TiPS- October I991 IVOI. 221

17 Olson, L. F., Troty, C. L. and Schaffer, W. M. (19BB) Theer. Popul. Biol. 33, 344-370

18 Harrison, R G. and Biswas, D. J. (1986) Nahrrr 321,394&l

19 Rapp, P. E. (1%) in Chaos (Holden, A. V., ed.), pp. 179-208, Manchester univemily Press

20 GoldberBec, A. L. and West, 8. J. (1987) Ann. NY Acad. Scf. SOJ, 195-213

21 Babbyantz, A. and Destenhe, A. (1986) Pmt. Neff Acnd. Sci. USA 83.35X1-3517

22 Gokibeqec, A. L., Rigney, D. R., Mietus, J., Autman, E. M. and

Greenwald, S. (1988) Experknfin 44, 983-987

23 &finkel, A. (1983) AJ~. I. Physiol. 245, R455-R466

24 Pool, R. (19B9) Scicncr 243,25-2B 25 van Roasum, J. M. and de Bie, 1. E.G. M.

(19B9) I. Pharmacokbtct. Biophnrm. 17, 365-392

26 Kaneko, K. (1989) Pltys. Rev. Ltt. 63, 219-222

27 Kaneko, K. (1990) Phys. Rev. Ltt. 65, 1391-1394

2B Cuyton, A. C., Coleman, T. C., Cowlay, A. W., Manning, R. D. and Norman,

383

R. A. (1974) Circ. RES. 35, 159-176 29 Sheiner, L. B., Stanski, D. R, Voseh, S.,

Miller, R. D. and Ham, J. (1979) Clin. P11ormnc0l. Thrr. 25.~371

30 Reid. J. L. and Meredith. P. A. (1990) Hyprrtmsio~t 16, 12-18

31 Grassberger, P. (1985) Nat1rrc 323, 609-611

32 Sugihara. G. and May, R M. (1990) Nahtrt 344,734-741

33 Wales, D. I. (1991) Natan 350,4B5-4% 34 Maddox, 1. (1990) Nature 347,421 35 Kauffman, S. A. (1991) Sri. Am. 265,

64-m

Amyloici deposition as the central event in the aetiology of Alzheimer’s disease John Hardy and David Allsop

While there may be many causes of Alzheimer’s disease (AD), the same pathological sequence of events, described here by John Hardy and David AIIsop, is Iikely to occur in all cases. The recent discoocry of a pathogenic mutation in the &amyloid precursor protein (APP) gene on chromosome 21 suggests that APP mismetabolism and pamyloid deposition are the primary events in the disease process. The occurrence of AD in Down syndrome is con&tent roith this hypothesis. The pathological cascade for the disease process is most likely to be: /3-amylotd deposition -B tau phosphorylation and tan Ie formation --) ncuronaf death. The derrelopment of a biochemical un l&s tanding of this pathologkal cascade will facilitate rational design of drugs to intetiene in this pro&s. Alzheimetr’s disease (AD) is the fourth major cause of death in the developed world after heart dis- ease, cancer and stroke. It is largely, but not exclusively, a dis- ease of the elderly and has been estimated to afflict l-6% of those aged wer 65 years. Demographic changes (the falling birth rate and ageing of the baby boom that foIlowed World War 1) mean that the prevaIence of the disease is increasing. TypicaIIy, AD begins insidiousIy with memory prob- lems, which become progress- ively worse until sufferers are bedridden, doubly incontinent, and have completely lost their presymptomatic persona.

CIearIy this disease is a major social and health care problem. There is no effective treatment, and for the majority of patients I. Hardy is Senior &chcr ft~ rhc Deperfrrn~ of BiocheMstry and Mofecalur Cmefics. SI Mary’s Hospftaf M&al S&al, Loedoa W2 IPC, UK. ad D. A/Imp b Lrchtnr iH Lr Dbisbn of Biochrmfstry, Schod o/ Biology and B&&m&try, Qnerr’s Unlaniry of Beljaat, Me&al Biology Crntre, 97 Lkbm had, Belfast B7Y 781. UK.

there is no certain means of diag- nosis other than brain biopsy. It is only recently that advances in molecular genetics have begun to shed some tighl on the causes of the disease, although it has long been recognized that pemons with trisomy 21 (Down syndrome) inevitably develop the fuII neuro- pathological changes of advanced AD by their fourth or fifth decade.

The finding of an amyioid pre- cursor protein (APP) gene mu- tation in familial AD (see beIow) now makes it clear that deficits in neurotmnsmittem, transmifter- metabolizing enzymes and recep- tors are a consequence of amyioid deposition (or abnormal APP processing) and not the initial cause of the disease. Therapeutic approaches to AD based on in- creasing the Ieveis of acetykhoIine in the brain have not met with great success, despite claims to the contrary, and drug develop ment in the future may turtl towards compounds designed to inhibit the formation of amyioid in the brain, alleviate any neuro-

toxic activity of abnormal APP fragments, or interfere with the sequence of events linking amy- ioid deposition and abermnt APP processing to neuritic alteration, neurofibriiiary tangle formation and ceil death.

If this type of therapeutic ap- proach could be applied in the early stages of the disease, before significant damage has accrued to neurons1 systems, then there is good reason to hope that this might halt or slow the progression of the disease. For these reasons, an understanding of the role of amyioid formation and senile plaque development in AD is of considerable importance to pharmacologists.

PathoIogIcaI fea4ures The pathology of the disease is

complex. There are three weii- known sites of abnormal fibrous protein deposits within the brains of AD patients. These are the senile plaques, the neurofibrillary tangles and the walls of cerebral blood vessels. There is also eaten- sive neuronai damage and loss. All of these features are also found, usually to a Iesser degree, in the ‘normaI’ elderly. Thus the relationship between AD and ‘nonnal’ ageing remains undear. There has been much debate about the relative importance of these different p&KtiOgiC~ features and how they relate to each other. This articie reviews the present understanding of the molecular pathokqty of these lesions and the aetiology of the disease.

Senile plarjnes The cIassicaI senik plaque con-

sists of a central core of r8diating amyloid fibriis surrounded by a rim of dystrophic neurites to- gether with reactive microgUa wd astrocytes. AmyIoid is a generic

TiPS - October I991 /Vol. 121

term used to describe a group of chemically heterogeneous pro- teins found in a number of dif- ferent diseases and tissues. All amyloid deposits are composed at the ultrastNctural level of straight, unbranching fibrils of 6-10nm diameter, and the deposits also share common properties of Congo red birefrtngence and resistance to proteolysis. These properties have been ascribed to a predominant cross-g structure in the constituent chains.

polypeptide

Amino acid sequencing of senile plaque amyloid A4 protein’ revealed it to be essentially the same as the cerebrovascular amy- loid f3 protein originally isolated from meningeal blood vessels by Glenner and Wonti. The 39-42 amino acid senile plaque or cer- ebrovascular amyloid peptide will be referred to here as F/A4.

Molecular cloning has indicated that f3/A4 is a small fragment of a much larger amyloid precursor protein (APP)3 (see Fig. 1). This precursor is predicted to have a single membrane-spanning region with a long extracellular N-terminal segment and a short intracellular C-terminal tail. The $/A4 sequence begins close IO the membrane on the extracellular side and ends part-way through

the putative transmembrane region.

There are at least five alternative transcripts of the APP gene, some of which contain an internal Kunitz-type pmtease inhibitor insert’. Significantly, the APP gene was localized by several groups to chromosome 21. Poss- ible functions of APP and its role in the pathogenesis and aetiology of AD are considered further below.

Over the past few years consider- able attention has been focused on a previously poorly described type of plaque-like lesion that cannot be detected easily using routine Congo ted, thioflavin or silver stains. These ‘diffuse’ plaques can be seen as an area of granular staining after an im- munohistochemical reaction with antibodies to fVA4. Most diffuse plaques show no assockttion with abnormal neurites or reactive glial cells. They are often the only form of senile plaque found in cases of Down syndrome below the age of 55 (Ref. S), and substan- tial numbers have even been de- tected in a 13 ear old Down syndrome patlen % . Ultrastructural observations of these lesions have revealed few or no amyloid fibrlls, leading to the proposal that they may be accumulations of ‘pre-

amyloid’ (gfA4, APP or fragments of APP not yet in a fibrous form)‘.

These obwrvatiom fqgest that diffuse plaquea are an early stage of plaque form&ion and that the deposition of p/A4 (or its pre- cursor) precedes any obvious neuritic pat-. Abundant dif- fuse plaques have &o been found in the brains of ex-boxers with dementia pugRistka*. In AD, diffuse pIaquea are usually more numerous and widespread throughout the CNS than typical senile plaques (e.g. they am often found in the spinal cord and cer- ebellum) implying that they can evolve into typical neuritic plaques only in specific brain regions9*‘o.

The dystrophic neurites sur- rounding the amyloid core of a mature senile plaque contain paired helical filaments (PHFs) - pairs of Hlaments 10 nm in dia- meter, wound into a left-handed helix with a cross-over approxi- mately every 80 nm. These PHFs also constitute the main structural element of the neuroftbdllary tangle”.

M~~jfbr&y tanglea and renropllth&

NeurofibrilIary tangles of PHFs are found inside dying neurons. There has been much debate and

TiPS - October 1991 /Vol. 12f

confusion concerning the bio- chemical make-up of these lesions. Direct protein chemical analyses of isolated PHF preparations have identified ubiquitin, @A4 and the microtubule-associated protein, tau, as potential cor&ituents, while immunohistochemical stud- ies have shown that neurofibril- lary tangles share epitopes with these proteins and with neuro- filament components and another miaotubule-a5sociated protein, MAP 2 (reviewed in Ref. 12).

It has been suggested that PHFs are composed of fVA4, but this remain5 highly controversial. It is now clear that some extracellular ‘tombstone’ tangle5 (where the surrounding neuron ha5 com- pletely degenerated) do indeed react with antibodies to $/A4 (Ref. 13). However, recent ultrastruc- tural observations suggest that this finding is due to the second- ary deposition of amyloid fib& on the surface of exposed tangles rather than any intrinsic amtri- bution of $/A4 to PHF structure”.

A report appearing to confirm that infmcallular tangle5 can be immuno&ined with antibodies to luAI” must al5o be treated with caution, since this finding could be due to APP within the

ibly trapped inside or onto the surface of the

tangle. APP Is synthesized in the neuronal cell body and normally undergoes fstt anterograde axonal transpr#. Large bundle5 of PHF in the cyh#asm might interfere with thb process and lead to an abnormal accumulation of APP within the pevikprya of affected cells. This might alsu explain the finding of Yamaguchi et al.“, who have repMtly described APP-like immunaactivity of intracellular tangle5.

Many of the chemical studies on tangle compagi tion suffer fnim the lack of sufficient quantitative data on protein recovery and yields of amino acid residue5 obtained during sequencing. Thus it is not always ciear if the reported amino acid sequences relate to major PHP component5 or minor con- taminants (e.g. co-purifying amy- loid fib&). The most convincing data suggc5t that PHFs consist, at least in part, of an abnormally phorphorylskd fra

Fm ent of one

or more isofonns 0 tau proteW8. However, it has bean claimed that this protein can account for only

about 10% of the mass of PHFs, and the remainder is unknown12. There is scope for further protein chemical analyses of PHFs.

A third site of PHF accumu- lation in the brain in AD is inside tau-immunoreactive, slender, ar- gyrophilic fibres found scattered throughout the all~cal and isocortical neuropil. Some of these ‘neuropil threads’19 or ‘curly fib&” have been shown to be the dendrites of tangle-bearing neurons”; however, they are con- sidered by Iha&’ to be dendritic growth cones attempting a mass- ive regenerative response.

Newonnl loss The distribution of neuronal

damage and loss ha5 never been thoroughly and formally quanti- fied; the clearest description of this pathology is given by Brun and colleagues (reviewed in Ref. 22). Two important points may be gleaned from the distribution of neuronal logs: the first is that the affected neurons do not share any particular transmitter or any other biochemical marker t&ad so far (deficits have been found in choline@, noradrenergic, seru- tonergic and dopaminergic 5y5- terns, a5 well as in amino acid and peptide neurotransmittam); the second is that the selectivity of neuronal loss appears to be anatomically determined. In par- ticular, the suggestion that the pathology spread5 along neuronal pathway5 is .a particularly attrac- tive way to explain the distri- bution of neuronal damage and loss=.

Aetiology of AM&m& disaa5a A large number of epidemio-

logical studies have consistently shown that family history is a risk factor for developing Ap. In addition there are numerous re= ports of families in which Alz- heimer’s disease is inherited a5 an autosomal dominant disorder (reviewed in Ref. 24). hnetic linkage studies showed that the pathogenic bcu5 for some families was on the proximal long arm of chromosome 21 (Ref. 2.5), although in other families them ~18 no evidence for a lesion on this chromo50m~6.

It is now clear that one locus on chromosome 21 is the APP gene, and that a mutation in this gene resulting in a substitution of iso-

385

leucine for valine at codon 717 (residue 642 in APP-695) (Fig. 1) give5 rise to the disease in &me families2’. An alternative mu- tation resulting in a substitution of glutamine for glutamic acid at codon 693 (residue 618 in APP- 695) results in the deposition of p/ A4 amyloid in the walls of cerebral blood vessels, invariably leading to death at an early age due to massive cerebral haemorrhafl. Patients with the latter disorder (cerebral amyloid angiopathy of Dutch type) sometimes also de- velop diffuse plaques in the brain but there are no significant num- bers of typical senile plaques or neurofibrillary tangles. Neither of these APP mutations has Leery found in the general population, and the APP7l7 mutation ha5 now been detected in English, North American and Japanese families with early onset AD29.

It is likely that further altemr- tive APP gene mutation5 remain to be discovered in otherkindreds with autosomal dominant familial AD. Linkage with other genetic markers on or outside chromo- 5ome 21 could be explained by mutations in regions of DNAcon- trolling APP gena expm55ion or in gene5 coding for protein5 having some influence on APP metab- olism (e.g. APP-proassing en- zymes or their inhibitors).

It should be stmsed that the proportion of case5 in which AD is acquired a5 a single-gene, auto- somal dominant disorder is very low and there is a substantial body of evidence indicating that the neuropathok@ai changes of AD can also be triggered (most likely in genetically swceptible individuals) by various environ- mental factors. Potentially import- ant factor5 indude aluminium ex- posure, head trauma (c.f. boxers’ brains8 noted above), and virus infections. However, the fact that an APP gene mu&&an can give rise to all of the neuropa@iogical hallmarks of AD without any pre- ceding underlying neunmrl path- ology or any othe! defect strongly suggests that amyioid mismetab- olism and deposition is the sami- nd event in the pathogenesis of all cases of AD.

Fan~ofAPPandnschm~ of~Alforln&a

APPisex/wau+ath&hknls in many dif arent h!Sues thmugh-

306

??

out the body. In most tissues, hanscript APP-770 is the most abundant, but in the CNS the tmnscrlpt lacking the protease inhibitor domain (APP-695) is predominant. This leaves a num- ber of important unanswered questions. What are the physio- k@al functions of the various APP Worms, and why is APP- 695 so abundant in brain tissue? From which tissue source(s) and isoform(s) of APP is the amyloid in the braIn derived? Where does the convemion of APP into amy loId take place and what are the molecular mechanisms involved? How do the APP gene mutations mentioned above lead to brain amyloidosis? These questions are presently the subject of intense investigation, and answers to some of them are only just be- ginning to emerge.

SecWedfaTms of APIJ Soluble, secretory derivatives of

Appcanbeproducedbyproteo- lytic cleavage within the R/A4 sequence close to the extraceIIular side of the membrane. These have been detected in cerebrospinal fhddss2’, the conditioned media

of various cell culture#‘, and in human serum=. Most of the se- cretory APP circulating in the blood probably orlghtates from plateletsssa. Human embryonic kidney cells transfected with cDNA constructs encoding full- length forms of APP-695 and APP- 751 release secretory forms of APP that terminate at residue GM5 bf fi/A4, leaving behind a C-terminal membrane-bound fragment com- mencing at Leul7 (Ref. 35; Fig. I). This suggests that an ‘APP secmtase’ deavage enzyme acts at either the Gin%-Lysl6 or Lysl& Let117 bonds, leaving Lysl6 to be excised by an aminopeptldase or carboxypeptidase, mspectively.

In AD an alternative or abnor- mal pathway of APP proteolysis (involving ckzavage at two sites on either side of B/A4) must liberate the amyloid peptide. Experiments with synthetic g/A4 peptides indicate that the EMted peptide could into a Ref. 12). At present tie identity of the proteolytlc enzymes involved in the ‘normal’ and ‘abnormal APP-proceasing pathways is un- known, although some prelimi-

TiPS - October 2991 Wol. 121

nary data on this subject have been reported (reviewed in Ref. 36).

Physiofogfcal @rctforrs of APP These appear to include pro-

tease inhibition and a roIe in cell adhesion and the regrdation of cell growth (Fig. 1). SecreWy forms of APP containing the Kunitz in- hibitor insert have the same N-terminal sequence, m&cular mass and properties as protease nexin II, a growth-regulating mol- ecule secreted by fibroblasts3’. Secretory APPlnexin 11 can form stable complexes with several serine proteases, such as trypsin and chymotrypsin, and in the blood this species conceivably plays a role in the physiOk@caI events assockrted with bbod dot- ting, by inhibiting appropriate coagulation pathway enzymes including coagulation factor XIa (Ref. 3@.

Allsop et al.” sugge&d that the structumI orga&JWn of APP resembIestherpidernugrowth factor pmcumor and ape&ted that the $/A4 peptide might show biologiul actfvity. SnbaequentIy, synthetic $/A4 peptI&s of various lengths were shown to exhibit neurotmphic or nemotoxIc ac- tivity (Fig. 1).

A recent report suggests that this property resides in r&dues &35 of @IAl, a region ahowIng homology with the tachykmln family d neumpeptida”. A g/A4 &35 peptIde proved to be trophlc at low concentmtlons to undifferentiated rat MppocampaI neurons but toxic at higher con- centrations to mature neurons.

These effects were mimicked by substance P antagonists and re- versed by substance P agonist& implying that they are medinted by a tachykmin recapt#. Sub- sequently, the neurotoxic potency of $!A4 was mported to be grea act3

*enhanced by nerve growth f . These reports await confirmation.

Intact APP promotes &I-sub- stratum adhesion of both neur- onal and non-neuronal cells. Such APP-mediated adhesion can be specific&y inhUed by anti- bodies directed against residues 1-16 of j3lA4, the presence of a ce Sik within this regIon4*. This might also exp&in the euIier findings of binding sites for a g/A4 8-17 pep

TiPS - October 1992 fVol. 121

tide in human adrenal gland and intheisIetsofLan5erhansofthe pancn&? These observations have recently been extended to indude high-affinity binding sites (with a & in the pkomobu ran@inntbninu.ThuSAPP seemsbbeamoIecukwithat z z i-72

&mwth ‘&a&tory site and celi adhe&mdte.Thesizeandnature inufuoofaIlofthephysioIogWIy active fragments of APP that express these pmpertiea is not Clear.

Amywlldepositiot¶ In Down syndrome, amyloid

deposition might be a cmae- quence of an excessive amount ofAPP(duetoa5O%increasein

ebral amyloid sn Dutch type lies wi t%Fpa

*Y of n six resi-

dues of the normal proteolytic ckava5e site, while the APP7l7 mutation in famihai AD is four residues past the C-terminus of the @/A4 re@on (Pig. 1). One simpk hypothe& is that these mutations somehow direcdy in-

theeffaetaoftheAPP493mutation isthatthe’abWmaPprocem@

EzzaLS?~k?E @/A4 pep&k with the ghuamate- to-&uamine substitution has a greater tendency to aggregate into

amyloid 5brils (i.e. is more ‘amy- loidogenic’) than the normal

reactive mate&I decorating neur- onalsomaanddendritesoflocal

peptide. pyramidal neurons (Pig. 2). This second explanation ap-

pearsnottoapplytotheAPP7l7 mutation since it lies just outside the VA4 region. However, the amyIoidfromfam5ialcasesofAD with this mutation has not yet beenisolatedandsequen&,and it is possibk that the jWA4 peptide inthesepatienbisextaukdbe- yond 42 residues to inch& the mutatedaminoadd.InthisreqanI, it may ba d5nificant that in other chemical classes of amyIoid where there is a point mutation (e.g. transthyretin, cystatin C) tie de- posited fibrus always contain the mutated residue. A major Problem in determining the mokcular ef- fktsoftheAPPaminoacidsub- stitutions wilI be that any alter- ationsinpmcess@naedordybe ma@aItoaccountfortheslow ztp of $lA4 amyloid over

ThefactthattheQpicalBtnile plaque is intimatehf assoWed with abnormaI neuritic eiements containing PHP that plaquesandtanqksdonotarise independently but that faemation of one leads to formation of the other.ThedkoveryofanAPP genemut8tbnina#soffaa&al ADwithb&hmilephqtramd neurofibrilky tanglea implicates theformerastheprimazyIeaion.

The o&in of the $/A4 amyioid is a contentious issue, with most advocates supporting either 8 *9s- cularoraneuronalsourceforthe APP. In the case of the senile plaque amyloid, the balance seemstobeshiftingdec5WIyin favourofamuuonalorigihAPPis expressedathighlevelsinneur- ens, and the neuritic elements of typical senile pIaques have been showntoreactwithantibodiesto various regions of H.

Observauons of dMuse (pm sumably edy) plaques have con- SirtcRtlyfrihdtofnnrlmy~

ZE asps” I fuse pkques in Down ‘syndrome and AD seem to be composed essentially of @/A4-immuno-

This conch&on is supported by the temporal sequence of events deducedfnnncasesofDownsyn- drome of different ages. Diffuse plaques are the hist detectable lesion in Down sy&omesfi whe=thcydeilrlVWncrrro- fiblillary tan$es, cerebroWc&u amyloidosis, or any discernible neuritic chan5e. Later, tyPical senileplaqueswiLamuroundin5 neuritic msponse appsu, and theseu- acam&Wdadbyneum- 5brilluyWandthedosay associated nemopil thre&. It is onlylater,aftertheageof!ID,that any goss atrophy and loss of neurons are evide#.

This alIows a probable patho- genicpat.hwayforADtobedc- line&d (pig. 3) runnin5 from 81 A4 amyloid formation and de- position8 5=fIh ==fibr5lary tan@formation,Bod~md clinical prwenhtion. PreaWabW boththediSNpWofryarptic connections~withrarik

them is also the poseibtug that

388 TiPS - October 1991 /Vol. 221

neumtoxic fraginents of APP can cause neuronal death indepen- dently of plaque and tangle for- mation.

The neurotrophicineur0toxic properties of p/A4 (or,other API’ fragments) may account for the secondary neuritic changes that take place around the developing senile plaque. At least some of the PHF-containing dystrophic neur- ites in the plaque periphery are likely to be connected to tangfe- bearing neuronal perikarya, prob- ably via neuropil threads although this has not been examined in detail.

modeiling (and APP turnover?) OECUIS to the greatest extent. Al- ternatively, Yankner et af.” have recently suggested that selective vulnerability in AD can be ex- plained by a neurotoxic inter- action between @IA4 and nerve growth factor. They have stated that $/A4 can induce the pro- duction of NGF receptors in cer- tain rat hippocampal neurons, which are then selectively vulner- able to the neurotoxic effects of p/A4.

Whatever the explanation for primary vulnerability, two factors may be important in the sub- sequent spread of the pathology: electrical signalling along neur- onal pathways, or the transfer of pathogenic molecules between neurons. In this respect, the fact that APP is transported along axons may be relevant’6.

A detailed molecular under- standing of the pathological cas- cade outlined above will permit the design of novel drugs to inter- vene at various points in this process. The recent finding that diffuse plaque formation can be induced in mice by -ion of APP in transgenic animalss” may provide one suitable model for testing some of these drugs.

138.699405 15 SpiRanti, M. C., Goadart, M., Jakaa, R.

and Ktu~, A. (19Xl) Pmt. NetI Acud. Sci. USA B7,3947-3951

16 Koo, E. H. d al. (1990) Proc. Nail Acad. Sri. USA B7,1561-1565

17 Yamagocbi, H. et PI. (1990) Brain Res. 537,318-322

1s Lea, V. M., B&in, 8. J., Dhroa, L and TroJanowaki, J. Q. (1991) !%eace 251, 675-67B

19 Braak, H., Bnrk, E.. Crundka-Iqbal, I. and lqbal, K. (19B6) Newwci. Irtt. 65, 351-355

As noted above, neurofibrillary tangles are thought to contain an abnormally phosphorylated frag ment of tau prptein. It is not yet rlear whether abnormal phos- phorylation precedes or follows tau deposition, but a likely se- quence of events is that exposure to abnormal fragments of APP, possibly p/A4 itself, leads to hyperphosphorylation of tau, which then polymerizes to form stable PHF. While this process may be reversible in its early phases, it becomes irreversible as other modifications of PHF (such as ubiquitination) occur.

20 Ibara, Y. (19BB) Brain Res. 09.1~144 21 Braak, H. and Braak, E. 09IlB) Neuro-

palhal. Appl. Neumbial, 14,3puI 22 Peareon, R. C. A. and Pow& T. P. S.

(1989) Annu. Rm. Neunwci. 2,340-372 23 Haston, L C. et PI. (19B981) Arch. Gen.

Peychiahy 3l,lO&lm 24 St Gaor~e-Hysbp, P. H. et al. (19B9)

Neumbiol. Aging lo,4174 25 St GaorBa-Hyabp, P. H. et al. (1990)

Natare 347,194-197 26 Scbdknbar& C. D. ft PI. (ISee) Science

241,15&15lO 27 Goak, A. et al. (1991) Nature 349,

704-706 2s Levy, E. ct al. (1990) Science 24%

1124-1126 29 Hardy, J. et al. (1991) Lancel 337,

134m343 30 Waidamann, A. et al. (1589) Cell 57,

115-126

One possible route towards ab- normal phosphoxylation is chronic disturbance of Ca*+ homeostasis mediated by the neumtoxic pmper- ties of $/A4. The precise molecular details explaining the connection between o/A4 amyloid deposition and neurofibrillary tangle for- mation is one important focus for further research.

31 Pabnart, M. R. ct al. (19B9) Pnw. Nafl Acad. Sci. USA 8s. b33b63u

32 Podky, M. 8. d al. (l#o) &o&em.

33 8% f??zzi?~?P. and Saitob, T. (l9W) &chest. B&phye. Res. CamlmoI. 170, m

34 Bush, A. I. etrl. (199ll)J. Bfa/. On. 265, l5977-l59B3

Acluunvledgmntnts We thank Carvell Williams for

helpful comments on the manu-

35 Each, F. S. el al. (1990) Science 24B, 1122-1124

36 Iabiora, S. (1991) /. Near&em. 56,

script.

Refenmces 37 Van No&and, W. B. cl al. (1989) N&n

341,546-549

Selectivity of cell death

1 titer& C. L. et al. (19B5) Pmt. N&t Acrd. Sci. USA 82.4245-4249

2 Clennar, C. C. and Wonb, C. W. (1984) Biachem. Biophys. Res, Commun. 120, 855-890

The above sequence of events provides a basic outline of the processes that lead to plaque and tangle formation and to cell death in AD. However, it is difficult to understand why the distribution of the pathology is not general&d but seems to be ceritred on the hippocampal/amygdala complex early in the disease, and then seems to spread out from there along neuronal pathways22.

3 Kang, J. cl al. (1987) N&we 3?S, 733-736 4 KitaBoctli, N., Takahashi, Y.,

Tokuabima, Y., Shiojid, S. and Ito, H. (19BB) Nature 331,530-532

5 Mann, D. M. A. (19B9) Neuropalhal. Appl. Neurobiof. 15,317329

6 Rumble, B. et al. (1989) New Engl. /. Med. 320,1446-1452

7 YamaBucbi, H., Nakarato, Y., Hini, S. and shoji, M. (19%) Brain Res. 508, 320-324

38 Smith, R P., H@&i, D. A. and Braze, G. J. (1990) S&we 24Bj lf?bll26

39 Aulop, D. et rl. (l%lB) Pmt. Nafl Acad. Sri. USA t&279%2794

40 Yankner, B. A., Ddy, L K. and Kimcbnar, D. A. (1990) Scfmcc 250,

41 E, 8. A., Caceres, A. urd Duffy, L K. (1990) Proc. Nati Acad. Sri. USAB7. 9020+2i

42 Cban, M. and Yankner, B. A. (1991) Neuroeci. L&t. l25,22%%

43 Ausop. D., Yamamok T., Kamatani, F., Miyazakt, N. and lab& T. (1991) Brain Res. 551,l-r)

s Roberta, C. W., Allsop, D. and B&on, C. (1990) J. Neural. Newosug. Psychiaf. 53,373-378

9 Bugiani, 0.. Ciaccone, C., FranBione, B.. Cbatti, 6. and Ta@ii+ni, F. (1989) Neunwi. wf. 103,263-26s

10 D~omori. K. eI 41. (1989) .4r. \. Pathol. 134.243-251

44 Isbli, T., Kamatanl, F., Ha~a, S. and Sate, M. (19B9) Neumpathol. Appf. Newobiol. 15,l35-147

45 Sboji, M. cl al. (1990) Brain Rts. 512, 164&8

TOO little is known about the distribution, biochemistry and functions of APP or neuronal net- works to formulate an hypothesis about the underlying basis for this selective vulnerability. It may be that the disease process starts in the hippocampus because this is an area in which svnabtic re-

46 Mann, D. M. A., Roy&on, M. C. and Ravindra, C. R. (19%) J. Neural. Sci. 99, 153-164

47 Yankner, 8. A. el al. (1989) Science 245, 4174

11 Kidd, M. (1963) Naktre 197,192-193 12 Crowtber, R. A. (1991) B&him. Biaphys.

Acta 1096,1-9 13 Ak%op, D., Hagr, S., Bruton, C., Iabii, T.

and Roboti, G. W. (1990) Am. /. Prlhol. 136.255-260 _... ..__ . . ~~.

48 Whitaon, J. S., sdkoa, D. J. and Cotman, c. w. (19B9) scieaee 245, MBB-14%

49 Wbiteon, J. S., Gtabe, C G., Bbintani, E, Abcar, A. and Cobnan, C W. (1990) New&. LcN. 110, SW-324

50 Qoon, D. et al. (1991) Nahtre 352, I . ~~~ ~- 14 Yamaychl, H. rt a/. (1991) Am. J. Palhof. 239-241