396 BIOCHEMICAL SOCIETY TRANSACTIONS be inferred both from the Ca' mobilization experiments (Connolly et ul., 1984) and from some metabolic observations (Nedergaard, 1981), such an increased level has not been directly demonstrated to date. 7. A Cu' -inducid opening oj K channels in the plusnia nienibranr. This would be a feedback step in this cascade, counteracting and stabilizing the membrane depolarization and allowing the K + gradient to approach the Nernst equilibrium. An 2-adrenergically stimulated release of K ( " , R b + )has been observed (E. Nanberg, J. Nedergaard & B . Cannon, unpublished work). Although the existence of a potential-gated K channel cannot be excluded at present, it is in agreement with the action of a Ca2+-dependent K + channel that the CaZ ionophore A23187 can stimulate K efflux from the cells (J. Nedergaard, E. Connolly, E. Nanberg & N. Mohell, unpublished work). The ionophore presumably acts by increasing cytosolic Ca2+,in parallel to what happens when the N a + / C a ' + antiport is stimulated physiologically. + + + + + + + Thp qicestiori of the iitiir~ersalitj~ of the phetionirnon We have here described a possible active regulatory role of the mitochondrial Na + / C a 2 antiport in brown adipose tissue. I t is possible that the antiport in this tissue is unusually important because of the very high abundance of mitochondria in the tissue, and because of the limited presence of other Ca' regulatory systems (endoplasmic reticulum). We would however propose that the significance of the antiport and the cascade system sketched could be more general. I t was, for example, initially believed that the mitochondria from liver did not possess the Na' /Ca2 antiport. but a series of investigators now claim its presence also here (Nedergaard & Cannon, 1980; Haworth et al., 1980; Heffron & Harris, 1981; Harris & Heffron, 1982; Goldstone & Crompton, 1982; J . Nedergaard, unpublished work). It may be argued that the activity of the antiport in liver is low, but functionally this is not the case, since also in incubation of liver mitochondria an elevation of the extramitochondrial CaZ level of about 1 PM can be induced by Na' (J. Nedergaard, unpublished work). I t may be added that extracellular N a + has been demonstrated to be essential for the mediation of r-adrenergic stimuli in liver (Haylett & Jenkinson, 1972a,b). Thus it is perhaps a general phenomenon that the mitochondrial Na /CaZ antiport is involved in the mediation of r,-adrenergic (and/or other hormonal) stimuli. + + + + + + This work was supported by the Swedish Natural Science Research Council. The authors would like to thank Barbara Cannon for valuable discussions. Al-Shaikhaly, M. H. M., Nedergaard. J . &Cannon, B. (1979) Proc. Nail. Acad. Sci. U . S . A . ,76. 2350 2353 Cannon. B. & Lindberg, 0. (1979) Methods Eiizrniol. 55. 65 78 Cannon, B., Nicholls, D. G. & Lindberg, 0. (1973) in Mc~chatii.sni.s in Bioenergetics (Azzone, G . F., Ernster. L., Papa, S., Quagliariello. E. & Siliprandi, N., eds.), pp. 357 363. Academic Press. New York Cannon, B., Sundin, U., & Romert, L. (1977) FEBS Lett. 74, 4346 Carafoli, E. (1981) in Mitochondria and Micwsonics (Lee, C. P., Schatz, G.. & Dallner, G.. eds.), pp. 357 374. Addison-Wesley, Reading, M A Carafoli. E., Tiozzo, R., Lugli. G., Crovetti, F. & Kratzing, C. (1974) J. Mol. Cell. Cardiol. 6. 3 6 1 ~371 Connolly, E., NAnberg, E. & Nedergaard, J . (1984) Eur. J. Bk~cheni.in the press Denton, R. M. & McCormack, J . G. (1981) FEBS Lcw. 119. I 8 Exton, J . H. (1980) Am. J. Phj~siol.238, E3 E l 2 Fain, J. N . & Garcia-Siinz, J . A. (1980) Lifi, Sci. 26. 1183- I194 Fink, S. A. & Williams, J . A. (1976) Am. J . Phwiol. 231, 700 706 Girardier, L. & Seydoux, J . (1971~)in Non.shirw"ng Thcwmgenc,.ris (Jansky. L., ed.), pp. 255-270, Academia, Prague Girardier, L. & Seydoux. J . (1971h) J. Physiol. (Paris)63, 147 186 Girardier, L.. Seydoux, J . & Clausen, T. (1968) J . Gcw. Physiol. 52. 925-940 Goldstone, T. P. & Crompton. M. (1982) Biochrni. J. 204. 3 6 9 ~371 Harris, E. J. & Heffron. J. J. A. (1982) Arch. Biochiwi. B i ~ ~ p h218, y~. 531-539 Haworth, R. A., Hunter, D. R. & Berkoff, H. A. (1980) FEBS Lcw. 110, 216-218 Haylett, D. G. & Jenkinson. D. H. (1972a)J. Physiol. 225, 721 750 Haylett, D. G . & Jenkinson. D. H. (1972h)J. P h y s i d . 225, 751 772 Heffron. J . J . A. & Harris, E. J . (1981) Biochem. J. 194, 925 929 Horowitz, J . M., Horwitz, B. A. &Smith, R. E. (1971) E.\pricwtia 27, 1419-1421 Lilien, 0. M. (1976) J. Urol. 116, 277-281 Lilien, 0. M., Krauss, D. J., Manke, C. G., Euser, B. A . & Subramanian, G . M. (1978) J . Urol. 119, 170-174 Locke, R., Rial, E., Scott, I. D. & Nicholls, D. G. (1982) Eur. J . Biochem. 129, 373-380 Mohell, N., Svartengren, J . &Cannon, B. (1983) Eur. J . Pharniacol. 92, 15-25 Nedergaard, J . (1981) Eur. J. Biochem. 114. 159-167 Nedergaard, J. (1983) Eur. J . Biocheni. 133, 185 191 Nedergaard, J . & Cannon, B. (1980) Actu Chcm. Scand. B34, 149 151 Nedergaard. J . & Lindberg, 0.(1982) Inter. RCT.c.vlfJ/.74. 187 286 Nicholls, D. G. (1978) Biochem. J . 176, 463 474 Nicholls, D. G. (1979) Biochim. Biophp. Acta 549, I 29 Nicholls, D. G. & Akerman, K . (1982) Biochini. Bii~phy.~. Acta 683. 57-88 Seydoux, J., Constantinidis, J., Tsacopoulos, M. & Girardier, L. (1977) J. Physiol. (Paris) 73, 985-996 Strieleman, P. J.. Edson. C. E. & Shrago, E. (1983) Fed. Proc. 42, I324 Electron flow in membranes ROBERT J. P. WILLIAMS Inorganic Chemistry Laboratory, Uniwrsit-v qj ' O.yjbrd, O.yfiird OX1 3 Q R , U . K . In 1961 two papers appeared describing the way external energy could be transduced to give internal biochemical energy, ATP (Mitchell, 1961; Williams, 1961). Both started from electron transfer in membranes. Detailed analysis of electron transfer in protein matrices (see Moore e t al., 1982) now makes it clear that the electron can hop from site to site covering about 1.5nm at each hop. The only problems that remain in this area concern the disposition of and the medium between sites. Given this hop distance it now Abbreviation used: Q, coenzyme Q seems very unlikely that electrons as such can travel across membranes in some of the particles which transduce energy, such as cytochrome oxidase. I say this since the construction of this protein generates haem a and the associated Cu well to the cytosolic side of the membrane so as to accept electrons from cytochrome c and it has only one other electron-transfer centre, the haem a , . Cu pair. The construction could hardly span a membrane and thus the idea of an electron path across a membrane, a central part of chemiosmosis (Mitchell, 1961) is likely to be incorrect in this case. A more plausible scheme is that the electron reactions are connected in the membrane to reactions, redox or otherwise, of protons also in the membrane so that energized protons (acids) are generated in the niemhrane (Williams, 1961). These protons then diffuse by restricted 1984 397 606th MEETING, CORK diffusion, i.e. in channels, firstly in the membrane and then either directly (locally) to the ATP synthase (Williams, 1961) or establish of necessity chemiosmotic gradients (Mitchell, 1961). Originally chemiosmosis had no proton diffusion channels. I show elsewhere (Williams, 1984) that very much evidence is now against the general equilibrations of chemiosmosis. This hypothesis appears now to be at fault at all four steps; (a) the electron path, (b) the proton path, (c) the direct field-driven synthesis of ATP, (d) the need for equilibrated aqueous gradients. The alternative of a more restricted electron flow, together with an energized proton flow in membranes and a local flow of protons through the ATP synthase accompanied by conformational change, is a more convincing description. It should be noted that the energized proton flow in a membrane (Williams, 1961), is of course a proton pump. I wish now to turn to the experimental situation relating to the recognized proton pumps, which are driven by electron flow in the mitochondria1 and thylakoid chains. (Note that the proton pump of bacteriorhodopsin, halorhodopsin and rhodopsin is driven instead by a conformational energy change used to generate the acidity in the membrane.) What are the problems? They all rest in the interaction between the electron and the proton flow. We can dissect away the independent parts of this flow in the membranes, that is of electrons alone (1-5nm hops) and of protons alone (by rotational migration along - O H or H,O chains; Williams, 1984). We are then left with the proton/electron intercommunication which occurs at three sites at least. They are in particles I, 111 and IV of mitochondria. In particle I the two-electron reaction is (see Williams, 1970): NADH (flavin H) + N A D (flavin) + 2e + 2H+ In particle 111 it is: QH2-+Q+2e+2H+ In particle IV it is: OH, -+ 10,+ 2e + 2H (FeOH, t FeO + 2e + 2H +) + (In all particles other, non-redox proton movements may also be driven.) Notice that these are all two-electron reactions of non-metals and not one-electron reactions of metals as is used in simple electron flow (Williams, 1961). It is generally true that non-metals react in two-electon steps with proton involvement while metals in low oxidation states, e.g. Fe(I1) and Cu(I), react by one-electron protonindependent steps in water. It is then the involvement of non-metal couples which brings the proton and the electron together in the membrane (Williams, 1961). One essential feature of the non-metal reactions is that if they are part of a catalytic transfer the non-metal group must cycle chemically. This applies to NAD (flavin) and Q but not to O,, which is destroyed. Again if during the reactions energy is to be conserved then the oxidation and reduction paths of the cycling unit must not be directly connected physically. Space must be used in the membrane to control the diffusion from the reaction centres of both the products e and H so that different e and H are used in oxidation and reduction respectively. I described this as the dislocation of reaction products (Williams, 1961), which has no membrane counterpart in chemiosmosis and is quite separate from subsequent translocation of either H + or e. movement of the electron in the membrane other than by transfer from metal site to metal site it is only through the swinging-arm motions of the coenzymes (Williams, 1961) in which H + and e move together as bound H atoms. When the electron goes to a metal the proton must find its own separate path. In particle IV the same situation arises since oxygen does not cycle. In effect the two-electron reactions of the high oxidation states of iron, e.g. FeO -+ Fe, are very similar to the two-electron reaction N A D + -+ NADH. Hydroxide leaves the first on reduction while protons leave the second on oxidation due to the different sense of the redox reaction, that is, reduction of oxygen (non-cycling) and oxidation of substrate H (non-cycling). When we turn to particle 111 there is a difficulty since the substrate of the enzyme is now in a reversible chemical system. The substrate, free QH,, first reacts with a bound Q in the particle and the generated free 0'-must later react with a second bound system to be re-reduced so as to complete the cycle. The electrons and the protons are here travelling together as bound H in the membrane. However, at some stage in the reaction protons and electrons must be separated if energy capture is to be achieved via energized protons. What happens? Almost everybody has tried his hand at devising schemes. Let us analyse them. The problems of particle III The first obvious point is that we need Q and QH, to communicate between different major protein redox particles. This lateral diffusion in the membrane of Q/QH, is essential if (1) redox equivalents once in the membrane are to be kept there and since (2) redox equivalents in the membrane must travel in a kinetically inert form and (3) must be a source of H + on oxidation. Quinones are ideally designed for this lateral two-dimensional diffirsion in the membrane avoiding any communication with the aqueous phases and only reacting in the membrane at chosen centres. The significance of this lateral diffusion has become strikingly obvious now that the lateral separation of particles in thylakoids (Fig. 1) is established. We d o not know the lateral separation in mitochondria. A major problem now concerns the possible transverse diffusion of Q/QH2, which is essential in chemiosmosis but is not necessary if the membrane at particle 111 has proton and electron transverse diffusion paths. [Note that if there is any rate-limiting step for transverse movement of redox W/A Local pH high \ 1 . + + The cycles In particle I the reactions of NADH or flavin H are those of fixed sites. The electron leaves in one direction and the proton in another. No problem arises since both these entities are replaced by neutral hydrogen atoms from the substrate which is consumed, e.g. malate. If there is any Vol. 12 Local pH high u, 0, Fig. 1. The flow problem in thylakoids Note the separation of the particles and the assumption that chemiosmosis is not required. Local circuits are shown. PSI, Photosystem I ; PS 11, photosystem 11. BIOCHEMICAL SOCIETY TRANSACTIONS 398 equivalents at the Q/QHzreaction sites then the two sides of the membrane are not at redox equilibrium and two Q/QH, pools, one on each side of the membrane at different redox potentials, are an energy store in and across the membrane, i.e. not a store of free protons but of bound redox equivalents in two pools of Q/QH,. In fact even if there were aqueous pH gradients in redox equilibrium with Q/QH, on both sides of the membrane then there would always be a substrate gradient in the membrane. One consequence is that Q/QH, gradients could be the essential feature of energy transduction and proton gradients could be secondary. Various observations on the relationships (not chemiosmotic) between proton gradients and ATP production could then be explained as properties of such energized Q pools within membranes.] Q cycles Q and QH, are not reactive. We therefore need special reaction sites not in particle I or IV. We now think that Qbound sites in particle 111 dislocate H + and e, e travels on via cytochrome 6, the Rieske centre, cytochromes c, and c (see Fig. 3). QH, must reform from different protons and different electrons. Now there is a known redox-potential drop in the particle which can be transduced to a proton energy for ATP synthesis. What happens to e, H , Q and QH, in the reactions of the particle? My first solution of this problem (Williams, 1970), is shown in Fig. 2 where I have + Fig. 2. The form of Q cycle used by Williams (1970) legitimately replaced X with Q for particle 111. I used a mobile QH radical to help the cycle between two reaction centres, nowadays written as coenzyme Qinand coenzyme Q,,*. Before stating my later worries over this scheme I draw attention to other types of scheme in which e and H + are moved together transversely as either QH, or QH, or Q is moved as Q or Q'-. The most elaborate are by Crofts et al. (1982) and Slater (1983) and are very difficult to test. There are some simpler ones, e.g. there are a variety by Mitchell (1976) which are elaborated in Mitchell & Moyle (1982). My problems with all the schemes are as follows. (i) If Q'- and/or QH (the radicals) can migrate they can carry e and H + in energized reactions forms and so break the dislocation (or chemiosmotic gradient) since these forms of Q do not need electron-transfer catalysts and have the same pK, as uncouplers. I doubt very much that such reactions are allowed to occur across membranes. (ii) Q and QH, need catalysts so their migration is apparently harmless. But can they migrate rapidly between the reaction centres in vivo if these are on opposite sides of highly rigid, organized and energized membranes (see Hauska & Hurt, 1982)? (iii) What is the construction of particle III? It has a wellestablished ability to pump protons and to move electrons between many centres. Where are they in space, and what are they for? Taking point (ii) first, there is no proof yet that Q or QH, can cross biological membranes at reasonable rates. The work of Hauska & Hurt (1982) shows that in artificial vesicle membranes the rate of Q/QH, transfer could be adequate but this is a far cry from movement in the rigidly structural convoluted membrane of Fig. 1. This problem needs immediate study. As far as the construction of particle I11 is concerned, point (iii), it is generally taken today that cytochrome c , , the Rieske protein, one cytochrome b and one bound Q site are on one side of the membrane (Fig. 3). If this is true the second cytochrome b is unlikely to be in direct electrontransfer contact since it is expected from sequence data to be on the other side of the membrane. There could be a second bound Q site towards this side of the membrane of course. In this case we can understand why, on energization, passing redox equivalents through the Q pool, we see the differential energization of b cytochromes, and of the two different bound Q sites. However, there is no need for Q to migrate between the two sides of the membrane and it may well be necessary to consider two Q/QH, pools and to postulate electron and proton channels for their communication (Fig. 3). The model has no Q cycles (see Papa, 1976; Wikstrom & Krebs, 1979). This paper directs attention to the movement of electrons, protons and bound H in membranes. It suggests that the nature of the Q/QHz pool can be such that it acts as a simultaneous system for energized e and H lateral transfer as H. Protons and electrons are transferred transversely in the membrane. The Q pools are energized and could drive ATP synthesis without any essential chemiosmotic communication (see Fig. 1 and Fig. 3). We shall not solve the problem without the structure of particle 111. + Fig. 3. A suggested form for the reactions of particle Iff in mitochondria N o Q cycle is used but movement of protons is shown in the membrane and lateral flow of Q and QH, is required. There is no flow of Q across the membrane. Crofts, A. R., Meinhardt, S. W. & Bowyer, J. R. (1982) in Function of Quinones in Energy Conseruing Systems (Trumpower, 9. L., ed.), pp. 477499, Academic Press, New York Hauska, G . & Hurt, E. (1982) in Function of Quinones in Energy Conserving Systems (Trumpower, 9. L. ed.), pp. 87-110, Academic Press, New York Mitchell, P. (1961) Nature (London) 191, 144-148 Mitchell, P. (1976) J . Theor. B i d . 62, 327-367 Mitchell, P. & Moyle, J. (1982) in Function of Quinones in Energy Conseruing Systems (Trumpower, 9. L. ed.), pp. 553-573, Academic Press, New York 1984 606th MEETING, CORK Moore, G. R., Huang, Z.-X., Eley, C. G. S., Barker, H. A., Williams, G., Robinson, M. N. & Williams, R. J. P. (1982) Furuduy Discuss. Chem. SOC.74, 31 1-329 Papa, S. (1976) Biochim. Biophys. Acta 456, 39-84 Slater, E. C. (1983) Trends Biochem. Sci. 8, 239-245 Wikstrom, M. K. F. & Krebs, K. (1979) Biochim. Biophys. Acta 549, 177-222 399 Williams, R. J. P. (1961) J. Theor. Biol. 1, 1-17 Williams, R. J. P. (1970) in Electron Trunsport and Energy Conservation (Tager, J. M., Papa, S., Quagliariello, E. Slater, E. C., eds.), pp. 7-23, Adriatica Editrice, Bari Williams, R. J. P. (1984) in The Enzymes of the Bio/ogicu/ Membrunes (Martonosi, A., ed.), 2nd edn., Plenum Press, New York, in the press Some implications of fixed-charge formation during electron-transport-chainactivity with the energized membrane they inhibit energy transduction and ATP synthesis and cause the release of protons into the medium (Higuti et al., 1978); the last-named authors comment that ‘the energy-dependent binding of ethidium In 1969 Azzi and co-workers concluded that the fluores- cation to the membrane inhibits ATP synthesis by cence changes of ANS which accompanied its reaction with neutralizing the negative charges created on the surface of mitochondria ‘indicated structural changes of the mito- the C-side of the membrane’. A similar explanation had chondrial membrane associated with energy conversion’ been given at much the same time for the action of Ca2+, (Azzi et al., 1969). Azzi (1969) reported a 20% decrease in and for the same reasons: the release of protons into the the amount of dye bound after adding succinate; from this, medium during Caz+uptake, and a reduction in synthesis coupled with the fact that a positively charged dye, which, since Ca2 is a permeant ion and its association with auramine-0, showed increased binding on energization, he a fixed-charge molecule must be supposed transient, was deduced that ‘the interaction between dyes and the seen as a ‘lag’ period, delaying rather than inhibiting the energized membrane is electrostatic’ and that ‘it is probably onset of synthesis (Archbold et al., 1979; McKay & important in the mechanism of ATP synthesis that an Malpress, 1980). An involvement of fixed charges in the asymmetrical charge distribution is associated with energy energizing process has also been assumed by Kell (1979), conservation in the mitochondria1 membrane’. Norden- arguing from the general inadequacy of bulk-phase electrobrand & Ernster (1971) extended Azzi’s studies, differen- chemical forces to account for observed phosphate tiating the non-energized and energized ANS binding and potentials. fluorescence changes by several different kinetic indices; It may seem surprising that the chemiosmotic hypothesis they suggested that the energy-dependent reaction reflected in particular should have remained uninfluenced by these ‘a charge separation in a special locus ... distinct from non- varied observations. The reason seems to have been the energy-dependent binding sites’, and that the energized delusive support which the hypothesis drew from a number state was to be considered as a ‘heterogeneous concept .. . of experiments which were either irrelevant to the purpose rather than a homogeneous entity such as that defined by a of identifying the nature of the protonmotive force, for bulk chemical or electrical potential across the mito- which they were nonetheless used, or which were interpreted with a free inferential logic which left much to be chondrial membrane’. It was, however, a more homogeneous view of the mito- desired. Archbold et al. (1975, 1979) attempted to draw chondrial energy transduction mechanism, in the form of attention to these remarkable lacunae in the arguments of a the chemiosmotic hypothesis, which gained the authority of major hypothesis, and two examples will be mentioned here a near consensus during the ensuing decade, a view that in general terms. (i) It is not correct to infer from the proton held no explicit function for the fixed-charge changes movements in the cation-exchange reactions of an oxygendetected by the dye-binding studies. The chief alternative pulse experiment that bulk-phase transmembrane forces hypothesis at this time, the ‘local energized proton’ will be operating in synthesis; nor, indeed, in the light of the exposition (Williams, 1978), also lacked any essential role fixed-chargechanges during respiratory activity is it right to for fixed-charge development, though the production of assume that transmembrane potentials are at any time, even negatively charged groups on energization could be the during permeant cation exchange, the primary mediating adventitious accompaniment of local proton release from force in energy transduction. (ii) To show that an artificially apoproteins within the membrane. imposed ApH+can lead to synthesis (e.g. Thayer & Hinkle, In contrast to these prominent and contending hypoth- 1975) provides no proof of the in uiuo mechanism of eses Archbold et al. (1974) regarded an increased negative synthesis; such experiments require the use of respiratoryfixed-charge presence in energized mitochondria as the inhibited particles and are therefore carried out in the probable explanation for their observations on the absence absence of the fixed-charge development which is intrinsic of proton movement into the bulk-phase during ATP to n o w 1 oxidative phosphorylation; they demonstrate no synthesis. Following Azzi’s model (1969) they envisaged more than that the energy input for ATP synthesis may be proton retention at the outer surface of the inner membrane presented successfully on an experimental basis in more as the direct result of an increased surface negative than one form. electrical field, an important factor not to be ignored by any Two consequences may be expected to follow from an energy transduction hypothesis (Archbold et al., 1975). increase in the negative surface charge of the coupling Confidence in the development of these views (Archbold et membrane: an increase in the surface electrical field giving al., 1979, 1 9 8 0 ~ ;Lappin & Malpress, 1980; Malpress, an enlargement of the diffuse double layer, and an increase 198la,b) owed much to the steadily accumulating evidence in the concentration of cations retained within the diffuse for fixed-charge generation in the energized mitochondrion double layer. The retentive force will be electrostatic and (Kamo etal., 1976; Quintanilha & Packer, 1977; Aiuchi and exerted chiefly on Ca2+ and H +.It is important to note that Kobatake, 1979; Archbold et al., 19806; Malpress, 1981~). as soon as we cease to regard the protonmotive force as a It was also noted that when positively charged dyes react generalized, dissipative bulk-phase property, and see it as localized and non-dissipative, the surface area involved in acid. energy transduction may become a mere fraction of the total Abbreviation used : ANS, 1-anilino-8-naphthalene-sulphonic FRANK H. MALPRESS Department of Biochemistry. Queen’s University, Belfast BT7 INN. N.Ireland, U.K. + VOl. 12