, C. A. or Sutcliffe, J. G. (1989) Identification of photoreceptor-specific mRNA encoded by the gene responsible for retinal degeneration slow (rds). Nature 338:7073. [MWK] Travis, C. H. , Sutcliffe, J. G. or Bok, D. (1991a) The retinal degeneration slow (rds) gene product is a photoreceptor disc membrane-associated glycoprotein. Neuron 6:61-70. [aSPD, MWK] Treisman, G., Bagley, S. & Gnegy, M. (1983) Calmodulin-sensitive and calmodulin-insensitive components of adenylate cyclase activity in rat striatum have differential responsiveness to guanyl nucleotides. Journal of Neurochemistry 41:1398-1406. [MMR] Tsuboi, S. Matsumoto, H., Jackson, K. W., Tsujimoto, K., Williams, T. & Yamazaki, A. (1994a) Phosphorylation of an inhibitory subunit of cGMP phosphodiesterase in Rana catesbiana rod photoreceptors: 1. Characterization of the phosphorylation. Journal of Biological Chemistry 269:15016-23. [aMDB] Tsuboi, S., Matsumoto, H. & Yamazaki, A. (1994) Phosphorylation of an inhibitory subunit of cGMP phosphodiesterase in Rana catesbiana rod photoreceptors: 2. A possible mechanism for the turnoffof cGMP phosphodiesterase without GTP hydrolysis. Journal of Biological Chemistry 269:15024-29. [aMDB, AY] Tully, T, Preat, T, Boynton, S. C. & Del Vecchio, M. (in press). Genetic dissection of consolidated memory in Drosophila melanogaster. Cell. [EDR] Tycko, R.. & Smith, S. O. (1993) Symmetry principles in the design of pulse sequences for structural measurements in magic angle spinning nuclear magnetic resonance. Journal of Chemical Physics 98:932— 43. [SOS] Udovichenko, I. P., Cunnick, J., Gonzalez, K. & Takemoto, D. J. (1994) Functional effect of phosphorylation of the photoreceptor phosphodiesterase inhibitory subunit by protein kinase C. Journal of Biological Chemistry 269:9850-56. [aMDB] Uhl, R., Wagner, R. & Ryba, N. (1990) Watching C proteins at work. Trends in Neuroscience 13:64-70. [aMDB] Ulshafer, R. J., Garcia, C. A. & Hollyfield, J. G. (1980) Sensitivity of photoreceptors to elevated levels of cGMP in the human retina. Investigative Ophthamology and Visual Science 19:1236-41. [aSPD] Ulshafer, R. J., Sherry, D. M., Dawson, R., Jr. & Wallace, D. R. (1990) Excitatory amino acid involvement in retinal degeneration. Brain Research 531:350-54. [MT] Unger, V. M., Hargrave, P. A. & Schertler, G. F. X. (1995) Lxxalization of the transmembrane helices in the three-dimensional structure of frog rhodopsin. Biophysics Journal 68: A330. [rPAH] Unwin, P. N. T. & Zampighi, G. (1980) Structure of the junction between communicating cells. Nature 283:545-49. [aPAH] Vaithinathan, R., Berson, E. L. & Dryja, T. P. (1993) Further screening of the rhodopsin gene in patients with autosomal dominant retinitis pigmentosa. Investigative Ophthalmology and Visual Science 34:1149. [aSPD] van der Steen, R., Groesbeek, M., van Amsterdam, L. J. P., Lugtenburg, J., van Oostrum, J. & de Grip, W. J. (1989) All E-10,20-methanoretinylopsin, light-stable rhodopsin. Synthesis and spectroscopy of all E-10,20methano- and all-E-retinoyl fluoride and their reaction with bovine opsin. Recueil des Travauz Chimiques des Pays-Bos 108:20-27. [aPAH] Van Nie, R., Ivanyi, D. & Demant, P. 91978) A new H-2 linked mutation, rds, causing retinal degeneration in the mouse. Tissue Antigens 12:106— 8. [MWK] van Soest, S., van den Bora, L. 1., Gal, A., Farrar, G. J., BleekerWagemakers, L. M., Westerveld, A., Humphries, P, Sandkuijl, L. A. or Bergen, A. A. B. (1994) Assignment of a gene for autosomal dominant recessive retinitis pigmentosa (RP12) to chromosome Iq31-q32.1 in an inbred and genetically heterogeneous disease population. Cenomics 22:499-504. [aSPD, AB] Vanderkooi, G. (1974) Organisation of proteins in membranes with special reference to the cytochrome oxidase system. Biochimica et Biophysica Acta 344:307-45. [aPAH] Villacres, E. C., Xia, Z.. Bookbinder, L. H., Edelhoff, S., Adler. D. A., Disteche, C. M. & Storm, D. R. (1993) Cloning, chromosomal mapping, and expression of human fetal brain type I adenylyl cyclase. Genomics 16:473-78. [aZX] Vuong, T. M. & Chabre, M. (1990) Subsecond deactivation of transducin by endogenous GTP hydrolysis. Nature 346:71-74. [aMDB] (1991) Deactivation kinetics of the transduction cascade of vision. Proceedings of the National Academy of Sciences USA 88:981317. [aMDB] Vuong, R. M., Chabre, M. & Stryer, L. (1984) Millisecond activation of transducin in the cyclic nucleotide cascade of vision. Nature 311:65961. [aMDB] Wagner, R., Ryba, N. & Uhl, R. (1988) Sub-second turnover of transducin CTPase in bovine rod outer segments. FEBS Letters 234:4448. [aMDB] (1989) Calcium regulates the rate of rhodopsin disactivation and the primary amplification step in visual transduction. FEBS Letters 242:24954. [aMDB] Wahlsten, D. (1991) Sample size to detect a planned contrast and a one degree-of-freedom interaction effect. Psychological Bulletin 110:58795. [DW] Walsh, M. P., Valentine, K. A., Ngai, P. K., Carruthers, C. A. & Hollenberg, M. D. (1984) Ca2+-dependent hydrophobic-interaction chromatography: Isolation of a novel Ca2+-binding protein and protein kinase C from bovine brain. Biochemistry Journal 224:117-27. [KT] Wang, N. & Rasenick, M. M. (1991) Tubulin-G protein interactions involve microtubule polymerization domains. Biochemistry 30:1095765. [MMR] Wang, N., Yan, K. & Rasenick, M. M. (1990) Tubulin binds specifically to the BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 519 Re/erences/Controversies in Neuroscience HI signal-transducing proteins, Csa and Cial. Journal of Biological Chemistry 265:1239-42. [MMR| Watanabe, S. I. & Murakami, M. (1991) Similar properties of the cGMPactivated channels between cones and rods in the carp retina. Visual Neuroscience 6:563-68. [aRSM] Watson, J. B., Battenberg, E. F., Wong, K. K., Bloom, F. E. & Sutcliffe, J. G. (1990) Subtractive cDNA cloning of RC3, a rodent cortex-enriched mRNA encoding a novel 78 residue protein. Journal of Neuroscience Research 26:397-408. [EDR] Watson, J. B., Sutcliffe, J. G. & Fisher, R. S. (1992) Localization of the protein kinase C phosphorylation/calmodulin binding substrate RC3 in dendritic spines of neostriatal neurons. Proceedings of the National Academy of Sciences USA 89:8581-85. [EDR] Wayman, G. A., Impey, S., Wu, Z., Kinsvogel, W., Prichard, L. & Storm, D. R. (1994) Synergistic activation of the type I adenylyl cyclase by Ca2+ and Gs coupled receptors in vivo. Journal of Biological Chemistry 269:25400-5. [arZX, TWA, MMR] Weber, B. H. F., Vogt, G., Pruett, R. C , Stfihr, H. & Felbor, U. (1994) Mutations in the tissue inhibitor of metalloproteinases-3 (TIMP3) in patients with Sorby's fundus dystrophy. Nature Genetics 8:352— 55. [rSPD] Wehner, R., Marsh, A. C. & Wehner, S. (1992) Desert ants on a thermal tightrope. Nature 357: 586-87. [aPAH] Weitz, C. J. & Nathans, J. (1992) Histidine residues regulate the transition of photoexcited rhodopsin to its active conformation, metarhodopsin II. Neuron 8:465-72. [aSPD] Weitz, C. J., Went, L. N. & Nathans, J. (1992a) Human tritanopia associated with a third amino acid substitution in the blue-sensitive visual pigment. American Journal of Human Genetics 51:444—46. [aSPD] Weleber, R. G., Carr, R. E., Murphey, W. H., Sheffield V. & Stone, E. M. (1993) Phenotypic variation including retinitis pigmentosa, pattern dystrophy and fundus flavimaculatis in a single family with a deletion of codon 153 or 154 of the pheripherin/RDS gene. Archives of Ophthalmology 111:1531-42. [AB, aSPD] Well, D., Blanchard, S., Kaplan, J., Guilford, P., Gibson, F., Walsh, J., Mburu, P., Varela, A., Levillers, J., Weston, M. D., Kelly, P. M., Kimberling, W. J., Wagenaar, M., Levi-Acobas, F., Larget-Piet, D., Munnich, A., Steel, K. P., Brown, S. D. M & Petit, C. (1995) Defective mysin VIIA gene responsible for Usher syndrome type IB. Nature 374:60-61. [rSPD] Wells, J., Wroblewski, J., Keen, J., Inglehearn, C , Jubb, C , Eckstein, A., Jay, M., Arden, G., Bhattacharya, S., Fitzke, F. & Bird, A. (1993) Mutations in the human retinal degeneration slow (rds) gene can cause either retinitis pigmentosa or macular dystrophy. Nature Genetics 3:21318. [aSPD] Welte, W. & Wacker, T. (1991). Protein-detergent micellar solutions for the crystallization of membrane proteins: Some general approaches and experiences with the crystallization of pigment-protein complexes from purple bacteria. In: Crystallization of membrane proteins, ed. H. Michel. CRC Press. [aPAH] Westcott, K. R., LaPorte D. C. & Storm, D. R. (1979) Resolution of adenylyl cyclase sensitive and insensitive to Ca2+ and CDR by CDR-sepharose affinity chromatography. Proceedings of the National Academy of Sciences USA 76:204-28. [WH, aZX] Weyand, I., Godde, M., Frings, S., Weiner, J., Muller, Altenhofen, W., Hatt, H. & Kaupp, U. B. (1994) Cloning and functional expression of a cyclicnucleotide-gated channel from mammalian sperm. Nature 368:85963. [rRSM] Whelan, J. P. & McCinnis, J. F. (1988) Light-dependent subcellular movement of photoreceptor proteins. Journal of Neuroscience Research 20:263-70. [JFM] Wilden, U., Hall, S. W. & Kuhn, H. (1986) Phosphodiesterase activation by photoexcited rhodopsin is quenched when rhodopsin is phosphorylated and binds the intrinsic 48 KDa protein of rod outer segments. Proceedings of the National Academy of Sciences USA 83:117478. [aMDB] Wilden, U. & Kuhn, H. (1982) Light-dependent phosphorylation of rhodopsin: Number of phosphorylation sites. Biochemistry 21:301422. [aMDB] Williams, D. S., Linberg, K. A., Vaughan, D. K., Fariss, R. N. & Fisher, S. K. (1988) Disruption of microfilament organization and deregulation of disk membrane morphogenesis by cytochalasin D in rod and cone photoreceptors. Journal of Comparative Neurology 272:161-76. [aMDB] Wohlfart, P., Haase, H., Molday, R. S. & Cook, N. J. (1992) Antibodies against synthetic peptides used to determine the topology and site of glycosylation of the cGMP-gated channel from bovine rod photoreceptors. Journal of Biological Chemistry 267:644-48. [aRSM] Wohlfart, P., Muller, H. & Cook, N. J. (1989) Lectin binding and enzymatic 520 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 deglycosylation of cGMP-gated channel from bovine rod photoreceptors. Journal of Biological Chemistry 264:20934-39. [aRSM] Wong, S. & Molday, R. S. (1986) A spectrin-like protein in retinal rod outer segments. Biochemistry 25:6294-6300. [aRSM] Wu, S. M., Qiao, X., Noebels, J. L. & Yang, X. L.: Localization and modulatory actions of zinc in vertebrate retina. Vision Research 33:261116. [MT] Wu, Z., Thomas, S. A., Xia, Z., Willacres, E. C , Palmiter, R. D. & Storm, D. R. (in press) Deficient spatial memory and altered LTP in type 1 adenylyl cyclase mutant mice. Proceedings of the National Academy of Sciences 92: 220-4. [arZX] Wu, Z., Wong, S. T. & Storm, D. R. (1993) Modification of the calcium and calmodulin sensitivity of the type I adenylyl cyclase by mutagenesis of its calmodulin binding domain. Journal of Biological Chemistry 268:2376668. [aZX] Xia, Z., Choi, E. J., Wang, F., Blazynski, C., & Storm, D. R. (1993) The type I calmodulin sensitive adenylyl cyclase is neural specific. Journal of Neurochemistry 60:305-11. [aZX] Xia, Z., Choi, E. J., Wang, F. & Storm, D. R. (1992) The type III calcium/calmodulin-sensitive adenylyl cyclase is not specific to olfactory sensory neurons. Neuroscience Letters 144:169—73. [aZX] Xia, Z., Refsdal, C. D., Merchant, K. M., Dorsa, D. M. & Storm, D. R. (1991) Distinct patterns for the distribution of mRNA for the calmodulinsensitive adenylyl cyclase in rat brain: Expression in areas associated with learning and memory. Neuron 6:431-43. [aZX] Yamagata, K., Goto, K., Kuo, C.-H., Kondo, H. & Miki, N. (1990) Visinin: A novel calcium binding protein expressed in retinal cone cells. Neuron 2:469-76. [aJBH, KT] Yamazaki, A., Hayashi, F., Tatsumi, M., Bitenski, M. W. & George, J. S. (1990) Interaction between the subunits of transducin and cyclic GMP phosphodiesterase in Rana catesbiana rod photoreceptors. Journal of Biological Chemistry 265:11539-48. [aM DB, AY] Yamazaki, A., Yamazaki, M., Tsuboi, S., Kishigami, A., Umbarger, K. O., Hutson, L. D., Madland, W. T. & Hayashi, F. (1993) Regulation of G protein function by an effector in GTP-dependent signal transduction. An inhibitory subunit of cGMP phosphodiesterase inhibits GTP hydrolysis by transducin in vertebrate rod photoreceptors. Journal of Biological Chemistry 268:8899-8907. [rMDB] Yarfitz, S. & Hurley, J. B. (1994) Transduction mechanisms of vertebrate and invertebrate photoreceptors. Journal of Biological Chemistry 269:1432932. [aMDB] Yau, K.-W. (1994) Phototransduction mechanism in retinal rods and cones: The Friedenwald lecture. Investigative Ophthalmology and Visual Science 35:9-32. [aMDB] Yau, K.-W. & Baylor, D. A. (1989) Cyclic GMP-activated conductance of retinal photoreeeptor cells. Annual Review of Neuroscience 12:289327. [arMDB, aRSM, LWH, RLH] Yau, K.-W & Nakatani, K. (1984a) Electrogenic Na-Ca exchange in retinal rod outer segment. Nature 311:661-63. [aRSM] (1984b) Cation selectivity of light-sensitive conductance in retinal rods. Nature 309:352-54. [aRSM] (1985a) Light-suppressible, cyclic GMP-sensitive conductance in the plasma membrane of a truncated rod outer segment. Nature 317:25255. [aRSM] (1985b) Light-induced reduction of cytoplasmic free calcium in retinal rod outer segment. Nature 313:579-82. [aRSM] Yeager, R. E., Heideman, W., Rosenberg, G. B. & Storm, D. R. (1985) Purification of the calmodulin-sensitive sensitive adenylyl cyclase from bovine brain. Biochemistry 24:3776-83. [aZX] Yellen, G., Jurman, M. E., Abramson, T. & MacKinnon, R. (1991) Mutations affecting internal TEA blockage identify the probable pore-forming region of K+ channels. Science 251:939-42. [aRSM] Yool, A. J. & Schwartz, T. L. (1991) Alteration of ionic selectivity of a K+ channel by mutation of the H5 region. Nature 349:700-4. [aRSM] Yoshida, T , Willardson, B. M., Wilkins, J. F., Jensen, G. J., Thornton, B. D. & Bitensky, M. W. (1994) The phosphorylation state of phosducin determines its ability to block transducin subunit interactions and inhibit transducin binding to activated rhodopsin. Journal of Biological Chemistry 269:24050-57. [BMW] Yoshikami, S., George, J. S. & Hagins, W. A. (1980) Light-induced calcium fluxes from outer segment layer of vertebrate retinas. Nature 287:395— 98. [TGW] Yoshimura, M. & Cooper, D. M. (1992) Cloning and expression of a Ca2+ inhibitable adenylyl cyclase from NCB-20 cells. Proceedings of National Academy of Sciences USA 89:6716-20. [aZX] Younger, J. P., McCarthy, S. T. & Owen, W. G. (1992) Modulation of cytosolic free calcium and changes in sensitivity and circulating current occur over References/Controversies the same range of steady-state adapting lights in rod photoreceptors. /nutsfigaiiw; Ophthalmology and Visual Science 33:1104. [aMDB] Yovell, Y. & Abrams, T. W. (1992) Temporal asymmetry in activation of Aplysia adenylyl cyclase may explain properties of conditioning. Proceedings of the National Academy of Sciences USA 89:6526-30. [TWA] Yovell, Y., Knndel, E. R., Dudai, Y. & Abrams, T. W. (1992) A quantitative study of the Ca2+/calmodulin sensitivity of adenylyl cyclase in Aplysia, Drosophila, and rat. Journal of Ncurochemistry 59:1736—44. [aZX] Yurkova, E. V., Demin, V. V. & Abdulaev, N. G. (1990) Crystallization of membrane proteins: Bovine rhodopsin. Biotnedical Science 1:58590. laPAH] Zankel, T , Ok, H., Johnson, R., Chang, C. W., Sekiya, N., Naoki, H., Yoshihara, K. & Nakanishi, K. (1990) Bovine rhodopsin with 11-cis-locked retinal chromophore neither activates rhodopsin kinase nor undergoes conformational change upon irradiation. Journal of the American Chemical Society 112:5387-88. [aPAH] Zhong, Y. & Wu, C. F. (1991) Altered synaptic plasticity in Drosophila in Neuroscience III memory mutants with a defective cyclic AMP cascade. Science 251:198201. [aZX] Zhukovsky, E. A. & Oprian, D. D. (1989) Effect of carboxylic acid side chains on the absorption maximum of visual pigments. Science 245:928-30. [aPAH] Zimmerman, A. L. & Baylor, D. A. (1986) Cyclic GMP-sensitive conductance of retinal rods consists of aqueous pores. Nature 321:70-72. (aRSM] Zimmmerman, A. L., Yamanaka, G., Eckstein, F., Baylor, D. A., & Stryer, L. (1985) Interaction of hydrolysis-resistant analogs of cyclic GMP with the phosphodiesterase and light-sensitive channel of retinal rod outer segments. Proceedings of the National Academy of Sciences USA 82:8813-17. [aRSM] Zong-Yi, L., Jacobson S. G. & Milam S. G. (1994) Autosomal dominant retinitis pigmentosa caused by the threonine-17-methionine rhodopsin mutation: Retinal histopathology and immunochemistry. Experimental Eye Research 58:397-408. [AB] Zozulya, S. & Stryer, L. (1992) Calcium-myristoyl protein switch. Proceedings of the National Academy of Sciences USA 89:11569-73. [a|BH, SK] BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 521 Cambridge - Bringing You the Psychologist's Best Friend The Psychologist's Companion A Guide to Scientific Writing for Students and Researchers Third Edition ^^^^HHB^^^^H Robert J. Sternberg "... [students/ will fin fl this book helpful because it contains good advice...[it] can be read with profit by those in fields other than psychology where research is reported in the same format or in a similar one...lively, informative, and concise. " -Contemporary Psychology "An extremely useful book for undergraduates that provides the basics for writing psychology papers." -Choice Sternberg reviews rules for effective prose in a variety of formats, debunks common misconceptions about writing, highlights commonly misused words, gives instruction on the preparation of tables, figures, and bibliographies, and explains the American Psychological Association guidelines for psychology papers. He has also updated the volume's references. Contents: THE PSYCHOLOGISTS COMPANION EDITION III A guide to scientific writing forstudentsand researchers ROBERT]. STERNBERG 1993 45123-X 45756-4 233 pp. Hardback Paperback $49.95 $14.95 • Eight common misconceptions about psychology papers • Steps in writing the library research paper • Steps in writing the experimental research paper • Rules for writing the psychology paper • Commonly misused words • American Psychological Association guidelines for psychology papers • Guidelines for data presentation • References for the psychology paper • Standards for evaluating the psychology paper • Submitting a paper to a journal • How to win acceptances from psychology journals: Twenty-one tips for better writing • Writing a grant proposal • Finding a publisher • Writing a lecture CAMBRIDGE 40 West 20th Street, New York, N Call toll-free 800-8"2-"423. UNIVERSITY PRESS M;isierC;ird/VISA accepted. Prices s BEHAVIORAL AND BRAIN SCIENCES (1995) 18, 523-599 Printed in the United States of America The sociobiology of sociopathy: An integrated evolutionary model Linda Mealey Department of Psychology, College of St. Benedict, St. Joseph, MN 56374.' Electronic mail: lmealey@psy.ug.edu.au Abstract: Sociopaths are "outstanding" members of society in two senses: politically, they draw our attention because of the inordinate amount of crime they commit, and psychologically, they hold our fascination because most ofus cannot fathom the cold, detached way they repeatedly harm and manipulate others. Proximate explanations from behavior genetics, child development, personality theory, learning theory, and social psychology describe a complex interaction of genetic and physiological risk factors with demographic and micro environmental variables that predispose a portion of the population to chronic antisocial behavior. More recent, evolutionary and game theoretic models have tried to present an ultimate explanation of sociopathy as the expression of a frequency-dependent life strategy which is selected, in dynamic equilibrium, in response to certain varying environmental circumstances. This paper tries to integrate the proximate, developmental models with the ultimate, evolutionary ones, suggesting that two developmentally different etiologies of sociopathy emerge from two different evolutionary mechanisms. Social strategies for minimizing the incidence of sociopathic behavior in modern society should consider the two different etiologies and the factors that contribute to them. Keywords: antisocial personality; criminal behavior; emotion; evolution; facultative strategies; game theory; moral development; psychopathy; sociobiology; sociopathy Sociopaths, who comprise only 3%-4% of the male population and less than 1% of the female population (Davison & Neale 1994; Robins et al. 1991; Strauss & Lahey 1984), are thought to account for approximately 20% of the United States prison population (Hare 1993) and between 33% and 80% of the population of chronic criminal offenders (Hare 1980; Harpending & Sobus 1987; Mednick et al. 1977). Furthermore, whereas the "typical" U.S. burglar is estimated to have committed a median five crimes per year before being apprehended, chronic offenders - those most likely to be sociopaths - report committing upward of 50 crimes per year and sometimes as many as two or three hundred (Blumstein & Cohen 1987). Collectively, these individuals are thought to account for over 50% of all crimes in the U.S. (Hare 1993; Loeber 1982; Mednick et al. 1987). Whether criminal or not, sociopaths typically exhibit what is generally considered to be irresponsible and unreliable behavior; their attributes include egocentrism, an inability to form lasting personal commitments and a marked degree of impulsivity. Underlying a superficial veneer of sociability and charm, sociopaths are characterized by a deficit of the social emotions (love, shame, guilt, empathy, and remorse). On the other hand, they are not intellectually handicapped, and are often able to deceive and manipulate others through elaborate scams and ruses, or by committing crimes that rely on the trust and cooperation of others, such as fraud, bigamy, and embezzlement. The sociopath is "aware of the discrepancy between his behavior and societal expectations, but he seems to be neither guided by the possibility of such a discrepancy, nor disturbed by its occurrence" (Widom © 1995 Cambridge University Press OUO-525X/95$9.OO+.1O 1976a, p. 614). This cold-hearted and selfish approach to human interaction at one time garnered for sociopathy the moniker "moral insanity" (Davison & Neale 1994; McCord 1983). Sociopaths are also sometimes known as psychopaths or antisocial personalities. Unfortunately, the literature reflects varied uses of these three terms (Eysenck 1987; Feldman 1977; Hare 1970; McCord 1983; Wolf 1987). Some authors use one term or another as a categorical label, as in psychiatric diagnosis or in defining distinct personality "types"; an example is the "antisocial personality" disorder described in the Diagnostic and Statistical Manual of the American Psychiatric Association (1987). Other authors use the terms to refer to individuals who exhibit, to a large degree, a set of behaviors or personality attributes that are found in a continuous, normal distribution among the population at large; an example of such usage is "sociopathy" as defined by high scores on all three scales of the Eysenck Personality Questionnaire: extraversion, neuroticism, and psychoticism (Eysenck 1977; 1987). Other authors make a distinction between "simple" and "hostile" (Allen et al. 1971), or "primary" and "secondary" psychopaths or sociopaths (Fagan & Lira 1980). These authors reserve the term "simple" or "primary" for those individuals characterized by a complete lack of the social emotions; individuals who exhibit antisocial behavior in the absence of this emotional deficit are called "hostile" or "secondary" psychopaths or sociopaths, or even "pseudopsychopaths" (McCord 1983). Other authors also make a typological distinction, using the term "psychopath" to refer to antisocial individuals who are of relatively high 523 Mealey: The sociobiology of sociopathy intelligence and middle to upper socioeconomic status and who express their aberrant behavior in impressive and sometimes socially skilled behavior which may or may not be criminal, such as insider trading on the stock market (Bartol 1984). These authors reserve the term "sociopath" for those antisocial persons who have relatively low intelligence and social skills or who come from the lower socioeconomic stratum and express their antisocial nature in the repeated commission of violent crimes or crimes of property. I will begin by using the single term "sociopath" inclusively. However, by the end of the paper I hope to convince the reader that the distinction between primary and secondary sociopaths is an important one because there are two different etiological paths to sociopathy, with differing implications for prevention and treatment. My basic premise is that sociopaths are designed for the successful execution of social deception and that they are the product of evolutionary pressures which, through a complex interaction of environmental and genetic factors, lead some individuals to pursue a life strategy of manipulative and predatory social interactions. On the basis of game theoretic models this strategy is to be expected in the population at relatively low frequencies in a demographic pattern consistent with what we see in contemporary societies. This strategy is also expected to appear preferentially under certain social, environmental, and developmental circumstances which I hope to delineate. In an effort to present an integrated model, I will use a variety of arguments and data from the literature in sociobiology, game theory, behavior genetics, child psychology, personality theory, learning theory, and social psychology. I will argue that: (1) there is a genetic predisposition underlying sociopathy, which is normally distributed in the population; (2) as the result of selection to fill a small, frequency-dependent, evolutionary niche, a small, fixed percentage of individuals - those at the extreme of this continuum - will be deemed "morally insane" in any culture; (3) a variable percentage of individuals who are less extreme on the continuum will sometimes, in response to environmental conditions during their early development, pursue a life strategy that is similar to that of their "morally insane" colleagues; and (4) a subclinical manifestation of this underlying genetic continuum is evident in many of us, becoming apparent only at those times when immediate environmental circumstances make an antisocial strategy more profitable than a prosocial one. 1. The model 1.1. The evolutionary role of emotion As the presenting, almost defining characteristic of sociopaths is their apparent lack of sincere social emotions in the absence of any other deficit such as mental retardation or autism (Hare 1980), it seems appropriate to begin with an examination of some current models of emotion. Plutchik (1980) put forth an evolutionary model of emotion in which he posits eight basic or "primary" emotions (such as fear, anger, and disgust) which predate human evolution and are clearly related to survival.2 524 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 According to the model, everyone (including sociopaths) experiences these primary emotions, which are crosscultural and instinctively programmed. "Secondary" and "tertiary" emotions, on the other hand, are more complex, specifically human, cognitive interpretations of varying combinations and intensities of the primary emotions.3 Because they are partly dependent upon learning and socialization, secondary emotions, unlike primary emotions, can vary across individuals and cultures. Thus, the social emotions (such as shame, guilt, sympathy, and love), which are secondary emotions, can be expected to exhibit greater variability. Griffiths (1990) points out that most of the important features of emotion argue for an evolutionary design: emotions are generally involuntary, and are often "intrusive" (p. 176); they cause rapid, coordinated changes in the skeletal/muscular system, facial expression, vocalization, and the autonomic nervous system; they are to a large extent innate, or at least "prepared" (see Seligman 1970); and they do not seem as responsive to new information about the environment as do beliefs. Griffiths argues that emotional responses to stimuli (he calls them "affectprograms" after Ekman 1971) are informationally encapsulated, complex, organized reflexes, which are "adaptive responses to events that have a particular ecological significance for the organism" (p. 183). That is, they are likely to be highly specialized reflexive responses elicited spontaneously by the presence of certain critical stimuli, regardless of the presence of possible mediating contextual cues or cognitive assessments. Nesse (1990) likewise posits an evolutionary model in which emotions are "specialized modes of operation, shaped by natural selection, to adjust the physiological, psychological, and behavioral parameters of the organism in ways that increase its capacity and tendency to respond to the threats and opportunities characteristic of specific kinds of situations" (p. 268). He attributes a particular role to the social emotions, a role he couches in the language of reciprocity and game theory. Presenting a classic Prisoner's Dilemma matrix, he notes which emotions would be likely to be associated with the outcomes of each of the four cells: when both players cooperate, they experience friendship, love, obligation, or pride; when both cheat or defect, they feel rejection and hatred; when one player cooperates and the other defects, the cooperator feels anger and the defector feels anxiety and guilt. Given that these emotions are experienced after a behavioral choice is made, how could they possibly be adaptive? Nesses explanation is based on the models of Frank (1988) and Hirshleifer (1987), which posit that ex post facto feelings lead to behavioral expressions that are read by others and can be used to judge a person's likely future behavior. To the extent that the phenomenological experience of emotion serves to direct a person's future behavior (positive emotions reinforce the preceding behavior whereas negative emotions punish and, therefore, discourage repetition of the behavior), the outward expression of emotion will serve as a reliable indicator to others of how a person is likely to behave in the future. Indeed, that there exist reliable, uncontrollable outward expressions of these inner experiences at all suggests that the expressions must be serving a communicative function (Dimberg 1988). Mealey: The sociobiology of sociopathy If, however, as in the case of the Prisoner's Dilemma, the most rational strategy is to be selfish and defect, why should the positive (reinforcing) emotions follow mutual cooperation rather than the seemingly more adaptive behavior of defection? Here lies the role of reputation. If a player is known, through direct experience or social reputation, always to play the "rational" move and defect, then in a group where repeated interactions occur and individuals are free to form their own associations, no player will choose to play with the known defector, who will thus no longer be provided the opportunity for any kind of gain - cooperative or exploitative. To avoid this social "shunning" based on reputation4 and hence, to be able to profit at all from long-term social interaction, players must be able to build a reputation for cooperation. To do so, most of them must in fact, reliably cooperate, despite the fact that cooperation is not the "rational" choice for the short-term. Frank (1988) and Hirshleifer (1987) suggest that the social emotions have thus evolved as "commitment devices" (Frank) or "guarantors of threats and promises" (Hirshleifer): they cause positive or negative feelings that act as reinforcers or punishers, molding our behavior in a way that is not economically rational for the short term but profitable and adaptive in situations where encounters are frequent and reputation is important. Frank (1988) presents data from a variety of studies suggesting that people do often behave irrationally (emotionally) in many dyadic and triadic interactions - sometimes even when it is clear that there will be no future opportunity to interact again with the same partner. These studies support the suggestion that in social situations, one's emotional response will often prevail over logic, the reason being that such behavior is, in the long run, adaptive under conditions when one's reputation can follow or precede one. (See also Alexander 1987; Anawalt 1986; Axelrod 1986; Caldwell 1986; Dugatkin 1992; Farrington 1982; Frank et al. 1993; and Irons 1991 for more on the role of reputation.) According to these models, emotion serves both as a motivator of adaptive behavior and as a type of communication: the phenomenological and physiological experience of emotion rewards, punishes, and motivates the individual toward or away from certain types of stimuli and social interactions, whereas the outward manifestations of emotion communicate probable intentions to others. Once such reliable communicative mechanisms have evolved, however, when communication of intent precedes interaction, or when one's reputation precedes one, the conditions of interaction become vulnerable to deception through false signaling or advance deployment of enhanced reputation (e.g., Caldwell 1986). Those who use a deceptive strategy and defect after signaling cooperation are usually referred to as "cheaters," and, as many authors have pointed out (e.g., Alexander 1987; Dennett 1988; Quiatt 1988; Trivers 1971), the presence of cheaters can lead to a coevolutionary "arms race" in which potential cooperators evolve fine-tuned sensitivities to likely evidence or cues of deception, while potential cheaters evolve equally fine-tuned abilities to hide those cues. 5 As long as evolutionary pressures for emotions as reliable communication and commitment devices leading to long-term, cooperative strategies coexist with counterpressures for cheating, deception, and "rational" shortterm selfishness, a mixture of phenotypes will result, such that some sort of statistical equilibrium will be approached. Cheating should thus be expected to be maintained as a low-level, frequency-dependent strategy, in dynamic equilibrium with changes in the environment which exist as counter-pressures against its success. This type of dynamic process has been modelled extensively by evolutionary biologists who use game theory: the topic I turn to next. 1.2. Game theory and evolutionarily stable strategies Game theory was first introduced into the literature of evolutionary biology by Richard Lewontin (1961), who applied it to the analysis of speciation and extinction events. It was later taken up in earnest by John Maynard Smith and colleagues (e.g. Maynard Smith 1978; 1974; Maynard Smith & Price 1973 [See also Maynard Smith: "Game Theory and the Evolution of Behaviour" BBS 7(1) 1984.]) who used it to model contests between individuals. Maynard Smith showed that the "evolutionarily stable strategies" (ESSs) that could emerge in such contests included individuals' use of mixed, as well as fixed, strategies. Alexander (1986) writes: "it would be the worst of all strategies to enter the competition and cooperativeness of social life, in which others are prepared to alter their responses, with only preprogrammed behaviors" (p. 171). The maintenance of mixed ESSs in a population can theoretically be accomplished in at least four ways (after Buss 1991): (1) through genetically based, individual differences in the use of single strategies (such that each individual, in direct relation to genotype, consistently uses the same strategy in every situation); (2) through statistical use by all individuals of a species-wide, genetically fixed, optimum mix of strategies (whereby every individual uses the same statistical mix of strategies, but does so randomly and unpredictably in relation to the situation); (3) through species-wide use by all individuals of a mix of environmentally contingent strategies (such that every individual uses every strategy, but predictably uses each according to circumstances); (4) through the developmentally contingent use of single strategies by individuals (such that each individual has an initial potential to use every type of strategy but, after exposure to certain environmental stimuli in the course of development, is phenotypically canalized from that point on to use only a fraction of the possible strategies). To Buss's fourth mechanism can be added a differential effect of genotype on developmental response to the environment, thus adding another mechanism: (5) genetically based individual differences in response to the environment, resulting in differential use by individuals of environmentally-contingent strategies (such that individuals of differing genotypes respond differently to environmental stimuli in the course of development and are thus canalized to produce a different set of limited strategies given the same, later conditions). Following the leads of Cohen & Machalek (1988); Harpending & Sobus (1987); Kenrick et al. (1983); Kofoed & MacMillan (1986); and MacMillan & Kofoed (1984), I BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 525 Mealey: The sociobiology of sociopathy would like to suggest an evolutionary model in which sociopaths are a type of cheater-defector in our society of mixed-strategy interactionists. I will be arguing that sociopathy appears in two forms, according to mechanisms 1 and 5 (above): one version that is the outcome of frequency-dependent, genetically based individual differences in use of a single (antisocial) strategy (which I will refer to as "primary sociopathy") and another that is the outcome of individual differences in developmental response to the environment, resulting in the differential use of cooperative or deceptive social strategies (which I will refer to as "secondary sociopathy"). To support this model, I will provide evidence that there are predictable differences in the use of cheating strategies across individuals, across environments, and within individuals across environments; this evidence will integrate findings from the fields of behavior genetics, child psychology, personality theory, learning theory, and social psychology. 2. The evidence 2.1. Behavior genetics For decades, evidence has been accumulating that both criminality and sociopathy have a substantial heritable component, and that this heritable component is to a large extent overlapping; that is, the heritable attributes that contribute to criminal behavior seem to be the same as those which contribute to sociopathy. Although there is no one-to-one correspondence between those individuals identified as criminals and those identified as sociopaths, (indeed, the definitions of both vary from study to study), it is clear that these two sets of individuals share a variety of characteristics and that a subset of individuals share both labels (Ellis 1990b; Gottesman & Goldsmith 1993; Moffitt 1987; Robins et al. 1991). 2.1.1. Studies of criminal behavior. The behavior-genetic literature on criminal behavior suggests a substantial effect of heredity across several cultures.6 Christiansen (1977a; 1977b), Cloninger & Gottesman (1987), Eysenck & Gudjonsson (1989), Raine (1993) and Wilson & Herrnstein (1985) review studies of twins which, taken as a whole, suggest a heritability of approximately 0.60 for repeated commission of crimes of property. (Heritability is a measure of the proportion of variance of a trait, within a population, that can be explained by genetic variability within that population; it thus ranges theoretically from 0.00 to 1.00, with the remaining population variance explained by variance in individuals' environment.) Adoption studies (reviewed in Cloninger & Gottesman 1987; Eysenck & Gudjonsson 1989; Hutchings & Mednick 1977; Mednick & Finello 1983; Mednick et al. 1987; Raine 1993; and Wilson & Herrnstein 1985) arrive at a similar conclusion.7 Several adoption studies were also able to demonstrate significant interactive effects not discriminable using the twin methodology. Cadoret et al. (1983); Crowe (1972; 1974); Mednick & Finello (1983); and Mednick et al. (1984) report significant gene-environment interactions, such that adoptive children with both a genetic risk (criminal biological parent) and an environmental risk (criminality, psychiatric illness, or other severe behav- 526 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 ioral disturbance in an adoptive parent), have a far greater risk of expressing criminal behavior than do adoptees with no such risk or only one risk factor, and that increased risk is more than simply an additive effect of both risk factors. In addition, Baker et al. (1989) report an interaction based on sex, in which females are more likely to transmit a genetic risk to their offspring than are males. 2.1.2. Studies of sociopathy. The literature on sociopathy suggests a pattern similar to that on criminality: Cadoret (1978); Cadoret & Cain (1980); Cadoret & Stewart (1991); Cadoret et al. (1987); Crowe (1974); and Schulsinger (1972) demonstrate a substantial heritability to sociopathy; Cadoret et al. (1990) found a gene-environment interaction similar to the one found for criminal behavior; and Cadoret & Cain (1980; 1981) found an interaction involving sex, such that male adoptees were more sensitive to the influence of environmental risk factors than were female adoptees. The similarity of the patterns described in these two domains is to some extent due to the fact that the diagnosis of sociopathy is often based in part upon the existence of criminal activity in a subject's life history. On the other hand, consider the following facts: (1) criminal behavior and other aspects of sociopathy are correlated (Cadoret et al. 1983; Cloninger & Gottesman 1987; Eysenck 1977; Morrison & Stewart 1971; Patterson et al. 1989; Wolf 1987); (2) criminal activity is found with increased frequency among the adopted-away children of sociopaths (Moffitt 1987); (3) sociopathy is found with increased frequency among the adopted-away children of criminals (Cadoret & Stewart 1991; Cadoret et al. 1990). This all suggests that criminality and sociopathy may share some common heritable factors. For this reason, early researchers and clinicians (e.g. Cadoret 1978; Schulsinger 1972) suggested using the term "antisocial spectrum" to incorporate a variety of phenotypes considered likely to be manifestations of closely related genotypes.8 The existence of this spectrum suggests a multifactorial, probably polygenic, basis for sociopathy and its related phenotypes. Using an analogy to "g," [See Jensen: "The Nature of the Black-White Difference on Various Psychometric Tests" BBS 8(2) 1985.] which is often used to refer to the common factor underlying the positive correlations between various aptitude measures, Rowe (1986) and Rowe & Rodgers (1989) use "d" to refer to the common factor underlying the various expressions of social deviance. 2.1.3. Sex differences and the "two threshold" model. Cloninger put forth a "two threshold" polygenic model to account for both the sex difference in sociopathy and its spectral nature (Cloninger etal. 1975; 1978). According to the model, sociopaths are individuals on the extreme end of a normal distribution whose genetic component is (1) polygenic and (2) to a large degree, sex limited. (Sexlimited genes, not to be confused with sex-linked genes, are those which are located on the autosomes of both sexes but are triggered into expression only within the chemical/hormonal microenvironment of one sex or the other. Common examples include beard and mustache growth in men, and breast and hip development in women.) If a large number of the many genes underlying sociopathy are triggered by testosterone or some other androgen, Mealey: The sociobiology of sociopathy many more men than women will pass the threshold of the required number of active genes necessary for its outward expression. According to the two-threshold model, those females who do express the trait must have a greater overall "dose" or "genetic load" (i.e., they are further out in the extreme of the normal distribution of genotypes) than most of the males who express the trait. This proposition has been supported by data showing that, in addition to the greater overall risk for males as opposed to females, there is also a greater risk for the offspring (and other relatives) of female sociopaths as compared to the offspring (and other relatives) of male sociopaths. This phenomenon cannot be accounted for either by sex linkage or by the differential experiences of the sexes. Besides providing a proximate explanation for the greater incidence of male sociopathy and crime, the twothreshold model also explains on a proximate level the finding that males are more susceptible to environmental influences than females. Somewhat paradoxically, although a male will express sociopathy at a lower "genetic dose" than is required for expression in a female, the heritability of the trait is greater for females, meaning that the environmental component of the variance is greater for males.9 The two-threshold model thus explains in a proximate sense what sociobiologists would predict from a more ultimate perspective. The fact that males are more susceptible than females to the environmental conditions of their early years fits well with sociobiological theory in that the greater variance in male reproductive capacity makes their "choice" of life strategy somewhat more risky and therefore more subject to selective pressures (Buss 1988; Mealey & Segal 1993; Symons 1979). Sociobiological reasoning thus leads to the postulate that males should be more sensitive to environmental cues that (1) trigger environmentally contingent or developmentally canalized life history strategies or (2) are stimuli for which genetically based individual differences in response thresholds have evolved. (Recall mechanisms 3, 4, and 5 for the maintenance of mixed strategy ESSs in a population). If the evolutionary models apply, then when, specifically, would sociopathy be the best available strategy? What would be the environmental cues that, especially for boys, would trigger its development? To answer these questions, I turn to the child psychology literature, with a special focus on studies of life history strategies, delinquency, and moral development. 2.2. Child psychology 2.2.1. Life history strategies. Beginning with Draper and Harpending's now-classic 1982 paper on the relationship between father absence and reproductive strategy in adolescents, there has been an increasing effort to view development as the unfolding of a particular life strategy in response to evolutionarily relevant environmental cues (Belsky et al. 1991; Crawford & Anderson 1989; Draper & Belsky 1990; Draper & Harpending 1982; Gangestad & Simpson 1990; MacDonald 1988; Mealey 1990; Mealey 6 Segal 1993; Moffitt et al. 1992; Surbey 1987). These models are based either implicitly or explicitly on the assumption that there are multiple evolutionarily adap- tive strategies and that the optimal strategy for particular individuals will depend both upon their genotype and their local environment. To date, most developmental life history models address variance in reproductive strategies (for example: age at menarche or first sexual activity, number of mating partners, and amount of parental investment), but this type of modeling can also be applied to the adoption of social strategies such as cheating versus cooperation. Perhaps the most oft-mentioned factor suggested as being relevant to the development of a cheating strategy, especially in males, is being competitively disadvantaged with respect to the ability to obtain resources and mating opportunities. Theoretically, those individuals who are the least likely to outcompete other males in a status hierarchy, or to acquire mates through female choice, are the ones most likely to adopt a cheating strategy (see, e.g., Daly & Wilson 1983; Gould & Gould 1989; Thornhill & Alcock 1983, regarding nonhuman animals, and Cohen & Machalek 1988; Kenrick et al. 1983; Kofoed & MacMillan 1986; MacMillan & Kofoed 1984; Symons 1979; Thornhill & Thornhill 1992; Tooke & Camire 1991, regarding humans). In humans, competitive disadvantage could be related to a variety of factors, including age, health, physical attractiveness, intelligence, socioeconomic status, and social skills. Criminal behavior, one kind of cheating strategy, is clearly related to these factors. Of the seven cross-cultural correlates of crime reported by Ellis (1988), three seem directly related to resource competition - large number of siblings, low socio economic status, and urban residency. The four others - youth, maleness, being of black racial heritage, and coming from a single-parent (or otherwise disrupted) family background - can be plausibly argued to be related to competition as well (see, e.g., Cohen & Machalek 1988; Ellis 1988; Kenrick et al. 1983; Wilson & Herrnstein 1985). Empirical data suggest that deficits in competitive ability due to psychosis (Hodgins 1992), intellectual handicap (Hodgins 1992; Moffitt & Silva 1988; Quay 1990a; Stattin & Magnusson 1991), or poor social skills (Dishion et al. 1991; Garmezy 1991; Hogan & Jones 1983; Simonian et al. 1991) are also associated with criminal behavior. Likewise, the competitive advantages conferred by high intelligence (Hirschi & Hindenlang 1977; Kandel et al. 1988; Silverton 1988; White et al. 1989; Wilson & Herrnstein 1985) or consistent external support (Garmezy 1991), can mitigate the development of criminal or delinquent behavior in those who are otherwise at high risk. Rape and spouse abuse, other forms of cheating strategy, appear to be related to the same life-history factors as crime (Ellis 1989; 1991a; Malamuth et al. 1991; Thornhill & Thornhill 1992). In fact, Huesmann et al. (1984), Rowe and Rodgers (1989), and Rowe et al. (1989) present evidence that there is a common genetic component to the expression of sexual and nonsexual antisocial behavior. Given the overlaps between rape, battering, and criminality in terms of life history circumstances, genetics, and apparent inability to empathize with one's victim, it would be parsimonious to postulate that they might be expressions of a single sociopathy spectrum. As such, these antisocial behaviors could be considered to be genetically influenced, developmentally and environmentally contingent cheating strategies, utilized when a BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 527 Mealey: The sociobiology of sociopathy male finds himself at a competitive disadvantage (see also Figueredo & McCloskey, unpublished manuscript). Along these lines, MacMillan and Kofoed (1984) presented a model of male sociopathy based on the premise that sexual opportunism and manipulation are the key features driving both the individual sociopath and the evolution of sociopathy. Harpending and Sobus (1987) posited a similar basis for the evolution and behavioral manifestations of Briquet's hysteria in women, suggesting that this syndrome of promiscuity, fatalistic dependency, and attention-getting, is the female analogue, and homologue, of male sociopathy. 2.2.2. Delinquency. Childhood delinquency is a common precursor of adolescent delinquency and adult criminal and sociopathic behavior (Loeber 1982; Loeber & Dishion 1983; Loeber & Stouthamer-Loeber 1987; Patterson et al. 1989; Robins & Wish 1977); in fact, childhood conduct disorder is a prerequisite finding in order to diagnose adult antisocial personality (American Psychiatric Association 1987). As in the literature on adults, an important distinction is frequently made between two subtypes of conduct disorder in children: Lytton (1990), for example, distinguishes between "solitary aggressive type" and "group type"; Loeber (1990) distinguishes between "versatiles" and "property offenders"; and Strauss & Lahey (1984) distinguish between "unsocialized' and "socialized." I will argue that these subtypes are precursors of two types of adulthood antisociality (with "solitary aggressive," "versatile," or "unsocialized" types leading to primary sociopathy and "group," "property offender," or "socialized" types presaging secondary sociopathy). I will also argue that the differing life history patterns of these two types of delinquents are reflections of two different evolutionary mechanisms for maintaining ESSs in a population: mechanism 1 and mechanism 5, respectively (see sect. 1.2). Although more than half of juvenile delinquents outgrow their behavior (Gottesman & Goldsmith 1993; Lytton 1990; Robins et al. 1991), the frequency of juvenile antisocial behaviors is still the best predictor of adult antisocial behavior, and the earlier such behavior appears, the more likely it is to be persistent (Farrington 1986; Loeber & Stouthamer-Loeber 1987; Lytton 1990; Robins et al. 1991; Stattin & Magnusson 1991; White et al. 1990). The mean age at which adult sociopaths exhibited their first significant symptom is between 8 and 9 years; 80% of all sociopaths exhibited their first symptom by age 11 (Robins et al. 1991). Over two-thirds of eventual chronic offenders are already distinguishable from other children by kindergarten (Loeber & Stouthamer-Loeber 1987). Thus, by evaluating the environments of juvenile delinquents, we can fairly reliably reconstruct the childhood environments of adult sociopaths. Studies of this sort consistently implicate several relevant environmental factors correlated with boyhood antisocial behavior: inconsistent discipline, parental use of punishment as opposed to rewards, disrupted family life (especially father absence, family violence, alcoholic parent, or mentally ill parent), and low socioeconomic status (Cadoret 1982; Farrington 1986; 1989; Loeber & Dishion 1983; Lytton 1990; McCord 1986; Offord et al. 1991; Patterson et al. 1989; Silverton 1988; van Dusen et al. 1983; Wilson & Herrnstein 1985). Aside from the fact that 528 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 all of these variables are more likely to exist when one or the other parent is sociopathic, and the child, hence, genetically predisposed to sociopathy, behaviorist and social learning models of the dynamics of early parentchild interactions (to be described in sect. 2.4.2) have been fairly convincing in explaining how antisocial behaviors can be reinforced under such living conditions. In line with the postulate that cheating strategies would most likely be used by individuals who are at a competitive disadvantage, Hartup (1989), Hogan & Jones (1983), Kandel et al. (1988), Loeber & Dishion (1983), Magid & McKelvie (1987), McGarvey et al. (1981), and Patterson et al. (1989) suggest that the common way in which high risk familial and environmental factors contribute to delinquency is by handicapping children with respect to their peers in terms of social skills, academic ability, and selfesteem. This noncompetitiveness then leads disadvantaged youths to seek alternative peer groups and social environments in which they can effectively compete (Dishion et al. 1991). If they are successful in the estimation of their new peer group, adopting this strategy may lead to "local prestige" (Rowe, personal communication) sufficient to commandeer resources, deter rivals, or gain sexual opportunities within the new referent group (see also Moffitt 1993). In other words, competitively disadvantaged youth may be trying to "make the best of a bad job" (Cohen & Machalek 1988, p. 495; Dawkins 1980, p. 344), by seeking a social environment in which they may be less handicapped or even superior. The correlates of delinquency in girls are essentially the same as those for boys, although delinquency is less common in girls (Lytton 1990; Robins 1986; White et al. 1990). Caspi et al. (1993) found that delinquency in girls, as in boys, is arrived at via two different developmental trajectories. One pattern includes a history of antisocial behavior throughout childhood and a tendency to seek out delinquent peers; based on previous research (White et al. 1990), this life history trajectory is thought to lead to persistent antisocial behavior in adulthood. The second pattern is exhibited by girls who have few behavior problems in childhood, but who, upon reaching menarche, exhibit more and more frequent antisocial behaviors. The antisocial behavior of girls who show this latter pattern is thought to be more a product of environmental influence than that of girls who follow the first trajectory, as this pattern is selectively exhibited by girls who (a) have an early age of menarche and (b) are in coeducational school settings. These girls, upon reaching early sexual maturity, start associating with older male peers and exhibiting some of the antisocial behaviors that are more often displayed by older boys than by their younger female peers (see Maccoby 1986); girls who follow this trajectory are expected to "outgrow" their antisocial activities. Although the two subsets of delinquent girls would be difficult to differentiate using a cross-sectional methodology, in accordance with the model presented here, Caspi et al. (1993) consider their differing developmental histories to be of theoretical importance for longitudinal studies and of practical importance for early intervention (see Moffitt 1993 for a similar scenario regarding boys.) 2.2.3. Moral development. Like the tendency to engage in antisocial behavior, an individual's tendency to engage in prosocial behavior seems to be fairly stable from an early Mealey: The sociobiology of sociopathy age (Rushton 1982). Yet the development of individual differences in behavior has not been as well studied as the presumably universal stages of cognition that underlie changes in moral reasoning. Kohlberg's (1964) stage model of moral development, for example, ties advances in moral thinking to advances in reasoning ability and attributes individual differences largely to differences in cognitive ability. Although it is clear that both moral reasoning and moral behavior covary with age (Rushton 1982) and may do so in a manner consistent with some evolutionists' thinking (e.g., Alexander 1987), cognitive models alone cannot explain the absence of moral behavior in sociopaths, who are not intellectually handicapped with respect to the normal population. Other developmental models posit the emergence of empathy and the other social emotions as prerequisites for moral behavior (see Zahn-Waxler & Kochanska 1988 for a review). Even very young children, it seems, are in a sense biologically prepared to learn moral behavior, in that they are selectively attentive to emotions - especially distress - in others (Hoffman 1978; Radke-Yarrow & Zahn-Waxler 1986; Zahn-Waxler & Radke-Yarrow 1982). Hoffman (1975; 1977; 1982), for example, suggests that the observation of distress in others triggers an innate "empathic distress" response in the child, even before the child has the cognitive capacity to differentiate "other" from "self." Accordingly, any instrumental behavior that serves to reduce the distress of the other also serves to relieve the vicarious distress of the child. Thus, very young children might learn to exhibit prosocial behavior long before they are able to conceptualize its effect on others. In Hoffman's model, the motivation behind early prosocial behavior is the (egocentric) need to reduce one's own aversive feelings of arousal and distress. As the child ages, the range of cues and stimuli that can trigger the vicarious distress increase through both classical and operant conditioning. Eventually, when the child develops the cognitive ability to "role play," or take on another's perspective, empathic distress turns to "sympathetic distress," which motivates the prosocial behavior that is more likely to be interpreted as intentional, altruistic, and moral. Hoffman's model of prosocial behavior dovetails nicely with Hirshleifer's (1987) "Guarantor" and Frank's (1988) "Commitment" model of emotion (see sect. 1.1): the reduction in anxiety that follows cooperative or prosocial behavior reinforces such behavior, whereas the increase in anxiety which, through stimulus generalization, follows acts or thoughts of antisocial behavior will punish and therefore reduce those acts and thoughts. Dienstbier (1984) reported an interesting series of studies testing the role of anxiety and emotional arousal on cheating. As expected, high arousal levels were associated with low cheating levels (and vice versa), but the subjects' attribution of the cause of high arousal was also important. When subjects were able to attribute their arousal to a cause other than the temptation to cheat, they found it much easier to cheat than when they had no other explanation for their arousal level. Subjects were also less willing to work to avoid punishment when they were able to attribute their arousal to an external cause rather than to an internal source of anxiety associated with the threat of punishment. Dienstbier concluded that when a situa- tion is perceived to be "detection-free," one's temptation to cheat is either resisted or not, depending on the levels of anxiety perceived to be associated with the temptation. The ability to act intentionally in either a prosocial or antisocial manner (or in the terms of game theory, cooperatively or deceptively), depends upon having reached a certain level of cognitive development at which it is possible to distinguish emotions of the self from emotions of others; that is, the child must pass from empathic responses to sympathetic responses (Dunn 1987; Hoffman 1975; 1977; 1984; Mitchell 1986; Vasek 1986). This transition begins to occur some time during the second year (Dunn 1987; 1988; Dunn et al. 1991; Hoffman 1975; 1982; Leslie 1987) when the child is beginning to develop what has come to be called a "theory of mind" (Premack & Woodruff 1978). [See also Gopnik: "How We Know Our Minds" BBS 16(1) 1993 and Tomasello et al.: "Cultural Learning" BBS 16(3) 1993.] Having a theory of mind allows one to impute mental states (thoughts, perceptions, and feelings) not only to oneself, but also to other individuals. Humphrey (1976; 1983) suggests that this kind of awareness evolved in humans because it was a successful tool for predicting the behavior of others. Humphrey claims that the best strategists in the human social game would be those who could use a theory of mind to empathize accurately with others and thereby be able to predict the most adaptive strategy in a social interaction (Byrne & Whiten 1988 call this aptitude "Machiavellian intelligence.") [See also Whiten & Byrne: "Tactical Deception in Primates" BBS 11(2) 1988.] Humphrey's model is something of a cognitive equivalent of the evolutionary models of emotion discussed in section 1.1; they can probably be considered complementary and mutually reinforcing. With regard to sociopathy, the question is whether a strategist can be successful using only the cognitive tool of a theory of mind, without access to emotional, empathic information which, presumably, sociopaths lack (Mealey 1992). In the next section I will argue that this is exactly what a sociopath does. 2.3. Personality theory The models of normative moral development presented above are helpful but clearly insufficient to explain sociopathy. Although some adoption studies and most longitudinal studies report significant effects of social and environmental risk factors on delinquency and criminality, the magnitude of that risk as a simple main effect is rather small. Despite repeated exposure to inconsistent and confusing reinforcement and punishment, most children who grow up with these risk factors do not turn out to be sociopathic, whereas some children who do not experience such risk factors do. Studies have repeatedly shown that the effect of the environment is much more powerful for children at biological risk than for others. What is it that makes high risk environmental features particularly salient for those individuals who have a certain predisposing genotype? 2.3.1. The role of gene-environment interactions. Stimulated by the work of Rowe and Plomin (Dunn & Plomin 1990; Plomin & Daniels 1987; Rowe 1983a; 1983b; 1990a; 1990b; Rowe & Plomin 1981), evidence is accumulating BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 529 Mealey: The sociobiology of sociopathy that, unlike what has been traditionally assumed, the most important environmental features and events that influence personal development are not those that are shared by siblings within a family (such as parenting style, socioeconomic status, and schooling), but rather are idiosyncratic events and relationships which are difficult to study systematically with traditional methods. Despite a shared home, individual children will encounter different microenvironments: their individual relationships with their parents will differ, and their experiences on a day to day, minute by minute basis will not overlap significantly. In addition, there will be some environmental differences which are due to genetic differences; children with different personalities, aptitudes, and body types will not only seek out different experiences (Caspi et al. 1987; Rowe 1990a; Scarr & McCartney 1983), but will also attribute different phenomenological interpretations to the same experiences (Dunn 1992; Rowe 1983a; 1990b). For these reasons, any two children will experience an (objectively) identical environment in different ways; there is, in some sense, no real validity to some of the operational measures we currently use to describe a child's environment (Plomin & Daniels 1987; Rowe & Plomin 1981; Wachs 1992). Although this may sound discouraging for those who seek to apply psychological research to the prevention of crime and delinquency - and most such efforts have, in fact, been fairly unsuccessful (Borowiak 1989; Feldman 1977; Gottschalk et al. 1987; Patterson et al. 1989) - there are reasons for optimism. Palmer (1983) suggests that the "nothing works" conclusion is valid only in the sense that no single intervention technique will be successful across the board, and that targeting different strategies to different individuals should prove more successful. More and more studies are suggesting that there are at least two developmental pathways to delinquency and sociopathy and that we need to address them separately (Caspi et al. 1993; Dishion & Poe 1993; Lytton 1990; McCord 1993; Moffitt 1993; Patterson 1993; Quay 1990b; Simons 1993; White et al. 1990). The evolutionary model presented here makes specific predictions about the likely differential success of various intervention and prevention strategies for individuals arriving at their antisocial behavior via different paths: although individuals of dissimilar genotypes may end up with similar phenotypes, different environmental elements and experiences may be particularly salient for them. (This is a corollary of mechanism 5 for the maintenance of ESSs presented in sect. 1.2.) As I will argue below, primary and secondary sociopathy seem to provide an excellent illustration of the development of similar phenotypes from different genotype-environment interactions. To the extent that we understand it now, primary sociopaths come from one extreme of a polygenic genetic distribution and seem to have a genotype that disposes them "to acquire and be reinforced for displaying antisociality" (Rowe 1990a, p. 122). That genotype results in a certain inborn temperament or personality coupled with a particular pattern of autonomic arousal which, together, seem to design the individual (1) to be selectively unresponsive to those environmental cues necessary for normal socialization and moral development and (2) to actively seek the more deviant and arousing stimuli within the environment. Secondary sociopaths, on the other hand, are not as 530 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 genetically predisposed to their behavior; rather they are more responsive to environmental cues and risk factors, becoming sociopathic "phenocopies" (after Raine 1993) or "mimics" (after Moffitt 1993) when the carrying capacity of the "cheater" niche grows. What are the predisposing constitutional factors that place some individuals at high risk? 2.3.2. The role of temperament. In a twin study, Rushton et al. (1986) found evidence of substantial heritability of self-reported measures of altruism, nurturance, aggressiveness, and empathy. Across twin pairs, altruism, nurturance, and empathy increased with age, whereas aggressiveness decreased; sex differences (in the expected direction) were found for nurturance, empathy, and aggression; for all measures, the environmental contributions were determined to be individual rather than familial. Methodological considerations do not allow full confidence in the numerical heritability estimates of this study, but Eisenberg et al. (1990) conclude that it reports true individual differences that are likely to be a result of genetic differences in temperament, specifically sociability and emotionality. More recently, two additional twin studies have confirmed the findings of Rushton et al. Emde et al. (1992) reported significant heritabilities for empathy, behavioral inhibition, and expressions of negative affect, while Ghodsian-Carpey & Baker (1987) found significant heritabilities on four measures of aggressiveness in children. Like the Rushton et al. study, both of these studies also reported sex differences, and both confirmed the relative importance of nonshared, as opposed to shared, environmental influences. A fourth twin study (Rowe 1986) used a different set of personality indices but went a step further in establishing the link between temperament and antisocial behavior. Rowe's analysis suggests that, especially for males, the inherited factors correlated with one's genetic risk of delinquency are the same as those that lead to the temperamental attributes of anger, impulsivity, and deceitfulness ("self-serving dishonesty with people with whom a person ordinarily has affectional bonds," p. 528). It is interesting to note that although Rowe found that common genetic factors related temperament and delinquency, environmental factors related academic nonachievement with delinquency. These findings provide evidence for the two-pathway model presented in section 1.2, in that such a geneenvironment interaction (1) would create at least two possible routes to sociopathy or criminality, one primarily heritable and one less so; and (2) in terms of the latter, less heritable pathway, would set the stage for developmentally and environmentally contingent individual differences in antisocial behavior. In addition, in line with previously mentioned studies and the proposed model, the environmental factors Rowe found to be statistically significant varied within families and were more significant for males than for females. Most of the research into the relationship between temperament, personality and sociopathy has been based on the extensive work of Hans Eysenck (summarized in Eysenck 1977 and 1983, Eysenck & Gudjonsson 1989, and Zuckerman 1989). Eysenck first postulated and then convincingly documented that sociopathy in particular Mealey: The sociobiology of sociopathy and antisocial behavior in general are correlated with high scores on all three of the major personality dimensions of the Eysenck Personality Questionnaire: "extraversion" (versus introversion), "neuroticism" (versus emotional stability), and "psychoticism" (versus fluid and efficient superego functioning - not synonymous with psychotic mental illness; Zuckerman (1989) suggests that this scale would be better called "psychopathy"). All three of these dimensions exhibit substantial heritability, and since psychoticism is typically much higher in males than females, it is a likely candidate for one of the relevant sex-limited traits that fits Cloninger's two-threshold risk model explaining the sex difference in expression of sociopathy. In trying to explain the proximate connections between temperament, delinquency, sociopathy, and criminal behavior, Eysenck and his colleagues devised the "General Arousal Theory of Criminality" (summarized in Eysenck & Gudjonsson 1989), according to which the common biological condition underlying all of these behavioral predispositions is the inheritance of a nervous system that is relatively insensitive to low levels of stimulation. Individuals with such a physiotype, it is argued, will be extraverted, impulsive, and sensation seeking, because under conditions of relatively low stimulation they find themselves at a suboptimal level of arousal; to increase their arousal, many will participate in high-risk activities such as crime (see also Farley 1986 and Gove & Wilmoth 1990). In general support of this model, Ellis (1987) performed a meta-analysis which found that both criminality and sociopathy were associated with a variety of indicators of suboptimal arousal, including childhood hyperactivity, recreational drug use, risk taking, failure to persist on tasks, and preference for wide-ranging sexual activity. Additional confirmation of the arousal model comes from Zuckerman, who found a similar pattern of behaviors associated with his measure of sensation seeking. (The following summary is derived from Daitzman & Zuckerman 1980; Zuckerman 1979; 1983; 1984; 1985; 1990; 1991; and Zuckerman et al. 1980). In addition to seeking thrill and novelty, sensation seekers describe "a hedonistic pursuit of pleasure through extraverted activities including social drinking, parties, sex, and gambling," "an aversion to routine activities or work and to dull and boring people," and "a restlessness in an unchanging environment" (Zuckerman et al. 1980, p. 189). In college students, sensation seeking is correlated with the Pd (Psychopathic Deviate) scale of the Minnesota Multiphasic Personality Inventory; among prisoners it can be used to distinguish primary psychopaths from secondary psychopaths and nonpsychopathic criminals (see also Fagan & Lira 1980). Zuckerman also shows that sensation seeking as a temperament appears at an early age (3-4 years), exhibits a high degree of heritability, correlates negatively with age in adults, and exhibits sex differences, with higher scores more often in males. Because it shows a relationship with both sex and age, sensation seeking (and its presumed underlying hypoarousal) may also be a good candidate for a trait that can explain the distribution and expression of sociopathy (see also Baldwin 1990). Gray (1982; 1987), and Cloninger (Cloninger 1987a; Cloninger et al. 1993) have proposed updated versions of the Eysenck model in which the three personality factors are rotated and renamed to more clearly correspond to known neural circuitry. Gray names the three systems: the approach or behavioral activation system, the behavioral inhibition system, and the fight/flight system; Cloninger names them "novelty-seeking," "harmavoidance," and "reward-dependence." The three factors explain the same variance in personality as Eysenck's original factors and have been shown to be independent and highly heritable (Cloninger 1987a). In addition to mapping more closely to known neural systems, these three factors are also proposed to correspond to differential activity of three neurochemicals: dopamine for behavioral activation (or novelty seeking), serotonin for behavioral inhibition (or harm avoidance), and norepinephrine for fight/flight (or reward dependence); see Charney et al. 1990; Cloninger 1987a; Depue & Spoont 1986; Eysenck 1990; and Raine 1993 for partial reviews. 2.3.3. The role of physiology. Using Cloninger's terminology, sociopaths are individuals who are high on novelty seeking, low on harm avoidance, and low on reward dependence. Thus, we should expect them to be high on measures of dopamine activity, low on measures of serotonin activity, and low on measures of norepinephrine activity; data suggest that they are. Zuckerman (1989) reports that sensation seeking is negatively correlated with levels of dopamine-betahydroxylase (DBH), the enzyme that breaks down dopamine, and that extremely low levels of DBH are associated with undersocialized conduct disorder and psychopathy. With respect to the two pathway model, boys with socialized conduct disorder (those with fewer, later-appearing symptoms and who are posited to be at risk for secondary, as opposed to primary, sociopathy) had high levels of DBH. In addition, extraverts and delinquents are reported to have lower than average levels of adrenaline (epinephrine) and norepinephrine under baseline circumstances; Magnusson (1985, as cited by Zuckerman 1989) reports that urinary epinephrine measures of boys at age 13 significantly predicted criminality at ages 18-25. High sensation seekers, criminals, and other individuals scoring high on measures of impulsivity and aggression also have significantly lower levels than others of the serotonin metabolite, 5-HIAA (Brown et al. 1982; Brown et al. 1979; Depue & Spoont 1986; Kruesi et al. 1992; Muhlbauer 1985; Raine 1993; Zuckerman 1989; 1990). These are not small effects: Raine (1993) reports an average effect size (the difference between groups divided by the standard deviation) for serotonin of 0.75, and for norepinephrine of 0.41; Brown et al. (1979) reported that 80% of the variance in aggression scores of their sample was explained by levels of 5-HIAA alone; Kruesi et al. reported that knowing 5-HIAA levels increased the explained variance of aggression at a two yearfollow up from 65% (using clinical measures only) to 91% (clinical measures plus 5-HIAA measures). Levels of monoamine oxidase (MAO) - an enzyme that breaks down the neurotransmitters serotonin, dopamine, epinephrine, and norepinephrine - are also low in antisocial and sensation-seeking individuals (Ellis 1991b; Zuckerman 1989). Individual differences in platelet MAO appear shortly after birth and are stable (Raine 1993; Zuckerman 1989; 1990); Zuckerman reports an estimated BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 531 Mealey: The sociobiology of sociopathy heritability of 0.86. Recently, a mutant version of the gene coding for MAO-A, the version of MAO specific to serotonin, has been identified in an extended family in which the males show a history of repeated, unexplained outbursts of aggressive behavior (Brunner et al. 1993; Morell 1993); urinalysis indicated that the MAO-A is not functioning normally in the affected men. Results of psychophysiological studies also report significant differences between sociopaths and others. Reviews of this literature can be found in Eysenck & Gudjonsson (1989), Mednick et al. (1987), Raine (1989), Raine (1993), Raine & Dunkin (1990), Trasler (1987), and Zuckerman (1990). Among the findings are: high sensation seekers and sociopaths are more likely than lows and normals to show orienting responses to novel stimuli of moderate intensity, whereas lows and normals are more likely to show defensive or startle responses; criminals and delinquents tend to exhibit a slower alpha (resting) frequency in their electroencephalogram (EEG) than age-matched controls; high sensation seekers and delinquents differ from lows and nondelinquents in the amplitude and shape of cortical evoked potentials; extraverts and sociopaths show less physiological arousal than introverts and normals in response to threats of pain or punishment and more tolerance of actual pain or punishment; and delinquents (though not necessarily adult criminals) tend to have a lower baseline heart rate than nondelinquents. The importance of the role of these psychophysiological factors as significant causes, not just correlates, of sociopathy is strengthened by evidence that (a) these measures of autonomic reactivity are just as heritable as the temperament with which they are associated (Gabbay 1992; Zuckerman 1989), and that (b) the same physiological variables that differentiate identified sociopaths, delinquents, and criminals from others can also significantly predict later levels of antisocial behavior in unselected individuals (Loeb & Mednick 1977 using skin conductance; Raine et al. 1990a using EEG, heart rate, and skin conductance; Raine et al. 1990b using evoked potentials; Satterfield 1987 using EEG; and Volavka et al. 1984 using EEG). As for the reports on neurochemistry, these effects are not small; Raine (1993) reports that for heart rate, the average effect size across ten studies was 0.84. Another important physiological variable in the distribution of sociopathic behavior is testosterone. Testosterone (or one of its derivatives) is a likely trigger for the sex-limited activation of genes required by the twothreshold model presented earlier. The mechanism of action of steroid hormones is to enter the nucleus of the cell and interact with the chromosomes, regulating gene expression. This differential activity of the genes leads to some of the individual, age, and sex differences we see in temperament, specifically, psychoticism, aggression, impulsivity, sensation-seeking, nurturance, and empathy (Ellis 1991b; Zuckerman 1984; 1985; 1991; Zuckerman et al. 1980). Variation in testosterone levels also parallels the age variation in the expression of sociopathic behavior and is correlated with such behavior in adolescent and adult males (Archer 1991; Dabbs & Morris 1990; Daitzman & Zuckerman 1980; Ellis & Coontz 1990; Gladue 1991; Olweus 1986; 1987; Rubin 1987; Schalling 1987; Susman et al. 1987; Udry 1990; Zuckerman 1985). Testosterone is thus likely to play a dual role in the development of 532 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 sociopathy, just as it does in the development of other sex differences: one as an organizer (affecting traits) and one as an activator (affecting states). Udry (Drigotas & Udry 1993; Halpern et al. 1993), unable to replicate his own 1990 study suggesting an activating effect of testosterone, has suggested that the correlation between testosterone and aggression might be due to a physiosocial feedback loop; he posits that boys with high, early levels of testosterone mature faster and, being bigger, are more likely to get into fights. Since levels of testosterone, adrenaline, and serotonin have been shown to fluctuate in response to social conditions (Archer 1991; Kalat 1992; McGuire et al. 1983; Olweus 1987; Raleigh et al. 1991; Raleigh et al. 1984; Schalling 1987), this sociophysiological interaction creates a positive feedback loop: those who start out with high levels of testosterone and sensation seeking (and low levels of adrenaline, serotonin, and MAO) are (1) more likely than others to initiate aggressive behavior and (2) more likely to experience success in dominance interactions, leading to (3) a greater probability of experiencing further increases in testosterone, which (4) further increases the likelihood of continued aggressive behavior. Another example of a sociophysiological feedback loop comes from Dabbs and Morris (1990), who found significant correlations between testosterone levels and antisocial behavior in lower class men but not in upper class men. They explained this by positing that upper class men are more likely, because of differential socialization, to avoid individual confrontations. If this is true, it would mean that upper class men are, because of their socialization, specifically avoiding the types of social encounters that might raise their testosterone (and, in turn, their antisocial behavior). This interpretation is supported by the finding (in the same study) that significantly fewer upper class than lower class men had high testosterone levels. Thus, it is possible that upper class socialization may mitigate the influence of testosterone. An alternative explanation - that the aggressive behavior associated with higher testosterone levels leads to downward social mobility - also suggests a recursive sociophysiological interaction. Raine (1988) has argued that since upper class children are less likely than lower class children to suffer the environmental risks predisposing one toward sociopathic behavior, when such behavior is seen in upper class individuals, it is likely to be the result of a particularly strong genetic predisposition. Evidence supporting this has been reported by three independent studies. Wadsworth (1976) found physiological indicators of hypoarousal among upper-class, but not lower-class, boys who subsequently became delinquent. Raine (Raine 1988; Raine & Dunkin 1990; Raine & Venables 1981; 1984) found indicators of hypoarousal in his upper-class antisocial subjects, but the reverse in his lower-class subjects. Satterfeld (1987) found that of his lower-class subjects, those in a biological high-risk group were seven times more likely to have been arrested than those in his control group, whereas among his middle- and upper-class subjects, the rate was 25 and 28 times, respectively. This outcome was a result of lower rates of criminal activity in the control groups of the middle- and upper-class subjects as compared to the lower-class controls; that is, almost all of those who had been arrested from the middle and Mealey: The sociobiology of sociopathy upper class were biologically at high risk, but this was not true for the lower class subjects. The implications of these findings are of tremendous import, as they suggest that (1) the effect of the social environment might be considerably larger than suggested by adoption studies and (2) there might be different etiological pathways to sociopathy and therefore different optimal strategies for its prevention or remediation, depending upon what kind of social and environmental background the individual has experienced. Adoption studies show that the environment clearly plays an important role in the etiology of sociopathy, but that its effects are different for individuals of different genotypes. As mentioned in section 2.3.1, some of this difference is likely to be a result of gene-environment correlations, in that different environments are sought by individuals of different genotypes; some will be a result of differences in the interpretation of the same environment by individuals of different genotypes; and some will be a result of differences in environment impinging upon people because of differences in their genotype (e.g. discriminating parental treatment of two children differing in temperament). In nonadoptive families, gene-environment correlations will be even stronger because parents with certain personality types will provide certain environments for their children. These differential effects of environment on individuals of varying genetic risk for sociopathy become readily apparent when we examine the effect of the interaction between physiotype and conditioning on the process of socialization. result of their behavior or an indirect result such as parental punishment. Despite continuing problems with operational definitions, recent research suggests that there might be distinguishable differences in learning between primary and secondary sociopaths, or children with unsocialized versus socialized conduct disorder (Gray 1987; Newman et al. 1992; Newman et al. 1985; Quay 1990b). Primary sociopaths, with their inability to experience the social emotions, exhibit deficits on tasks which typically induce anxiety in others, specifically, passive avoidance tasks, approach-avoidance tasks, and tasks involving punishment, but they can learn well under other conditions (Newman et al. 1992; Patterson & Newman 1993; Raine 1993; Raine et al. 1990b). Secondary sociopaths and extraverts, on the other hand, have normal levels of anxiety and responses to punishment, but they may be especially driven by high reward conditions (Boddy et al. 1986; Derryberry 1987; Newman et al. 1990). Primary sociopaths, with diminished ability to experience anxiety and to form conditioned associations between antisocial behavior and the consequent punishment, will be unable to progress through the normal stages of moral development. Unlike most children who are biologically prepared to learn empathy, they are contraprepared to do so, and will remain egoistic unable to acquire the social emotions of empathy, shame, guilt, and love. They present at an early age with "unsocialized" conduct disorder. Secondary sociopaths, with normal emotional capacities, will present, generally at a later age, with "socialized" conduct disorder (Loeber 1993; Patterson 1993; Simons 1993). What socialization processes contribute to their development? 2.4.1. Conditioning. There is evidence that individuals with a hypoaroused nervous system are less sensitive than most people to the emotional expression of other individuals and to social influences in general (Eliasz & Reykowski 1986, Eysenck 1967 as cited in Patterson and Newman 1993). They are also less responsive to levels and types of stimuli that are normally used for reinforcement and punishment (Eliasz 1987); as a result, they are handicapped in learning through autonomic conditioning although they exhibit no general intellectual deficit (e.g., Eysenck 1977; Gorenstein & Newman 1980; Hare & Quinn 1971; Lytton 1990; Mednick 1977; Newman et al. 1985; Raine 1988; Ziskind et al. 1978; Zuckerman 1991). One of the posited consequences of this learning deficit is a reduced ability to be socialized by the standard techniques of reward and punishment that are used (especially in the lower classes and by uneducated parents) on young children. In particular, hypoaroused individuals have difficulty inhibiting their behavior when both reward and punishment are possible outcomes (Newman 1987; Newman & Kosson 1986; Newman et al. 1985; Patterson & Newman 1993; Zuckerman 1991); in situations when most people would experience an approachavoidance conflict, sociopaths and extraverts are more likely to approach (see also Dienstbier 1984). Because of their high levels of sensation seeking, children with a hypoaroused nervous system will be more likely than other children to get into trouble and when they do, they will be less likely to be affected by, and learn from, the consequences, whether those consequences are a direct 2.4.2. Social learning. In sect. 2.2.1, it was noted that a cheating strategy is predicted to develop when a male (especially) is competitively disadvantaged, and that criminal behavior (especially in males) is clearly related to factors associated with disadvantage. These factors are: large numbers of siblings, low socioeconomic status, urban residency, low intelligence, and poor social skills. How, in a proximate sense, do these variables contribute to the development of secondary sociopathy? Path models suggest a two-stage process involving a variety of cumulative risk factors (Dishion et al. 1991; Loeber 1993; McGarvey et al. 1981; Moffitt 1993; Patterson et al. 1991; Simons 1993; Snyder et al. 1986; Snyder & Patterson 1990).10 In the first stage, disrupted family life, associated with parental neglect, abuse, inconsistent discipline, and the use of punishment as opposed to rewards, are critical (Conger 1993; Feldman 1977; Luntz & Widom 1993; McCord 1986; Patterson et al. 1989; Simons 1993; Snyder et al. 1986; Wilson & Herrnstein 1985). Poor parenting provides the child with inconsistent feedback and poor models of prosocial behavior, handicapping the child in the development of appropriate social, emotional, and problem-solving skills. This pattern is found most frequently in parents who are themselves criminal, mentally disturbed, undereducated, of low intelligence, or socioeconomically deprived (Farrington 1986; McCord 1986; McGarvey et al. 1981), leading to a cross-generational cycle of increasing family dysfunction (e.g., Jaffe et al. 1992; Luntz & Widom 1993). 2.4. Learning theory BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 533 Mealey: The sociobiology of sociopathy In the second stage, children with poor social skills find themselves at a disadvantage in interactions with age mates; rejected by the popular children, they consort with one another (Dishion et al. 1991; Hartup 1989; Kandel et al. 1988; Loeber & Dishion 1983; Patterson et al. 1989; Snyder et al. 1986). In these socially unskilled peer groups, which will also include primary sociopathic or unsocialized conduct disorder children, delinquent, antisocial behavior is reinforced and new (antisocial) skills are learned (Maccoby 1986; Moffitt 1993). Antisocial behavior may then escalate in response to, or as prerequisite for, social rewards provided by the group, or as an attempt to obtain the perceived social (and tangible) rewards which often accompany such behavior (Moffitt 1993). As the focus of the socialization process moves outside the home, parental monitoring becomes more important (Conger 1993; Dishion etal. 1991; Forgatch 1991; Simons 1993; Snyder et al. 1986; Snyder & Patterson 1990), as does the availability of prosocial alternatives for the socially unskilled adolescent (Apter 1992; Farrington 1986; Moffitt 1993). The development of secondary sociopathy appears to depend much more on environmental contributions than does primary sociopathy. Since it is secondary sociopathy which, presumably, has increased so rapidly and so recently in our culture, what can social psychologists contribute to our understanding of the sociocultural factors involved in its development? 2.5. Social psychology 2.5.1. Machiavellianism. First, the use of antisocial strategies is not restricted to sociopaths. The majority of people who are arrested are not sociopathic and many people exhibit antisocial behavior that is infrequent enough or inoffensive enough to preclude arrest. Some antisocial behavior is even considered acceptable if it is expressed in socially approved circumstances. Person (1986), for example, relates entrepreneurism to psychopathy, whereas Christie (1970) notes that people who seek to control and manipulate others often become lawyers, psychiatrists, or behavioral scientists; Jenner (1980) also claims that "subtle, cynical selfishness with a veneer of social skills is common among scientists" (p. 128). Christie (see Christie & Geis 1970) developed a scale for measuring this subclinical variation in antisocial personality; he called it the "Machiavellianism" or "Mach" scale. One's Mach score is calculated by compiling answers to Likert-format queries of agreement or disagreement with statements like "humility not only is of no service but is actually harmful," "nature has so created men that they desire everything but are unable to attain it," and "the most important thing in life is winning." Adults who score high on the Mach scale express "a relative lack of affect in interpersonal relationships," "a lack of concern with conventional morality," "a lack of gross psychopathology," and "low ideological commitment" (Christie & Geis 1970, pp. 3-4); children who score high on Machiavellianism have lower levels of empathy than their age mates (Barnett & Thompson 1984). High Machs have an "instrumental cognitive attitude toward others" (Christie & Geis 1970, p. 277) and, because they are goal oriented as opposed to person oriented, they are more successful in face-to-face bargaining 534 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 situations than low Machs. High Machs "are especially able communicators, regardless of the veracity of their message" (Kraut & Price 1976). In a related vein, high Machs, like sociopaths, are more resistant to confession after cheating than are low Machs, and they are rated as being more plausible liars (Bradley & Klohn 1987; Christie & Geis 1970); like sociopaths, high Machs are often referred to as "cool." According to Christie, "If Machiavellianism has any behavioral definition. . . selfinitiated manipulation of others should be at its core" (p. 76). One can thus easily think of Machiavellianism as a low-level manifestation of sociopathy. It even shows a sex difference consistent with the two threshold model (Christie & Geis 1970), an age pattern consistent with age variation in testosterone levels (Christie & Geis 1970), significant positive correlations with Eysenck's psychoticism and neuroticism scales (Allsopp et al. 1991), and a correlation with serotonin levels (Madsen 1985). In one study, Geis & Levy (1970) found that high Machs (who were thought to use an "impersonal, cognitive, rational, cool" approach with others), were much more accurate than low Machs (who were thought to use a "more personal, empathizing" approach) at assessing how other "target" individuals answered a Machiavellian attitudes questionnaire. Even more interesting is the result (from the same study) that the high Machs achieved their accuracy by using a nomothetic or actuarial strategy: they guessed that everyone was at about the average level, without discriminating between individuals based on differences they had had an opportunity to observe during a previous experimental session. In addition, their errors tended to be random, which would fit with reports by Eliasz & Reykowski (1986) and Damasio et al. (1990), who found that hypoaroused and antisocial individuals are less attentive to social and emotional cues than others. Low Machs, on the other hand, used an idiographic approach, and although they successfully differentiated between high scorers and low scorers, they grossly underestimated the scores of both, guessing at a level that was more reflective of their own scores than those of the population at large. This study suggests two things: (1) Basing one's playing strategies on an "impersonal, cognitive, rational, cool" approach to others might be more accurate in the long run than using a "personal, empathizing" approach (at least in those situations where cooperative long-term partnerships are not possible) and (2) the errors made by those who use the personal, empathizing approach are of the kind more likely to result in playing the cooperation strategy when the cheating strategy would be more appropriate (rather than vice versa). Thus, the personal, empathizing approach is likely to make one susceptible to being exploited by others who use the impersonal cognitive approach; indeed, high Machs outcompete low Machs in most experimental competitive situations (Christie & Geis 1970; Terhune 1970). As I have argued elsewhere (Mealey 1992), the common assumption that an empathy-based approach to predicting the behavior of others is better than a statistical approach is not necessarily correct; this belief may itself be an emotion-based cognitive bias. To have such a bias may be beneficial, however, for the same reason that emotional commitment biases are beneficial: in situations where voluntary, long-term coalitions can be formed, the Mealey: The sociobiology of sociopathy personal, empathizing (and idealistic) low Machs might outperform the more impersonal, cognitive (and realistic) high Machs, since low Machs would be more successful than high Machs in selecting a cooperator as a partner. Although two studies (Hare & Craigen 1974 and Widom 1976a) report on the strategy of sociopaths in Prisoner's Dilemma-type settings, in both studies the sociopaths were paired with one another; thus, we do not have a measure of the strategy sociopaths use against partners of their own choosing or in situations with random, rotating partners. 11 I would predict that in such settings, sociopaths, like Geis & Levy's high Mach subjects, would be less proficient than others in distinguishing between high and low Mach partners, and would thus be at a disadvantage in iterated games with a chosen partner; on the other hand (again like high Mach subjects), they should perform at better than average levels when playing with randomly assigned, rotating partners. Widom (1976b) found that when asked to guess how "people in general" would feel about different social situations sociopaths guessed that others would feel essentially the same way that they do, whereas control subjects guessed that others would feel differently. As in the Geis & Levy study, both groups were wrong, but in different ways: the sociopaths underestimated their differences from others, whereas the control subjects substantially overestimated their differences from others, suggesting that sociopaths (like high Machs) were using a nomothetic approach to prediction, whereas controls (like low Machs) were using an idiographic approach. Machiavellianism and the related propensity to use others in social encounters has generally been looked upon as a trait. An alternative perspective, however, acknowledges both the underlying variation in personality and the situational factors that are relevant to an individual's behavior at any given moment (e.g., Barber 1992). In line with mechanism 5 for maintaining ESSs (presented in sect. 1.2), Terhune (1970) says "actors bring to the situation propensities to act in a certain general way, and within the situation their propensities interact with situational characteristics to determine their specific behavior" (p. 229).12 This brings us to the last question: beyond the constitutional and environmental variables that contribute to the development of individual differences in personality and antisocial behavior, what can social psychology tell us about the within-individual situational factors that encourage or discourage cheating strategies, and how can these be explained? 2.5.2. The role of mood. Although mood and emotion are not identical concepts, they are clearly related. 13 Mood might be thought of as a relative of emotion which clearly varies within individuals but is perhaps less an immediate response to concrete events and stimuli and more a generalized, short- to mid-term response to the environment. As such, the role of mood must be addressed by any model that relies so heavily on the concepts of emotion, emotionality, and emotionlessness, as determinants of behavior. Positive mood and feelings of success have been demonstrated to enhance cooperative behavior (Cialdini et al. 1982; Farrington 1982; Mussen & Eisenberg-Berg 1977). If, as Nesse (1991) has argued, positive mood is a reflection not only of past success, but also of anticipation of future success, the facilitation of cooperation by positive mood could be seen as part of a long-term strategy by individuals who feel they can afford to pass up possible short-term gains for the sake of establishing a cooperative reputation. Sad affect and feelings of failure can also affect strategy in social interactions. To the extent that sadness and feelings of failure follow losses of various sorts, individuals in these circumstances should be expected to be egoistic and selfish. In children, this is typically what is found (Baumann et al. 1981; Mussen & Eisenberg-Berg 1977). In some children, and more consistently in adults, on the other hand, sadness and feelings of failure can facilitate prosocial behavior. Mussen & Eisenberg-Berg (1977) suggest that this is a result of a deliberate effort: to enhance one's (diminished) reputation among others; Baumann et al. (1981) and Cialdini et al. (1982) suggest that it is a result of a deliberate effort to relieve negative affect based on prior experience that prosocial behavior often has a positive, self-gratifying effect. If sadness is profound, that is, if one is depressed and experiencing the cognitive biases and selective attention associated with depression (Mineka & Sutton 1992; Nesse 1991; Sloman 1992), one would be expected to desist from all social interaction, being neither antisocial nor prosocial, but asocial (Nesse 1991; Sloman 1992). In this view, the lethargy and anhedonia associated with depression could be considered to be facultative lapses in the emotions or moods which typically motivate a person toward social interaction. Hostility can also lead to cognitive biases and selective attention to relevant social stimuli. Dodge and Newman (1981) show that aggressiveness in boys is associated with the over-attribution of hostile intent to others. The authors conclude that such attributions lead to increased "retaliatory" aggression by the hostile individuals, fueling a cycle of true hostility and retaliation by all parties. It is also abundantly clear that anger and hostility, once expressed, do not lead to catharsis, but to amplified feelings and outward expressions of that anger (Tavris 1982). Guilt, which often follows selfish behavior, typically results in an increase in subsequent prosocial behavior (Cialdini et al. 1982; Hoffman 1982); Hoffman calls this "reparative altruism." Guilt can easily be seen as one of Hirshleifer's (1987) or Frank's (1988) emotional commitment devices, compelling one to perform prosocial behavior as a means of reestablishing one's tarnished reputation. Cialdini et al. (1982) also report that prosocial behavior increases after observing another's transgression. They explain this phenomenon within the context of what they call the "negative relief model: prosocial behavior is performed as a means of alleviating negative feelings in general (including direct or vicarious guilt, sympathy, distress, anxiety, or depression). Like Hoffman's, this model postulates that the reinforcing power of (relief provided by) prosocial behavior is learned during childhood. Since guilt, anxiety, and sympathy are social emotions that primary sociopaths rarely, if ever, experience, there is no reason to expect that they might moderate their behavior so as to avoid them. On the other hand, there is no reason to expect that sociopaths do not experience fluctuations in mood (such as depression, optimism, or BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 535 Mealey: The sociobiology of sociopathy anger) in response to their changing evaluation of their prospects of success and failure. To the extent that we can manipulate the sociopath's mood, therefore, we might be able to influence his behavior. 2.5.3. Cultural variables. Competition, in addition to being one of the most important variables in determining long-term life strategy choices, is also one of the more important situational variables influencing the choice of immediate strategy. Competition increases the use of antisocial and Machiavellian strategies (Christie & Geis 1970) and can counteract the increase in prosocial behavior that generally results from feelings of success (Mussen & Eisenberg-Berg 1977). Some cultures encourage competitiveness more than others (Mussen & EisenbergBerg 1977; Shweder et al. 1987) and these differences in social values vary both temporally and crossculturally. Across both dimensions, high levels of competitiveness are associated with high crime rates (Farley 1986; Wilson & Herrnstein 1985) and Machiavellianism (Christie & Geis 1970). High population density, an indirect form of competition, is also associated with reduced prosocial behavior (Farrington 1982) and increased antisocial behavior (Ellis 1988; Robins et al. 1991; U.S. Department of Justice 1993; Wilson & Herrnstein 1985) - especially in males (Wachs 1992; see sect. 3.2.1 and references therein for ultimate, game theoretic explanations why this might occur; see Draper 1978; Foster 1991; Gold 1987; Siegel 1986; and Wilson & Daly 1993 for a variety of proximate explanations). Fry (1988) reports large differences in the frequency of prosocial and antisocial behaviors in two Zapotec settlements equated for a variety of socioecological variables; the one major difference - thought perhaps to be causal - was in land holdings per capita, with the higher levels of aggression found in the community with the smaller per capita land holdings. Last, but not least, is the relatedness or similarity of the interactors to their partners in an interaction. Based on models of kin selection and inclusive fitness, individuals should be more cooperative and less deceptive when interacting with relatives who share their genes, or relatives who share investment in common descendents. Segal (1991) reported that identical twins cooperated more than fraternal twins in playing the Prisoner's Dilemma. Barber (1992) reported that responses on an altruism questionnaire were more altruistic when the questions were phrased so as to refer to relatives (as opposed to "people" in general), and that Machiavellian responses were thereby reduced. Rushton (1989; Rushton et al. 1984) presents evidence that people also cooperate more with others who are similar to them even though not genetically related. There are a variety of plausible evolutionary explanations for this behavior (see Mealey 1984; Pulliam 1982; and BBS commentary on Rushton 1989). 3. Integration, implications, and conclusions 3.1. Integration: Sociopathy as an ESS leads to two types of sociopaths 3.1.1. Primary sociopathy. I have thus far argued that some individuals seem to have a genotype that disposes 536 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 them "to acquire and be reinforced for displaying antisociality" (Rowe 1990a, p. 122). The genotype results in a certain inborn temperament or personality, coupled with a particular pattern of autonomic hypoarousal that, together, design the child to be selectively unresponsive to the cues necessary for normal socialization and moral development. This scenario is descriptive of mechanism 1 (sect. 1.2) of maintaining ESSs in the population; it describes the existence of frequency-dependent, genetically based individual differences in employment of life strategies. I accordingly suggest that there will always be a small, cross-culturally similar, and unchanging baseline frequency of sociopaths: a certain percentage of sociopaths - those individuals to whom I have referred as primary sociopaths - will always appear in every culture, no matter what the sociocultural conditions. Those individuals will display chronic, pathologically emotionless, antisocial behavior throughout most of their lifespan and across a variety of situations, a phenotype that is recognized (according to Robins et al. 1991, p. 259) "by every society, no matter what its economic system, and in all eras".14 Since it is a genetically determined strategy, primary sociopaths should be equally likely to come from all kinds of socioeconomic backgrounds; on the other hand, since they constitute that small group of individuals whose physiotype makes them essentially impervious to the social environment almost all sociopaths from the upper classes will be primary sociopaths.15 Of course, because they are not intellectually handicapped, these individuals will progress normally in terms of cognitive development and will acquire a theory of mind. Theirs, however, will be formulated purely in instrumental terms, without access to the empathic understanding that most of us rely on so much of the time. They may become excellent predictors of others' behavior, unhandicapped by the vagaries and "intrusiveness" of emotion, acting, as do professional gamblers, solely on nomothetic laws and actuarial data rather than on hunches and feelings. In determining how to "play" in the social encounters of everyday life, they will use a pure costbenefit approach based on immediate personal outcomes, with no "accounting" for the emotional reactions of the others with whom they are dealing. Without love to "commit" them to cooperation, anxiety to prevent "defection, " or guilt to inspire repentance, they will remain free to continually play for the short-term benefit in the Prisoner's Dilemma. 3.1.2. Secondary sociopathy. At the same time, because changes in gene frequencies in the population would not be able to keep pace with the fast-changing parameters of social interactions, an additional, fluctuating proportion of sociopathy should be a result of mechanism 5 for maintaining ESSs, which allows for more flexibility in the ability of the population to track the frequencydependent nature of the success of the cheating strategy. Mechanism 5 (genetically based individual differences in response to the environment, resulting in differential use by individuals of environmentally contingent strategies) would explain the development and distribution of what I have referred to as secondary sociopathy. Secondary sociopathy is expressed by individuals who are not extreme on the genetic sociopathy spectrum, but who, because of their exposure to environmental risk factors, pursue a life Mealey: The sociobiology of sociopathy strategy that involves frequent, but not necessarily emotionless cheating. Unlike primary sociopaths, secondary sociopaths will not necessarily exhibit chronic antisocial behavior, because their strategy choices will be more closely tied to age, fluctuation in hormone levels, their competitive status within their referent group, and changing environmental contingencies. Since secondary sociopathy is more closely tied to environmental factors than to genetic factors, secondary sociopaths will almost always come from lower class backgrounds and their numbers could vary substantially across cultures and time, tracking environmental conditions that favor or disfavor the use of cheating strategies. The existence of this second etiological pathway to sociopathy explains the fact that cultural differences are correlated with differences in the overall incidence of antisocial behavior (Ellis 1988; Farley 1986; Gold 1987; Robins et al. 1991; Wilson & Hermstein 1985). It also explains why, as the overall incidence of sociopathy increases, the discrepancy in the ratio of male to female sociopaths decreases (Robins et al. 1991): since secondary sociopathy is less heritable than primary sociopathy (according to this model), the effect of sex-limited genes (like that of all the genes contributing to the spectrum) should be less important for the development of secondary sociopathy, resulting in less of a sex difference. Rased on this model, I would predict that, unlike what we find for primary sociopathy (see sect. 2.1.3), we should find no differential heritability between the sexes for secondary sociopathy (even though there will still be a sex difference in prevalence). 3.2. Implications of the two-pathway model Terliune (1970) suggests that choice of strategy in experimental game situations (and, presumably, real-life settings as well) depends upon two things: (1) cognitive expectations regarding others (i.e., a theory of mind) and (2) motivational/emotional elements such as hopes and fears. Since primary sociopaths have a deficit in the realm of emotional motivation, they presumably act primarily upon their cognitive expectations of others; to the extent that they do act upon emotions, it is most likely to be upon mood and the primary emotions (like anger and fear) rather than upon the social and secondary emotions (like love and anxiety). Thus, the extent to which a society will be able to diminish the antisocial behavior of primary sociopaths will depend upon two things: (1) its influence on the sociopath's cognitive evaluation of its own reputation as a player in the Prisoner's Dilemma, and (2) the primary emotion- or mood-inducing capacity of the stimuli it utilizes in establishing the costs and benefits of prosocial versus antisocial behavior. Manipulations of these two variables will also influence the numbers of secondary sociopaths by changing the size of the adaptive niche associated with antisocial behavior. In addition, since the development of secondary sociopathy is more influenced by the social environment than is the development of primary sociopathy, and since secondary sociopaths are not devoid of social emotions, changing patterns in the nurturing and socialization of children and in the socialization and rehabilitation of delinquents and adult criminals is an additional, viable possibility for reducing the overall prevalence of antisocial behavior. 3.2.1. Minimizing the impact of primary sociopaths: Society as a player in the Prisoner's Dilemma. Sociopaths' immediate decisions are based in part on their ability to form a theory of mind, and to use those expectations of others' behavior in a cost benefit analysis to assess what actions are likely to be in their own self-interest. (This is true for both primary and secondary sociopaths.) The outcome of such analyses is therefore partially dependent on the sociopath's expectations of the behavior of other players in the game. I would argue that an entire society can be seen as a player and that the past behavior of that society will be used by the sociopath in forming the equivalent of a theory of mind to predict the future behavior of that society. Like an individual player, a society will have a certain probability of detecting deception, a more-or-less accurate memory of who has cheated in the past, and a certain proclivity to retaliate or not, based upon a cheater's past reputation and current behavior. Since the sociopath is using a rational and actuarial approach to assess the costs and benefits of different behaviors, it is the actual past behavior of the society which will go into his calculations, rather than risk assessments inflated from the exaggerated fears or anxieties that most people feel in anticipation of being caught or punished. Thus, to reduce antisocial behavior, a society must establish and enforce a reputation for high rates of detection of deception and identification of cheaters, and a willingness to retaliate. In other words, it must establish a successful strategy of deterrence. Game theory models by Axelrod and others have shown that the emergence, frequency, and stability of social cooperation is subject to an abundance of potential deterrent factors (Axelrod 1984; Axelrod & Dion 1988; Axelrod & Hamilton 1981; Boyd 1988; Boyd & Richerson 1992; Dugatkin & Wilson 1991; Feldman & Thomas 1987; Heckathorn 1988; Hirshleifer & Coll 1988; Nowak & Sigmund 1993; Vila & Cohen 1993). Among these are: group size (as it decreases, cooperation increases); nonrandom association of individuals within the population (as it increases, cooperation increases); the probability of error in memory or recognition of an individual (as it decreases, cooperation increases); the effect of a loss on a cooperator (as it decreases, cooperation increases); the effect of a gain on a defector (as it decreases, cooperation increases); the frequency of punishment against defectors (as it increases, cooperation increases); the cost of punishment for the punished (as it increases, cooperation increases); and the cost of punishment for the punishers (as it decreases, cooperation increases).16 Recent game-theoretic models are coming closer and closer to the complexity of real-world, human social interactions on a large scale by examining the role of culture and technology in expanding society's collective memory of individual players' past behavior, broadcasting the costs and benefits of cooperation and defection, and the development and application of new socialization, deception-detection, and punishment techniques (see especially Dugatkin 1992; Hirshleifer & Rasmusen 1989; Machalek & Cohen 1991). These models begin to provide useful strategies for the real-world prediction and reduction of cheating strategies and antisocial behavior. (See also Axelrod 1986; Bartol 1984; Ellis 1990a; Eysenck & Gudjonsson 1989; Farrington 1979; Feldman 1977; MaBEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 537 Mealey: The sociobiology of sociopathy chalek & Cohen 1991; and Wilson & Herrnstein 1985 for some nonquantitative models and tests that incorporate some of these variables in their explanation of the socialization, punishment, and deterrence of crime.) Since neither secondary nor primary sociopaths have a deficit in the ability to perform accurate cost-benefit analyses, increasing the probabilities of criminal detection, identification, and punishment can also reduce crime; a society must therefore establish a reputation for willingness to retaliate. (The National Research Council [1993] reports that a 50% increase in the probability of incarceration for any single crime reduces subsequent crime twice as much as does doubling incarceration duration [p. 294]). Harsher penalties can also be deterring, but only if they are reliably meted out. Another key is in making the costs of cheating salient. Generally speaking, antisocial and uncooperative behaviors increase as the costs become more diffuse or removed in time, and prosocial and cooperative behaviors decrease as the benefits become more diffuse or removed in time (Bartol 1984; Low 1993; Ostrom 1990). For primary sociopaths, this is even more so, since their sensation-seeking physiotype makes them particularly unable to make decisions based on nonimmediate consequences. Although able to focus attention on interesting tasks for short periods, the sociopath cannot perform well under conditions of delayed gratification (Pulkkinen 1986) and is more motivated to avoid immediate costs than by threats or avoidance of future punishments (Christie & Geis 1970; Forth & Hare 1989; McCord 1983; Raine 1988; 1989). Costs associated with social retaliation must therefore not only be predictable, but swift, and the swiftness itself must also be predictable. Another factor the sociopath will use to "compute" the potential value of an antisocial action is the cost-benefit ratio of the alternatives (Piliavin et al. 1986). For the sociopath, money and other immediate tangible rewards are more motivating than social reinforcers (such as praise) or promises of future payoff, and visual stimuli are more salient than auditory stimuli (Chesno & Kilmann 1975; Forth & Hare 1989; Raine 1989; Raine & Venables 1987; Raine et al. 1990b; Zuckerman 1990). Thus, alternatives to crime must be stimulating enough and rewarding enough to preferentially engage the chronically hypoaroused sensation seeker. This will be a difficult task to achieve, but it will be more successful if we can effectively distinguish primary from secondary sociopaths. Primary sociopaths, with their emotional, but not intellectual deficit, will be competent on some tasks on which secondary sociopaths, with deficits in social skills, emotion regulation and problem solving, will not. Possibilities might include: novelist, screenplay writer, stunt man, talk show host, disk jockey, explorer, treasure hunter, race car driver, or skydiving exhibitionist. Given that primary sociopaths will always be with us in low numbers, it would be a wise social investment to create - even on an individual basis, if necessary - a number of exciting, highpayoff alternatives for them, in order to minimize the number who may otherwise cause pain and destruction. Distinguishing between primary and secondary sociopaths is also critical for decisions about confinement and rehabilitation. Quinsey & Walker (1992) cite examples where recidivism rates went up for psychopaths, but down for nonpsychopaths, after they were exposed to the 538 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 same kind of "treatment." Recidivism is much greater in primary sociopaths than in secondary sociopaths (Hare et al. 1992), and sometimes the only response is prolonged incapacitation (until they literally "grow out of it"). A recent international meeting of experts concluded that "treatment" programs dealing with primary sociopaths should be "less concerned with attempts to develop empathy or conscience than with intensive efforts to convince them that their current attitudes and behavior (simply) are not in their own self-interest" (Hare 1993, p. 204). 3.2.2. Minimizing the prevalence of secondary sociopathy: Society as a socializing agent and mood setter. Given that secondary sociopaths have a different life history and are more responsive to environmental influences than primary sociopaths, social changes can be designed to minimize not only their impact, but their incidence. Loeber (1990) argues that each generation in our society is being raised with an increasing number of environmental risk factors, leading to increasing generation-wide deficits in impulse control. He makes specific suggestions to screen for high-risk children and institute early intervention, noting that different interventions are likely to be more or less effective given different risk factors in the child's or adolescent's life history (see also U.S. Department of Justice 1993). One possible intervention is parent training (see Magid & McKelvie 1987 and Dumas et al. 1992 for reviews and programmatic suggestions). Laboratory experiments show that antisocial behaviors can be reduced and prosocial behaviors reinforced by appropriate use of modelling, induction, and behavioral modification techniques (Feldman 1977; Gelfand & Hartmann 1982; Grusec 1982; Kochanska 1991; 1992; Mussen & Eisenberg-Berg 1977; Radke-Yarrow & Zahn-Waxler 1986; Rushton 1982). Recent longitudinal studies in natural settings suggest that the positive effects of good parenting, especially parental warmth and predictability, may be long lasting (Kochanska 1991; 1993; Kochanska & Murray 1992; McCord 1986). The cause and effect relationship between parental behavior and child behavior, however, is not likely to be one-way. Children of different gender, temperaments, and even social classes, respond differentially to different socialization techniques (Dienstbier 1984; Kochanska 1991; 1993; Kochanska & Murray 1992; Lytton 1990; McCord 1993; Radke-Yarrow & Zahn-Waxler 1986), and, to some extent, difficult children elicit poor parenting (Bell & Chapman 1986; Buss 1981; Eron et al. 1991; Lee & Bates 1985; Lytton 1990; Snyder & Patterson 1990). It is easy for parents of difficult children to lose heart, and in so doing, become even less effective (Patterson 1992). For example, studies cited in Landy & Peters (1992) found that mothers of aggressive children, like other mothers with a low sense of personal power, tend to give weak, ineffectual commands to their children. This lack of "goodness of fit" between parental style and the needs of the child is probably an important factor in the exacerbation of conduct disorder (Landy & Peters 1992; Lee & Bates 1985; Moffitt 1993; Wachs 1992). Parents need help in identifying high-risk children, and they need instruction in how to take a practical, assertive approach with them (see Garmezy 1991; Magid & Mealey: The sociobiology of sociopathy McKelvie 1987), while using a more inductive, empathic approach with their other children (see Kochanska 1991; 1993; Kochanska & Murray 1992). Social workers, health care providers, and employees of the criminal justice system also need to be able to distinguish between children with different risk factors and life histories and to respond accordingly. Palmer (1983) argues that agents should be individually matched with each client/offender based on style and personality characteristics to prevent high Mach and sociopathic offenders from taking advantage of low Mach employees. At a broader level, many sociocultural aspects of modern society seem to contribute to antisocial behaviors and attitudes (Moffitt 1993; National Research Council 1993). As a society gets larger and more competitive, both theoretical models (sect. 3.2.1) and empirical research (sect. 2.4.2) show that individuals become more anonymous and more Machiavellian, leading to reductions in altruism and increases in crime. Social stratification and segregation can also lead to feelings of inferiority, pessimism, and depression among the less privileged, which can in turn promote the use of alternative competitive strategies, including antisocial behavior (Magid & McKelvey 1987; Sanchez 1986; Wilson & Daly 1993). Crime may be one response to the acquisition of an external locus of control (Raine et al. 1982) or learned helplessness. Learned helplessness and other forms of depression have been associated with reduced levels of serotonin (Traskman et al. 1981); since reduced levels of serotonin have also been shown to be related to increased aggression, it is likely that physiological changes mediate these psychological and behavioral changes. The neurochemical pathway involved in learned helplessness (identified by Petty & Sherman 1982) appears to be the same one identified by Gray (1982; 1987) and Cloninger (1987a) with mediation of behavioral inhibition/harm avoidance, and by Charney et al. (1990) with anxietymediated inhibition. Crime may also function to obtain desirable resources, increase an individual's status in a local referent group, or provide the stimulation that the more privileged find in more socially acceptable physical and intellectual challenges (e.g., Apter 1992; Farley 1986; Farrington 1986; Lyng 1990; Moffitt 1993). According to Apter, "the vandal is a failed creative artist," a bored and frustrated sensation seeker who "does not have the intellectual or other skills and capacities to amuse or occupy himself" (1992, p. 198). Thus, in addition to making the costs of antisocial behavior greater, strong arguments can be made for providing early social support for those at risk, and for developing alternative, nonexploitative, sensation-seeking ventures that can meet the psychological needs of disadvantaged and low-skilled individuals. 3.3. Conclusions A review of the literature in several areas supports the concept of two pathways to sociopathy: (la) "Primary sociopaths" are individuals of a certain genotype, physiotype, and personality who are incapable of experiencing the secondary, "social" emotions that normally contribute to behavioral motivation and inhibition; they fill the ecological niche described by game theorists as the "cheater strategy," and, as the result of frequency-dependent selection, will be found in low frequency in every society. (lb) To minimize the damage caused by primary sociopaths, the appropriate social response is to modify the criminal justice system in ways that obviously reduce the benefits and increase the costs of antisocial behavior, while simultaneously creating alternatives to crime that could satisfy the psychophysiological arousal needs of the sociopath. (2a) "Secondary sociopaths" are individuals who use an environmentally contingent, facultative cheating strategy not as clearly tied to genotype; this strategy develops in response to social and environmental conditions related to disadvantage in social competition and will thus covary (across cultures, generations, and even within an individual lifetime), with variation in immediate social circumstances. (2b) To reduce the frequency of secondary sociopathy, the appropriate social response is to implement programs that reduce social stratification, anonymity, and competition, intervene in high-risk settings with specialized parent education and support, and increase the availability of rewarding, prosocial opportunities for at-risk youth. Since the genetics and life histories of primary and secondary sociopaths are so different, successful intervention will require differential treatment of different cases; we thus need to encourage the widespread adoption of common terminology and diagnostic criteria. ACKNOWLEDGMENTS I would like to thank Mr. Rainer Link, who helped me get started on this project, and who collaborated with me on the first version and first public presentation of the model (Link & Mealey 1992). I would also like to extend thanks to the many individuals who provided useful comments during the revision process: J. D. Baldwin, David Buss, Patricia Draper, Lee Dugatkin, Lee Ellis, Hans Eysenck, David Farrington, Hill Goldsmith, Henry Harpending, James Kalat, John Loehlin, Michael McGuire, Randy Nesse, Jaak Panskepp, David Rowe, Sandra Scarr, Nancy Segal, Chuck Watson, David S. Wilson, and four anonymous BBS reviewers. NOTES 1. As of January 1, 1996 Linda Mealey will be at the Dept. of Psychology, University of Queensland, Brisbane, Australia 4702. Email: lmealey@psych.uq.edu 2. Plutchik's eight primary emotions are: anger, fear, sadness, disgust, surprise, joy, acceptance, and anticipation. Others posit a few more (Izard 1977; 1991) or fewer (Ekman 1971; Panksepp 1982) but what is basically agreed is that primary emotions are those which can be found in other mammals, are hard-wired in the brain, are reflexively produced in response to certain stimuli, are associated with certain, sometimes speciesspecific, physiological responses (e.g., piloerection, changes in heart rate, facial expressions), and, in humans, are found crossculturally and at an early age (see Ortony & Turner 1990 for a dissenting opinion). Note that the "social emotions," including love, guilt, shame, and remorse, do not meet the above criteria, and are not considered to be primary emotions by most authors (see Izard 1991 for another perspective). Although distinctly human, the social emotions seem to involve a critical element of learning, and, central to the argument I will be making, are not panhuman. 3. According to Plutchik, cognitive processes themselves evolved "in the service of emotions," "in order to make the evaluations of stimulus events more correct and the predictions BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 539 Mealey: The sociobiology of sociopathy more precise so that the emotional behavior thatfinallyresulted would be adaptively related to the stimulus event" (1980, p. 303). This model of the relationship between emotion and cognition is somewhat similar to Bigelow's (1972), which postulates that intelligence evolved as a result of the need to control the emotions (especially the aggressive emotions), in the service of sociality, and Humphrey's (1976; 1983), which claims that selfawareness evolved because it was a successful toolforpredicting the behavior of others. 4. See Draper (1978) and the 1986 special issue of Ethology and Sociobiology on ostracism for further discussion of the role of shunning with specific reference to human societies; see Hirshleifer & Rasmusen (1989) for a game theoretic model of shunning; and see Nathanson (1992) for the importance of the social emotion shame. 5. The wealth of literature on strategies that people use to detect deception in interpersonal interactions, as well as the technologies that have been developed in order to further enhance that ability in less personal social interactions, are indicators of the importance we bestow on such ability (see Zuckerman et al. 1981, Mitchell & Thompson 1986, and especially Ekman 1992). 6. Although the data are overwhelming, the particular articles cited in this section should not be considered to be independent reports, since most of the reviews cited overlap substantially in their coverage and many authors or teams report their findings more than once in a series of updates. Interested readers should direct themselves to the most recent publications; however, older publications do contain some information not presented in the updates and are thus included for thoroughness and ease of reference. 7. Twin study methods yield estimates of what is termed "broad heritability," which includes both "additive" genetic factors (i.e., the summed effect of individual genes on the phenotype) and "non-additive" genetic factors (i.e., the phenotypic effects of dominance interactions between homologous alleles on paired chromosomes, and the epistatic interactions between nonhomologous genes throughout the genome). Adoption study methods, on the other hand, yield estimates of what is termed "narrow heritability," which is only the additive genetic component. The additive component is that which can be selected for (or against) as it is transmitted from generation to generation, whereas the nonadditive effect is unique to each individual genotype and is broken and reshuffled with every episode of sexual recombination. Because of this difference, twin studies typically yield higher heritability estimates than adoption studies. [See also Plomin & Daniels: "Why Are Children in the Same Family So Different from One Another?" BBS 10(1) 1987; Plomin & Bergeman: "The Nature of Nurture" BBS 14(3) 1991; and Wahlsten: "Insensitivity of the Analysis of Variance to Heredity-Environment Interaction" BBS 13(1) 1990.] Another difference between the twin methodology and the adoption methodology is that twin studies generally provide heritability coefficients which estimate the proportion of the total explained variance accounted for by genetic factors, whereas adoption studies provide heritability coefficients which estimate the proportion of the total variance (including measurement error) that is accounted for by additive genetic factors. Since measurement error is so large when assessing criminality, adoption studies tend to yield both smaller and more varied heritability estimates than do twin studies. A third difference is that twin studies yield heritability estimates for members of a particular generational cohort, usually tested at the same age, whereas adoption studies necessarily regress measures from one generation to another. This leads to two problems in interpreting heritability estimates derived from adoption studies that are not germane to twin studies. The first is that heritability can change across generations - even in the absence of genetic change - due to changes in the environment; this effect cannot be assessed in either twin or adoption 540 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 studies, but is only a limitation of generalizability for the former, whereas it is conflated in the latter. The second is that heritability can also be different at different ages. Huesmann et al. (1984), for example, report that the correlation between children's aggression level and their parents' aggression level when measured at the same age, is greater than the correlation between the child's own aggression level at one age and at a later age. This phenomenon also results simply in limited generalizability of heritability estimates derived from twin studies, but yields conflated estimates from adoption studies. The heritability of 0.6 reported herein is an estimate of broad heritability as derived directly from twin studies; similar estimates can also be calculated indirectly from adoption study data after accounting for measurement error, but cohort effects cannot be separated out. See Loehlin (1992) for a general discussion of twin and adoption methodologies and Emde et al. (1992) and Raine (1993) for further discussions of the relevance of methodological considerations as they pertain to interpretation of the specific studies summarized herein. 8. There is also evidence that at least one form of alcoholism belongs to the sociopathy spectrum: Type II alcoholism, which is also much more prevalent in men than women and seems to be transmitted in the same way (Bohman et al. 1981; Cadoret 1986; Cloninger et al. 1978; 1981; McGue et al. 1992; Stabenau 1985; Zucker & Gomberg 1986). Type II alcoholism is characterized by early onset, frequent violent outbursts, EEG abnormalities, and several of the personality attributes that are often seen in sociopathy: impulsivity, extraversion, sensation seeking, aggressiveness, and lack of concern for others (Cloninger 1987b; Tarter 1988). 9. The interesting phenomenon of differential heritability of traits across the sexes can occur, as in this case, as a result of differential (sex-limited) expression of the same genes or, as it does with Type I alcoholism (a milder, nonviolent form), as a result of differential environmental experiences of the sexes (Cloninger et al. 1978). Since heritability is measured as a proportion, the value of a heritability estimate will change whenever the numerator (variance in a trait due to genetic variance) or the denominator (total variance in the trait changes). Since the denominator (total variance) is composed of both genetic and environmental variance, changes in either will change the heritability. This method of defining heritability also explains some other apparent paradoxes, such as how two populations (e.g., racial groups or two successive generations of a single group) could have exactly the same genotypic variation with respect to a trait, but because of differences in their environments, exhibit differential phenotypes and differential heritability of the trait. 10. Like the behavior-genetic studies cited in section 2.1, these studies provide overwhelming data, but should not be considered as independent reports, because many overlap or update earlier work. Methodologically, although path models and the longitudinal studies from which they are derived have excellent ecological validity, they are correlational; although they improve upon cross-sectional designs by noting which factors precede others developmentally, they cannot completely sort out cause and effect - especially in the earliest stages of parent-child interactions. 11. The strategy of sociopaths against one another, although not a test of the current model, is still interesting in its own right. In the Hare and Craigen (1974) modified Prisoner's Dilemma, the majority of sociopaths, in their turn, chose from amongfive"plays," the choice that minimized their own pain (an electric shock) for that trial, but that maximized their partner's. Since partners took turns in selecting from the samefive"plays," this strategy actually maximized pain over the long run. The alternative, pain-minimizing strategy, involved giving both oneself and one's partner a small shock - a choice that most subjects declined to use. This result seems to confirm the sociopath's inability to consider anything other than the immediate consequences of an act, as well as the ineffectiveness of delayed Commentaryi'Mealey: The sociobiology of sociopathy punishment or threat of punishment as a deterrent. In the Widom (1976a) study, sociopaths did not, in general, "defect" more often than the controls, but in the condition when subjects were informed of their partner's move on the previous trial, sociopaths were much more likely than controls to "defect" after a mutual cooperation. On this measure, at least, the sociopaths seemed to demonstrate an inability to "commit" to an ongoing cooperative relationship. [See also Caporael et al.: "Selfishness Examined" BBS 12(4) 1989.] 12. Terhune (1980) reports that personality is the most important factor for strategy choice within the setting of single-trial Prisoner's Dilemma interactions. In multiple-trial interactions, however, when players have the opportunity to learn one another's dispositions, situational factors are more important for determining play (see Frank et al. 1993). This is consistent with the idea that the establishment of reputation is a key goal, even for players who on a single trial would choose not to cooperate. For more on the idea that establishing a certain reputation within one's referent group is a conscious goal and how that might play a role in the development of antisocial behavior, see Hogan and Jones (1983) and Irons (1991). 13. For some of the debate on this issue see the series of comments and replies following Nesse (1991) in the electronic journal Psycoloqtty. The comments specifically addressing the relationship between mood and emotion are: Morris (1992), Nesse (1992a), Plutchik (1992), and Nesse (1992b). 14. While searching for data to test this prediction, I came across only the Robins et al. (1991) reference in support of it, and one reference in an introductory psychology text (Wade & Tavris 1993) against it. The latter stated that antisocial personality disorder "is rare or unknown in small, tightly knit, cooperative communities, such as the Inuit, religious Hutterites, and Israelis raised on the communal plan of the kibbutz" (p. 584). Contact with Dr. Tavris allowed me to follow up on the sources from which the latter statement was derived (Altrocchi 1980; Eaton & Weil 1953; and Montagu 1978). My conclusion (which is shared by Dr. Tavris in personal communication) is that the absence or rarity of sociopathy in these small, tightly knit societies is not a result of the creation of a social system in which sociopaths never develop; rather, it is that secondary sociopaths do not develop (keeping total numbers at the low baseline) and that primary sociopaths emigrate. Small, closely knit societies have all the properties that game theoretic models indicate will reduce (but not eliminate) the incidence of the cheater strategy (see sect. 3.2.1). One of the most important of these features is size per se; the cheater strategy cannot be used repeatedly against the same interactors and remain successful (see sect. 1.2). Thus, in small societies, sociopaths are likely to do their damage, acquire a reputation, and leave - to avoid punishment and move on to greener pastures. This "roving strategist" model (Dugatkin 1992; Dugatkin & Wilson 1991; Harpending & Sobus 1987) allows for both the evolution and the maintenance of a low baseline of successful sociopaths even in small groups (like those in which we presumably evolved). 15. Despite being a genetically based strategy, because primary sociopathy is the endproduct of the additive and interactive effects of many genes, we will not be able to predict or identify individual sociopaths by knowledge of their genotype. We will, however, be able to predict which children will be at risk, given their genetic background, the same way we predict which children will be at risk given their familial and sociocultural background. We will also be alerted to the need to differentiate between diagnoses of primary sociopathy and secondary sociopathy (and our consequent approaches to them) based upon knowledge of an already identified sociopath's genetic and environmental background. 16. Axelrod (1986) and Boyd and Richerson (1992) also consider the extension of punishment not only to cheaters, but to those cooperators who do not, themselves, punish cheaters. The presence of this strategy can lead to an ESS of practically any behavior, regardless of whether there is any group benefit derived from such cooperation. Clearly this extension of the model has some analogues with totalitarian regimes and ingroups of a variety of sorts. Open Peer Commentary Commentary submitted by the qualified professional readership of this journal will be considered for publication in a later issue as Continuing Commentary on this article. Integrative overviews and syntheses are especially encouraged. Testing Mealey's model: The need to demonstrate an ESS and to establish the role of testosterone John Archer Department of Psychology, University of Central Lancashire, Preston, Lanes, PR1 2HE, England. j.archer@uk.ac.uclan.p1 Abstract: Two specific aspects of Mealey's model are questioned: (1) the application of the concept of Evolutionarily Stable Strategy to all alternative strategies, including those that involve reduced lifetime reproductive success; and (2) the evidence for the dual role of testosterone, which is based mainly on studies of a modulating effect on aggression. The case presented for sociopathy being an ESS (evolutionarily stable strategy) is a subset of the general case where a minority of individuals are successful because the majority behave in a different way. It is true that those who do not follow the rules of the majority may have evolved to behave like this, forming a genuine equilibrium or ESS in the population, in which case the two forms will have equivalent lifetime reproductive success (Dominey 1984). This would fit the case argued for primary sociopathy. An alternative is that the minority strategy is the best available in the current circumstances; in this case, lifetime reproductive success may be much lower than that of the dominant strategy (Dominey 1984) and is not an ESS. This would appear to fit the case argued for secondary sociopathy. To pursue the argument that primary sociopathy is a genuine ESS, a cross-cultural comparison would seem to be necessary. Cross-cultural psychologists have described different cultures in terms of individualism-collectivism (Triandis et al. 1988). In a traditional collectivist society there will presumably be a greater value placed on a reputation for cooperation, so that cheats will be shunned. In hunter-gatherer societies this would presumably apply to a greater extent, and such circumstances will be nearer the social arrangements of evolving hominids. In contrast, among individualist modern industrial societies, with looser family networks and many more interactions with strangers, the opportunities for cheating - Machiavellianism and sociopathy - will be much greater. It would therefore be instructive to compare societies that differ in terms of the extent to which cheats would be expected to prosper. We would need to show that there is indeed a stable proportion of sociopaths in traditional societies, and that they are on average as reproductively successful as nonsociopaths, for the ESS explanation to be established. Mealey suggested a dual role for testosterone. It is first described as "a likely candidate for the role of trigger of the sexlimited activation of genes required by the two-threshold model" (sect. 2.3.3., para. 7). This action is described as an organizer role that affects traits, and contrasted with a second BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 541 Commentaryl Mealey: The sociobiology of sociopathy role for the hormone, as "an activator (affecting states)." Some clarification is required. When considering hormonal influences, the term "organizer" usually implies a neonatal influence (of testosterone). However, Mealey used an analogy with secondary sexual characteristics such as beard and breast development: genes for these are present in both sexes but are only activated by the appropriate hormonal environment. I presume from this analogy that the term organizer was intended to refer to an interaction between particular genes and testosterone at puberty to produce a certain pattern of behavior, namely, a threshold effect as opposed to a modulating effect, which varied according to the level of testosterone. This would, however, be difficult to reconcile with the claim that the tendency to engage in antisocial behavior is stable from early childhood. In support of a dual role for testosterone, Mealey cited studies of adults (mainly men) linking testosterone with aggressive behavior, and with other forms of antisocial behavior and sensation seeking. The evidence on these topics, of which most concerns aggressiveness, is almost entirely correlational. It therefore fails to address the issue of a threshold effect, only a modulating one. Overall, there is a correlation between various measures of human aggressive behavior and testosterone level of 0.38 (Archer 1991), providing suggestive evidence for a modulating role for the hormone. Since successful aggressive and competitive behavior can lead to increased testosterone levels, alternative interpretations are possible, in particular, that testosterone levels are largely the result of status-related and other activities (Kemper 1990). One difficulty with suggesting any major causal role for testosterone in aggression — even with the addition of positive feedback to take account of the two-way link - is that aggressiveness as a trait shows stability from childhood to adulthood (e.g., Olweus 1979). Thus, the onset of testosterone secretion at puberty must interact with an already-established disposition. Animal studies have shown that past learning associated with aggression can override the effect of testosterone on aggressiveness. To understand the sequential effects of all possible influences, including hormonal ones, on a particular behavioral trait requires a developmental perspective (Archer 1994). This is more complex than the causal model adopted by Mealey. Mealey stated that, "Variation in testosterone levels also parallels the age variation in the expression of sociopathic behavior and is correlated with such behavior in adolescent and adult males" (sect. 2.3.3., para. 7). The phrase "such behavior" is misleading as it apparently widens the scope of the discussion, but the references that followed mostly concerned aggressiveness. It is not clear how this is related to other attributes known to be associated with testosterone, such as impulsiveness and sensation seeking, or indeed to sociopathy itself. To test the specific hypothesis put forward by Mealey we would need to establish that testosterone is primarily associated with sociopathy rather than with aggressiveness or some other disposition. Abstract: Mealey's excellent target article rests on several assumptions that may be questioned, including the overarching assumption that sociopathy reflects the failure of a small minority of males to cooperate with the larger group. I suggest that violent competition in ancestral bands - and not "cheating" in the "game" of cooperation - was the primary evolutionary precursor of sociopathy. Today's violent sociopath is far more a "warrior hawk" than a failed cooperator. primary and secondary sociopathy, and in generating thoughtful and convincing recommendations for reducing sociopathic behavior and crime in general. Her excellent target article, however, rests on several implicit assumptions that can be questioned. Among these are the notion that both ancestral and current forms of social organization are overwhelmingly cooperation-based, that sociopathic outputs emanate primarily from rational cost-benefit calculations, that intelligence is not a major mediator of such calculations, and that game theory is the appropriate metaphor for understanding and predicting sociopathic behavior. Space does not allow me to challenge all these issues, so I will focus primarily on Mealey's assumption that sociopathy is largely a failure of a small minority of human males to cooperate with the larger group. I will proceed on the basic assumption that violent competition in ancestral bands - and not "cheating" in the "game" of cooperation - was the primary evolutionary precursor of sociopathy. It is likely that aggressive intermale competition for status, females, and other resources was prevalent in the EEA (environment of evolutionary adaptation), and some degree of predatory violence was required in the seek and kill aspects of hunting large game animals. Moreover, it is likely that violent interband competition was present in the earliest phases of human evolution and a given group's number of healthy, adventurous, and potentially violent young men was a major key to survival. The absence of a contingent of "dawn warriors" (Bigelow 1969) in readiness would be disastrous in the event of attack by an invading party composed entirely of such men. Although it is fashionable to ascribe intergroup conflict, crime, and male violence in general to the exigencies of modernity, the common ancestor to the primate and human line has been described as sexually dimorphic, socially aggregated into closed networks, and prone to hostile intergroup encounters among males (Wrangham 1987). In this scenario, the issue of "cheaters" in the game of cooperation is transformed into the game of "hawks" and "doves" (Maynard Smith 1982), where doves always lose when one strategy is pitted against the other (Archer 1988). Within groups, doves fare better against doves than hawks do against each other, and a frequency-dependent, evolutionary stable point is theoretically possible where the competing classes are roughly equal in their respective fitnesses. But the fact remains that doves always lose in head-tohead competition with hawks. The only solution is for the cooperating majority of doves to be protected by a subclass of "warrior hawks" that is capable of repelling invasion by a similar contingent from another group. The price that each group pays is having to tolerate, within groups, the disruptiveness, callousness, bullying, and manipulativeness of the warrior class. What might be the defining characteristics of these warrior hawks? Mealey has done an excellent job of enumerating them in detail: coldness, detachment, manipulativeness, egocentrism, absence of social emotions, lack of empathy, fearlessness, deceptiveness, life history of "predatory social interactions," orientation to the present, strong need for respect and reputation among peers, extraversion, frequent "cheating," adventurousness and risk taking, impulsiveness, pleasure seeking, resistance to socialization, moral deficiency, minimal responsiveness to reward/punishment, approach orientation, high Machiavellianism, and proneness to anger and aggression. These are exactly the traits necessary on the battlefield, where an inherent "delight in destruction" fuses with a desire to eradicate the hated enemy, and the warrior's "hunting impulses are released to seek the most dangerous of all beasts" (Gray 1970, p. 149). In deadly combat with hated and feared enemies, the traits of the warrior reach fruition and the most brutal warrior hawks have the advantage over their less "sociopathic" adversaries; it is in peacetime within groups that the suite of warrior traits becomes problematic. Mealey has performed a great service in reviewing current knowledge on sociopathy, in carefully distinguishing between As Bigelow (1972) tells us, intergroup conflict has been with us since the dawn of humanity, and indeed, the success of the The sociopath: Cheater or warrior hawk? Kent G. Bailey Department of Psychology, Virginia Commonwealth University, Richmond, VA 23284-2018 542 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 Commentaryi'Mealey: The sociobiology of sociopathy human line was predicated on potential for aggressive group response, when necessary, within the context of in-group solidarity, cooperation, and intelligent self-control. It is particularly important that a social group keep its young, aggressive, and impulsive males - that is, potential warrior hawks - under control. From this perspective, violent, predatory crime in modern contexts reflects a loss of control over potential warriors who come to treat in-group members, even their neglectful and abusive family members, as outsiders and enemies to exploit, abuse, and even destroy in extreme cases. This they do with a coldness and callousness that is impossible for the cooperative doves in the group to understand. Moreover, cooperative, lawabiding doves often further their own exploitation by providing sympathy and forgiveness to those who prey upon them without empathy or remorse. At several points, Mealey refers to sociopathy as abnormal and pathological, and psychiatry has traditionally classified it as a disorder. Certainly, the behavior of the sociopath is undesirable and destructive in peaceful and cooperative contexts, but the issue of abnormality is problematic. The sociopath does not see himself as abnormal, psychiatric symptoms are minimal or absent, and any pathology is social and interactive rather than internal (MacMillan & Kofoed 1984). Reasoning backwards to the EEA, the warrior traits that underlie violent crime today were not only normal in ancestral contexts but were necessary for survival in an atmosphere of occasional interband conflict. Presumably, in the small group atmosphere dominated by older males past their warrior prime, young males were limited in the antisocial options available to them in the in-group context. As Mealey implies in her argument, such natural controls over male adventurism and exploitation are absent today in our crowded, poverty-ridden, and socially decaying urban environments, and it is no surprise that rootless and discouraged young men will sometimes fall back on the ancient warrior hawk option. Mealey is right that sociopathy does involve cheating in cooperative contexts, but the overriding concern of most people in modern society is not with lying, misrepresentation, and loss of material resources; rather, our foremost concern is fear of violent crime in the form of gang warfare and drive-by shootings, carjackings, muggings, burglary and forcible entry, stranger rape, child abuse/ molestation, and various other threats to life and limb. Our concern is heightened by knowledge that the "warriors without portfolio" in our midst are becoming more numerous, more violent, more addicted to substances that exacerbate their violent traits, and increasingly alienated from family, community, cultural tradition, educational opportunities, and other constraining influences. For many of today's warrior hawks, anyone not in their socially marginal peer group is the "enemy" and the status of hero is conferred upon those with the most "wins" on the urban battlefield. Continua outperform dichotomies John D. Baldwin Department of Sociology, University of California at Santa Barbara, Santa Barbara, CA 93106. baldwin@alishaw.uscb.edu Abstract: Mealey's data do not support her dichotomous model of primary and secondary sociopathy; this data supports the view that there is a continuum of degrees of sociopathy, from zero to the maximal manifestation. There are multitudes of factors that can contribute to sociopathy - from both genetic and environmental sources - and the countless different mixes of them can produce multiple degrees and variations of sociopathic behavior. Mealey does a good job detailing many facets of sociopathy, including the role of sensory stimulation as a powerful reinforcer for people with hypoaroused nervous systems. The role of empathy - or lack thereof - is of central importance, too. I applaud her attention to both nature and nurture, though I detect a favoritism for nature instead of a balanced respect for both nature and nurture (see below). In the introduction, Mealey states that she hopes "to convince the reader that the distinction between primary and secondary sociopaths is an important one because there are two different etiological paths to sociopathy." I argue that her data do not support a dichotomous model as well as they support the view that there is a continuum of degrees of sociopathy, from zero to the maximal manifestation. Mealey correctly notes that there can be multiple genetic and multiple environmental inputs that operate together, and it seems to me that this approach clearly suggests that different mixtures of genetic and environmental inputs can allow people to fall at many different positions on a continuum of sociopathy. First, in section 2.1.3, Mealey argues that sociopathy is polygenetic and that each individual can inherit a different "dose" or "genetic load" of the genes that predispose one for sociopathy. In these pages, it is artificial to argue for a twothreshold model. Her suggested thresholds are merely hypothetical constructs that must be introduced if she hopes to create a dichotomous model rather than a continuum model. I believe that the many different possible genetic loadings of the polygenetic inputs can lead to numerous different genotypes with a continuum of different levels of predisposition to sociopathy. Second, in section 2.4.2, Mealey lists multiple environmental risk factors that can increase the chances that a person becomes a sociopath. Later we learn that there are also multiple environmental protective factors - such as good parenting, exciting prosocial jobs, and so on - that can steer high sensation-seeking individuals away from antisocial behavior. Why not simply create a model that integrates nature and nurture in a straightforward (and more parsimonious) manner, stating that the higher a person's genetic load of predisposing genes is, the more likely the person is to learn sociopathic behavior if exposed to risky environmental factors without adequate protective factors. If sociopathy emerges as a combination of various genetic loadings and exposure to countless developmental risk and protective factors, it is easy to postulate a continuum of sociopathy with individuals distributed mostly near the low-end of the continuum. Since only a small number of individuals would receive both a heavy genetic load and experience the maximal number of risk factors needed to produce maximally sociopathic behavior, I would expect to see a skewed bell-shaped curve that peaks near the low-end of the sociopathy continuum. Thus, extreme sociopathy (which Mealey calls "primary") would be rare - and the most intractable. Numerous levels of intermediate sociopathy would exist, being more common and more treatable than the few extreme cases. Seeing a continuum of possibilities (based on both nature and nurture acting in concert) allows for a finer grained and more sensitive analysis of sociopathy than does Mealey's dichotomous model. The logic Mealey uses to create the dichotomy also tends to elevate genetic causes to a special status based on her a priori assumption that primary sociopaths can only be understood in terms of "a genetically determined strategy" (sect. 3.1.1). Mealey often writes with an "either-or" logic, saying, for example, that some people are emotionless and others are not. Perhaps the dichotomous model (of primary and secondary sociopathy) has emerged as an artifact of either-or logic. A multiple factor model - that recognizes both multiple polygenetic causes and countless environmental risk and protective factors - helps one avoid either-or logic and think in terms of complex nature-nurture interactions with a continuum of outcomes ranging from zero to maximal manifestations of any behavior. Nondichotomous models also sensitize us to look for the many types of socialization that can lead to high Mach scores, calcuBEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 543 Commentary I'Mealey: The sociobiology of sociopathy lated thinking, and deficits in empathy. Since each component of sociopathy is only partially correlated with the others, many other types of socializations can lead to variable levels of each one. I believe we need to examine each individual separately, to identify sensitively as many of the developmental risk and protective factors as possible. If some individuals appear to be very susceptible to becoming sociopathic - and do so in a protective environment with little exposure to risk factors - we can hypothesize that they have a strong genetic loading of the predisposing genes. If physiological markers or measures are available to identify the level of genetic load, so much the better. In section 3.1.1, and elsewhere, we see Mealeys tendency to think of "primary" sociopaths as professional gamblers, emotionless cheaters, and incorrigible criminals. Let us not forget that they might also become TV evangelists, lawyers, politicians, driven scientists, and others who ruthlessly pursue their career without being thought of as criminals - though those who know them might note a calculatingness and social insensitivity in their actions. In section 3.2,1 see no treatment benefits from dichotomizing sociopathy. At present, we cannot change a person's genetic load for the behavior. All we can do for both of Mealey's "primary" and "secondary" sociopaths is try to make proactive and reactive environmental interventions: first, we can proactively try to reduce the risk factors and increase the protective factors that affect the development of the behavior; or, second, for those who already exhibit sociopathic behavior, we can use punishment or the threat of punishment for antisocial behavior along with rewards for socially acceptable behavior. Thinking of sociopathy in terms of a continuum (not just primary or secondary types) alerts us to the multiple variables we need to attend to when proacting or reacting to any given individual's sociopathic possibilities. Sociopathy, evolution, and the brain Ernest S. Barratt and Russell Gardner, Jr. Department of Psychiatry and Behavioral Sciences, University of Texas Medical Branch, Galveston, TX 77555-0443. ernest.barratt@utmb.edu Abstract: We propose that Mealey's model is limited in its description of sociopathy because it does not provide an adequate role for the main organ mediating genes and behavior, namely, the brain. Further, on the basis of our research, we question the view of sociopaths as a homogeneous group. Although Mealey deals at length with sociobiology and impulsive mechanisms of genes and their resultant behaviors, she does not deal in an evolutionary sense with the organ through which the genes and behavior must be mediated: the brain. She discusses neurochemistry related to personality traits and correctly describes psychophysiological factors "as significant causes, not just correlates" of sociopathy. Yet she does not discuss from an evolutionary viewpoint, relevant neuropsychological constructs and selected brain structures that relate to sociopathy (or psychopathy or antisocial personality disorders). Since impulsiveness and impulse control are key constructs in the definition of behavioral control, and since these constructs have been related to selected brain areas, a discussion of the development of the brain and related behavioral control or inhibitory mechanisms would seem to be an appropriate and essential part of the development of her model. For example, damage to the orbital frontal cortex almost invariably results in impulse control problems. The personality trait of impulsiveness has also been related to frontal cortex functions (Barratt 1993). Damasio et al. (1990) have written on "acquired sociopathy" stemming from brain damage in the orbital frontal region. 544 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 Thus, cross-species comparisons of frontal cortex functions related to behaviors similar to those of psychopaths should provide data to strengthen and extend her argument. This is not to argue that sociopathy is a brain disease; but caution should be used in relying on population statistics and conceptual models such as the Prisoner's Dilemma as ultimate and sufficient explanations of complex constructs like psychopathy. Some personality traits such as impulsiveness may have had (and may still have) adaptive functions among persons with intact brains. Acting "automatically" or without reflection may be helpful in emergency situations. For example, it was estimated in a recent air crash that the pilot had less than five seconds to observe and respond appropriately to a warning light. Over the span of animal evolution, acting quickly to protect territory or other natural goods has always been at a premium. Perhaps this was true long before mammals were primates, or for that matter, before vertebrates were mammals. The population biology of sociobiologists often overlooks the length of time that Darwinian mechanisms may have been operating. In modern society, to react quickly and consistently without thinking can be a handicap when coupled with other personality traits (e.g., the high impulsiveness and low anxiety seen in psychopaths) or in situations that call for more reflection. Again, we suggest that Mealey's model should include more of a role for the main organ through which the genes and behavior must be mediated, namely, the brain. In a study of impulsive aggression in our laboratory (currently being prepared for publication), we studied prison inmates, all of whom satisfied the criteria for antisocial personality as defined in DSM-III-R and measured by a standard interview (PDI-R). These inmates were all young male recidivists. Based on formal disciplinary reports of aggressive acts, we classified the subjects into impulsive aggressive and premediated aggressive groups. Again, each aggressive act was classified using a standard interview. The measures of these aggressive acts were normally distributed along a continuum from completely premeditated to completely impulsive. We had hypothesized that the inmates who committed impulsive aggressive acts would have higher levels of impulsiveness and anger/hostility personality traits than those who committed the premeditated acts. Both groups of inmates had significantly higher levels of impulsiveness and anger/hostility than noninmate controls (matched for sex, age, education level, and race). The two inmate groups did not differ significantly in these personality traits. They did differ significantly, however, in cognitive psychophysiological measures of information processing during the performance of a simple choice task. The results implicated the frontal cortex and several other cortical areas. A frequency analysis of the EEGs (electroencephalograms) suggested that anticonvulsant medication might be helpful in controlling impulsive aggression among the inmates. The data to date are consistent with this suggestion. Within Mealey's model, these inmates would also satisfy the criteria for sociopaths. Yet, there were two distinct groups among them that expressed aggression in different ways and had significant differences in brain functions. Mealey's model suggests primary sociopaths are a homogeneous group. The above data raise questions about such homogeneity. You can cheat people, but not nature! John Barresi Department of Psychology, Dalhousie University, Halifax, Nova Scotia B3H 4J1, Canada. |barresl@ac.dal.ca Abstract: The psychological mechanisms implicated in psychopathy do not limit their activity to those behaviors that support a cheater strategy in social games. They result in a number of other clearly maladaptive behaviors that do not directly involve other individuals. Thus, any gains Commentaryi'Mealey: The sociobiology of sociopathy that might arise from the use of a cheater strategy in social situations are probably lost elsewhere. The primary sociopath, or psychopath (Cleckley 1941; Hare 1986), exhibits behaviors easily interpreted as a cheater strategy in a reciprocal altruistic game. As such, the psychopath seems to have a marginally adaptive strategy in social interactions with other humans. But can the psychological mechanisms implicated in psychopathy limit their activity just to those behaviors that support the cheater strategy in social interactions? I think not. As I will show shortly, the same psychological mechanisms also cause the psychopath to engage in a number of behaviors not directly involved in social interactions with others that are clearly maladaptive. Thus, the real question is whether there is a net gain for the psychopath when some of his behaviors that do not involve other agents in social games are maladaptive as compared to nonpsychopaths, while other behaviors that do involve other agents are adaptive as a low-frequency cheating strategy in reciprocal altruistic games. This is a more difficult question to answer; however, I would guess that the psychopath obtains at best a net gain of zero from the costs that go along with the benefits of this strategy. Whatever the psychopath gains from reciprocators in direct competition through the use of a cheater strategy, he loses to these very same people through the indirect competition with them in situations not involving social games. According to Mealey, the gains from cheating that the primary sociopath obtains occur because he takes a short- rather than long-term view of social interaction, choosing immediate gain over long-term gain through reciprocal altruism. She suggests that this occurs because the primary sociopath lacks the social emotions that bind individuals to each other, producing cooperation through time. It is the social emotions such as shame, guilt, sympathy, and love that are an essential part of the psychology that maintains the human social adaptation of reciprocal altruism, or of a cooperative strategy over multiple Prisoner's Dilemma-type social situations. The psychopath, lacking these emotions, is inclined not to cooperate over the long term with others, but prefers to cheat by taking the dominant defect strategy in Prisoners Dilemma-like social situations. 1 do not wish to debate with Mealey over whether the primary sociopath, or psychopath, lacks these social emotions or whether the lack of these emotions might provide an advantage to the psychopath through leading him to defect in Prisoner's Dilemma-type social situations. Indeed, the psychopath exhibits other related behaviors that seem to optimize the success of this strategy, such as his tendency to wander from one population to another so as to reduce multiple interactions with the same individuals (Harpending & Sobus 1987). However, there is every reason to think that the psychopath lacks not only the relevant social emotions but also other emotions that inhibit self-interested behavior even when other individuals are not involved. For example, it has long been known that psychopaths do not exhibit the usual physiological correlates of anxiety or fear when they know that they are about to be shocked (e.g., Hare 1986) and that this apparent emotional deficit reduces their capacity to learn responses that avoid both shocks and certain social punishments (Lykken 1957; Schmauk 1970). Such a general deficit in passive avoidance learning relative not only to other humans but to many other organisms (e.g., Mineka & Zinbarg 1991) can hardly be viewed as an adaptive consequence of psychopathy. It is a cost that must be balanced by the success of cheating in social situations. But there are further difficulties associated with the lack of emotions in psychopaths that must also be considered, in particular, their lack of prudent behavior concerning future consequences of current activity. Even when other individuals are not involved, the psychopath fails to show concern about his own future. Especially in approach-avoidance conflict situations, the psychopath is unwilling to delay immediate gratifica- tion for a better outcome later (e.g., Newman et al. 1992). Although he is quite capable of calculating what will happen in the future, he fails to take a sufficient interest in it to be motivated by the consequences of present actions for his future self. William Hazlitt pointed out long ago that it is the same sympathetic imagination that "must carry me out of myself into the feeling of others . . . by which I am thrown forward as it were into my future being and interested in it. I could not love myself, if I were not capable of loving others" (Hazlitt 1805/1969, pp. 2 - 3 ; Martin & Barresi 1995). In the case of the psychopath, just as he fails to show sympathy for other people's interests, he also fails to show sympathy for the interests of his future self. And again we can ask whether the direct payoffs from the cheater strategy offset these losses due to this lack of sympathy by the psychopath for his own future self. [See also Logue 1988; Rachlin 1995.] My suspicion is that if the psychology of the psychopath is at all successful as a low-frequency strategy for human social interaction, it has developed through a form of genetic drift, where gains that are made through cheating are offset by the losses in noncheating situations. Turned around, what this means is that the success of the cheating strategy allows an intrinsically maladaptive behavioral disorder, psychopathy, to exist in higher frequency than it normally would. This possibility makes one wonder whether there are other psychopathologies that are maintained at fairly high frequencies through such "secondary gains" of the pathology. Conversion hysteria might be a prime example of just such a psychopathology. Secondary sociopathy and opportunistic reproductive strategy Jay Belsky Department of Human Development and Family Studies, College of Health and Human Development, Penn State University, University Park, PA 16802. jxb@psuvm.psu.edu Abstract: Mealey s analysis of secondary sociopathy has much in common with Belsky, Steinberg, and Draper's (1991) evolutionary theory of socialization. Both draw attention to the potential influence of early rearing in the promotion of a cold, detached, manipulative, and opportunistic style of relating to others and, in so doing, raise the question of whether secondary sociopathy represents a facultative reproductive strategy. The assertion that a cold, detached, manipulative, opportunistic style of relating to others is, in some cases, the result of contextual stresses and rearing practices during childhood has much in common with an evolutionary theory of socialization advanced by Belsky et al. (1991). Building upon many of the same sociobiological foundations as Mealey, these theorists argued that humans have evolved to be responsive to their rearing circumstances in the service of reproductive goals (i.e., fitness). When parental care is harsh, rejecting, insensitive, and inconsistent, as it is likely to be when economic resources are limited and family stress is heightened, children are predisposed to develop insecure attachments to their parents, perceive others as untrustworthy and relationships as unenduring; in consequence, they develop an opportunistic (including aggressive, especially in males) advantage-taking style of relating to others (a sociopathic style?). This developmental trajectory, Belsky et al. argued, was part of a reproductive strategy designed to favor growth and mating over parenting, and thus was hypothesized to be associated with earlier timing of puberty, earlier onset of sexual activity, unstable pair bonds, and limited parental investment (see Fig. 1). Like Mealey, Belsky et al. (1991, p. 650) highlighted BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 545 Commentaryi'Mealey: The sociobiology of sociopathy TYPE I I TYPE I A. FAMILY CONTEXT Spousal harmony Adequate $ resources B. CHILDREARING Infancy / Early Childhood Sensitive, supportive, responsive Positively aflectionate Marital discord High stress Inadequate $ resources Harsh, rejecting insensitive Inconsistent Insecure attachment Mistrustful internal working model Opportunistic interpersonal orientation <t 9 Aggressive Noncompliant Anxious Depressed Early maturation / puberty Earlier sexual activity Short-term, unstable pair bonds Limited parental investment C. PSYCHOLOGICAL / BEHAVIORAL DEVELOPMENT D. SOMATIC DEVELOPMENT E. REPRODUCTIVE STRATEGY Secure attachment Trusting internal working model Reciprocally rewarding interpersonal orientation Later maturation / puberty Later sexual activity Long-term, enduring pair bonds Greater parental investment Figure 1 (Belsky). Developmental pathways of divergent reproductive strategies (Type I: Opportunistic; Type II: Mutually beneficial; Belsky et al. 1991, p. 651). genotype-environment interactions, specifically underscoring "differential susceptibility to environmental experience" in an attempt to explain why an association between contextual stress and an opportunistic reproductive strategy might not always obtain (for a related argument, see Wilson 1994). In fact, in discussing environmentally induced reproductive strategies, Belsky et al. further pointed out that some genotypes might be more reactive to specific environmental influences than others: Whereas some individuals may be genetically predisposed to respond to contextual stress and insensitive rearing by maturing early and engaging in relatively indiscriminate sexual behavior (and other manifestations of sociopathy), others may be genetically predisposed to respond to sensitive rearing by deferring sexual behavior and establishing pair bonds in adulthood. (1991, p. 650) When applied to Mealey's analysis, this observation raises the question of whether a sociopathy-inducing environment would foster (secondary) sociopathic behavior among all genotypes. To the extent that the answer to this question is no, it suggests that genotype-environment interaction is probably more complex than Mealey's analysis of sociopathy implies. Once it is acknowledged that (secondary) sociopathy as detailed by Mealey or the (highly correlated?) opportunistic reproductive strategy outlined by Belsky et al. may be the result of contextual conditions and rearing practices and that individuals may be differentially susceptible to these environmental influences, a number of interesting questions arise. We can ask first whether secondary sociopaths mature earlier than individuals with similar genotypes who are exposed to dramatically different rearing conditions. An affirmative answer to this query would be consistent with the notion that the secondary sociopaths of Mealey's analysis are likely to be the same individuals who are central to Belsky et al.'s opportunistic reproductive strategy. The explicit linking of these two sets of ideas highlights a second important issue, one that Belsky et al. raised but could not resolve. As applied to the topic of secondary sociopathy, it takes the following form: Is secondary sociopathy best conceptualized as a continuous or a categorical construct? To the extent that varying exposure to harsh, insensitive, and inconsistent rearing systematically fosters varying degrees of sociopathic 546 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 behavior, one must wonder whether Mealey is describing a specific form of psychopathology or a far more general behavioral and even developmental phenomenon. A third important issue with regard to secondary sociopathy concerns developmental timing and plasticity. Will secondary sociopaths, as a result of early rearing experiences, be resistant, like primary sociopaths, to many environmental efforts to modify their behavioral proclivities? In fact, might there be a point in development after which deflection from a sociopathic developmental trajectory would be less, rather than more, likely? The answer to these queries may themselves depend on the reply to the last question raised in this commentary about secondary sociopaths: Do the early experiences that promote secondary sociopathy induce in the victims of such rearing the same neurochemistry - high dopamine activity, low serotonin activity, and low norepinephrine activity - that characterizes sociopaths in general? That is, are primary and secondary sociopaths as similar in terms of brain chemistry as they might be in their overt behavior? To the extent that brain chemistry is influenced by early developmental experiences, the prospect arises that processes set in motion early in life may become selfsustaining for neurochemical reasons as much as for socialexperiential ones. And to the extent that neuronal pruning further characterizes processes of brain development pertinent to the understanding of secondary sociopathy, reservations might have to be raised about the prospect of ameliorating this psychopathological syndrome. Group differences ^ individual differences C. S. Bergeman and A. D. Seroczynski Department of Psychology, University of Notre Dame, Notre Dame, IN 46556. cbergema@lrlshmvs.cc.nd.edu Abstract: Mealey's etiological distinction between primary and secondary sociopathy blurs the delineation between individual and group differences. She uses physiological evidence to support her claim of genetic influences, neglecting variability within social classes, fre- Commentary/Mea\ey: The sociobiology of sociopathy quency of delinquent behavior in upper and middle classes (measured by self-report), and discontinuity of criminal behavior across the life span. Finally, Mealey's proposals for differential intervention fall short of a future agenda, which should tailor to individual needs, not social classes. Mealey presents a provocative model of the development of sociopathy, which blends research from a wide variety of disciplines. Unfortunately, the assumptions upon which the theory is based are often unsubstantiated; we will attempt to outline a few of the discrepancies. First, Mealey defines two types of sociopathy - primary and secondary. She contends that primary sociopathy is due to a strong genetic endowment and accounts for most of the sociopathic individuals found in the upper and middle socioeconomic classes. Secondary sociopathy, on the other hand, is "not as clearly tied to genotype" (sect. 3.3, para. 4) and can be explained by environmental risk factors such as high population density, competition for limited resources, poor parenting skills, and other factors often associated with low social class. Mealey contends that secondary sociopaths will almost always come from lower-class backgrounds, whereas primary sociopaths are found almost exclusively in the middle and upper classes. Unfortunately, to support this distinction Mealey has drawn the mistaken conclusion that physiological responses or differences are necessarily genetic. In other words, to support her contention that primary and secondary sociopathy have different underlying etiologies, Mealey has assumed that if physiological characteristics (which have been shown to have a genetic component) relate to behavioral differences between social classes (stronger relationship in upper-class individuals), then those differences must be due to genetic influences in upperand middle-class individuals and environmental factors in the lower class. For example, she cites research (Raine & Venables 1981; 1984; Satterfeld 1987) that indicates that physiological response differences between social classes have been associated with differences in antisocial behavior in children. Because these physiological variables (heart rate, skin conductance, EEGs) have been shown to have a strong genetic component, it is assumed that this is the "genetic link." These physiological differences found between social classes might not necessarily be genetically determined. Environmental variables such as familial stressors or systematic desensitization to violence might account for the observed physiological differences. A later work by Raine et al. (1990) found no relationship between socioeconomic status and psychophysiological measures, even though there was a relationship between central and autonomic measures of arousal at age 15 and criminality at age 24. Raine et al. conclude that "social and academic factors do not appear to be important mediators of the crime-arousal relationship" (p. 1006). A second problem is Mealey's lack of consideration of variability within social classes. A multitude of studies across the social sciences have focused on average group differences and have paid little attention to the relative amount of variability within the groups. This neglect of variability can inhibit both conceptual and empirical development in the study of behavioral phenomena. The problem is that average group differences are not necessarily related to individual differences, thus we will learn little about the description or explanation for individual differences in the development of sociopathy by studying group differences. In addition, research has indicated that the differences between individuals within a group are far greater than average differences between groups (Plomin 1990). If all we know about individuals is whether they come from the upper, middle, or lower socioeconomic class, we will know very little about their sociopathic tendencies. For example, the causes of individual differences in the development of sociopathological behavior could be entirely influenced by genetic factors; however, the average differences between social classes could be environmental in origin. Mealey violates this premise by assuming that delinquents from high-SES (socioeconomic status) families don't come from "bad homes.' Even if Mealey's theory is "right on the mark," and genetic factors are responsible for the development of primary sociopathy in middle- and upper-class families, namely, the extent to which the trait runs in families for genetic reasons, we would expect it to relate to parenting behavior and family life. That is, high-SES families of sociopathic kids should exhibit similar deviant tendencies. To date, no research has specifically tested the differences in genetic and environmental influences between high- and Iow-SES groups. A third concern with this model is that Mealey views unsocialized and socialized juvenile delinquents as precursors of primary and secondary sociopathy, respectively. One problem with this line of reasoning is that it relies on research that defines delinquent behavior based on police records and public criminal files. It is not surprising that these differences could be associated with social class, considering that there is an increased likelihood that lower-class individuals will be caught and convicted of crimes due to an uneven distribution of police or biased police handling (Sampson 1986; Staples 1987). Over 15 years ago, Tittle et al. (1978) refuted the argument that social class is tied to criminality. More recent work by Rowe (1983) has indicated that the use of self-report measures of delinquency show no social class differences. A fourth issue is that much of the work focuses on children and adolescents. What is the expectation for the other three quarters of the life-span? Research has indicated that less than half of juvenile delinquents become repeat offenders and the majority do not continue their criminal behavior in young adulthood (Farrington 1979; West & Farrington 1977). Moreover, some people do not commit their first crime until adulthood (Guttridge et al. 1983). In addition, research has indicated that criminal behavior in adulthood shows a stronger genetic relationship than does juvenile delinquency in adolescence (Plomin 1991). The Mealey model does not go far enough to explain the complexity of this behavior across the entire life span. Finally, Mealey states that "since the genetics and life histories of primary and secondary sociopaths are so different, successful intervention will require differential treatment of different cases" (sect. 3.3, para. 6). She has failed, however, to show that the genetic and environmental influences relating to these two types of sociopathy are different. She also makes the assumption that if trait is genetic, then it is immutable and nothing can be done to change the behavior, or that the best we can do is provide socially acceptable outlets for primary sociopaths. The result is her suggestion that interventions should be directed toward secondary sociopathy and not primary sociopathy. The one issue on which we do agree with Mealey is th'at a single intervention is unlikely to work for all individuals. But, rather than to direct differential interventions toward members of different social classes, we suggest the need to identify the individual maladaptive behaviors and to tailor intervention strategies accordingly. That is, if poor parenting strategies are identified, then the intervention should be directed toward parental education. Likewise, if social skill deficits or drug abuse are problems, then let us direct intervention at these issues. In conclusion, what the target article points out is the woeful lack of research in this area. Future research should be directed toward assessing genetic and environmental influences on primary and secondary sociopathy and the extent to which physiological indices associated with sociopathic behavior share the same underlying genetic and environmental influences. It BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 547 Commentaryl Mealey: The sociobiology of sociopathy should attempt to determine what intervention strategies are effective with each type of sociopathy, regardless of social class. Putting cognition into sociopathy R. J. R. Blair and John Morton MRC Cognitive Development Unit, London WC1H OBT, and Department of Psychology, University College, London WC1E6BT, England. ucjtrjb@ucl.ac.uk; John@cdu.ucl.ac.uk Abstract: We make three suggestions with regard to Mealey s work. First, her lack of a cognitive analysis of the sociopath results in underspecified mappings between sociobiology and behavior. Second, the developmental literature indicates that Mealey's implicit assumption, that moral socialisation is achieved through punishment, is invalid. Third, we advance the use ofcausal modelling to map the developmental relationships between biology, cognition, and behaviour. The target article is an interesting attempt to relate sociobiological theory, the physiological data on various antisocial populations, and criminal behaviour. During the attempt, Mealey joins the growing band of authors (e.g., Fagan & Lira 1980; Moffitt 1993) who distinguish between primary and secondary sociopaths. The groups display similar behaviour, which we might expect to be controlled by roughly equivalent cognitive structures. Finally, we are in total agreement with at least one of Mealey's descriptions of sociopaths, that they will acquire a theory of mind without access to empathic responding. Indeed, we have shown that psychopaths perform well on complex theory-of-mind tasks in the absence of physiological arousal responses to distress cues (Blair 1994). Our major problem with Mealey's work is the lack of a cognitive analysis; there is little reference to the kinds of cognitive structures that might mediate sociopathic behaviour. Indeed, some of the core features of her model, for example cheating strategies, are not considered at a cognitive level at all. We believe that if such an analysis had been carried out, some of the difficulties in Mealey's argument might have become clear. For example, Mealey claims that rape and spouse abuse are "genetically influenced, developmentally and environmentally contingent cheating strategies" (sect. 2.2.1, para. 4) that have been selected for. It seems unlikely that these behaviours per se are part of the phenotype; that motor programs for antisocial behaviours, such as spouse abuse, are laid down in the genotype. Thus, Mealey's position requires that some more general cheating mechanism has been selected for - what she calls the cheating strategy - which appears, from her argument, to correspond to some form of inborn personality trait. Presumably, given Mealey's position on cost-benefit analysis in the sociopath, this corresponds at the cognitive level to some form of value-specifying mechanism that undervalues prosocial actions and overvalues antisocial actions. Of course, even if this reading of her theory is correct, an account of how the cheating strategy leads to the development of particular cheating strategies such as spouse abuse still needs to be formulated. Our second problem concerns the claim that sociopaths suffer from a hypoaroused nervous system. First, it should be noted that the data concerning this claim are, at best, equivocal (e.g., Patrick et al. 1993). Second, and far more important, the theoretical importance of this claim is unclear. Mealey makes the assumption, as others in the sociopath/psychopath literature have done before her (e.g., Eysenck 1964; Patterson & Newman 1993), that moral socialisation is achieved through the use of punishment. The difficulty with this assumption is that there is no evidence for it. Indeed, the evidence that exists indicates that moral socialisation is best achieved through exposure to and focus on the victims of moral transgressions (see, for review, Hoffman 1977). Moreover, much of the exposure and focus on 548 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 victims appears to be independent of primary caregivers and occurs in interactions with teachers and peers (Nucci & Nucci 1982). Finally, it should be noted that we applaud the fact that Mealey is attempting to model a disorder developmentally and at several levels simultaneously. Too often, models of sociopathy have ignored development (e.g., Gorenstein & Newman 1980). However, we suggest that this developmental multiple level modelling must be formalised to increase clarity. One way to increase clarity is to use causal models; formal models of developmental interlevel relationships - genetics, physiology, social, cognitive, and behaviour (Morton & Frith, in press). Causal modelling has been used to model the development of several disorders, including autism (Morton & Frith, in press), dyslexia (Morton & Frith 1993), and psychopathy (Blair, in press). These models require the clear representation of the cognitive underpinning of behaviour and are a particularly powerful tool for representing the fact that behaviours could arise from different developmental histories. With specific reference to criminal behaviour, high levels of violence are associated with a variety of different groups. With delinquents, this violence has been linked to disinhibition caused apparently by frontal lobe deficits and/or maldevelopment (e.g., Moffitt 1993). Although the high level of aggression in psychopaths has also been attributed to a particular frontal lobe deficit (Gorenstein & Newman 1980), psychopaths do not seem to be any more deficient in general frontal lobe functioning than other criminal groups (see, for a review, Kandel & Freed 1989). The second crucial developmental relationship that causal models clearly express is that the behavioural consequences of particular structures are mediated by the rest of the cognitive system. This phenomenon is hinted at in the target article; Mealey suggests that the genotype for sociopathy that is expressed in males as sociopathy may be expressed in females as Briquet's Syndrome. However, she offers no consideration of what differences in the cognitive structures of males and females might mediate this distinction. In conclusion, we appreciate that other researchers are now using a multiple level, developmental approach to sociopathy. We only suggest that without cognition sociopathy cannot be fully understood. Sociopathy or hyper-masculinity? Anne Campbell Psychology Department, Durham University, Science Laboratories, Durham DH1 3LE, England, a.campbell@durham.ac.uk Abstract: Definitional slippage threatens to equate secondary sociopathy with mere criminality and leaves the status of noncriminal sociopaths ambiguous. Primary sociopathy appears to show more environmental contingency than would be implied by a strong genetic trait approach. A reinterpretation in terms of hypermasculinity and hypofemininity is compatible with the data. Mealey's criteria for distinguishing primary and secondary sociopaths are that the former show a greater genetic predisposition, less environmental contingency of their behaviour, an absence of secondary emotions, and more marked hypoarousal. But when we turn to the data, this distinction, critical to policy and program development, starts to dissolve. Primary sociopaths become equated with chronic criminal offenders (sect. 2.2.2, para. 2), despite Mealey's initial statement that sociopaths may constitute as few as 33% of this population (Introduction, para. 1). In accounting for secondary sociopaths, who are more canalised by early developmental experiences, she asserts that studies of juvenile delinquents can be used as a reliable guide to their childhood environments. If secondary sociopaths are simply criminal offenders who begin their careers as juvenile Commentary/ Mealey: The sociobiology of sociopathy delinquents (and most do), then sociopathy becomes coterminous with criminality. Although the DSM-III requires criminality as a defining feature of sociopathy, Mealey herself criticises it on these very grounds. If the term sociopath is instead used (as Mealey prefers) as descriptive of a particular constellation of behavioural and affective traits that have adaptive advantage, we should expect sociopaths to thrive wherever there is competition and individual disadvantage, be it on Wall Street or in the Bronx. Mealey recognises this but asserts confusingly both that noncriminal sociopaths are "subclinical" (sect. 2.5.1, para. 2), implying a rather weak behavioural manifestation of the trait, and that "almost all sociopaths from the upper-classes will be primary sociopaths" (sect. 3.1.1, para. 1), implying a rather strong manifestation. The distinction between simple criminality and sociopathic criminality is an important one for Mealey's argument, because it leads to different predictions about the age-crime curve. Gottfredson and Hirschi (1990) have argued that high crime rates during the teen years and early twenties transcend nation, offence, sex, and race - in line with Daly and Wilson's (1988) evolutionary account. However, Blumstein et al. (1988) point out that a small subset of chronic offenders show a constant rate of offending throughout their criminal careers with no evidence of a decrease with age. Mealey suggests that these chronic offenders are likely to be sociopaths who "consistently use the same strategy in every situation" (sect. 1.2, para. 2) and hence we should expect to see a similar invariance in their offending rates, but this is not found. Hare et al. (1992) computed mean number of convictions per year free for sociopaths and nonsociopaths. Both groups show a similar increase and decrease in convictions over the age span 16 to 46. This suggests that, unlike chronic offenders, sociopaths' offending follows the same age trajectory as that of other criminals. The age-relatedness of criminal behaviour by both sociopaths and other criminals suggests that their behaviour may be more environmentally contingent than Mealey suggests. I offer the following proposal. Cheating, as discussed by Mealey, implies a strategy in which an individual exploits others' cooperation selfishly by a failure to respond in kind. As she notes, most of us do not cheat because it is a risky strategy, both in the short term (immediate detection and retaliation) and the long (social shunning). Yet whereas cheating is only one form of risk taking, the converse is not true, making riskiness the superordinate construct. Daly and Wilson (1988) have amply documented the evolutionary relevance of risky behaviour to males in their early reproductive years, arguing that the greater preference for risk by males is part of our evolved psychology. Mealey rightly observes that women show less cheating (sociopathy), but they also show less risk taking overall, as measured by, for example, driver deaths and interpersonal violence. This suggests that there might be a continuum of psychological readiness to take risks (including cheating), with the male mean "shifted up" relative to that of females. Two psychological dimensions powerfully discriminate between males and females. Initially called masculinity and femininity, they are now more widely referred to as instrumentality and expressivity (Spence 1985). Instrumental traits include being individualistic, aggressive, and willing to take risks, embodying a competitive view of relationships and a readiness to assert oneself at the expense of others. Expressive traits include being gullible, sensitive to the needs of others, and compassionate, embodying an interdependent view of relationships and a willingness to sacrifice oneself for others. While Mealey posits sociopathy as the underlying continuum behind cheating, I suggest that instrumentality and expressivity may be twin continua lying behind riskiness, and that sociopaths may be viewed as a case of simultaneous hyper-masculinity and hypofemininity. Like sociopathy, instrumentality and expressivity have a substantial genetic component (Bouchard & McGue 1990; Mitchell et al. 1989; Rowe 1982). Just as sociopaths exceed normals, men exceed women on both the psychoticism and the sensation-seeking scale. Neurochemically, male rats and primates have lower serotonin levels than females, and estrogen reverses the inhibitory effect of dopamine on the pituitary. The direct excitatory effect of estrogen is even greater than that of norepinephrine, and estrogen also inhibits cortical responsivity to the inhibitory transmitter adenosine (see Hoyenga & Hoyenga 1993, for references and their proposal of relating masculinity to aggression). In evolutionary terms, men possessing masculine qualities of independence and aggression probably had a fitness advantage whereas the simultaneous presence of feminine characteristics of nurturance and empathy may have mediated social bonding and cooperation. Although sociopaths show no change over the life course in their lack of concern for others (Hare et al. 1992), their willingness to engage in risky behaviour diminishes after the peak reproductive years when, in evolutionary terms, it is of most value. In primary sociopaths, we may be seeing the far end of the male instrumental continuum completely untempered by the softening effects of expressivity. Cheaters never prosper, sometimes H. Lome Carmichael Department of Economics, Queen's University, Kingston, Ontario, Canada K7L 3N6. carmykle@qucdn.queensu.ca Abstract: In the Frank (1988) model, a small increase in the number of cheaters will soon be reversed. It is not clear that this prediction holds for sociopathy. There are also many attractive evolutionary models that do not admit a small, stable proportion of cheaters. Hence, without definitive evidence about the character of early human society, we cannot conclude that sociopathy has an evolutionary origin. In reviewing earlier issues of BBS for examples of fine commentary, I was struck by the virulence of the debates surrounding sociobiological papers like this one. To some proponents, the fact that human life evolved on this planet is sufficient to relegate all opposition to the realm of divine myth (e.g., Thornhill & Thornhill 1992, especially the Authors' Response). Of the opponents' position, sometimes the phrase "society is to blame" seems altogether too weak to be an effective parody. Indeed, suggestions to the contrary are sometimes branded harmful and worthy of suppression (e.g., the comment by Dupre (1992) in the above collection). Sociobiology has the attraction that it is at least a well-defined theory. It has assumptions that can be expressed clearly, that are amenable to logical manipulation, and that can generate clear predictions. This, in an area as complex as human behavior, makes it an easy target. No such model will ever be completely true. Rather, one hopes that a simple model will capture aspects of reality that are particularly useful and relevant to the questions at hand. Unfortunately, the target article fails to meet even this weaker standard. Linda Mealey has convinced me that primary sociopathy is present in individuals at a very early age and develops regardless of parenting, and that secondary sociopathy requires a combination of inherent vulnerability and bad luck. I am willing to accept that these traits are heritable, although I do not think it proven. The rings on this target article center elsewhere - on its main hypothesis that sociopathy exists for a particular reason illuminated by simple evolutionary theory. Even if we accept that evolutionary processes are determinate, I wager that most of us could create a plausible society that, had it existed for 100,000 years or so, would have selected for just about any trait purported to exist in the species today. In BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 549 Commentary/Mealey. The sociobiology of sociopathy the absence of a definitive historical record, therefore, the empirical strength of the current paper rests entirely on the charm, as a metaphor for human interactions, of the random matching one-shot Prisoner's Dilemma model with imperfect identification of types. But this is not the only charming model available, and some of the others (including the same one with different parameter values) do not admit a stable proportion of cheaters, or do not admit cheaters at all. Also, if taken seriously, the model chosen makes predictions about sociopathy that do not seem to be true. I will address these criticisms in reverse order. Recall that if a mixed strategy profile is evolutionarily stable, then if a particular strategy A increases (or decreases) in frequency, the payoffs to all the strategies must change so that A is relatively disadvantaged (or advantaged). Natural selection will then ensure that strategy A goes back to its proper place. If we are to take the current application of Frank's (1988) model seriously, then the proportion of sociopaths in a society should follow a similar dynamic. But does it? Would primary sociopaths in an ancient society with "too many" of their own kind have had difficulty gaining access to resources and mating partners? We just do not know, and the target article is not helpful. Perhaps cooperators would become more diligent, but if cheaters ganged up under a charismatic Attila one suspects large numbers would be an advantage. The dynamic for secondary sociopathy is discussed in the paper, but things seem to go the wrong way. One has to be careful here - if I move from Kingston to New York City and as a result my kids are more likely to become sociopathic, this could be because environmental differences lead New York to have a higher proportion of cheaters. Rather, suppose that a small group of young sociopaths move to Kingston, all else the same. According to the Frank/Mealey model, Kingston children will now be less likely to become sociopathic. I have no evidence; but like most parents, I think not. Frank's (1988) model is a variant of the round robin, infinitely repeated Prisoner's Dilemma introduced by Axelrod (1984) in his celebrated competition. But there are other models that have equal "charm' and are therefore equally (un)likely to capture the essence of the conditions facing primitive homo sapiens. Here are two examples: Consider a model where individuals match up randomly, play a one-shot Prisoner's Dilemma, and then have the choice of continuing or terminating the match (Carmichael & MacLeod 1994; Stanley 1993). If the match ends, both parties go back, anonymously, to the matching market. If not, the partners may stay matched until death, continuing to play a Prisoner's Dilemma each period. For a modern image, think of the matching market as a large, dimly lit singles' bar. Even though cooperators play a repeated Prisoner's Dilemma, the strategy "tit for tat" does very poorly. It is quickly invaded by cheaters who defect at the first opportunity and then move on to a new match. An interesting evolutionarily stable strategy is for cooperators to offer (and demand) an exchange of gifts at the beginning of any new match (Carmichael & MacLeod 1993). Cheaters in this society would have to buy a succession of gifts, and this effectively screens them out. This model makes quite a few predictions about the form of the gifts that must be used. l bargainers will be vulnerable to the "terrible twos" strategy of demanding almost everything, backed up with the emotional threat to ensure that otherwise the hyena gets everything. Faced with such an opponent, a rational bargainer cuts his losses, takes what is offered, and moves on. Of course an entire society of two-year-olds does very poorly, and can be invaded by a group whose members fight if they do not receive at least half the spoils. This strategy is evolutionarily stable - it quickly reaches agreement with itself and can do no worse than any invader it meets. 2 Again, in this simple model, there is no room for cheaters. Readers will recognize the "bourgeois" strategy of Maynard Smith (1982), but there are some new twists. In particular, if people are of two types, there are many equilibria where one type does better than the other. If men fight whenever they get less than one-third and women fight whenever they get less than two-thirds, for example, this is evolutionarily stable. Equilibria like these require that one's notion of territory be socially determined. There must be a way for early experience and teaching to establish and coordinate internal notions of what one deserves out of life. Sociobiology can therefore account for the existence of systemic discrimination, and society may indeed, at least partly, be to blame. The point, of course, is not that these are better models of reality than the one used in the target article, but they do seem equally plausible, and they have implications that are at least as attractive and intriguing. Perhaps they each capture relevant but separate aspects of reality. If so, Mealey's conclusion - that evolution is unable to rid society of a small proportion of cheaters - is not robust. (The rule of emotions in these models, by contrast, is robust.) Cheaters do prosper, no doubt. But until we have excellent evidence about the exact nature of early human society, or until we can show that in any sensible evolutionary model there will survive a small proportion of cheaters, sociobiology will not be able to tell us why. Here is another one (Carmichael 1994). Suppose we retain the one-shot matching framework of Frank but change the game from a Prisoner's Dilemma to a bargain. People meet and have to decide how to divide the spoils of some joint venture (the carcass of some animal, perhaps). If they can agree quickly on a division, then all is well. If they cannot, the spoils disappear, dragged away by a hyena. Strategies that do well in these bargains will proliferate into the future. An intraspecies arms race might develop, where "bargaining ability" grows over time. Rational and unemotional In a target article of exceptional scholarship and originality, Mealey has put forward an interesting new interpretation of sociopathy. Given the vast range of material covered by the article and the limited space available for my commentary, I shall confine my comments to the specific game theoretic model that underpins Mealey's interpretation. I shall argue that it cannot do what is required of it, and I shall suggest an alternative. Like most earlier theorists who have used game theory to explain the evolution of social behavior, starting with Maynard 550 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 NOTES 1. Among other things, gifts should have low use value, be overpriced, and should be hard to recycle as gifts in a subsequent match. Cut flowers and chocolates work - house plants and money do not. 2. Unless, of course, the invader has a weapon. The arms race in this model is real. Prisoner's Dilemma, Chicken, and mixedstrategy evolutionary equilibria Andrew M. Colman Department of Psychology, University of Leicester, Leicester LE1 7RH, England, amc@lelcester.ac.uk Abstract: Mealey's interesting interpretation of sociopathy is based on an inappropriate two-person game model. A multiperson, compound game version of Chicken would be more suitable, because a population engaging in random pairwise interactions with that structure would evolve to an equilibrium in which afixedproportion of strategic choices was exploitative, antisocial, and risky, as required by Mealey's interpretation. Commentary/Mealey: (a) Smith (1974; Maynard Smith & Price 1973), Mealey relied on a two-person game, specifically the Prisoner's Dilemma game. A generalized payoff matrix for any two-person, two-choice game can be represented as follows: CD C RS DTP The row and column players each choose between two pure strategies, C and D, the payoffs shown in the matrix are those to the row player. Thus the payoff to the row player following a C choice is ft or S, depending on whether the column player chooses C or D, respectively, and following a D choice it is Tor P, depending on whether the column player chooses C or D, respectively. In the Prisoner's Dilemma game, C represents cooperation and D defection, and by definition T > R > P > S, so that the row player receives the best payoff by choosing D (defect) while the column player chooses C (cooperate), the second-best payoff by choosing C while the column player chooses C, and so on. Although it is customary to show only the row player's payoffs, the game is the same from the column player's point of view, so that the column player also gets the best payoff by choosing D while the row player chooses C, and so on. The standard two-person model is of limited value in deterining evolutionary processes. We need to establish what will ippen in an entire population in which individuals interact ith one another in pairwise games with this strategic structure, :suming that the payoffs represent units of Darwinian fitness he lifetime reproductive success of the individual players) and lat the propensity to choose C or D is heritable. For this uqjose, we need to construct a multiperson compound game dolman 1982, pp. 163-66, 243-49), in which it is assumed that very player plays the same average number of two-person ;ames either with each of the others or with a random sample of he others. Considering the situation from a single player's viewpoint, suppose that the number of other players is n and the number of Dther players choosing C is c. The total payoff to a player choosing C, denoted by P(C), and the total payoff to a player choosing D, denoted by P(D), are then defined by the following payoff functions: P(C) = Re + S(n - c), P(D) = P(n - c). The total payoff to a player adopting a mixed strategy is just a weighted average of P(C) and P(D). The values of the P(C) and P(D) payoff functions at their endpoints are found by setting c = 0 and c= n. Thus, if none of the other players chooses C (i.e., c = 0), the payoff to a solitary C chooser is Sn and the payoff to a D chooser is Pn. If all of the other players choose C (i.e., c = n), then a C chooser gets Rn and a solitary D chooser is Tn. It is clear that in the case of the Prisoner's Dilemma game Tn can be interpreted as the temptation to be the sole D chooser, fin the reward for collective cooperation, Pn the punishment for collective defection, and Sn the sucker's payoff for being the sole C chooser. Figure l(a) shows clearly that, in the case of the Prisoner's Dilemma game (with T > R > P > S), the P(D) payoff function strictly dominates the P(C) payoff function, which means that a D choice pays better than a C choice irrespective of the number of others choosing C. The evolutionary optimal strategy is therefore not frequency-dependent, and the population will (regrettably) evolve to a stable equilibrium in which every player chooses D in every two-person encounter. This means that the Prisoner's Dilemma game cannot provide a basis for I& The sociobiology of sociopathy P(D) (b) P(Q P(D) P(C) Figure 1 (Colman). Multiperson compound games based on 2 X 2 matrices. Panel (a) on the left is multiperson Prisoner's Dilemma; (b) on the right is multiperson Chicken. The P(C) and P(D) functions indicate the payoffs to a player choosing C or D when c of the other players choose C. Dashed circles indicate stable equilibria. Mealey's interpretation of sociopathy, in which the "cheater strategy" (the D choice) corresponds to various criminal, delinquent, and generally antisocial or predatory forms of behavior that she claims exist at a low frequency, in every society and are maintained through frequency-dependent Darwinian selection. A more appropriate game theoretic model might be a compound version of the game of Chicken, which Maynard Smith (1976; 1978) and Maynard Smith and Price (1973) call the HawkDove game. This game is defined by the inequalities T> R> S > P, and the P(C) and P(D) payoff functions are shown graphically in Figure l(b). In this case, the population will evolve to a mixed-strategy equilibrium point, where the two payoff functions intersect. To the left of the intersection, when relatively few of the others choose C (c is small), the C function lies above the D function, which means that the fitness payoff from a C choice is higher than from a D choice, so the number of C choosers will increase relative to D choosers and the outcome will move to the right as c increases. To the right of the intersection, exactly the reverse holds: D choosers will increase relative to C choosers and the outcome will move to the left as c decreases. At the intersection, and only there, the strategies are best against each another and are in equilibrium, and any deviation from the mixture at that point will tend to be self-correcting. By setting the parameters (values of the payoffs T, R, S, and P) appropriately, the intersection point, and thus the proportion of "predatory" D-choices, can be made as small as required. It appears, therefore, that the Prisoner's Dilemma game cannot underpin an evolutionary explanation of sociopathic behavior, but that a multiperson compound game version of Chicken, in which cheating is at least frequency-dependent,, might be more promising. Chicken is the archetypal dangerous game, because a player can outdo a co-player only by cheating (choosing D) while the co-player behaves cautiously (by choosing C), and any such attempt to get the best payoff (T in the payoff matrix above) involves a necessary risk of the worst possible payoff (P). The interpretation of criminal, delinquent, and generally antisocial behavior in terms of strategic choices seems more natural in the game of Chicken. (For a more detailed discussion of the strategic properties of Chicken and some observations on its application to antisocial and criminal behavior, see Colman 1982, pp. 98-104; 1995, sect. 9.6.) ACKNOWLEDGMENT Preparation of this commentary was facilitated by Grant No. L122251002 from the Economic and Social Research Council as part of the Framing, Salience and Product Images project. BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 551 Commentaryi'Mealey: The sociobiology of sociopathy The sociopathy of sociobiology Wim E. Crusio G6n6tique, Neurog6n6tique et Comportement, URA 1294 CNRS. UFR Biom&dicale, University Paris V Ren6 Descartes, 75270 Paris Cedex 06, France, crusio@clti2.fr Abstract: Mealey's evolutionary reasoning is logically flawed. Furthermore, the evidence presented in favor of a genetic contribution to the causation of sociopathy is overinterpreted. Given the potentially large societal impact of sociobiological speculation on the roots of criminality, more-than-usual caution in interpreting data is called for. Mealey's target article provides an enormous compilation of data on vastly different subjects, varying from social psychology to evolutionary theory. Admirable as the amount of work involved with gathering, digesting, and discussing this large body of literature may be, the result is unfortunately not very enlightening. The line of reasoning has become buried under an avalanche of uncritically enumerated research findings, so that after some time the reader loses sight of exactly what the author is trying to say, based on what evidence, and what the assumptions are on which the author's interpretations are based. In consequence, I am not quite sure what Mealey's main thesis is. As far as I see, she is trying to make two, not necessarily related, points. First, that sociopaths can be divided into two subtypes: primary and secondary sociopaths. Second, that this subdivision has come into existence because of selective forces, exerted during evolution on a genetic substrate underlying sociopathy. I trust that other commentators who are more competent on this subject than I will comment on the first point. I will direct my scrutiny mainly at the second one, but not without noting that neither evolution nor genetics appear to be very relevant for Mealey's arguments regarding the first point, and they clearly are irrelevant to the exclusively environmental recommendations presented as following from the model (sects. 3.2 and 3.3). To start with the evolutionary part, Mealey asserts that the two forms of sociopathy came into existence as a result of two different evolutionarily stable strategies (ESSs; sect. 1.2). Although it does not become clear what role natural selection is playing here, the description of these ESSs clearly shows the logical impossibility of having two different ESSs for one single phenotype within one single population. If two different mechanisms are at work, then this implies that primary and secondary sociopathy are two different, independent phenotypes. Yet, in the last paragraph of section 2.3.1 it becomes clear that this is not what the author has in mind. I would like to see some clarification of this serious contradiction and an explanation of how natural selection may have acted in order to result in two different ESSs for one single phenotypical continuum. In addition, might it not be necessary to consider whether and how the increasingly dramatic changes in living conditions over the last few thousands of years (and even more so in the last few centuries) may have influenced the nature of the natural selection involved, in the process perhaps rendering all this evolutionary speculation rather pointless? The last paragraph of section 2.3.1 poses a number of other problems, too. It is argued that some persons become primary sociopaths because of their genotype. Others become secondary sociopaths because of a smaller genetic load combined with certain environmental causes (see also n. 14). Perhaps this is so. Unfortunately, the evidence provided to support these ideas is highly inadequate. Let me give a few examples. A number of physiological factors are discussed, for which primary sociopaths apparently differ from other people; this supposedly suggests a genetic basis for primary sociopathy. But physiological variables are phenotypes, too; there is nothing inherently different about them that would suggest they are somehow more "heritable" 552 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 than other types of characters. Still, this appears to be implicitly assumed. Take the finding that upper class, but not lower class, boys who subsequently became delinquent showed physical signs of hypoarousal (sect. 2.3.3, last para.), which is presented as evidence for the idea that the upper-class individuals became delinquent as the result of a particularly strong genetic predisposition. Only when hypoarousal would be a phenotype with a heritability equal to unity and perfectly correlated with sociopathy would this conclusion be warranted. The finding that almost all upper-class subjects that had been arrested came from a biological high-risk group (sect. 2.3.3, last para.) is not conclusive either. As "biological high-risk group" probably means that subjects descended from parents with an antisocial background, the familial environment of these subjects is very likely to have differed from that of control subjects too. Mealey herself seems to recognize the importance of environmental influences, even for primary sociopaths, given the recommendations at the end of her article. The only reasonably solid evidence provided for any genetic basis for either form of sociopathy is heritability estimates, derived from twin and adoption studies, 1 of around 0.60 (but see BBS commentary on Wahlsten, 1990, for a discussion of the scientific value of such estimates). However, even when rather high, this value still tells us that genotype and environment explain about equal parts of the interindividual variation observed. Summarizing, the genetic component of Mealey's model is nothing more than just some elaborate speculation. If the target article were dealing, for example, with the mating behavior of black widow spiders or some other subject without direct societal impact, the author might have indulged in some speculation without much possible harm. But in the present case we are dealing with people, not spiders. Scientific speculation, however interesting in itself, may easily be interpreted by politicians, journalists, or others as stated scientific fact - the more so if the speculative nature of an author's writings is not clearly indicated as such. Sociobiologists appear to be rather prone to this kind of oversight, hence the title of my comment. ACKNOWLEDGMENTS The preparation of this commentary benefitted from support by the CNRS (URA 1294), UFR Biomedicale (University Paris V Rene Descartes), DRED, and the Fondation pour la Recherche M6dicale. NOTE 1. It should be noted that in all these studies at least the early part of childhood environment is confounded with possible genetic effects. Together with possible prenatal influences, this perhaps explains in part why human behavior-genetic analyses so often render heritability estimates (in the broad sense, see Mealey's n. 6) that are much higher than anything reported from animal research. Note also that experimental animals are generally bred and reared in a rigorously standardized environment in order to maximize genetic effects relative to environmentally induced variations (i.e., boosting heritability). A neuropsychology of deception and self-deception Roger A. Drake Department of Psychology, Western State College, Gunnison, CO 81231. rdrake@wsc.colorado.edu Abstract: As more criminals are imprisoned, other individuals change their behavior to replace them, as predicted by the "floating niche" theory of strategic behavior. The physiological correlates of sociopathy suggest that research in cognitive neuroscience can lead toward a solution. Promising pathways include building upon current knowledge of self-deceit, the independence of positive and negative emotions, the Commentary/Mealey: lateralization of risk and caution, and the conditions promoting prosocial behavior. Mealey provides a useful summary of research from a variety of disciplines on sociopathology. One of her best insights is the differentiation between primary and secondary sociopaths and, therefore, differences in the way they must be treated, to protect ourselves from their deception and predation. Of particular interest is her idea of a floating niche for deceit and exploitation. This fits with the reality that although the United States has a million individuals in prison, crime has not decreased. Trivers (1985) predicted this in his discussion of deception, when he said that, "In species with multiple mimics [of honesty], the frequency of mimics is controlled by the relative frequency of their models" (p. 420). This statement implies that as deceivers - who are mimics of honesty - become too common, there is greater detection of their dishonesty. But if some deceivers are removed, for example by imprisonment, other individuals will shift to a strategy of deception, because it will prove to be a more effective strategy than honesty. Mealey argues that this produces a massive practical problem, because of the great harm that sociopaths and other deceivers do to others. Trivers (1985) further argues that the most effective deceivers will be those who can deceive themselves. This reduces the possibility of cues of anxiety giving them away. Research by Drake and Sobrero (1987) indicates that relative activation of the right hemisphere of the brain produces behavior that is independent of a person's previously measured traits and attitudes. This recommends a fruitful direction for research into a neurological basis for self-deception. Mealey offers solutions to the predation of primary sociopaths on a trusting population that are based principally on game theory (introduced in sect. 1.2). She proposes that society can reduce antisocial behaviors in primary sociopaths by establishing a reputation for detection, identification, and retaliation (sect. 3.2.1). In other words, in an analogy with the Prisoner's Dilemma game, she advocates that the outcomes for deception must be harsh and must be publicized. The problem with this approach is that primary sociopaths are rarely deterred by fear or threat (Patrick 1994), but they do approach pleasure. This implies that one direction for research should be toward a greater knowledge of the separation of positive from negative emotions. This is an established concept within psychology (Cacioppo & Berntson 1994; Diener & Emnions 1985; Warr et al. 1983). Such research can profitably gain from integration with the neuropsychological evidence for distinctive regions of the brain associated with positive versus negative emotions (Ahem & Schwartz 1985; Gainotti et al. 1993; Merckclbach & van Oppen 1989; Schiff& Lamon 1989; Van Strien & Morpurgo 1992; Wittling & Rosehmann 1993). Mealey describes physiological correlates and perhaps causes of sociopathology (in sect. 2.3.3), yet physiological solutions to predatory deception arc never mentioned. Her proposals are mostly extensions of current government efforts, such as parental education, reduction of social stratification, and exciting alternative careers. These approaches are not novel. Perhaps the field of cognitive neuroscience could contribute more through both basic and applied research on the differences in brain structure, hormones, and neurotransmitters that differentiate the sociopath. These may eventually lead to treatments for the lack of fear and other negative emotions that typify the predatory sociopath. The biobehavioral scientific community can further contribute with basic and applied research on the neuropsychology of caution and risk (Drake & Ulrich 1992; Miller & Milner 1985) and on the conditions for the promotion of prosocial behaviors (Levine et al. 1994). Mealey has provided us with a strong outline of the extent of the problem, and has helped to The sociobiology of sociopathy point in the directions research should take us toward possible solutions. The role of emotion in sociopathy: Contradictions and unanswered questions Nancy Eisenberg Psychology Department, Arizona State University, Tempe, AZ 85287-1104. atnhe(« asuvm.inre.asu.edu Abstract: Emotion is critical in Mealey s conceptual arguments. However, several of her assertions about the role of emotion in sociopathy are problematic. Questions are raised regarding the link between lack of anxiety and low levels of secondary emotions such as love and sympathy, the argument that sociopaths are low in anxiety but high in neuroticism, and the designation of anxiety as a secondary emotion. In Mealey's arguments, emotion plays a crucial role; Mealey linked the unique characteristics of primary psychopaths to genetically based emotional mechanisms. A critical point in the target article is that "guilt, anxiety, and sympathy are social emotions that primary sociopaths rarely, if ever experience," (sect. 2.5.2, para. 8) although fluctuations in mood (e.g., anger, depression, and optimism) are experienced by primary (and secondary) sociopaths. A distinction is made between primary emotions, such as fear, anger, and disgust, and secondary emotions, which are viewed as more dependent upon learning and socialization (note 1). Primary sociopaths are viewed as experiencing primary emotions but as being very low in the experience of secondary, social emotions such as shame, guilt, sympathy, and love. In contrast, secondary sociopaths apparently are viewed as capable of experiencing normal amounts of both types of emotional reactions. Although Mealey's assertions are often supported by some empirical literature, several conceptual issues can be raised. One question concerns why primary sociopaths seem to have a diminished capacity to experience secondary social emotions. In note 1, secondary emotions are described as involving a critical element of learning. Given that primary sociopaths are not described as lacking in cognitive capacities or in the capacity for primary emotions, it is unclear why they seldom "learn" to experience such emotions. Is Mealey hypothesizing a biological mechanism in sociopathy that affects proneness to learn secondary emotions? Lack of anxiety is used as an explanation for low levels of guilt, but lack of anxiety does not explain low levels of love, sympathy, and some other emotions. A related point is Mealey's designation of anxiety as a secondary social emotion (sect. 2.5.2, para. 8). This is a critical point, given that proneness to anxiety is viewed as a crucial element in conditionability and in the development of conscience (i.e., negative emotional arousal) when one "cheats" or otherwise deviates. However, anxiety is an emotion closely tied to the primary emotion of fear (Plutchik 1984, p. 215) and is quite different from secondary social emotions such as love, empathy, sympathy, and true guilt. If primary sociopaths are described as experiencing primary emotions, such as fear, at normal levels, it is not obvious why they would not be as prone to anxiety as other people. On a related issue, Mealey noted that sociopaths score high on neuroticism as assessed by Eysenck (1976). Yet the neuroticism scale contains items pertaining to anxiety (worry) and guilt. Thus, the link between neuroticism and sociopathy would argue against sociopaths being immune to anxiety and hence being difficult to socialize. Mealey contradicts herself by saying that primary sociopaths are low in anxiety and then implying (sect. 2.3.2, para. 4) a link between neuroticism and primary sociopathy. BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 553 Commentary I Mea\ey. The sociobiology of sociopathy In actuality, the link between neuroticism and primary sociopathy may not be consistent or strong. Eysenck (1976) argued that primary sociopaths are high in extroversion (E) and psychoticism (P), with average scores on neuroticism (N), whereas typical secondary psychopaths are high in E and N, with average P scores. In other words, rather than being high on anxiety, worry, and other negative emotions linked to neuroticism, primary sociopaths appear to be high on hostility and to lack social feelings (elements of P). Perhaps primary sociopaths are actually low on some aspects of neuroticism (e.g., guilt, anxiety) and higher on others (e.g., proneness to experience irritability and emotional lability). But if primary psychopaths are not especially low in anxiety, the argument that they are difficult to socialize due to lack of the inability to condition would not seem plausible. In brief, several points regarding anxiety are inconsistent in the target article. My resolution of the inconsistencies is as follows. It is likely that anxiety is not comparable to the secondary social emotions of guilt, shame, empathy, and love. If this is true, then Mealey's assertions that primary sociopaths tend not to experience secondary social emotions but do experience primary emotions seem reasonable. In addition, primary psychopaths are probably average in general proneness to anxiety and not particularly high in neuroticism. The reason that primary sociopaths are difficult to socialize is not that they are low in anxiety per se; it is because they are low in the tendency to experience social emotions, such as love, shame, and empathy, that are likely to result in anxiety when people are in social situations in which needs linked to social emotions are in jeopardy (e.g., in socialization situations in which children might elicit disapproval or rejection from their parents). If primary sociopaths also do not seem to condition well in situations that do not involve social emotions and motivations, it may be due to other factors, such as their hypoaroused nervous system (which, as noted by Mealey, may impair learning), or their tendency to approach rather than inhibit in approach/ avoidance situations. This brings me to one final issue concerning the apparent low baseline physiological arousal of sociopaths and their tendency to be sensation seekers. Larsen and Diener (1987) proposed that people high in affective intensity use emotion for stimulation to compensate for a chronically low level of baseline arousal. If this were true, one would expect people high in primary sociopathy to be high on scores of affect intensity. Yet, at least for adults (results for children are more complex), we have found that emotional intensity is associated with sympathy (a secondary social emotion; Eisenberg et al. 1994). Thus, interesting questions for future consideration concern the role of baseline arousal in the tendency to experience primary and secondary emotions intensely and whether intensity of emotional experience is linked to sociopathy. Extending arousal theory and reflecting on biosocial approaches to social science Lee Ellis Department of Sociology, Minot State University, Minot, ND 55701. ellls@warp6.cs.misu.nodak.edu Abstract: This commentary extends arousal theory to suggest an explanation for the well-established inverse correlation between church attendance and involvement in crime. In addition, the results of two surveys of social scientists are reviewed to reveal just how little impact the biosocial/sociobiological perspective has had thus far on "mainstream" social science. Mealey's target article provides a useful summary of how neurological and hormonal factors help to orchestrate human varia- 554 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 tions in criminal and antisocial behavior within an evolutionary context. The evidence that has been amassed is consistent with my work on possible neurohormonal (Ellis 1987a; 1990a) as well as evolutionary (Ellis 1988; 1990b) underpinnings of crime and delinquency. Here two issues will be addressed briefly. First, I propose that information reviewed in the target article can be extended to explain a controversial observation, namely, that religious people are less criminal than nonreligious people. Second, I allude to two recent surveys that remind us of how little the research reviewed by Mealey has thus far influenced the way most social scientists think. Mealey appropriately notes that arousal theory has garnered substantial empirical support in recent years (sect. 2.3.2). According to arousal theory, for neurological reasons, persons who are most prone toward criminal or antisocial behavior tend to be under-aroused (or what I call suboptimally aroused; Ellis 1987a). Thus, from the standpoint of arousal theory, criminal and antisocial behavior is an expression of sensation seeking and boredom avoidance rooted in a nervous system that prefers stimulus intensity and novelty that are greater than the environment normally provides. [See also Zuckerman "Sensation Seeking: A Comparative Approach to a Human Trait" BBS 7(3) 1984.] In addition to predicting that people who are suboptimally aroused will be more likely to commit crimes, arousal theory suggests that criminally prone persons will attend religious services infrequently. This follows from noting that most church services are anything but exciting. As a result of infrequent church attendance, criminally prone persons should also come to hold relatively unorthodox religious views (Ellis 1987b; in press). These deductions are supported by numerous studies showing a significant inverse correlation between criminality and religiosity (reviewed in Ellis 1985; in press). If the hypothesized inverse association between religiosity and criminal/antisocial behavior continues to be supported, one could even postulate a new evolutionary explanation for religion. Perhaps religious institutions, rituals, and dogmas have evolved in part to help individuals and their parents screen out antisocial individuals from those being considered as prospective marriage partners. Females especially, who are prone toward investing most heavily in offspring production, may use religion as a filtering device in mate selection. In so doing, they should minimize their risk of marrying someone who is extremely antisocial. After reading Mealey's synthesis of the literature, it is hard not to be impressed by how strong the evidence now is for the involvement of biological factors in criminal and antisocial behavior. Nonetheless, a perusal of most criminology texts will show that the sympathetic attention of most criminologists to such factors is still minimal. Also, two surveys have indicated how social scientists continue to resist the evidence that Mealey reviews. The first survey asked 182 criminologists in 1986 what theories they thought had the most merit in explaining criminal behavior (Ellis & Hoffman 1990). Arousal theory was not mentioned by a single respondent, and less than 20% of the criminologists were sympathetic toward viewing either genetic or neurological factors as significant causes of criminal behavior. More recently, 164 of my fellow sociologists were asked to estimate how much of the variance in criminality could be attributed to biological, as opposed to social environmental, factors (Sanderson & Ellis 1992). The mean percentage attribution was 13% for serious crime and 9% for minor crime and delinquency. The median percents were even lower. Overall, Mealey's article shows that major progress has been made over the past two decades in identifying biological contributors to criminal and antisocial behavior. Nevertheless, much remains to be accomplished in the way of disseminating this information to the social science community at large. Commentary/Mealey: Sociopathy and sociobiology: Biological units and behavioral units Carl J. Erickson Department of Psychology - Experimental, Duke University, Durham, NC 27708. carl@psych.duke.edu Abstract: Behavioral biologists have long sought to link behavioral units (e.g., aggression, depression, sociopathy) with biological units (e.g., genes, neurotransmitters, hormones, neuroanatomical loci). These units, originally contrived for descriptive purposes, often lead to misunderstandings when they are reified for purposes of causal analysis. This genetic and biochemical explanation for sociopathy reflects such problems. The ill-defined field of sociobiology has been a mixed blessing. At its best it has been a heuristic force, shaping new questions and providing new insights into sex ratios, altruism, and a variety of behavioral phenomena. At its worst it has fostered a simplistic set of inferences in which utility means adaptation cans evolution means natural selection means heritability means genes. Mealey's analysis reflects both the strengths and the weaknesses of sociobiology s contribution. The research on infanticide provides a useful model from which to examine Mealey's views on sociopathic behavior. Infanticide, in both animals and humans, has often been dismissed as maladaptive, pathological behavior; however, sociobiologists, encouraged by a renewed interest in sexual selection, have argued that infanticide might serve biological fitness. Twenty years of research has strengthened that supposition. Mealey suggests that sociopathy, like infanticide, has utility for those individuals who exhibit it, and she notes that it is important for society to recognize this fact in coming to grips with the behavior. Such perspectives are welcome and deserve encouragement, but the comparison to infanticide also illuminates some of the limitations of her argument. Sociobiological interpretations work best when they relate to very specific behavior patterns having a direct bearing on reproductive success. Even then they can be dismissed as "just so" stories unless they carry enough detail and eliminate enough alternatives to make the adaptationist argument inescapable. For example, Vom Saal (1985) finds that the frequency of infanticide among male mice increases dramatically following ejaculation, and it remains high until just before their own young are born. It then drops to a low level until their young disperse, at which time it rises again. These patterns are so complex, so challenging to alternative explanations, and yet so consistent with an adaptationist viewpoint that it is difficult to escape the sociobiological conclusion. Moreover, when males kill the infants of other males, there is an obvious impact on inclusive fitness. An evolutionary explanation is therefore tenable, if not compelling. On the other hand, with the possible exception of rape, sociopathic behavior has no obvious connection to inclusive fitness. To be sure, a competitive edge in the most general sense is implied when it occurs at low frequency in the larger population, but how this advantage translates into a gain in inclusive fitness (a necessity for the evolutionary argument) remains unclear. Mealey suggests (sect. 3.1.1) that sociopathic behavior actually takes two forms: (1) a primary, "inborn" pattern and (2) a secondary, "environmentally contingent" form. Although only the first of these is explicitly presented as a "genetically determined strategy," both are presented as "genetically based." Such preformationistic positions have drawn a barrage of criticism over the past twenty years (cf. Lewontin et al. 1984), and even the most traditional ethologists and sociobiologists now concede that although genes make a difference, they do not govern any behavior in such deterministic fashion. Increasingly, developmental biologists and psychologists recognize that the developing organism is in a sense "self-organizing," and that there is no "blueprint" for behavior in the genome. Even if one accepts Mealey's categories of primary and secondary sociopathy (which The sociobiology of sociopathy I am reluctant to do), there still would be no logical reason to assume that genes are any more involved in one than the other. Moreover, even if the patterns vary in the frequency-dependent way Mealey describes (and there appears to be little evidence for this), it would be virtually impossible to distinguish the secondary sociopathy of an evolutionary adaptation from the coping behavior of a remarkably flexible human cognitive process. Brain mechanisms for the latter have evolved, to be sure, but the specificity of their role is very different from those that sociobiologists envision. Speaking more generally I might add that hormones, neurotransmitters, brain nuclei, and so on also make a difference in behavior, but again, they do not control behavior in the strong sense implied by Mealey's characterization. Testosterone's action is a case in point. Mealey's exposition would lead one to believe that higher levels of this hormone make male adolescents bigger, stronger, and more aggressive. The research literature remains unclear on testosterone's relationship to aggression, but it is quite clear on its relationship to body size. Testosterone terminates the growth of the long bones. Young boys who take anabolic steroids are short, and the castrati of 18th- and 19th-century Italy were excessively tall. Mealey suggests that aggression and testosterone are mutually stimulating and generate a positive feedback loop, but the relationship is far more complex than she suggests. True, subnormal levels of testosterone may indeed correlate with subnormal levels of behavior, but once serum testosterone reaches normal levels or above, the relationship breaks down. A panoply of internal and external contextual conditions modulate testosterone's influence. Liver activity (which may reflect drug and alcohol use), carrier proteins, receptor numbers, and past experience are among the many factors that complicate the relationship between blood testosterone level and aggressive or sexual behavior. Moreover, many of these factors have rate-limiting effects, preventing excessive testosterone from having excessive effects on behavior. In short, neither testosterone nor any other behaviorally relevant chemical plays a determining role in any behavior. Generally speaking, biological units (genes, hormones, neurotransmitters, or chunks of brain) do not have a one-to-one causal relationship with specific behavior patterns. Mealey musters an impressive volume of literature in her analysis of sociopathy. Much of it contributes to a better understanding of this intriguing behavior, but the sociobiological perspectives are largely untestable and add little. Moreover, by suggesting that the behavioral variants are genetically determined, this formulation encourages the unfortunate conclusion that developmental studies are irrelevant when, in fact, they are critical. Psychopathology: Type or trait? H. J. Eysenck Professor Emeritus of Psychology, University of London, London SE5 8AF, England Abstract: Mealey proposes two categorical classes of sociopath, primary and secondary. I criticize this distinction on the basis that "type" constructs of this kind have proved unrealistic in personality taxonomy and that dimensional systems capture reality much more successfully. I suggest how such a system could work in this particular context. Mealey's discussion of the sociobiology of sociopathy is remarkably complete and succinct; it covers much ground and establishes its primary contentions admirably. I have little to criticize, other than the main point of the paper, namely, establishing the two categorical classes of sociopaths - primary and secondary (with criminals who are not sociopathic as a third type, presumably). BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 555 Commentary/Mealey: The sociobiology of sociopathy I have always argued that the psychiatric commitment to a categorical sphere of diagnosis is fundamentally wrong and has to be replaced by a dimensional system, if we are to be governed by factual evidence (Eysenck 1960; 1970). The low reliability of categorical diagnoses, resulting from overlapping classifications, is notorious, and the claim that the intersituation has been remedied by successive editions of the APA Diagnostic and Statistical Manual has not been found to be justified (Kirk & Kutchins 1992). Even here, sanity is breaking in; DSM-IV agrees that a dimensional approach is scientifically preferable, although refusing to use it in actuality because of old, established habits of medical diagnosis. Mealey suggests an absolute contrast between primary and secondary sociopaths but reports no evidence that these could be separated diagnostically with any acceptable reliability. From experience, I would be surprised if agreement between professional observers would be greater than 0.3 or thereabouts; clearly insufficient for scientific or practical purposes. I believe a much better picture would be one in which individuals were represented by points in a hollow, three-dimensional globe, the three diameters of which would represent Extraversion, Neuroticism, and Psychoticism (Eysenck & Gudjonsson 1989); the group of primary and secondary sociopaths would then appear as clusters of points in the E + N + P + octant, but certainly not as two quite separate and distinct clusters. The differences Mealey notes would locate primary sociopaths further out toward the periphery, and possibly closer to P than secondary sociopaths, who might be closer to the centre, and nearer to £ and N. But all differences would be dimensional, with no absolute demarcations. And, clearly, a system of diagnosis referring each point in this globular universe to the three dimensions, as a three-digit number, would be more reliable and more valid than a verbal type of construct, as suggested by Mealey. It would be possible, of course, to rotate these three reference axes to accommodate the theories of Gray and Cloninger, as Mealey suggests. I believe that both have made important contributions to the psychophysiological interpretation and understanding of personality-related behaviour, but I am not convinced that the evidence presented by them is adequate to suggest the rotation of the three primary axes. The latest study of the Cray system (Carver & White 1994) suggests to me that the activity of the BAS system aligns it with extraversion, that of the BIS system with neuroticism. But however that may be, and even if future research should force some degree of rotation, the principle of dimensional diagnosis would remain unaffected. If nature has not in fact produced some 300 separate psychiatric illnesses, as DSM suggests it has, then no human effort to force diagnoses into this Procrustean bed is likely to be successful. Nor will efforts to force behaviours into a two-type system be any more successful. Primary and secondary psychopaths may occupy different positions in our globe, but there are innumerable gradations observable in the points lying between pure examples of these points. If the notion of categorical disease entities breaks down completely when such time-honoured groups as schizophrenics and manic-depressives are concerned (Kendell & Brockington 1980), what hope is there for primary and secondary sociopaths? Primary sociopaths are said to be mainly activated by genetic factors, secondary ones by environmental ones, but surely no one would argue that the ratio of these two factors is not infinitely variable, and at present incapable of measurement. All the genetic and environmental correlates and possible causes of both conditions (and criminality as well) are continuously variable; they cannot conceivably give rise to two categorically distinct groups. Indeed, it is not even clear just how Mealey would diagnose her primary sociopaths phenotypically. She says that "there will always be a small, cross-culturally similar and unchanging baseline frequency of sociopaths and a certain percentage of sociopaths . . . will always appear in every culture, no matter what the socio-cultural conditions" (sect. 3.1.1, 556 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 para. 1). But how would we ever test such an hypothesis, in the absence of methods of diagnosing the genetic make-up of a given individual? And is it really suggested that each and every one of these primary sociopaths in fact had the identical genetic makeup? Surely there would always be a more-or-less and a gradual fading into the genetic make-up of the secondary sociopath, with no clear-cut boundary between them. And it is of course quite doubtful whether the proportion of sociopaths would indeed be identical from group to group. Rushton (1994) has reported evidence of very marked differences in criminality between racial groups; it is not unlikely that this portends similar differences in the number of primary sociopaths. Mealey also seems to suggest that treatment, whether preventive or later in life, should be categorically different for primary and secondary sociopaths. This again seems to go counter to fact. Unless diagnosis was well above the 0.70 level, any scheme of allocating treatment differentially to primary and secondary sociopaths would be doomed to failure. Considering that the great majority would fall between the two extreme pure groups, what should we do with them? Again, a graded, dimensional system would seem much better able to accommodate the facts of human diversity. Fortunately, it should not be difficult to translate the numerous psychological, hormonal, and physiological characteristics of Mealey's two types into a dimensional language, thus transforming an unnatural typological system into a much more natural dimensional one. The epigenesis of sociopathy Aurelio Jose Figueredo Department of Psychology, University of Arizona, Tucson, AZ 85721. ajf@arizvms/ajf@ccit.arlzona.edu Abstract: Mealey distinguishes two types of sociopathy: (1) "primary," or • obligate, and (2) "secondary," or facultative. Either sociopathy evolved twice, or one form is derived from the other, e.g., through: (1) genetic assimilation generating polymorphism in the relative strength of biases favoring the development of otherwise facultative strategies, or (2) independently heritable but strategically relevant characteristics biasing the optimal selection of facultative strategies. Mealey's target article represents an important contribution to the study of sociopathy. She is to be congratulated on a theoretically insightful synthesis and creative reinterpretation of a wideranging assemblage of scientific findings on the topic. This kind of work is a sure sign that evolutionary psychology is rapidly maturing in its ability to incorporate and accommodate otherwise fragmented and disparate empirical content from the traditional social sciences by providing a meaningful functional context and a powerful interpretive framework. Mealey's major claim is that sociopathy can be productively seen as an evolved adaptive strategy for intraspecific social parasitism, or "cheating." Although this identity remains far from conclusively established, the case for functional equivalence is too compelling to dismiss as mere coincidence. A less persuasive secondary claim, however, is the proposed distinction between "primary" and "secondary" sociopathy in both phenomenology and etiology. It is about this second and more minor claim that I harbor certain reservations. Although I have never worked explicitly on sociopathy, I have recently been involved in various collaborative research projects dealing with the deviant social and sexual strategies of competitively disadvantaged males. A major and recurring problem in this line of research concerns whether the deviant behavior patterns represent what behavioral biologists distinguish as "alternative" versus "conditional" adaptive strategies (Alcock 1989; cf. Mayr 1974). An "alternative" strategy can be defined as an adaptive strategy that is obligate for the individual, Commentary/Mealey: but for which the genetic predisposition is polymorphic within the population. A "conditional" strategy can be defined as an adaptive strategy that is facultative for the individual - contingent for its phenotypic expression upon environmental circumstances - but for which the genetic preparedness is monomorphic throughout the population. In a study of domestic violence, Figueredo and McCloskey (1993) found that a structural equations model, based on 21 manifest indicators that were theoretically related to the depressed mate quality of the perpetrator, predicted 60% of the variance in spouse abuse and an associated 25% of the variance in child abuse. In a clinical sample of juvenile sex offenders, Russell et al. (in preparation) found that a series of moderated multiple regressions predicted a sequential progression from psychosocial deficits to sexual deviance to nonsexual criminality and, finally, to sexual criminality, with psychosocial deficits functioning as significant upregulators of that sequential progression. In both of these path models, the development of deviant strategies was driven by individual failure at mainstream social and sexual strategies, seemingly favoring the "conditional" strategy hypothesis. In an explicitly behavior-genetic study, however, Rowe et al. (in press) found that the single most important predictor of juvenile delinquency, a measure reflecting the relative allocation of reproductive effort into mating effort (as opposed to parental effort), was highly heritable. In this study, individual failure at mainstream social and sexual strategies seemed more a consequence of this behavioral deviance than its initial precondition, seemingly favoring the "alternative" strategy hypothesis. Regrettably, we have never measured sociopathy per se in any of these studies but it would be surprising if it were not among the major contributors to these deviant behaviors. Nonetheless, Mealey's proposal that sociopathy is discriminate into two distinct types, based primarily on degree of genetic preparedness (cf. Garcia & Ervin 1968; Garcia et al. 1974; Seligman 1970; Seligman & Hager 1972), would greatly mitigate the apparent discrepancies between these various findings. I am nevertheless reluctant to accept that formulation. For one thing, conditional and alternative strategies are not necessarily mutually exclusive mechanisms. Certain microevolutionary processes, such as genetic assimilation (Waddington 1957) may produce developmental biases favoring the development of certain phenotypes over others in response to specific environmental contingencies. Thus, even within the context of conditional strategies, genetic polymorphism in the relative strength of these developmental biases could be generated in a population of otherwise facultative strategists. In addition, it is possible that the chance individual possession of independently heritable (but strategically relevant) characteristics may bias the selection of adaptive strategies (i.e., "reactive heritability'). For example, an individual with a sexually attractive phenotype will be more effective at pursuing certain mating strategies than others and would, thus, be rationally expected to select them preferentially. In the absence of more conclusive evidence I am personally biased toward the belief that it is more parsimonious to postulate a continuous trait than a discontinuous typology. This is not to say that such discrete typologies do not exist, as there are numerous nonhuman animal examples (Alcock 1989). In this case, however, it may be necessary to postulate that sociopathy arose not once but twice in human behavioral evolution; otherwise it is necessary to derive one form secondarily from the other, as the mechanisms proposed above purport to do. These and other plausible alternative hypotheses should be more fully explored, if not exhausted, before accepting the reality of a more complex state of affairs. Nevertheless, I must acknowledge that Mealey's theory is at least consistent with all of the data in this line of research that I have had direct involvement with, and holds the further promise of reconciling certain seemingly discrepant results. The sociobiology of sociopathy "Just So" stories and sociopathy Andrew Futtermana and Garland E. Allen" "Department of Psychology, College of the Holy Cross, Worcester, MA 01610; bDepartment of Biology, Washington University, St. Louis, MO 63130. futterman@hcacad.holycross.edu; allen@biodept.vmstl.edu Abstract: Sociobiological explanation requires both a reliable and a valid definition of the sociopathy phenotype. Mealey assumes that such reliable and valid definition of sociopathy exists in her "Integrated Evolutionary Model.' A review of psychiatric literature on the diagnosis of antisocial personality disorder clearly demonstrates that this assumption is faulty. There is substantial disagreement among diagnostic systems (e.g., RDC, DSM-III) over what constitutes the antisocial phenotype, different systems identify different individuals as sociopathic. Without a valid definition of sociopathy, sociobiological theories like Mealey's should be viewed as entirely speculative. Sociobiologists who study human behavior are confronted with a daunting task. They seek to provide a compelling evolutionary explanation for the emergence of behavioral phenotypes such as altruism, sociopathy, alcoholism, homosexuality, or any other trait that seems at first glance to be individually maladaptive. In order to do so, they must demonstrate that the phenotype advances the fitness of individuals, or relatives who demonstrate it. Darwinian fitness is claimed first by demonstrating that the behavioral phenotype in question is stable and consistent across environments, and then, by demonstrating that this behavioral phenotype in some manner confers an adaptive advantage over rival traits. In circumstances in which they cannot demonstrate consistency of phenotype or little is known about genetic and developmental mechanisms, sociobiologists often introduce evolutionary "models" to "explain" the trait's persistence. Another term for these models is "just so" stories, taken from the Rudyard Kipling book in which animal curiosities are fancifully and allegorically explained for the amusement of children (e.g., "How the Leopard Got His Spots"). However troubling or important these behaviors may be, such behavior-genetic explanations amount to little more than speculations that may sound reasonable but for which there is little or no evidence. In the case of the "just so" story to explain the persistence of sociopathy developed by sociologist Linda Mealey, not only are the mechanisms and developmental processes underlying the sociopathy trait not specified, but the stability and consistency for the sociopathy phenotype itself has not even been demonstrated. And unless it can be demonstrated empirically that a sociopathy phenotype remains stable over a range of environments, a study of its genetic base becomes impossible. One reason for this is that the classification scheme for sociopathy is not firmly fixed so that we should have any confidence that descriptions of the trait are either reliable or valid. Unlike the classification of Kettlewell's moths as speckled, dark, or white in rural and industrial Britain - a classification scheme that is broadly understood relative to selection effects and remains largely unchanged and unquestioned after 90 years - the validity of the classification scheme for sociopathy has changed frequently, and the properties that comprise the syndrome have been and still are assessed in a multitude of different ways. Moreover, unlike judgments of moth color, identification of sociopathy requires judgments of properties such as "lack of remorse," "impulsivity," "reckless disregard for others," and other social behaviors (American Psychiatric Association 1994) that are subjective and of highly variable description. Thus, that Mealey treats sociopathy "inclusively" should not prompt us to concede that she has identified sociopathy in any meaningful way or that she or other independent observers can assess the disorder reliably. As Kitcher (1985, p. 124) rightly points out: "Those who engage in studies of evolution of behavior need to do some work before they can reach the point from which Kettlewell began." [See also BBS, multiple book review of Kitcher's BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 557 Commentary/Mealey: The sociobiology of sociopathy Vaulting Ambition BBS 10(1) 1987.] Unless sociobiologists like Mealey can demonstrate that sociopathy is defined consistently and in meaningful ways, namely, that her definition meets the standards of an evolutionary, stable strategy, she has not reached "the point from which Kettlewell began." Her whole theory is built like a house of cards - if the bottom card falls, the whole edifice collapses. Recent research on the validity of psychiatric diagnosis clearly suggests that sociopathy - or antisocial personality disorder (ASP) as the disorder is known - is not a demonstrably reliable or valid diagnosis. For one thing, ASP has received four different "standard" definitions in the past 15 years - the Research Diagnostic Criteria (RDC, Spitzer et al. 1978), the Diagnostic and Statistical Manual, Third edition (DSM-III, American Psychiatric Association 1980), Third edition-Revised DSM-III-R, APA, 1987), and Fourth edition (DSM-IV, APA 1994). Over twodozen changes have been made in the definition of the ASP syndrome from the RDC to DSM-IV. Although some of these changes are apparently minor — for example, the inclusion of an additional 12th symptom "initiation of fights" prior to age 15 (DSM-III) opposed to the 11th symptom listed in the RDC), others are more theoretically significant and suggest a different conception of ASP (e.g., the inclusion of the adult symptom "lack of remorse" in DSM-III-R). It is important to note that the different classification schemes do not typically identify the same people as antisocial. Prevalence rates (the percent of those identified as ASP) can differ by as much as 800% using different definitions in the same sample. Moreover, agreement rates for different methods of classification of ASP are generally poor (e.g., interview-based versus self-report versus family history methods; Kosten & Rounsaville 1992; Zimmerman & Coryell 1990). Thus, it is not true that one researcher's RDC ASP will necessarily be like another researcher's DSM-IV ASP, or that sociopathy assessed with one method interviewer will be like sociopathy assessed by another. The upshot here is that when we focus on sociopathy, we are literally seeing a moving target. Difficulties also exist in demonstrating the reliability and validity of particular symptom ratings, for example, "lack of remorse" or "impulsivity," that comprise ASP diagnoses. Existing data clearly show that items assessing core traits of sociopathy have very low inter-rater agreement and are poorly correlated with other criteria for the ASP syndrome (cf. Carroll et al. 1993). Moreover, researchers have different views of what constitutes necessary features for ASP (and for that matter, most personality disorders; Livesley et al. 1987). Thus, even at the level of prototypic symptoms, sociopathy remains a muddy and an ill-conceived notion. So where are we? Without a clear idea of what the sociopathy is, sociobiological theories of sociopathy such as Mealey's develop a mental terrain that is only tangentially connected to available data. That rarely stops the press from uncritically accepting and presenting such theories as "state of the art." What is worse, such representations present a hopelessly simplistic and naive picture of genetics and human behavior that serves as an explanation for mounting economic and social problems in modern American society. If increase .in violence, crime, and other so-called sociopathic behaviors can be attributed to innate biological defects, then current social practices and policies are not to blame. The victims themselves are to blame; no matter that more jobs and benefits have been cut in the past three years than in the preceding decade, that pay and vacation time have been reduced, and that working hours and workplace stress have both increased. The evolutionary model of sociopathy focuses the problem back on the individual, and the status quo remains unchanged. That more powerful forces in our society - those elites on Wall Street and in Washington who manage our economic and political system - solicit and give wide publicity to such theories is, of course, Machiavellian. That some sociobiologists and psychologists flock to provide such 558 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 explanations based on minimal or hypothetical data, borders on irresponsibility. The primary/secondary distinction of psychopathy: A clinical perspective Gisli H. Gudjonsson Department of Psychology, Institute of Psychiatry, Denmark Hill, London SE5 8AF, England. Abstract: In this brief commentary the author concentrates on the treatment perspectives of Mealey's model. The main weakness of the model is that it does not provide a satisfactory theoretical connection between treatment and different types of target behavior. Even within the primary-secondary distinction, there are large individual differences that should not be overlooked in the planning of treatment. Mealey has made a formidable attempt to produce an "integrated evolutionary model' of psychopathy. Her main conclusion is that there are two fundamentally different etiologies or pathways to psychopathy - "primary" and "secondary" - which have distinct prevention and treatment implications. The primary/secondary distinction is, of course, not a new idea, and Mealey's main contribution is to integrate these concepts into a comprehensive evolutionary model. In this brief commentary I shall focus on the clinical perspectives of Mealey's model, particularly in relation to treatment issues. I agree with Mealey that a proper theoretical framework is essential for the classification, diagnosis, prevention, and treatment of psychopathy. Mealey's model postulates that only "secondary psychopaths" are likely to be responsive to treatment interventions. In contrast, "primary psychopaths" require a firm response from the criminal justice system while simultaneously providing them with constructive alternative activities to crime. How this might be achieved in practice is not specified. The main weakness of the model is that it fails to provide a clear theoretical connection between treatment techniques and different types of target behavior. The goals identified and the techniques recommended for modifying the undesirable behavior, including the psychopathy itself, seem rather vague. The most important message is that because the background histories of psychopaths are so variable, the programs used have to suit the needs of the individual psychopath. From a clinical perspective, the extensive work of Blackburn (1994) and Copas et al. (1984) on psychopaths in British hospital settings is important, but there is no reference to it in Mealey's target article. I believe this is a serious omission, since some of the ideas in Mealey's article overlap with those of British authors and do not represent distinct "new" knowledge. The British work of Blackburn and Copas and his colleagues indicates that psychopaths seen in clinical settings most typically resemble secondary psychopaths. However, even within the classification of secondary psychopathy, not all subjects so diagnosed are equally treatable. For example, the study of Copas et al. (1984) showed that as far as therapeutic community treatment is concerned, the anxious and intropunitive psychopaths responded most favourably to treatment programs, whereas treatment was least effective with the anxious psychopaths with extrapunitive personality. These findings suggest that even within the secondary psychopath category there are likely to be individual differences in responsiveness to treatment which are related to personality. Indeed, it may be unwise, for the purposes of treatment programs, to view secondary psychopaths as a homogeneous group. In their 1984 paper, Copas and his colleagues also expressed important treatment implications from their findings for "secondary" and "primary" psychopaths, along very similar lines to those Commentaryi'Mealey: The sociobiology of sociopathy recommended by Mealey. That is, the secondary psychopaths were as a group more responsive to treatment that assisted them to overcome their high anxiety and acting-out behavior, whereas the primary psychopaths, in contrast, needed an approach that confronted them with the consequences of their behavior. Primary psychopaths, according to Mealey's model and previous empirical research cited in her paper, are deficient in social emotions such as shame, guilt, and empathy, making them susceptible to continued irresponsible and criminal behavior. In addition, according to Mealey's model, this lack of propensity for social emotion in the primary psychopath is crucial in differentiating them from secondary psychopaths, and it makes them relatively unresponsive to environmental influences, such as those found in most treatment programs. Unlike primary psychopaths, secondary psychopaths do experience some feelings of guilt and anxiety. Indeed, in our own work (Cudjonsson & Roberts 1983; 1985) we found that both male and female secondary psychopaths obtained significantly higher guilt and anxiety scores on the Eysenck Personality Inventory and the Mosher True-False Guilt Inventory, respectively. Therefore, not only do secondary psychopaths report feelings of anxiety and guilt, unlike primary psychopaths, but their scores significantly exceed those typically reported by normal subjects. In addition, we found that, unlike normal subjects, the secondary psychopaths reported constantly experiencing strong feelings of shame and guilt that were unrelated to specific situational transgression. That is, these feelings seemed to be a reflection of their generalized poor self-concept and negative preoccupation rather than a consequence of specific situational transgression. This may explain why feelings of guilt and shame in secondary psychopaths appear ineffective in preventing future transgressions (i.e., feelings of guilt do not show the typical increase in subsequent prosocial behavior as predicted in Mealey's model). The implication is that treatment with anxious and guilt-ridden secondary psychopaths should focus on improving their self-concept as well as on helping them cope with their high levels of anxiety and acting-out behavior. Implications of an evolutionary biopsychosocial model Harmon R. Holcomb III Department of Philosophy, University of Kentucky, Lexington, KY 40506-O027. holcomb.ukcc.vky.ech Abstract: Mealey's work has several interesting implications: It refutes the charge that sociobiology paints a cynical portrait of human nature and adopts a one-sided reductionisin; it exemplifies a general theoretical scheme for constructing evolutionary biopsychosocial models of human behavior; and it has the practical effect of promoting and informing early intervention in children at risk for psychopathic disorder. Mealey's evolutionary biopsychosocial model of psychopathy refutes the standard charge that sociobiology paints a cynical view of human nature. As products of natural selection, humans are motivated to do whatever it takes to perpetuate their genes. This is commonly interpreted to mean that we are egoistic manipulators who make cold-hearted calculations (or act as if we do) toward the end of reproductive success. By inferring psychological states (egoistic manipulators) directly from evolutionary processes (individual selection), the standard charge fails to distinguish proximate and ultimate levels of description. Mealey's lesson is this: first distinguish, then integrate these levels of description. If the standard charge were correct, we would all be psychopaths and the sociobiology of psychopathy would apply to all humans. On the contrary, Mealey's sociobiological study reveals that psychopaths arise from one of two very specific, developmentally distinct etiologies that emerge from two specific evolutionary mechanisms, showing how and why psychopaths differ from the rest of us. Whereas critics of sociobiology often contrast "genetic," "biological," and "evolutionary" with "psychological" and "social," evolutionary theory applies to human behavior only when species-typical and local biological, psychological, and social (cultural) circumstances are taken into account. Instead of a onesided reductionisin, sociobiology adopts an integrated causal model of human behavior. Mealey's integration of biological, psychological, and social factors within evolutionary theory helps us see how we can change human behaviors by changing factors in any of these three categories; sociobiology need not "provide a biological excuse for the status quo." Although the specific mix of biopsychosocial factors Mealey exhibits for psychopathy is unique, it illustrates a general scheme for constructing evolutionary biopsychosocial models of human behavior. Can she tell us more, in general, about how such models apply the evolutionary conception of human actors as fitness strategy (or adaptation) executors, and how such models interrelate biological, psychological, and social factors as inputs to development? By.going beyond heritability studies to focus on development, we can better discern how genetic factors are pertinent to a practical understanding of human behavior. Mealey cites research into psychopathy using quantitative genetics, which analyzes correlations gained from studying variations among individuals. But heritability studies have nothing to say about the causes of traits in particular individuals or about whether the same individuals treated differently or raised in different environments would develop the same or different traits. For a causal analysis pertinent to social policy directed at changing psychopathic behavior, developmental analysis is necessary. Given the two developmentally distinct types of psychopaths, Mealey's reasonable suggestion is that deterrence will work for primary psychopaths and that reduction of risk factors will work for secondary psychopaths. However, it would be a mistake to assume that what is special to psychopaths is the only route to effective intervention. Children who turn into killers often report that they acted out of fear, hatred, or self-defense; both psychopathic and nonpsychopathic killers are responding to perceived threats to their fitness. They feel trapped and have learned that violence is their only way to remove these threats. Teaching people to identify other ways to solve problems - how to deal with problems without crime, manipulation, and violence - is just as applicable to psychopaths as to everyone else. Deterrence and reduction of risk factors may very well work better when supplemented with problem-solving guidance than when used alone. There are limits to what environmental intervention will achieve. We cannot teach primary psychopaths how not to manipulate people; they do not have a "moral conscience" that could get the message. Suppose we tried to teach them how to manipulate people without breaking the law. A limit is imposed by the fact that many laws are enacted to prevent people from manipulating others, which is why psychopaths are so often criminals. Indeed, standard diagnostic evidence for antisocial behavior concerns behaviors often regulated by law. The Diagnostic and Statistical Manual of the American Psychiatric Association (1987) used at least four of the following criteria to indicate antisocial behavior after the age of 15: (1) inability to sustain consistent work behavior, (2) failure to conform to social norms with respect to lawful behavior, (3) irritability and aggressivity, as indicated by repeated physical fights or assaults, (4) repeated failure to honor financial obligations, (5) failure to plan ahead, or impulsivity, (6) no regard for truth, as indicated by repeated lying, use of aliases, or "conning" others, (7) recklessness regarding one's own or others' personal safety, as indicated by driving while intoxicated or recurrent speeding, (8) inability to function as a responsible parent, (9) failure to sustain a monogamous relationship for more than one year, and, (10) lacking remorse (feeling justified in having hurt, mistreated, or stolen from another). To re- BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 559 Commentary/Mealey: The sociobiology of sociopathy duce drastically the criminal behavior of psychopaths, we would have to change the laws to make extreme forms of these behaviors legal, something we will never do. The law is biased against psychopaths by design; laws are made for and by nonpsychopaths. Health care professionals already use combinations of biological, psychological, and social traits to diagnose health conditions. An easy way to make evolutionary biopsychosocial models practically useful is to use them to inform such diagnostic criteria. Mealey's thesis provides an example. First, the manual cited above reserves a diagnosis of "antisocial personality disorder" for adults (18 or over) and uses its typical childhood signs to diagnose "conduct disorder." A developmental approach suggests that the diagnosis of antisocial personality disorder should be expanded in scope to cover both children and adults. Second, the diagnostic criteria should be augmented to include the many symptoms of psychopathy cited in the variety of studies Mealey reviews. Third, the diagnostic criteria should distinguish diagnoses of primary and secondary sociopathy. Genes, hormones, and gender in sociopathy Katharine Hoyenga Department of Psychology, Western Illinois University, Macomb, IL 61455. hoyengak@ccmall.wlu.bgu.edu Abstract: Although serotonin, testosterone, and genes contribute to sociopathy, the relationships are probably indirect and subject to modifiers (e.g., present only under certain conditions of rearing and temperament). Age at menarche may be a marker variable as well as a causal factor. Since the genders differ in all four areas, sex differences in sociopathy represent a very complex interaction of these factors. Overall, I find Mealey's arguments persuasive and her theory provocative. I do have a few quibbles; those of us who look for biosocial covariates must take into account not only the consistent but also the discrepant results. First, the genetic evidence is not quite as clear or as consistent as implied. As Mealey pointed out, adoption studies find evidence for the heritability of property crimes only, not violent crimes. In addition, although heritability estimates of twin studies are often higher because of nonadditive genetic influences, identical twins also have stronger tendencies to imitate each other's violence (Carey 1992; Rowe 1990), which would spuriously inflate the heritability estimates derived from twin studies. Second, the relationship between testosterone (T) and violence is also not very consistent (Hoyenga 1993; Hoyenga & Hoyenga 1993a). I think that some of the variability in results, and weakness of the effects, can be attributed to genetic and developmental variability. People differ in their sensitivity to T, and before T can be related to aggression, certain other traits such as personality traits (with genetic and developmental components) might also have to be present. Overall, aggression may be more sensitive to postpubertal than to perinatal T, but failures to replicate are more numerous than positive results. Age at menarche may be a marker variable, as well as serving as a causal factor as described by Mealey. Age at menarche has genetic components and is also sensitive to disruptions in family life (Hoyenga & Hoyenga 1993b). Thus, early menarche might be a marker for mechanism 1 (primary), as well as being a causal component for mechanism 5 (secondary). Although Mealey covered some aspects of the role of temperament in the development of sociopathy, most of the work summarized in that section concerned psychopathological traits, and sociopathy can also be related to extreme forms of normal personality traits. For example, in a factor analytic study done with the Psychopathy Checklist, Revised, two factors were found (Harpur et al. 1989). The first corresponded almost perfectly to the interpersonal personality trait of hostile dominance, which is a blend of dominance and lack of nurturance. The second factor 560 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 found in the Harpur study - the one most closely related to acts of violence - did not seem to be a personality trait. Similarly, Foreman (as described by Hare et al. 1991) found that PCL-R total scores were also positively correlated with staff ratings of dominance and negatively correlated with nurturance. Although the relationship between impulsive violence and low serotonin is strong and often replicated, there are inconsistencies here as well. Some people tend to respond to an increase in brain serotonin with an increase rather than with the expected decrease in hostility and anger. This somewhat atypical reaction may be found more often in anxiety neurotics (Germine et al. 1992; Kahn et al. 1988). Also, in both humans (the Madsen [1985] reference cited by Mealey) and many nonhuman primates, dominance is positively correlated with brain serotonin levels (e.g., Brammer et al. 1991; Raleigh & McGuire 1986; Steklis et al. 1986). It seems that in some primate males, becoming dominant depends on successfully affiliating with several females; males with high aggression cannot affiliate, but males with high serotonin show more afRliative behaviors and so can increase their dominance status in the group. Sex differences in sociopathy may reflect something more complicated than implied by a two-threshold model. For example, females of a variety of species reliably have more brain serotonin or more active brain serotonergic systems than do males (Hoyenga & Hoyenga 1993a; 1993b). In addition, because of socialization and sex limitation, the sexes differ in the development of the relevant temperamental traits, including the traits of dominance and nurturance mentioned above (Hoyenga 1993). Thus, gender-related differences in sociopathy may reflect the simultaneous actions of multiple interacting factors. Al Capone, discrete morphs, and complex dynamic systems Douglas T. Kenrick and Stephanie Brown Department of Psychology, Arizona State University, Tempe, AZ 85287-1104. atdtk@asuacad.bitnet; assrb@asuacad.bltnet Abstract: We consider four mechanisms by which apparent discontinuities in the distribution of antisociality could arise: (1) executive genes or hormonal systems, (2) multiplicative interactions of predisposing factors, (3) environmental tracking into a limited number of social roles, and (4) cross-generational gene—environment interactions. A more explicit consideration of complex self-organizing dynamic systems may help us understand the maintenance of antisocial subpopulations. Even normal people occasionally consider antisocial acts such as homicide (Kenrick & Sheets 1994). Most of us keep such inclinations under restraint, though, and cannot empathize with characters like Al Capone, who invited three men to dinner, then had them tied to their chairs and beat them with a baseball bat. Capone showed the signs of primary sociopathy: personal charm, an early career of violence, and little remorse, fear, or sympathy. Though such cases fit Mealey's distinction, much of the evidence she reviews suggests a continuous distribution from well-socialized individuals through "secondary sociopaths" and "primary sociopaths." Is a discontinuity between primary sociopaths and the rest of the population plausible? What could produce a discontinuous distribution of antisocial behavior? We consider four mechanisms that could contribute to discontinuity in antisociality. 1. Executive genes or hormonal systems. If people vary continuously in levels of testosterone, adrenaline, serotonin, and dopamine, and if one hormonal system is independent of another, one expects a continuous rather than discontinuous distribution of antisociality. However, internal organization could yield a smaller number of discrete settings. Mealey mentions testosterone's developmental organizing role. In a related vein, gender differences result in two discrete morphs, Commentary/Mealey: The sociobiology of sociopathy not continuous variation in size of reproductive glands. Selflimiting discrete organizations may occur within each sex, as in certain fish species, when some individuals of the same sex grow into completely different adult morphs (Gross 1984). 2. Multiplicative interactions of predisposing factors. Apparent typologies or quasi-typologies could arise from multiplicative interactions of underlying physiological dimensions (see Table 1). Imagine several individuals with settings of either 2 or 4 on three underlying dimensions. Assuming additive combination, an individual with a 4 on each dimension would be twice as antisocial as one with all 2s (12 vs. 6). Assuming multiplicative combination, however, the highest setting would be eight times as antisocial as the lowest (64 vs. 8), yielding a distribution appearing more like a typology. Even assuming continuously distributed characteristics, multiplicative interactions enhance dissimilarity between highs and lows (see Fig. 1), and allow for antisociality even with a high environmental threshold (as found for the upper classes). 3. Environmental tracking into a limited number of social roles. Small and continuous differences between children amplify over the lifespan. A child with a small initial edge in basketball, music, or chess gets training and grows into a professional many times better than grade-school cronies. Mealey discusses how children having common troubles with authorities band together and amplify one another's antisociality. Environmental tracking would capitalize on initial genetic proclivities. Just as animals with tools such as talons have different cost/benefit weightings for aggression; an outlaw strategy could yield a more favorable cost/benefit ratio for attractive, lowfemotion people. 4. Cross-generational gene-environmental interactions. Societal forces may artificially select extremes of exploitativeness. Over time, laws punishing cheaters could increase differentiations between cheaters and cooperators, since half-hearted or inept exploitation will be punished, and the middle levels become less populous. Antisocial behaviors may also be maintained by reward structures. If physically attractive high-risk individuals choose one another as mates, the package of Cleckley characteristics is increasingly maintained by speciation-Iike processes. Note that ruthless promiscuous individuals who disregard social convention include spies and military heroes as well as criminals. Al Capone became incredibly wealthy and politically powerful, and in Sicily, mafiosi received esteem from the local populace as well as advantageous marital and extramarital mating opportunities (Serviado 1976). Payoffs for "antisocial" behavior come not only from cheating and backstabbing: people respect and are attracted to "fearlessness." thing but your success. Conventional individuals sometimes band together to lynch or tar and feather those who take more than their share of reproductive and material resources. On this view, it is fairer to envision dynamic conflict between different mating strategies and jettison the term "antisocial" for those playing one strategy. Complex dynamic systems from simple underlying properties. Mealey uses the term "dynamic equilibrium" but does not discuss emerging notions about complex self-organizing dynamic systems. Interesting things emerge when the Prisoner's Dilemma is expanded to larger matrices. We modelled primitive 6 x 6 networks and found that, when all individuals in an imaginary "neighborhood" were set to a similar strategy (act aggressive if 50% of neighbors were aggressive on the previous iteration), self-maintaining pockets of mutual aggression and mutual cooperation emerge, with occasional pockets of peripheral instability. Individuals set to more aggressive thresholds remain peaceful in unaggressive neighborhoods but catalyze the spread of pockets of aggressiveness. Such models are being used to map complex interactions between genes, between neurons in cognitive systems, and between species in ecosystems. How do cognitive dynamics map onto internal physiological systems and inputs from the larger social environment? Perhaps underneath the seemingly continuous variation of the social world are only a (relatively) few interlocking mechanisms. Though the proliferation of scientific findings sometimes moves us toward information overload, the insights of evolutionary science, dynamic systems theory, and cognitive neuroscience could be aiming us toward a metasynthesis of Copernican magnitude. At our present level of understanding of these systems, though, interventions such as Mealey suggests may be presumptuous and even dangerous. More certain punishments for antisociality, for instance, could further widen the gap between conventional and unconventional individuals. Draconian measures exact a high cost on marginally conventional and generally nondangerous individuals, while selecting the ranks of the Capones for even more deviousness and ruthlessness. (A) ADDITIVE MODEL Oscar Wilde noted, however, that people forgive you anyLOW Table 1 (Kenrick & Brown). A depiction of eight HWH Physiological Risk Factors hypothetical individuals varying on three (B) MULTIPLICATIVE MODEL dimensions relevant to antisocial behavior Resulting antisociality Traits Anxiety Sensation seeking Behavioral inhibition Additive model Multiplicative model 2 2 2 4 2 4 4 4 2 2 4 2 4 2 4 4 2 4 2 2 4 4 2 4 6 8 8 8 10 10 10 12 8 16 16 16 32 32 32 64 Physiological Risk Factors Note: Higher scores indicate more proclivity toward antisocial behavior. Figure 1 (Kenrick & Brown). How multiplicative interactions of underlying factors can produce an apparent typology. If several underlying factors contributed to antisocial behavior, and combined additively, the results would be a fairly continuous distribution, with many individuals at or near the threshold, as in (A). However, multiplicative interaction can enhance differences, so that some individuals would clearly be above most environmental thresholds, and others clearly below, as in (B). The threshold for engaging in antisocial behavior could vary as a function of environmental inputs, such as population size, availability of resources, and normative pressures. BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 561 Commentary/Mealey: The sociobiology of sociopathy An evaluation of Mealey's hypotheses based on psychopathy checklist: Identified groups David S. Kossona and Joseph P. Newmanb "Department of Psychology, Finch University of Health Sciences /Chicago Medical School, North Chicago, IL 60064; "Department of Psychology, University of Wisconsin - Madison, Madison, Wl 53706. jpnevvman@facstaff.wisc.edu Abstract: Although Mealey's account provides several interesting hypotheses, her integration across disparate samples renders the value of her explanation for psychopathy ambiguous. Recent evidence on Psychopathy Checklist-identified samples (Hare, 1991) suggests primary emotional and cognitive deficits inconsistent with her model. Whereas high-anxious psychopaths display interpersonal deficits consistent with Mealey's hypotheses, low-anxious psychopaths' deficits appear more sensitive to situational parameters than predicted. We applaud Mealey's attempt to integrate diverse literatures and to elucidate the significance of the historical distinction between primary and secondary psychopathy. However, although several of her ideas are quite interesting, much is lacking in the empirical basis for her proposals regarding primary and secondary psychopathy. Mealey acknowledges the problems associated with failures to distinguish psychopathy from criminality and antisocial personality disorder, but she proceeds to integrate data without regard to this distinction. There is now more than a decade of research on the Psychopathy Checklist (PCL; Hare 1980; 1991), a measure that assesses psychopathy based on individuals' personality as well as antisocial conduct. The PCL has impressive reliability and validity (Hare 1991; Kosson et al. 1990) and greater specificity than the American Psychiatric Association (1987) diagnosis of Antisocial Personality Disorder (see Hare et al. 1991). Moreover, applying taxometric methods to the PCL, Harris et al. (1994) have presented evidence that PCL ratings identify a discrete category or taxon. At least two partially correlated dimensions of psychopathy have been identified (Harpur et al. 1989). Chronic, impulsive antisocial behavior defines one dimension (viz., Factor 2); a second (viz., Factor 1) addresses the shallow, callous exploitation of others, regarded by many as the personality core of the disorder. To the extent that Factor 1 is essential for distinguishing psychopaths from antisocial personality and criminality, reviews like Mealey's that rely heavily on studies of criminals to bolster propositions about psychopaths are apt to be misleading. Case studies distinguish psychopaths from ordinary criminals (e.g., Cleckley 1976) and most research on psychopaths includes nonpsychopathic inmates as a control group. Moreover, psychopaths constitute only about 20% to 33% of male prison inmates (Hare 1991), so that differences between psychopaths and other criminal groups are obscured by studies on criminals as a whole. Indeed, most of the evidence Mealey uses to characterize primary psychopaths is probably more relevant to Antisocial Personality Disorder (i.e., Factor 2) than to psychopathy. Below, we examine Mealey's assertions about the psychological characteristics of primary and secondary psychopaths using, except as noted, PCL-defined psychopaths. Due to space constraints, we limit our review to the domains of emotional and cognitive function. 1. Emotional function In psychopaths. Emotional deficits play a central role in Mealey's account of primary psychopathy: psychopaths are deficient in social but not in primary emotions or moods. Preliminary studies of emotional processing in PCLdefined psychopaths provide some support for this position. Unlike nonpsychopaths, psychopaths' lexical decisions are not faster for emotional words (Williamson et al. 1991), and viewing unpleasant (vs. pleasant) slides does not increase their startle responses (Patrick 1994; Patrick et al. 1993). Moreover, these deficits are particularly related to Factor 1. 562 BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 With regard to emotional responsiveness, findings are inconclusive. Psychopaths display less autonomic arousal than nonpsychopaths during fear imagery (Patrick et al. 1994). Initial studies using film inductions provide evidence that psychopaths also show less disgust (Forth 1992) and report less affect to negative emotion film segments (Patterson 1990) but are not hyporesponsive on other measures. Thus, these studies suggest limited deficits for primary emotions. Few studies bear on Mealey's hypotheses regarding social emotions with PCL-identified psychopaths. However, extant studies provide little evidence that emotional deficits lead primary psychopaths "to continually play for the short-term benefit." (sect. 3.1.1, para. 2) Studies using the Prisoner's Dilemma game indicate that neither psychopaths (meeting Cleckley's 1976 criteria) nor low-anxious psychopaths (meeting the PCL or MM PI criteria) play more selfishly than nonpsychopaths, although high-anxious psychopaths do (Hare & Craigen 1974; Whitehill 1987; Widom 1976a). Sutker (1970) found psychopaths (defined by agency diagnoses and MMPI scores) as likely as controls to forfeit reward to prevent shocks to a confederate. In addition, studies of social sensitivity suggest that anxious or neurotic but not low-anxious psychopaths are deficient (e.g., Rime et al. 1978; Widom 1976b; Patterson 1990). Thus, although anomalies in psychopaths' emotional processing and responsiveness have been observed, there is little evidence that these underlie problems in interpersonal cooperation, and interpersonal deficits are largely limited to high-anxious psychopaths. 2. Integrating emotion and self-regulation. On the other hand, studies evaluating psychopaths' responsiveness to feareliciting stimuli while pursuing immediate rewards show, as Mealey notes, performance deficits in psychopaths. But in contrast to Mealey's view that such findings suggest emotional deficits per se, their situation-specificity is more consistent with deficits in information-processing integration (Newman & Wallace 1993). In comparison to low-anxious controls, low-anxious psychopaths display inhibitory problems in situations involving both reward and punishment incentives. Moreover, their "insensitivity" to punishment stimuli appears to reflect difficulty in shifting attention and interrupting ongoing behavior, which impairs their ability to profit from environmental cues (Kosson & Harpur, in press; Patterson & Newman 1993). Moreover, high-anxious psychopaths do not display deficits under these circumstances. They appear more able to interrupt goal seeking in response to punishment, whereas low-anxious psychopaths' responsiveness to punishment contingencies may depend on situational factors that make punishment salient prior to their initiating goal-directed behavior (e.g., Newman et al. 1990; Newman et al. 1992). This distinction resembles Mealey's assertion that secondary psychopaths respond to environmental contingencies, whereas a priori expectations dictate primary psychopaths' behavior. 3. Cognitive function In psychopaths. The proposal that psychopaths often adopt a cognitive (vs. affective) style of processing information was also stated by Newman and Wallace (1993). However, Mealey's proposal for accurate cost-benefit analyses implies intact cognition. Although psychopaths often perform as well as nonpsychopaths, subtle cognitive deficits are evident in specific situations. Becent evidence suggests specific deficits in divided attention, including reduced breadth of attention and deficient shifts of attention (Howland et al. 1993; Kosson & Harpur, in press). Moreover, these occur while subjects are actively engaged in goal-directed behavior and are related to both dimensions of psychopathy (Kosson 1995). Biases in appraisal have also been reported. Psychopaths display both hostile, attributional biases (Sterling & Edelmann 1988; Serin 1991; cf, Millon 1981) and optimistic biases, with the latter linked to their excessive goal seeking (Patterson 1990; Siegel 1978). Commentary/Mealey: The sociobiology of sociopathy 4. Toward a classification of secondary psychopaths. Compared with previous proposals for secondary psychopaths, Mealey's characterization of this group is unusual. Most proposals for "secondary" psychopathy explain antisocial behavior as secondary to some internal causal factor: neurotic conflict or anxiety (e.g., Karpman 1961; Lewis 1991), underlying schizotypy or schizoid tendencies (Hodgins 1994; Raine & Venables 1984), or inadequacy or passivity (e.g., Craft 1965). By contrast, Mealey's emphasis on competitive disadvantage, like Quay's (1986) category of "socialized delinquent," implicates individuals who are not emotionally or interpersonally impaired but develop antisocial values via social learning. Basic relations between competitive disadvantages, primary psychopathy, and other candidates for secondary psychopathy are also unclear. If anxiety and schizotypy are competitively disadvantageous, should antisocial behavior attributable to such factors be considered secondary psychopathy? Do low- and high-anxious psychopathy reflect separate genetic diatheses or a common (psychopathy) diathesis whose expression is moderated by level of anxiety? Answers to such questions appear fundamental to Mealey's prescriptions for change. In contrast to Mealey's general review, our more specific analysis of PCL-defined psychopaths suggests that they are characterized by deficits in both emotional and cognitive functioning that are more sensitive to situational parameters than Mealey predicts. Whereas some differences between low- and high-anxious psychopaths appear consistent with Mealey's distinction between primary and secondary psychopaths (e.g., integration of emotion with self-regulation), others do not (e.g., interpersonal sensitivity). In addition, correlations between self-reported anxiety and Factor 2 (Hare 1991) suggest that highanxious psychopaths should be well represented in the genetic studies of criminality that Mealey reviews. Given the apparent social deficits also displayed by those individuals, we imagine she would regard them as primary psychopaths. Indeed the possibility that Mealey s account could explain the behavior of high-anxious psychopaths remains an intriguing one which appears worthy of study with well-defined groups. Fatherless rearing leads to sociopathy David T. Lykken Department of Psychology, University of Minnesota, Minneapolis, MN 55455-0344. lykkeOOI@staff.tc.umn.edu Abstract: Endorsing Mealey s analysis, it is pointed out that increasing rates of crime and violence are due to increasing proportions of children being reared in circumstances radically different from the extendedfamily environment to which we are evolntionarily adapted, that is, they are reared without fathers. Mealey's valuable paper was especially interesting to me because, reasoning in terms of evolutionarily stable strategies, she arrives at a taxonomy of the family of antisocial personalities that is very similar to that which I derived from quite a different model (Lykken 1995). We can suppose that our species is adapted to live reasonably amicably in extended-family bands or tribal groups as they probably did in the Pleistocene. One of our species-specific characteristics is the capacity to become socialized, which means: (1) to develop a conscience that leads us to avoid antisocial behavior; (2) to develop prosocial inclinations such as empathy, altruism, respect for authority, and cooperativeness; and (3) to accept adult responsibilities, including those of parenting and pulling our own weight in the communal effort. However, like our capacity for language, this disposition toward socialization needs to be elicited and practiced during early childhood, and this training, in ancestral as in present-day traditional societies, was a responsibility shared by the parents and the extended family. Because we are adapted to grow up and live in such groups, extended-family child rearing is very effective in socializing youngsters and therefore intramural crime in traditional societies today is infrequent. The only chronic offenders are those individuals whose innate temperaments make them very difficult to socialize - the people Mealey calls "primary sociopaths" but whom I refer to as "psychopaths" in the diagnostic tradition that goes back to Cleckley (1941) and earlier. In her important study of mental illness in non-Western, primitive societies, the anthropologist Jane Murphy found that the Yupic-speaking Eskimos in northwest Alaska have a different name, kunlangeta, for this type of individual. The man who, for example, repeatedly lies and cheats and steals things and does not go hunting and, when the other men are out of the village, takes sexual advantage of many women — someone who does not pay attention to reprimands and who is always being brought to the elders for punishment. One Eskimo among the 499 on their island was called kunlangeta. When asked what would have happened to such a person traditionally, an Eskimo said that probably "somebody would have pushed him off the ice when nobody else was looking. " (Murphy 1976, p. 1026) Because traditional methods of socialization are so effective in tribal societies, where the extended family rather than just a particular parent-pair participate in the process, the kunlangeta probably possesses inherent peculiarities of temperament that make him unusually intractable to socialization. When a modern man and woman, often without any prior experience in dealing with young children (experience they would have had growing up in a traditional society), undertake the demanding task of parenting on their own, the failure rate (i.e., the crime rate) will inevitably be higher. I use "sociopath" to designate antisocials whose lack of socialization is primarily due to incompetent parenting. Because parenting is difficult, the failure rate is higher still for single parents. About 70% (!) of institutionalized delinquents grew up in homes without a father (Lykken 1995). My "sociopaths' are roughly equivalent to those whom Mealey calls "secondary sociopaths. " It is important to realize that sociopath and psychopath (Mealey's secondary and primary sociopaths) are endpoints on a continuum. With the best parents and home environments, the only antisocial offspring will be those who are the most fearless, aggressive, impulsive, and so on - psychopaths with truly hardto-socialize temperaments. In the worst home environments, a large fraction of all offspring will remain unsocialized. Over the broad middle range of parental competence and environmental risk factors, the incidence of antisocial offspring will be a product-function of parental incompetence (or indifference or parental sociopathy) and the child's innate proclivities. A complicating factor is that the worst parents are likely to contribute hard-to-socialize genetic tendencies as well. This line of thinking explains why the rate of juvenile violent crime has increased 100% since 1972 and is six to eight times higher among blacks than whites in the United States. The proportion of births to unwed mothers in the U.S. increased from about 5% in 1960 to about 30% today. In 1925, 85% of black families in Harlem were headed by fathers. By 1965, when Daniel Patrick Moynihan wrote his famous memorandum on the break-up of the black family, the rate of black illegitimacy had risen to 25%. Among urban blacks today, it is a rare youngster who is being reared by both biological parents. Since 1965, the increasing rates of arrest of juveniles in the U.S. for crimes of violence, plotted separately for blacks and whites, very closely parallel the increasing proportions of the two 12-17-year-age groups that were born out of wedlock (Lykken 1995). It is also important to realize that penal rehabilitation, psychotherapy, and other efforts at remediation do not work; the only solution to the increase in numbers ofjuvenile sociopaths is BEHAVIORAL AND BRAIN SCIENCES (1995) 18:3 563 CommentarylMealey: The sociobiology of sociopathy prevention. Parental training, as Mealey proposes, is a good investment for those few high-risk parents who can or want to be trained. But, for the majority of high-risk children, programs of professionalized foster care, of all-day care on the successful kibbutzim model, of single-sex boarding schools, and the like will be required. If each sociopath costs society at least $3 million in his lifetime (Westman 1994), such expensive programs may be justified. But, until the source of nascent sociopaths children born to incompetent parents - is substantially reduced (my proposal is parental licensure), this large, expensive, and dangerous problem will stay with us. Sociobiology, sociopathy, and social policy Richard Machalek Department of Sociology, University of Wyoming, Laramie, WY 82071. machalek@uwyo.edu Abstract: Evolutionary analysis suggests that policies based on deterrence may cope effectively with primary sociopathy if the threat of punishment fits the crime in the cost/benefit calculus of the sociopath, not that of the public. On the other hand, policies designed to offset serious disadvantage in social competition may help inhibit the development of secondary sociopathy, rather than deter its expres- Many social scientists in policy-oriented fields, especially criminology and sociology, have long feared that sociobiology necessarily leads to fatalism and hopelessness about antisocial behaviors. They reason, incorrectly, that adopting an evolutionary approach to human behavior means resigning oneself to the view that "biology is destiny." Mealey's analysis of sociopathy demonstrates clearly that this is not true. Instead, sociobiology can provide unique insights and help us clarify our options as we form policy responses to social problems such as those created by sociopathy. Consider what it means to replace the more conventional medical view of sociopathy with one informed by evolutionary reasoning. As the term itself suggests, the medical model attributes sociopathy to a "pathogen," in this case an emotional deficit that may be genetically rooted and physiologically expressed. This results in behavior that is described as "antisocial" (and thus "pathological"), because it is emotionless and uncompromisingly selfish. Although few among us would quarrel with characterizing sociopathy as pathological (except, perhaps, those who celebrate the rise of a Nietzschean, postmodern culture), evolutionary theory takes us beyond mere diagnostic descriptors and prompts us to ask whether such antisocial behaviors may, in some fundamental sense, be advantageous to those who express them. This is what Mealey does by exploring a conceptualization of sociopathy as a "life strategy" (Introduction, para. 6). In this view, the sociopath is often a highly rational but morally bankrupt strategist motivated uncompromisingly by self-interest. Framing sociopathy in evo