Basement 3 9080 02246 1377 )28 DEWEY [? 1414 oi MIT Sloan School of Management Working Paper 4228-01 December 2001 A MATHEMATICAL MODEL OF THE IMPACT OF NOVEL TREATMENTS ON THE Afi BURDEN IN THE ALZHEIMER'S BRAIN, CSF AND PLASMA Lawrence M. Wein, David © 2001 by Lawrence M. Wein, David Short sections of text, not to permission provided that L. Craft, L. Craft, Dennis exceed two paragraphs, full credit including Dennis J. J. Selkoe Selkoe. All rights reserved. may be quoted © notice is without explicit given to the source. This paper also can be downloaded without charge from the Social Science Research Network Electronic Paper Collection: http://papers.ssm. com/abstract_id=295076 A Mathematical Model of the Impact of Novel Treatments on the A(3 Burden David L. CSF and Plasma in the Alzheimer's Brain, Craft 1 , Lawrence M. Wein 2 *, Dennis J. Operations Research Center, MIT, Cambridge, 2 3 Sloan School of Management, MIT, Cambridge, Selkoe 3 MA 02139 MA 02139 Center for Neurologic Diseases, Harvard Medical School, Brigham and Women's Hospital, Boston, "Person to whom MA 02115 correspondence should be addressed; email: lwein@mit.edu 1 Abstract With the advent AD, of novel therapies for there is a pressing need for biomarkers that are easy to monitor, such as the amyloid-beta (A/3) levels in the cerebrospinal fluid (CSF) and plasma. the explanatory power of To gain a better understanding these biomarkers, we formulate and analyze a partmental mathematical model for the A/3 accumulation reveals that the total A/3 burden in the brain tity called the polymerization ratio, which is is CSF in the brain, and plasma throughout the course of Alzheimer's treatment. Our analysis dictated by a unitless quan- the product of the production and elongation rates divided by the product of the fragmentation and rates. of cora- loss In this ratio, the production rate and loss rate include a source and sink term, respectively, related to the inter-compartmental transport. Our results suggest that production inhibitors are likely to reduce the A/3 levels in all three compartments. In contrast, agents that ingest monomers polymers, or that increase fragmentation or block elongation, A/3 burden in the brain, but increase - CSF and plasma may produce little change in may - or even transiently A/3 levels. Hence, great care must be taken interpreting these biomarkers. off of also reduce when Introduction Amyloid-beta (A/3) polymerization and plaque deposition are central to the 1 pathogenesis of Alzheimer's disease (AD) . Several novel treatments, such as the enhancement of A/3 clearance by an A/3 vaccine tion by a 7-secretase inhibitor 6 5, , 2-4 and the reduction of A/3 produc- have shown promise in preclinical studies. agents 7 , it will still be important to estimate the A/3 burden therapies enter clinical While and other cognitive tests will provide the ultimate assessment of the efficacy of these in the brain. As these a key challenge will be to infer treatment-induced trials, changes in the A/3 burden in the brain from the A/3 compartments, such as the cerebrospinal fluid more levels in easily monitored (CSF) and the plasma. These infer- ences need to be based on a solid understanding of the A/3 kinetics throughout the three interrelated compartments. Although the transport of A/3 between the various compartments has been studied in recent years 8-12 , the impact of treatment on the A/3 kinetics in these three compartments has not been elucidated. In previous work 13 , we formulated and analyzed a mathematical model that tracks the dynamics of A/3 production mentation sults in the and loss, and polymer elongation and by considering a three-compartment model, where A/3 the brain, CSF and that provide we brain during the course of treatment. Here some plasma. is frag- generalize these re- transported between Simple formulas and numerical results are presented insights into system behavior, and that can be used key transport, production and clearance parameters as new clinical to estimate data becomes available. Mathematical model We focus on A/342 in (the 42 amino-acid form of A/3), which parenchymal plaques only AJ4n or A542 . 14 . If is the primary ingredient we assumed that each brain polymer consisted of then the model could also be applied to A/34 o- which produces 8, cerebrovascular deposits 10 ; however, we do not pursue this avenue here, and for ease of notation refer to A/342 simply as A/3. which likely to is be in a number and also only consider extracellular A/3, dynamic equilibrium with burden primarily interested in the total A/3 to the We size of plaques, we intracellular A/3. in the three restrict our attention to A/3 polymerization, and do not attempt to capture the downstream processes formation. While different parts of the brain is safely we model the brain much faster is an infinite CSF and , system of nonlinear ordinary and galaxy formation, crystallization the polymerization and depolymerization processes and are model less is , as a single general than many , plasma. differential equations that 17 are related to a large class of models used to study actin formation chemistry 16 15 than transport across the blood-brain barrier (BBB) assume homogenous mixing within the Our model 18 levels compartment. Finally, because diffusion through extracellular spaces homogeneous we can of fibrillization or plaque have markedly different A/3 these levels tend to change proportionately and hence in the brain Because we are compartments, as opposed and cloud formation we employ considered in the literature, 19 , polymer " 20 . While are specialized to the main novelty AD of our the incorporation of multiple compartments with sources (production) and sinks (loss). For ease of understanding, nearly For i= 1,2 we let all of our mathematical notation 6,(0 be the concentration of A/3 i-mers at time very few oligomers have been found in the appears in the is CSF and plasma CSF 21 or plasma, mnemonic. t. Because we assume that A/3 only as monomers, and denote their time-dependent concentrations by c(t) and p(t), respectively. The differential equations specify the time rate of change of these concentrations. denoted by b,(t), c(t) and p(0, and are given by + h(t) Vb brain A/? production -c6i(*)(26 1 (t) /f>(*) vJ-^ monomers ' ={p ' r c} ^ = bi(t) {p c} of (i)-mers ^ degradation) to CSF and plasma csf + c6i(t)6(_i(t) brain A/? elongations loss^T! by < monomers monomers from plasma and all ' ^ „ & \\ fragmentations ? + 53&i(t)) /bi+ i(t) fragmentation (i + l)-mers elongation (i - l)-mers of -/(l-^I{i=2})M*). e6i(t)6((t) ( 2) ' elongation of 7-mers = c(t) CSF + Pc Yl A/? r pp plasma A/5 E r ^ meis -cW , monomers to brain and plasma + - production r xp x(t) Yl - Yl VP(0 s _^ jrijW monomers brain and • loss > v monomers from (3) loss brain and plasma J~ ' fe^S »={M^ monomers from p{t) of -x W - f={M^ production fragmentation to brain and csf CSF (4) The model is also depicted in Figure 1. In equation (1), A/? monomers in the brain are produced by cleavage of the amyloid precursor protein (APP) at the constant rate pb . These monomers live for l^ 1 time units on average before being via cell internalization or protease degradation. Similar production appear only by in equations (3) monomer and (4). and lost, e.g., Equations (l)-(2) assume that elongation occurs addition with elongation constant e (i.e.. e&i (f )6,_i(f) is that (i-l)-mers elongate to i-mers), and that fragmentation occurs only by break-offs at rate / (i.e., terms loss i-mers fragment into an (i - l)-mer and a the rate monomer monomer at rate The fbi(t)). factor 2 in front of b\(t) in equation (1) arises because the elongation of a dimer requires 2 monomers. The indicator variable otherwise; the factor \ in front of I{i=2} due to the is 1 if i fact that a = 2 and equals dimer can only two potential fragmentation location, whereas larger i-mers possess fragment in one equals 1 {,=2} sites. An alternate modeling approach, which direct clearance of monomers i-mers off of is pursued later (e.g., the monomer The remaining terms is typically in the CSF—»plasma an than obeying Michaelis-Menten kinetics j is it denoted by r l3 for i,j . diffusion and active or carrier-mediated and plasma—>brain transport are nated by active transport, we assume that that monomer fragmentation model describe the intercompartmental transport, some combination of passive Although transport. is to allow This alternative would result in the omission of clearance. 13 ingestion rate, and gives qualitatively similar results ment is fragmentations" term in equation (1) and the re-interpretation of /as "all which the paper, to represent microglial ingestion of polymers), rather than requiring a two-step procedure of followed by in =b,c 22 ; all domi- transport rates are first-order, rather the rate from compartment or p. likely Our main i to compart- reason for this simplification eases the parameter estimation task. However, the A/3 levels in these three compartments do not undergo many orders-of-magnitude changes during the course of treatment; hence, the saturation effect may be minor in the practically relevant range. Finally, Similarly, bound we note that most CSF plasma A/3 A,J binds to gelsolin 25 . is 23 bound, primarily to albumin Moreover, many only free A/3 (i.e., ' 24 . not yet understood whether it is or free A/3, or both, gets transported across the further, occurs of BBB 11 To . confuse matters laboratory and clinical studies that report plasma A/3 values quantify 23 . free Here, and we implicitly assume that linear total A/3 are in direct proportion relationship between the total A/3 and its (i.e., 22 ) unsaturated) binding and that there is a linear clearance rate out of the plasma. There- when we model fore monomers free entering the plasma, is it understood that a proportion of them bind to plasma proteins but that this binding does not affect the linear degradation and transport laws applied to the total A/3. Results Steady-state solution The steady-state analysis of equations (l)-(4) consists of setting the sides of these left equations to zero and solving for the steady-state A(3 levels b t c and p; the details , of the derivation are omitted. first To present our determine the effective production and augment the actual manner, we results in a transparent loss rates for each compartment, which by incorporating monomer exchange with the other two rates compartments. Because only monomers are transported between compartments, by symmetry effective it index the three compartments by suffices to production and survival probability that a compartment h2 = r 2i + r 23 We 1. compartment loss rates for monomer entering h3 = r 31 + + r 32 I3, production and loss rates for compartment probabilities s 2 and By S3. Figure 1, For compartment also define the total exit rate + U and 1. 2 1, i i and = 3, 2, 3, and derive the we let s t eventually makes from compartments respectively. we need To r 31 to derive the to compartment 1 it compartment to r x ( ^ ) as unknown r23 survival f<\ h2 , r 32 / fi \ "3 «3 partment 2 makes to calculate the effective , h2 (5) states that 2 it and 3 follows that 1, it r21 For example, equation be the the probability that a 1 equals the probability that plus the probability that (^p) and from there eventually makes monomer it to it first compartment it entering com- goes directly goes to compartment 3 1 (S3). The solution to equations (5)-(6) is 2 ''32 '"31 h3 V With the ' H3 / H 2 /t3 \ r 21 \ /, h2 h3 ^23^32 yi--i^) h h3 ) (8) . 2 survival probabilities in hand, the effective production and loss rates are given by Pi i1 Note that the of 1 monomers is increased by the compartment We 1 1 compartments, whereas the 1 - s terms, to the other 1, c for 2 ). (10) 1 is enhanced by the survival effective loss rate of compartments but never return. and and p for 3 (or, loss rates in the brain, pb by symmetry, p for 2 and subscripts of equations (9) and (10). Similarly, the effective rates for respectively) are given by substituting compartment subscripts We are now in compartment by the monomers that are transported from i.e., define the effective production substituting b for (9) =l +r 12 (l-s 2 )+r 13 {l-s3 production rate of compartment effective in other =Pi +P2S2 +P3S3, for 2 and 3 c (p, respectively) for 1 and lb , by c for 3) in the CSF (plasma, and the other two in equations (9)-(10). a position to present the steady-state solution. To do so, we define the polymerization ratio r = M (11) hf which is a unitless quantity that succinctly captures the four key processes of (effec- tive) production, elongation, (effective) loss of equations (l)-(4) to zero and solving for and fragmentation. Setting the finite where r > 1. side the steady-state concentrations reveal that there are two regimes: a steady-state (or subcritical) regime where r supercritical regime left < 1 and a In the former case, the A/3 levels eventually attain equilibrium values, where &! = &, (12) bi = 26ir i_1 1 , c = 2,3,..., (13) = £, (14) 'c (15) P=fBy equations (12)-(13), the total A/3 concentration in the brain ber of A/3 molecules, whether they exist as is denoted by b = YlvLi *&»> ' monomers the total (i.e., or as part of polymers), steady-state A/3 levels in the plasma (p), the and increasing in which s There are several noteworthy features of our steady-state solution. are linear num- CSF (c), First, the and the brain monomers (&i) compartment production terms, and independent of the elongation rate e and the fragmentation rate /; these latter two parameters only z-mer concentration for affect the brain also decreasing in loss rates. is more By > All steady-state concentrations are 2. influence of transport parameters on the A/? levels subtle, as explained in the Discussion. the brain A/3 burden goes to infinity as the polymerization ratio r (16), approaches i,j), The i If 1. we assume no transport across compartments then the solution coincides with our one-compartment results steady A/3 levels in relevant one, AD patients 26 " 28 suggest that the r and that the slow A/3 accumulation < 1 13 regime is less increasing over the years. Consequently, the paper focuses on the r > 1 regime is is than = The . in the brain results steady-state situation, where the polymerization ratio r a brief discussion of the r ru (i.e., 1 < for all relatively the clinically from a quasibut 1 is slowly regime, and deferred until later. Post-treatment kinetics The impact (i.e., of treatment can be assessed by changing the appropriate parameter production, loss, transport, elongation or fragmentation rate) in the model and substituting To analyze the assumptions. simulations 13 it into equations (14)-(16) to find the post-treatment steady state. A/3 kinetics immediately after treatment, The first assumption, which that the total number that the total is simplifying based on the observation from numerical is of brain oligomers changes number than the number of monomers, we make two much more slowly of oligomers in the brain, Yylvbii remains constant immediately after treatment; we denote this quantity by B2 , which equals 2^(y^:) by equation monomer level after much more This new level, 61, is we 1 assumption, the brain approximated by setting the of equation (1) to zero and solving equations values in Table this treatment quickly reaches a new level and thereafter changes gradually. presentation, Under (13). (1), (3) and (4) for b x . left side For ease of present the solution for b x under the special case of the parameter (i.e., 61 = = r bp -{l b r cb = r pc = p c = pp = lc = 0): + eB2 + J(lb + eB2 f + 8e(p + fB 2 b ) ) (I'J • 4e In equation (17), B2 is number the pre-treatment of oligomers and the four poly- merization parameters represent the post-treatment values. Because the rate of tercompartmental exchange (Table 1) faster is than the changes in- in the brain A/? polymers after treatment, we make the second simplifying assumption that the steady-state relationships in equations (14)-(15) actually hold for treatment. Therefore, (again, we predict that the assuming the parameters in Table CSF and plasma all times after levels rapidly change to 1) rtJ>i -Mc ' respectively, r cv+ l t'p pb and then gradually approach (18) r cp rbcbi + PP ' * their post-treatment steady states 10 (19) Parameter estimation 10 parameters, which are listed along with their values in Table Our model has The value of the A/? minutes, which group is monomer loss rate / in Table 1 corresponds to a close to the crude experimental value of 38 in the brains of APP transgenic mice 6 . Protofibrils 29 , 1. half-life of 41.6 minutes reported by one fibrils 30, 31 and plaques 32 appear to grow primarily via A/3 monomer addition, and we use an elongation rate Table e in 1 taken from a synthetic of other estimates for fibrils we set the CSF We and protofibrils , CSF 31 29 . which , is within a factor of two For lack of data to the contrary, loss rate lc and the plasma production zero. assume that only three and assume there a circular is of the six possible exchanges have a positive trans- CSF—>brain, we ignore brain—>plasma, plasma^CSF and port rate. In particular, flow given by brain—>CSF—>plasma^brain. To we note that these inclusions and omissions, tify analysis fibril production rate p c the pp eq ual to rate 30 CSF—>brain jus- and brain—^plasma transport are likely to be by passive diffusion and these transport rates are probably much BBB ics, smaller than the consistent with a specific transport port is CSF—>plasma difficult the brain—>CSF flow is which passes through the transport is quite rapid is is 10 , diffusion included in the model. CSF study of rats 9 8, , 8 9 ' . whereas plasma—>CSF trans- drained/transported into the taken from studies of guinea pigs based on a mechanism that dwarfs passive 33 because proteins do not readily pass through ultrafiltration neuronally-produced A/3 is rate, > according to a saturable mechanism that follows the Michaelis-Menten kinet- Similarly, rpb plasma— brain transport CSF Our estimate and the value of mice 34 , . Also, and hence of the transport rate for the transport rate r cp 10 . This leaves four unknown parameter values, which are determined by solving four equations (using equations (12) and (14)-(16)): the total A/?42 level in the brain b is 1975 xlO 3 pM (averaged over 5 different cortical regions of wet brain tissue of patients with a clinical dementia rating (CDR) 11 score of 5.0 (severe dementia) 35 , assuming that the density of wet cortical A/342 level c Ad is 115 pM that consists of Table 1, 36 ; tissue the plasma A/342 level monomers, ^, is 1.3% 38, 39 . is is equal to that of water); the 29 ; and the CSF fraction of brain These estimates, which are given lead to the polymerization ratio estimate r Parameter pM 37 = 0.84. in set equal to zero (details not shown), the monomer directly proportional to each other in Figures the CSF and plasma levels in the brain 2b and drop to the values predicted The hours after Note that in this case the values in equations (18) CSF are A/3 concentrations in and in equations (18) treatment, and then slowly approach (19) within their post-treatment steady states. the post-treatment steady-state values. Hence, the approach their post-treatment steady-state A/3 2c. and and (19) are very similar to CSF and plasma compartments levels much faster than the brain compartment. The post-treatment steady states represent an 18-fold reduction the brain, a 1.67-fold reduction in the Upon the CSF the discontinuation of treatment, similar kinetics occur: monomer levels in all three in and a 1.67- fold reduction in the plasma. a rapid change in compartments, followed by a slow return to the pre-treatment steady state. Figure 3 depicts the impact of an agent that enhances fragmentation by 100%. While the steady-state A/? burden in the brain drops 15.6-fold, the CSF and plasma A/? levels experience a transient rise and then return to the pre-treatment levels. The return to pre-treatment levels follows from the fact that the fragmentation rate does not alter the steady-state brain only interaction with the monomer CSF and plasma the production inhibitor in Figure different 2, We now is levels in the CSF and is deleted from equation (1) monomer an A/3 vaccine 2 and the fragmentation rate ingestion off of oligomers. plasma. As a "all is fragmen- interpreted Figure 4 shows the impact of that increases the ingestion rate by 100%. levels rise the brain's similar for all three compartments. consider an alternative to equations (l)-(4), in which the term is the values in equations (18) and (19) are quite approach to steady state as the rate of and plasma equation (12), which compartments. In contrast to the case of than the post-treatment steady-state result, the rate of tations" level in In this case, the CSF monotonically to post-treatment steady states that are 2.5% higher than the pre-treatment steady states, even though the steady-state brain A/3 burden decreases 8.6-fold. 13 Finally, sible we investigate the impact of changing the transport rates. improvements pos- reducing the A/3 burden in the brain are to increase the for brain—>CSF transport Two rate, r bc and to decrease the plasma—>brain , ure 6 shows that increasing r bc by rate, r pb 100% causes a modest 17% decrease in the brain A/3 level after one year (the post-treatment steady state level represents a duction), while almost doubling the A/3 levels in the CSF and a 100-fold reduction in r pb reduces the brain A/3 burden by Fig- . 24% re- plasma. In contrast, less then 0.2%. Supercritical regime We now turn to the case where the polymerization ratio r satisfies r mathematical analysis of this case shows that the A/? burden without bound, eventually increasing linearly at rate p b (r - > in the brain 1. A grows l)/r (Figure 5a). The polymer concentrations in the brain are not geometrically distributed as in the r < 1 where each z-mer successively achieves case, but are uniformly distributed, the concentration = 6j f/e, b z plasma concentrations attain brain concentration is = 2//e for finite levels in unbounded. This is i > the r Interestingly, the 2. > 1 case, CSF and even though the total because the brain A/3 burden grows by accumulating larger polymers, not more monomers. Discussion Despite the existence of some mathematical models for A/3 fibrillization and plaque growth, this paper appears to represent the ical model to of the body. CSF and assess the effect of We AD first treatment on A/3 attempt to use a mathemat- levels in various compartments provide simple formulas for the steady-state A/3 levels in the brain. plasma, both before and after treatment. Our solution reveals that there are two possible regimes, depending the brain, r = Pf which , is upon the value of the polymerization ratio in the product of the effective production rate and elon- 14 gation rate divided by the product of the effective loss rate and the fragmentation The rate. and effective production and loss rates account not just but also loss in the brain, the plasma and CSF. When for actual and sinks due to transport for sources the polymerization ratio We A/3 levels are achieved throughout the body. is less than believe that this 1, is production and from to steady-state the clinically relevant regime, in light of the slow accumulation of A/3 throughout the body. r > 1, If then the A/3 burden in the brain grows indefinitely (by accumulating larger polymers while maintaining a constant monomer early at rate pb (r - whereas the A/3 l)/r, level), levels in the a steady state. This unusual state of affairs in the r tant assumptions in CSF and plasma 1 case is lin- remain in due to two impor- our model: only monomers pass between compartments and no polymerization occurs in the with the plasma and CSF CSF and plasma. Consequently, the brain A/? interacts only via A/3 monomers. CSF and plasma A/3 levels provide a level in the brain, but not necessarily Consequently, from a biomarker viewpoint, reliable indirect estimate of the A/3 of the total A/3 > eventually increasing burden monomer in the brain. An extreme example of CSF and plasma above, where the easily-monitored levels this is the r > 1 case remain constant and do not hint at the unbounded A/3 accumulation in the brain. However, our analysis also has implications for the monitoring and assessment of treatment. Agents that inhibit the production of A/3 monomers 5 6 ' monomer or increase the are likely to cause significant reductions in the A/3 levels in with the levels CSF and plasma compartments more polymers 2 , all loss rate in the brain three compartments, attaining their post- treatment steady-state quickly than the brain. In contrast, agents that ingest monomers off of reduce the elongation rate, or increase the fragmentation rate of A/3 polymers in the brain, the steady-state may cause only minor changes - including increases - in CSF and plasma on the total A/3 burden primary toxic moiety levels, in the brain. 38 42 ' , which are not indicative of their impact However, then the plasma A/3 and 15 soluble A/3 if CSF monomers A/9 levels are the may be excellent biomarkers for efficacious all Alzheimer's treatments, and production inhibitors than other agents that have impairment of cognitive in less may be more impact on A/3 monomers. The prevention mice by an A/? vaccine 7 suggests that this may not be the case. Our analysis also reveals The A/3 throughout the body. more how the transport parameters affect the distribution of relationship of A/3 levels to transport parameters For example, a drug that increases removal of A/? from the brain by subtle. increasing the transport rate to the will increase is the CSF CSF mechanism (a possible which chosen in this will fall, and the CSF which (cr cp ), paper (based on A/342 ), the plasma than the brain 2 ) A/3 level but the change in plasma A/3 level can be positive or monomers by negative, depending on the relative contributions to plasma (^i r 6P ), of the Elan vaccine With the parameters will rise. level rises the brain because the CSF rather the primary extra-compartmental source of plasma A/3. However, is in the case of A/34 o, where it is estimated that r^ is 10 times larger than 43 r;, c , our analysis indicates that plasma A/34 o levels would decrease as a consequence of such a therapy. While many of the model's parameter values are imprecisely known and the model rather simple is and bound stem CSF in the brain was done in transport, linear relationship between free A/3), the qualitative nature of our results should in large part in the (e.g., first-order from the empirical observation 21 be robust because they that very few A/3 polymers reside or plasma, suggesting that A/3 polymerization occurs almost exclusively and A/3 polymers are not HIV research 44, 45 , easily transported out of the brain. mathematical analysis such as with data generated by perturbation of to derive estimates of human some parameters that this However, as can be combined A/3 compartments by novel agents are difficult to measure in vivo, thereby uncovering the primary flow dynamics of A/3 in the body. More generally, our model provides a systematic framework with which to interpret upcoming trials of novel agents for Alzheimer's disease. 16 human clinical Acknowledgment This research was partially supported by the Singapore-MIT Alliance by the Foundation for Neurologic Diseases (DJS). 17 (LMW) and References [1] Selkoe, D. disease. [2] J. biology into therapeutic advances in Alzheimer's cell Nature 399 (Supp), A23-A31 (1999). Schenk, D., Barbour, R., Dunn, W., Gordon, G., Grajeda, H., Guido, T., Hu, K., Z., Huang, Johnson- Wood, K., Khan, K., Kholodenko, D., Lee, M., Liao, J. Lieberburg, Vandevert, [3] Translating I., Motter, R., Mutter, C, Walker, L., Soriano, F., Shopp, G., Vasquez, N., Wogulis, M., Yednock, T., Games, D. S., & Seubert, P. Immunization with amyloid-/? attenuates Alzheimer-disease-like pathology in the Bard, PDAPP mouse. Nature 400, 173-177 (1999). Cannon, C, Barbour, F., Guido, T., Hu, K., Huang, D., Lee, M., Lieberburg, J., R., Burke, R.-L., Johnson-Wood, Games, Khan, K., D., Grajeda, H., K., Motter, R., Nguyen, M., Soriano, I., N., Weiss, K., Welch, B., Seubert, P., Schenk, D. & Kholodenko, F., Vasquez, Yednock, T. Peripherally administered antibodies against amyloid /3-peptide enter the central nervous system and reduce pathology in a mouse model of Alzheimer disease. Nature Medicine 6 916-919 (2000). [4] Weiner, H. C, L., Issazadeh, Lemere, C. A., Maron, R., Spooner, E. T., Grenfell, T. S., Hancock, W. W. & Selkoe, D. J. J., Mori, Nasal administration of amyloid-beta peptide decreases cerebral amyloid burden in a mouse model of Alzheimer's disease. Ann. Neurol. 48, 567-579 (2000). [5] Wolfe, M. Rahmati, S., T., Xia, W., Moore, C. L., Leatherwood, D. D., Ostaszewski, B. L., Donkor, I. O. k Selkoe, D. J. Peptidomimetic probes and molec- ular modeling suggest Alzheimer's 7-secretase is aspartyl protease. Biochem. 38, 4720-4727 (1999). 18 an intramembrane-cleaving [6] M. The next generation Felsenstein, K. of AD therapeutics: the future now. is Abstracts from the 1th annual conference on Alzheimer's disease and related disorders, Abstract 613 (2000). [7] Janus, C, Pearson, Chisti, M. A., J., Home, McLaurin, Mathews, J., French, P., Heslin, D., Mercken, M., Bergeron, C, Fraser, P. J., P. E., St. M., Jiang, Y., Schmidt, Mount, H. T. George-Hyslop, J., P. S. D., Nixon, R. A.. k Westaway, D. A/3 peptide immunization reduces behavioral impairment and plaques in a model [8] of Alzheimer's disease. Nature 408, 979-982 (2000). Zlokovic, B. V., Ghiso, J., Mackic, J. B., McComb, J. G., Weiss, M. H. k Frangione, B. Blood-brain barrier transport of circulating Alzheimer's amyloid P. [9] Comm. Biochem. Biophys. Res. Martel, C. L., Mackic, B., J. 197, 1034-1040 (1993). McComb, J. G., Ghiso, J. k Zlokovic, B. V. Blood-brain barrier uptake of the 40 and 42 amino acid sequences of circulating Alzheimer's amyloid [10] Ghersi-Egea, k J.-F., Fenstermacher, in guinea pigs. Neuroscience Letters 206, 157-160, 1996. Gorevic, P. D., Ghiso, J. J., Frangione, B., Patlak, C. S. D. Fate of cerebrospinal fluid-borne amyloid /i-peptide: rapid clearance into blood and appreciable accumulation by cerebral arteries. J. [11] Neurochemistry 166, 880-883, 1996. Poduslo, J. F., Curran, G. L., Haggard, J. J., Biere, A. L. k Selkoe, D. J. Permeability and residual plasma volume of human, Dutch variant, and rat amyloid /3-protein 1-40 at the blood-brain barrier. Neurobiol. Dis. 4, 27-34, 1997. [12] M. Mackic, J. B., Weiss, Bading, J., Frangione, B. k W., Kirkman, E., Ghiso, J., Calero, M., Zlokovic, B. V. Cerebrovascular accumulation and H., Miao, increased blood-brain barrier permeability to circulating Alzheimer's amyloid 19 8 peptide in aged squirrel monkey with cerebral amyloid angiopathy. J. Neu- rochem. 70, 210-215, 1998. [13] Craft, D. L., Wein, L. burden & M. Selkoe, D. J. The impact of novel treatments on KB Alzheimer's disease: insights from a mathematical model. Submitted in for publication, 2001. [14] Lemere, C. A., Blusztajn, Selkoe, D. APO E in J. J. K., Yamaguchi, H., Wisniewski, T., Saido, T. C. & Sequence of deposition of heterogeneous amyloid /^-peptides and down syndrome: implications for initial events in amyloid plaque formation. Neurobiol. Dis. 3, 6-12, 1996. [15] Oyler, G. A., Duckrow, R. B. <fe Hawkins, R. A. Computer simulation of the blood-brain barrier: a model including two membranes, blood flow, facilitated and non-facilitated [16] Gravina, Ho, S. A., L. H., Suzuki, diffusion. J. Neurosc. N. L., &: Eckman, C. Younkin, S. B., Meth. 44, 179-196, 1992. Long, K. E., Otvos, L. Jr., Younkin, G. Amyloid 8 (A/3) in Alzheimer's Disease Brain: biochemical and immunocytochemical analysis with antibodies specific for forms ending at A/340 or A/?42(43). Journal of Biochemistry 270, 7013-7016 (1995). [17] Oosawa, F. & Kasai, macromolecules. [18] Flory, P. NY, [19] J. J. M. Actin. theory of linear and helical aggregations of Mol. Biol. 4, 10-21, 1962. Principles of polymer chemistry. Cornell University Press, Ithaca, 1953. von Smoluchowski, M. Drei vortrage koagulation von kolloidteilchen. [20] A Z. fiber diffusion, brownsche bewegung und Phys. 17, 557-585 (1916). von Smoluchowski, M. Versuch einer mathematischen theorie der koagulationskinetic kolloider losungen. Z. Phys. 92, 129-168 (1917). 20 [21] Walsh, D. M., Tsang, B. P., Rydel, R. E., Podlisny, M. B. k Selkoe, D. The J. oligomerization of amyloid /^-protein begins intracellular^ in cells derived from human [22] Lutz, R. in [23] brain. Biochemistry 39, 10831-10839, 2000. J., Dedrick, R. L. vivo approach to k Zaharko, D. membrane S. transport. Physiological pharmacokinetics: an Pharmac. Ther. 11, 559-592 (1980). Kuo, Y.-M., Emmerling, M. R, Lampert, H. C, Hempelman, T. A., Woods, A. S., Cotter, R. k J. S. R., Kokjohn, Roher, A. E. High levels of circulating A/342 are sequestered by plasma proteins in Alzheimer's disease. [24] Selkoe, D. human [25] E. k transported on lipoproteins and albumin in Biere, A. L., Ostaszewski, B., Stimson, E. R., J. plasma. Chauhan, V. solin, Amyloid /^-peptide J. B. T., Maggio, J. Chem. 271, 32916-32922 (1996). Biol. P. S., is Hyman, Ray, I., Chauhan, A. k Wisniewski, H. M. Binding of a secretory protein, to amyloid /^-protein. Biochem. Bwphys. Res. gel- Comm. 258, 241-246, 1999. [26] Hyman, B. T. Marzloff, Kk plaques or amyloid burden Arriagada, in P. V. The lack of accumulation of senile Alzheimer's disease suggests a dynamic balance between amyloid deposition and resolution. J. Neuropathol. Exp. Neurol. 52, 594-600 (1993). [27] Arriagada, P. V., Growdon, J. H., Hedley-Whyte, E. T. k Hyman, Neurofibrillary tangles but not senile plaques parallel duration B. T. and severity of Alzheimer disease. Neurology 42, 631-639 (1992). [28] Berg, L., McKeel, D. W., Miller, ical J. P., Baty, J. k Morris, J. C. Neuropatholog- indexes of Alzheimer's disease in demented and nondemented persons aged 80 years and older. Arch. Neurol. 50 349-358 (1993). 21 [29] Harper, J. Amyloid D., Wong, protofibrils: S. S., Lieber, an in C. M. model vitro k for Lansbury, P. T. Jr. Assembly of A/? a possible early event in Alzheimer's disease. Biochemistry 38, 8972-8980 (1999). [30] Lomakin, On A., Chung, D. Benedek, G. S., B., Kirschner, D. A. the nucleation and growth of amyloid /3-protein fibrils: Teplow, D. B. detection of nuclei USA Natl. Acad. Sci. and quantitation of rate constants. Proc. k 93, 1125-1129 (1996). [31] Lomakin, A., Teplow, D. B., Kirschner, D. A. k Benedek, G. B. Kinetic theory of fibrillogenesis of amyloid /3-protein Proc. Natl. Acad. Sci. USA 94, 7942-7947 (1997). [32] Tseng, B. P., Esler, W. C. B., Stimson, E. R., Ghilardi, P., Clish, H. V., Mantyh, P. W., Lee, J. P. k Maggio, J. J. R., Vinters, E. Deposition of monomeric, not oligomeric, A/3 mediates growth of Alzheimer's disease amyloid plaques in human [33] brain preparations. Biochemistry 38, 10424-10431 (1999). Betz, A. L., Goldstein, G. barriers. W. k Katzman, Chapter 30 of Basic neurochemistry: molecular, aspects, 4th Ed., ed. Siegel, G. J., [34] R. Blood-brain-cerebrospinal fluid Calhoun, M. E., Burgermeister, Raven P., Press, Ltd., Phinney, A. cellular, New York, 1989. Stalder, L., M., Wiederhold, K.-H., Abramowski, D., Sturchler-Pierrat, Staufenbiel, M. k Jucker, M. Neuronal overexpression of and medical M., Tolnay, C, Sommer, mutant amyloid B., pre- cursor protein results in prominent deposition of cerebrovascular amyloid. Neurobiology 96, 14088-14093, 1999. [35] Naslund, P. J., Haroutunian, V., Mohs, R., Davis, K. k Buxbaum, in the J. L., Davies, P., Greengard, D. Correlation between elevated levels of amyloid /3-peptide brain and cognitive decline. JAMA 22 283, 1571-1577 (2000). [36] Galasko, D., Chang, Thomas, R., L., Motter, R., Clark, C. M., Kaye, Kholodenko, D., Schenk, D., Lieberburg, Basherad, R., Kertiles, Boss, M. A. L., & I., J., Knopman, D., Miller, B., Green, R., Seubert, P. High cerebrospinal fluid tau and low amyloid /?42 levels in the clinical diagnosis of Alzheimer disease and [37] relation to Apolipoprotein Scheuner, D., Eckman, D., Hardy, J., C, [38] Selkoe, D. McLean, C. genotype. Arch. Neurol. 55, 937-945 (1998). Jensen, M. Song, X., Citron, Hutton, M., Kukull, W., Larson, M., Peskind, E., Poorkaj, L., E & C, P., Schellenberg, S. Nature Medicine Younkin, A., Cherny, R. A., Beyreuther, K., Bush, A. I. & Fraser, F. E., M, Suzuki, N., Bird, T. Levy-Lahad, E., Viitanen, Tanzi, R., Waco, W., Lannfelt, 2, 864-870, 1996. W., Fuller, S. J., Smith, M. J., Masters, C. L. Soluble pool of A/? amyloid as a determinant of severity of neurodegeneration in Alzheimer's disease. Ann Neurol 46, 860-866 (1999). [39] Wang, J., Dickson, D. W., Trojanowski, J. Q. k Lee, V. M.-Y. The levels of soluble versus insoluble brain A/3 distinguish Alzheimer's disease from normal and pathologic aging. Experimental Neurology 158, 328-337 (1999). [40] Wolfe, M. &; S., Selkoe, D. Xia, W., Ostaszewski, B. L., Diehl, T. J. Two S., Kimberley, W. T., transmembrane aspartates in presenilin-1 required for presenilin endoproteolysis and 7-secretase activity. Nature 398, 513-517 (1999). [41] Ye, Y., Lukinova, N. & Fortini, M. E. Neurogenic phenotypes and altered Notch processing in Drosophila Presinilin mutants. Nature 398, 525-529 (1999). [42] Lue, L.-F., Kuo, Y.-M., Roher, A. E., Brachova, Kurth, J. H., Rydel, R. E. & Rogers, J. L., Shen, Y., Sue, Soluble amyloid /? as a predictor of synaptic change in Alzheimer's disease. 853-862 (1999). 23 L., Beach, T., peptide concentration Am. J. Pathol. 155, [43] Shibata, M., Yamada, Holtzman, D.M., B., S., Kumar, S.R., Calero, M., Bading, Miller, C.A., Strickland, D.K., Ghiso, J. Clearance of Alzheimer's amy loid-ss( 1-40) k J., Zlokovic, B.V., peptide from brain by related protein-1 at the blood-brain barrier. J. Clin. Invest. Frangione, LDL receptor- 106(12), 1489-1499 (2000). [44] Wei X., Ghosh S.K., Taylor J.D., Bonhoffer, S., human immunodeficiency [45] Ho D.D., Neumann M. Rapid turnover M.E. Johnson VA, Emini, E.A., Deutsch, Nowak, M.A., Hahn, B.H. virus type 1 infection. A.U., Perelson A.S., of plasma virions and Nature 373, 123-126 (1995). 24 k in Nature 373, 117-123 (1995). Chen W., Leonard J.M. CD4 P., Lifson, Shaw, G. Viral dynamics lymphocytes in k Markowitz HIV-1 infection. Production Loss Production P. \ t 1, Figure 1: A pictorial depiction of the 25 mathematical model. (a) z 1.5 //M 1 0.5 Brain total (c) CSF (a) z 1.5 (JM 1 0.5 Brain total (c) CSF (a) 0.6 (JM 0.4 0.2 Brain total (c) CSF (a) Brain total (c) CSF 250 12 10 200 8 150 (J.M pM 6 100 4 50 2 40 20 60 80 100 (b) Brain monomer 50000 40000 pM 30000 20000 10000 40 60 time (days) 40 (d) 60000 20 20 60 time (days) time (days) 100 Plasma 80 100 (a) /UM 1.8 (c) Brain total pM CSF Z.Z- Aj&j Date Due Z/jq ^ MIT LIBRARIES II I II I 3 9080 02246 1377