P. D., Schmidt, S. D., Wang, K. et al. (1995). Science 269,973-977. H. H., Ratovitsky, T., Podlisny, M. B., Selkoe, D. J.,. Seeger, M., Gandy, S. E., Price, D. L. et al.
Suspect Proteins in Neurodegeneration
Presenilins - in search of functionality E. H. Karran, D. Allsop, G. Christie, J. Davis, C. Gray, F. Mansfield and R. V. Ward Neurosciences Research, SmithKline Beecham Pharmaceuticals, New Frontiers Science Park, Harlow, Essex CM I 9 SAW, U.K.
Alzheimer’s disease (AD) is the most common cause of dementia in the aged population and is the fourth leading cause of death in the western world. T h e disease is characterized clinically by the progressive loss of memory and intellect. Pathologically, this cognitive decline is associated with the appearance in the brain of extracellular amyloid plaques and intracellular abnormalities of tau protein leading to the formation of neurofibrillary tangles. Amyloid, otherwise known as AD, is a 39-43 amino acid peptide that is excised from a larger type I transmembrane protein, the amyloid precursor protein (APP), by the sequential action of two proteolytic activities, known as P-secretase and y-secretase. T h e heterogeneity in the size of AP is predominantly due to alternative C-terminal cleavages by y-secretase. T h e identities of the APP secretases, and the function of APP, remain to be discovered, although both growth factor and neuroprotective roles for the secreted extracellular domain (known as secreted APPa) have been described [1,2]. There is now a large body of evidence from a variety of sources that supports the so-called ‘amyloid hypothesis’ - that the production and subsequent deposition of AP is a key aetiological event in the development of the disease [3]. T h e most persuasive data have been derived from the genetics of AD. A recent twin study showed pairwise concordance rates for AD of 19% and 5% for monozygotic and dizygotic twins respectively, emphasizing the considerable contribution of genetic factors to disease manifestation [4]. In the past, AD has been rather arbitrarily divided into early onset AD (sufferers below the age of 60 years) and late onset AD (sufferers over the age of 60 years). Early onset AD is sometimes taken to be synonymous with familial AD (FAD) in which, according to various criteria, the disease is judged to run in families. A further complication is that some of the familial cases are subcategorized into autosomal dominant Abbreviations used: AD, Alzheimer’s disease; APP, amyloid precursor protein; FAD, familial AD; EOFAD, early onset FAD; PS, presenilin; PS-1, presenilin-1; PS-2, presenilin-2.
49 I
FAD, where the propensity to develop the disease segregates in the appropriate manner. Notwithstanding these difficulties in definition, about 5% of all AD cases can be classified as autosomal dominant, early onset familial AD (EOFAD). One breakthrough in AD research, and one of the cornerstones of the amyloid hypothesis, was the discovery that in a small number of families, mis-sense mutations located either side of the AP sequence within the APP gene led to EOFAD. Subsequently, it was realized that mutations to the APP gene, located on chromosome 21 at 21q21.1, actually accounted for very few of the EOFAD cases, with approximately 70% of EOFAD cases being linked to a gene on chromosome 14 (although recent data suggest that this figure was a considerable over-estimate [S]). This led to the discovery, by positional cloning, of the gene encoding presenilin-1 (PS-1) which is located on chromosome 14 (14q24.3). Database searching led to the discovery and subsequent identification of a homologous gene, presenilin-2 (PS-2), located on chromosome 1 (lq41.2), that was responsible for EOFAD in the so-called Volga German families. T o date, 46 EOFAD mutations at 36 codons have been described for PS-1, and two mutations at two codons have been described for PS-2 (Table 1). In general, PS-2 FAD mutations cause AD with a later age of onset than PS-1 mutations [6], and as a consequence it is possible that PS-2 FAD mutants remain to be discovered among patients with late onset AD.
Presenilin cell biology PS-1 is a 467 amino acid protein with 10 hydrophobic domains [7]. PS-2 is a 448 amino-acid protein that is 67% identical to PS-1 [8]. Most of the diversity between the two proteins resides in the N-terminal and hydrophilic loop domains: the hydrophobic domains are very well conserved. Several studies have been performed to elucidate the membrane topology of PS-1. This information is critical to understanding the functionality of PS-1 (and by implication, PS-2) as it delineates the orientation of the various regions of the protein across the membrane which, in turn, dic-
I998
Biochemical Society Transactions
492
tates the availability of protein domains to make either lumenal or cytosolic protein-protein interactions. Given the current data, the best model, based on studies performed with PS-1 and its Caenorhabditis elegans homologue sel-12, is that
Table I Mutations in PS- I and PS-2 ~_____
Mutation
Domain
PS- I Ala79Val Val82Leu Val96Phe Tyr I I SCys, Tyr I I5His Pro I I7Leu Glu I ZOLys, Glu I20Asp Asn I 35Asp Met I39Thr, Met l39Val, Met I3911e He I43Phe, He l43Thr Met I46Val, Met I46Leu His I63Tyr, His I6311e, His I63Arg Glu I84Asp Gly209Val lle2 I3Thr Ala23 I Val, Ala23 I Thr Leu235Pro Ala246Glu Leu25OSer Ala260Val Leu262Phe Cys263Arg Pro264Leu Pro267Ser Arg269H is, Arg269Gly Leu282Arg Ala285Val Ser29OCys (exon 9 spliced out) Glu3 I8Gly Gly378Glu Gly384Ala Leu392Val Cys4 I OTyr Ala26Pro Pro436Ser
N terminus TMI* TM I HL I HL I HL I TM2 TM2 TM2 TM2 TM3 HL3 TM4 TM4 TM5 TM5 TM6 TM6 TM6 TM6 TM6 HL6 HL6 HL6 HD7 HD7 HD7/large HL Large HL Large HL TM7 TM7 TM8 TM8 HD9
PS-2 Asn I 4 I He Met239Val
TM2 TM5
*TM, transmembrane domain; HD, hydrophobic domain; HL,
hydrophilic loop.
Volume 26
PS-1 adopts an eight transmembrane configuration, where the hydrophilic loop and the N- and C-termini are oriented towards the cytoplasm (Figure 1) [9,10]. Most studies have suggested that the protein resides predominantly within the endoplasmic reticulum [ 11,121, although there have been reports that PS-1 can be present at the cell membrane [13] and in the dendrites and cell bodies of neurons [14,15]. Within the brain, PS-1 expression is mainly neuronal [ 161. One key feature of the presenilins (PSs) is that they normally undergo endoproteolysis in cells [17-191 which converts PS-1 from a 46 kDa holoprotein to 28 kDa and 18 kDa N- and C-terminal fragments and PS-2 from a 55 kDa holoprotein to 35 kDa and 20 kDa N- and C-terminal fragments. Although an early report suggested that PS-1 FAD mutants prevented the cleavage of PS-1 [17], subsequent work has demonstrated that PS-1 mutants are proteolytically processed normally, with one very interesting exception. A PS-1 FAD mutation to a splice acceptor site results in an in-frame deletion of the amino acids encoded by exon 9 [20] which removes the PS-1 cleavage site, leading to the production of an intact, but truncated, holoprotein [18]. The cleavage of PS-1 is predominantly at Met298/Ala299 [21], although evidence is now accruing to suggest that the precise cleavage site is developmentally regulated, with the generation of a larger N-terminal and a smaller C-terminal fragment [ 151 as neurons become fully differentiated. These fragments are similar in size to those generated by the action of caspases [22]. PS-2 was reported to be cleaved at several sites: Lys306/Leu307 [23] and also at caspase cleavage sites [22]. PS processing is a highly regulated process, and is therefore very likely to be biologically important. Thus, over-expression of PS leads to a build up of holoprotein, indicating that the endoproteolytic step is rate-limiting in these circumstances. In turn, the levels of .the cleaved PS fragments are also regulated, such that the overexpression of human PS-1 in a mouse cell line leads to replacement of the endogenous PS fragments with human equivalents [24]. How PS processing relates to its functionality has not been elucidated, and neither has the wider question of whether the holoprotein or the cleaved fragments, or both, play important biochemical roles.
Suspect Proteins in Neurodegeneration
Functionality of presenilins
APP processing
Investigations into the role of the presenilins, and more particularly their involvement in the pathogenesis of AD, currently fall into a number of experimental areas. Whatever hypothesis is posited for their functionality must account for the fact that mutations throughout PS-1 - lying on both sides of the endoproteolytic cleavage site - all produce EOFAD. Thus, the functional form of PS most probably involves either the holoprotein or a complex of the N- and C-terminal fragments.
With regard to the aetiology of AD, the most striking observation relates to the effects on APP proteolysis mediated by PS mutations. A seminal report [25] demonstrated that fibroblasts derived from patients carrying PS-1, PS-2 and APP mutations all produced elevated levels of the longer, more fibrillogenic form of amyloid. This observation has now been confirmed by a number of workers with cell lines and also in mice carrying PS-1 mutants [26-281. How mutations in the PS proteins mediate changes to the
Figure I Membrane topology of presenilins Amino acids that are changed by FAD mutations are shown as black circles. The region of PS- I deleted by the splice acceptor site mutation is shown as untilled circles. Arrows indicate endoproteolytic cleavage sites.
Presenilin-1
Presenilin-2
I998
493
Biochemical Society Transactions
494
y-secretase cleavage site remains unanswered, although some important experimental findings have been recently published. Firstly, it has been demonstrated that both PS- 1 and PS-2 can be co-immunoprecipitated with APP [29,30]. Interestingly, it was the Nglycosylated, immature APP molecule that coimmunoprecipitated with the PS holoprotein. This is consistent with the interaction occurring within the endoplasmic reticulum. Another very significant finding is that in embryonic neurons derived from PS-1 knockout mice, y-secretase activity was decreased by about 80% [31]. This argues in favour of PS-1 playing a role in regulating y-secretase activity: indeed, it could be that a macromolecular complex consisting of presenilin, APP and y-secretase forms across the endoplasmic reticulum membrane and controls the fate of APP. Further, it could be postulated that all of the mutations in PS are able to change subtly the conformation of the protein in the membrane and thereby alter the apposition of APP to y-secretase - in turn altering the position of the cleavage. If this were true, an extension of this hypothesis would be that APP FAD mutations at the C-terminus of the A/) sequence in APP would likewise alter the conformation of the PS-APP-y-secretase complex with a similar outcome. Of interest in this regard is the observation that, in differentiated neurons, would appear to be produced predominantly within the endoplasmic reticulum [321.
Cell signalling Lin-12 is a C. elegans homologue of the notch receptor family that is involved in cell fate determination. Levitan and Greenwald made a key observation when they were seeking C. elegans mutants that reverted a multivulva phenotype caused by expression of a hypermorphic lin-12 mutant [33]. From this screen, two mutants of a gene named sel-12 (sel means suppressor or enhancer of lin-12) were identified: sel-12 is the closest homologue of PS, possessing about 50% identity. Sel-12 mutants alone caused an egglaying deficiency. Subsequent work in this system has revealed that human PS-1 and PS-2 are able to restore normal function (as measured by rescue of the egg-laying defect), in contrast with the FAD PS-1 point mutants, which had greatly reduced activity [34,35]. T h e exception was the exon 9 splice acceptor site mutation, which was able to restore most of the functionality of PS. It has
Volume 26
also been confirmed that, like the PS proteins, sel-12 is normally endoproteolytically cleaved within cells in a similar position to that of the PSs [lo]. C. elegans remains an attractive model system with which to explore the functionality of the presenilins. However, considerable work needs to be performed before the effects of PS FAD mutants in C. eleguns can be linked biochemically with their effects in mammalian cells to alter the production of AP42. T h e association between notch and PS can also be inferred from the similar phenotypes induced by PS-1, notch-1 and Dlll knockout mice (Dlll is a notch ligand). PS-1 homozygous knockout mice die either in utero or shortly after birth. They suffer from intra-cranial haemorrhage, neuronal loss and severe malformation of the skeleton [36,37], as well as defective somite segmentation, which is also seen in notch-1 and Dlll knock-out mice [38,39]. In PS-1 knock-out mice, the levels of notch-I and D111 expression were markedly reduced [37]. Additionally, it has been shown in early mouse embryogenesis that the temporal and anatomical expression of notch1 and the PSs is very similar, with levels being high firstly in the neuroepithelium followed by the ventricular zone from which brain cells are derived. In the adult, PS was detected strongly in the cerebellum and hippocampus, as was notch-1 [401* Thus, there is a strong case to be made for an involvement between notch signalling and the PS proteins, at least during development. However, the relevance of these findings to the pathology of AD remains to be determined. T h e winglesslwnt signalling pathway, which like notch is involved in a range of cell development decisions, has also been demonstrated to interact with PS-1. Using the PS-1 loop region as a ‘bait’ in a yeast two-hybrid system, Zhou et al. isolated a protein, d-catenin, that is homologous to the Drosophila Armadillo protein which is a downstream mediator of the wingless/wnt signalling pathway [41]. An interaction between the loop region of PS-1 and b-catenin was demonstrated directly by co-immunoprecipitation, and interestingly 6-catenin is strongly and predominantly expressed in brain. T h e wnt pathway has been reported to down-regulate the notch signalling pathway via dishevelled [42], a wnt signalling molecule upstream of b-catenin. Thus, a pathway may exist where PS-1 connects, via dcatenin, the notch and wnt pathways. As PS-1 has been demonstrated to control in some way
Suspect Proteins in Neurodegeneration
the APP cleaving activity of y-secretase, it could be that binding of d-catenin to the loop region of PS-1 also serves to regulate y-secretase activity. A further fascinating observation is that glycogen synthase kinase, which is implicated in the phosphorylation of tau protein that may lead to neurofibrillary tangles in AD, is also part of the wnt signalling pathway [43]. It would be a highly significant finding if the two principal pathologies of AD could be linked within a common cellular signalling pathway.
Apoptosis Although the contribution and importance of apoptosis to AD pathology has yet to be clarified, some important experimental observations have been made that strongly imply a role for the PSs in programmed cell death. A seminal observation was made by Vito et al. who picked u p a cDNA they termed ALG-3 in a mouse T-cell hybridoma ‘death-trap’ screen [44]. ALG-3, which acted as an inhibitor of apoptosis, was in fact the C-terminal 103 amino acids of the mouse homologue of PS-2. Subsequent work demonstrated that over-expression of PS-2 in PC12 cells augmented programmed cell death in response to apoptotic stimuli [45] and that the N141I PS-2 FAD mutant increased basal apoptosis [46]. T h e s e apparently contradictory results were resolved by work showing that the ALG-3 fragment acted as a dominant inhibitor of apoptosis, and further that the C-terminal fragment of PS-2 that is generated by caspase-3 cleavage within the loop region of the protein is also anti-apoptotic. Also, there is evidence of alternative mRNA transcripts that encode PS-2 proteins similar in size to ALG-3 [47]. T h u s , there is considerable evidence that PS-2 plays a regulatory role in apoptosis. Other workers have shown that during apoptosis, both PS-1 and PS-2 are cleaved within the loop region at caspase substrate consensus sequences [48], with the N141I PS-2 mutant being more susceptible to cleavage.
Summary T h e discovery of the PS proteins, the complexities of their biochemistry and their potential involvement in signalling pathways and in apoptosis have galvanized research into AD. T o date, the aspect of the functionality of the PSs most relevant to the pathology of AD is the effect of PS FAD mutants to increase the proportion of produced from cells. This, coupled to the
observation that y-secretase cleavage is considerably reduced in neurons derived from PS-1 knockout mice, argues strongly that PS plays a very direct role in the proteolytic processing of
APP. 1 Araki, W., Kitaguchi, N., Tokushima, Y., Ishii, K., Aratake, H., Shimohama, S., Nakamura, S. and Kimura, J. (1991) Biochem. Biophys. Res. Commun. 181, 265-271 2 Mattson, M. P., Cheng, B., Culwell, A. R., Esch, F. S., Lieberburg, I. and Rydel, R. E. (1993) Neuron 10, 243-245 3 Hardy, J. and Allsop, D. (1991) Trends. Pharmacol. Sci. 12, 383-388 4 Raiha, I., Kaprio, J., Koskenvuo, M., Rajala, T. and Sourander, L. (1996) Lancet 347, 573-578 5 Cruts, M., Vanduijn, C. M., Backhovens, H., Vandenbroeck, M., Wehnert, A., Serneels, S., Sherrington, R., Hutton, M., Hardy, J., Stgeorgehyslop, P. H., Hofman, A. and Vanbroeckhoven, C. (1998) Hum. Mol. Gen. 7, 43-51 6 Tanzi, R. E., Kovacs, D. M., Kim, T. W., Moir, R. D., Guenette, S. Y. and Wasco, W. (1996) Neurobiol. Dis. 3, 159-168 7 Sherrington, R., Rogaev, E. I., Liang, Y., Rogaeva, E. A., Levesque, G., Ikeda, M., Chi, H., Lin, C., Li, G., Holman, K. et al. (1995) Nature (London) 375, 754-760 8 Levy-Lahad, E., Wasco, W., Poorkaj, P., Romano, D. M., Oshima, J., Pettingell, W. €I., Yu, C., Jondro, P. D., Schmidt, S. D., Wang, K. et al. (1995) Science 269,973-977 9 Doan, A., Thinakaran, G., Borchelt, D. R., Slunt, H. H., Ratovitsky, T., Podlisny, M. B., Selkoe, D. J., Seeger, M., Gandy, S. E., Price, D. L. et al. (1996) Neuron 17, 1023-1030 10 Li, X. and Greenwald, I. (1996) Neuron 17, 1015- 1021 11 Kovacs, D. M., Fausett, H. J., Page, K. J., Kim, T.-W., Moir, R. D., Merriam, D. E., Hollister, R. D., Hallmark, 0. G., Mandini, R., Felstenstein, K. M. et al. (1996) Nature Med. 2, 224-229 12 Walter, J., Capell, A., Grunberg, J., Pesold, B., Schindzielorz, A., Prior, R., Podlisny, M. B., Fraser, P., George-Hyslop, P. S., Selkoe, D. J. et al. (1996) Mol. Med. 2, 673-691 13 Dewji, N. N. and Singer, S. J. (1997) Proc. Natl. Acad. Sci. U.S.A. 94, 9926-9931 14 Cook, D. G., Sung, J. C., Golde, T. E., Felsenstein, K. M., Wojczyk, B. S., Tanzi, R. E., Trojanowski, J. Q., Lee, V. M. Y. and Doms, R. W. (1996) Proc. Natl. Acad. Sci. U.S.A. 93, 9223-9228 15 Capell, A., Saffrich, R., Olivo, J. C., Meyn, L., Walter, J., Grunberg, J., Mathews, P., Nixon, R., Dotti, C. and Haass, C. (1997) J. Neurochem. 69, 2432-2440
I998
495
Biochemical Society Transactions
496
16 Lee, M. K., Slunt, H. H., Martin, L. J., Thinakaran, G., Kim, G., Gandy, S. E., Seeger, M., Koo, E., Price, D. L. and Sisodia, S. S. (1996) J. Neurosci. 16, 7513-7525 17 Mercken, M., Takahashi, H., Honda, T., Sato, K., Murayama, M., Nakazato, Y., Noguchi, K., Imahori, K. and Takashima, A. (1996) FEBS Lett. 389, 297-303 18 Thinakaran, G., Borchelt, D. R., Lee, M. K., Slunt, H. H., Spitzer, L., Kim, G., Ratovitsky, T., Davenport, F., Nordstedt, C., Seeger, M. et al. (1996) Neuron. 17, 181-190 19 Ward, R. V., Davis, J. B., Gray, C. W., Barton, A. J. L., Bresciani, L. G., Caivano, M., Murphy, V. F., Duff, K., Hutton, M., Hardy, J. et al. (1996) Neurodegeneration 5,293-298 20 Perez-Tur, J., Froelich, S., Prihar, G., Crook, R., Baker, M., Duff, K., Wragg, M., Busfield, Fr., Lendon, C., Clark, R. F. et al. (1995) NeuroReport 7,297-301 21 Podlisny, M., Citron, M., Amarante, P., Sherrington, R., Xia, W., Zhang, J., Diehl, T., Levesque, G., Fraser, P., Haas, C. et al. (1997) Neurobiol. Dis. 3, 11006-1 1010 22 Loetscher, H., Deuschle, U., Brockhaus, M., Reinhardt, D., Nelboeck, P., Mous, J., Grunberg, J., Haass, C. and Jacobsen, H. (1979) J. Biol. Chem. 272,20655-20659 23 Shirotani, K., Takahashi, K., Ozawa, K., Kunishita, T. and Tabira, T. (1997) Biochem. Biophys. Res. Commun. 240,728-73 1 24 Thinakaran, G., Harris, C. L., Ratovitski, T., Davenport, F., Slunt, H. H., Price, D. L., Borchelt, D. R. and Sisodia, S. S. (1997) J. Biol. Chem. 272, 28415-28422 25 Schemer, D., Eckman, C., Jensen, M., Song, X., Citron, M., Suzuki, N., Bird, T. D., Hardy, J., Hutton, M., Kukull, W. et al. (1996) Nature Med. 2,864-870 26 Duff, K., Eckman, C., Zehr, C., Yu, X., Prada, C.-M., Perez-tur, J., Hutton, M., Buee, L., Harigaya, Y., Yager, D. et al. (1996) Nature (London) 383,710-713 27 Borchelt, D. R., Thinakaran, G., Eckman, C. B., Lee, M. K., Davenport, F., Ratovitsky, T., Prada, C. M., Kim, G., Seekins, S., Yager, D. et al. (1996) Neuron 17, 1005-1013 28 Citron, M.,Westaway, D., Xia, W. M., Carlson, G., Diehl, T., Levesque, G., Johnsonwood, K., Lee, M., Seubert, P., Davis, A., (1997) Nature Med. 3, 67-72 29 Weidemann, A., Paliga, K., Durnvang, U., Czech, C., Evin, G., Masters, C. L. and Beyreuther, K. (1997) Nature Med. 3, 328-332 30 Xia, W. M., Zhang, J. M., Perez, R., Koo, E. H. and Selkoe, D. J. (1997) Proc. Natl. Acad. Sci. U.S.A. 94, 8208-821 3
Volume 26
31 Destrooper, B., Saftig, P., Craessaerts, K., Vanderstichele, H., Guhde, G., Annaert, W., Vonfigura, K. and Vanleuven, F. (1998) Nature (London), 391,387-390 32 Cook, D. G., Forman, M. S., Sung, J. C., Leight, S., Kolson, D. L., Iwatsubo, T., Lee, V. M. Y. and Doms, R. W. (1997) Nature Med. 3, 1021-1023 33 Levitan, D. and Greenwald, I. (1995) Nature (London) 377, 351-354 34 Levitan, D., Doyle, T. G., Brousseau, D., Lee, M. K., Thinakaran, G., Slunt, 14. H., Sisodia, S. S. and Greenwald, I. (1996) Proc. Natl. Acad. Sci. U.S.A. 93, 14940-14944 35 Baumeister, R., Leimer, U., Zweckbronner, I., Jakubek, C., Grunberg, J. and Haass, C. (1997) Genes Funct. 1, 149-159 36Shen, J., Bronson, R. T., Chen, D. F., Xia, W. M., Selkoe, D. J. and Tonegawa, S. (1997) Cell 89, 629-639 37 Wong, P. C., Zheng, H., Chen, H., Becher, M. W., Sirinathsinghji, D. J. S., Trumbauer, M. E., Chen, H. Y., Price, D. L., Vanderploeg, L. H. T. and Sisodia, S. S. (1997) Nature (London) 387, 288-292 38 Hrabe de Angelis, M., McIntyre 11, J. and Gossler, A. (1997) Nature (London) 386, 717-721 39 Conlon, R. A., Reaume, A. G. and Rossant, J. (1995) Development 121, 1533-1545 40 Berezovska, O., Xia, M. Q., Page, K., Wasco, W., Tanzi, R. E. and Hyman, B. T. (1997) J. Neuropathol. Exp. Neurol. 56, 40-44 41 Zhou, J. H., Liyanage, U., Medina, M., Ho, C., Simmons, A. D., Lovett, M. and Kosik, K. S. (1997) NeuroReport 8,2085-2090 42 Axelrod, J. D., Matsuno, K., Artavanis-Tsakonas, S. and Perrimon, N. (1996) Science 271, 1826-1832 43 Cox, R. T. and Peifer, M. (1998) Curr. Biol. 144, 140- 144 44 Vito, P., Lacana, E. and D’Adamio, L. (1996) Science 271, 521-525 45 Deng, G. M., Pike, C. J. and Cotman, C. W. (1996) FEBS Lett. 397, 50-54 46 Wolozin, B., Iwasaki, K., Vito, P., Ganjei, J. K., Lacana, E., Sunderland, T., Zhao, B. Y., Kusiak, J. W., Wasco, V. and Dadamio, L. (1996) Science 274, 1710-1713 47 Vito, P., Ghayur, T. and D’Adamio, L. (1997) J. Biol. Chem. 272,28315-28320 48 Kim, T. W., Pettingell, W. H., Jung, Y. K., Kovacs, D. M. and Tanzi, R. E. (1997) Science 277, 373-376
Received 17 March 1998