Towards an Understanding of the Genetics of Alzheimer's Disease

Corinne L. Lendon and Alison M. Goate

Alzheimer's disease (AD) is the most common dementia in the elderly. It affects around 5% of the population over 65 years of age, and this figure rises rapidly with increasing age, to approximately 10–20% of those over 80 years old (29). Although the majority of those suffering from AD are elderly, the age of onset can be much younger: some cases as young as 35 have been documented. Clinically, AD is an insidious fatal dementia. The first symptom is usually a loss of short-term memory followed progressively by worsening memory loss, deterioration of mental ability, impairment of visuospatial skills, impairment of the perception and association of language, and progressive physical disability. Death occurs about 10 years after the onset of symptoms, mainly as a result of opportunistic infections. The latter years of a patient's life invariably require constant care, and in many cases they are spent institutionalized. As a result of its profound debilitating nature and the large number of individuals affected (4 million in the United States alone), AD constitutes a major health problem in aging populations.

AD is associated with a characteristic neuropathology: extracellular neuritic plaques and intraneuronal neurofibrillary tangles mainly within the temporal cortex. Histological detection of neuritic plaques and neurofibrillary tangles is used to diagnose "definite" AD when accompanied by a clinical history of "probable" AD dementia (29). The major proteinaceous component of the plaque core is b-A4, an aggregated form of a 39- to 43-amino-acid fragment of the b-amyloid precursor protein (b-APP) (17). b-A4 deposits occur diffusely and/or as dense cores surrounded by dystrophic neurites that comprise the extracellular neuritic plaques. Neurofibrillary tangles (NFTs) are largely composed of paired helical filaments containing an abnormally phosphorylated form of the microtubule associated protein, tau. The greatest numbers of these inclusions are found in the temporal cortex and hippocampus.

b-A4 deposition is a feature of "normal" aging. However, the number of plaque cores observed are fewer than in AD. Premature b-A4 deposition occurs in several other disorders; for example, AD-like neuropathology occurs in Down's syndrome (DS) subjects who live beyond their third or fourth decade (37). Studies of the brains of subjects with DS over a range of ages indicate that b-A4 deposition is an early event and that NFTs form decades after the first signs of b-A4 deposition. Cognitive decline has also been reported in aging DS individuals, suggesting that they may also dement. NFTs and diffuse deposits of b-A4 are also found in the brains of victims with dementia pugilistica, a rare disorder seen in boxers and battered wives where sufferers had received repeated head trauma (32). b-A4 deposition also occurs in the cerebral vasculature and as diffuse plaques in the brain parenchyma of individuals with the rare disorder hereditary cerebral hemorrhage with amyloidosis—Dutch type (HCHWA-D) (27). Cerebral hemorrhage occurs at an average age of 52 in these apparently healthy, normotensive subjects who do not normally develop symptoms of dementia (27).

The amyloid deposited in AD brains was isolated from cerebral vascular plaques (17) and shown to be a 4.2-kb polypeptide with a partial b-pleated sheet structure. Oligonucleotides corresponding to part of the peptide sequence were used to screen cDNA libraries. The cDNAs isolated were much larger than expected and appeared to code for an amyloid precursor protein (APP). Mapping in a somatic cell hybrid panel revealed that the APP gene mapped to a single locus in the middle of the long arm of chromosome 21 (Ch 21) between 21q11.2 and 21q22.1 (see Fig. 1) (24). The human APP gene contains 19 exons and covers 400 kb of genomic DNA (45).

AD was first described by Alois Alzheimer in 1907. The disorder was later reported to occur at high frequency in some families, suggesting a genetic component (30). In these families, AD was shown to follow an autosomal dominant pattern of inheritance with apparent 100% penetrance. The age of onset differs markedly between pedigrees but is quite consistent between affected members of the same family (66). Many epidemiological studies have shown that a positive family history is a consistent risk factor for AD (22). However, the vast majority of cases of AD do not show a clear pattern of inheritance. It is difficult to determine what proportion of cases are genetic in etiology for two reasons. Firstly, among late-onset cases some family members die of other causes before the age of onset of AD, which may lead to a failure to recognize a case as genetic. Secondly, the high frequency of AD in the elderly means that familial clustering could occur by chance quite often even when genetic factors are not involved.

Other risk factors that have been suggested to be involved in AD include head trauma, thyroid disease, and aluminum intake. Only in the case of head trauma, however, does there appear to be any significant correlation with disease (65). It has been suggested that a combination of genetic and environmental factors must predispose to AD because about a third of AD patients have affected first-degree relatives (see Basic Concepts and Techniques of Molecular Genetics, Genetic Strategies in Preclinical Substance Abuse Research, and Abuse and Therupatic use of Benzodiazepines and Benzodiazepine-Like Drugs for related topics and background).

One of the aims of the geneticist is to find the defective genes causing hereditary disorders. Two factors directed workers to search for an AD locus on Ch 21: (i) b-A4, the principal proteinaceous material of the senile plaque core, is a product of the much larger APP which is encoded by a gene which maps to Ch 21 and (ii) DS subjects, who invariably develop AD-like neuropathology, are trisomic for all or part of Ch 21. It is thought that an extra copy of a normal APP gene on Ch 21 in these individuals leads to overexpression of APP and the premature deposition of b-A4 (47). This is supported by findings of overexpression of APP in mouse trisomy 16, the mouse homologue to human Ch 21 (3).

These observations led to the suggestion that the APP gene on Ch 21 was a good candidate gene for the AD locus in the inherited forms of the disease. Support for this hypothesis could be drawn from the analogous situation of the prion dementias where prion protein is found in plaque deposits. Six mutations which co-segregate with the various forms of the prion dementias have been found in the human prion protein gene (PrP), which encodes the prion peptide deposited in Creutzfeldt–Jakob disease (CJD) and Gerstmann–Sträussler–Scheinker syndrome (GSS) (43).

The first indication of linkage to Ch 21 was reported by St. George-Hyslop et al. (60) in 1987. They described linkage to Ch 21 in four large AD pedigrees with histologically confirmed familial Alzheimer's disease (FAD). Each family had affected members over 6–8 generations, with a mean age of onset between 39.9 and 52 years. The putative location of a defective gene was deduced to be near two markers on the long arm of Ch 21. This was not in the 21q22 region, the so-called "Down's obligate region," but was closer to the centromere between 21q11.2 and 21q21, in the region of the APP gene. The highest lod scores, 4.25 and 4.06, occurred as two peaks on either side of markers D21S1 and D21S11. The story seemed well on its way to a tidy conclusion, especially when duplication of the APP gene was described in two cases of sporadic AD and not in seven controls (13). However, not only did other sporadic cases fail to reveal a gene duplication, but recombinants were reported between the APP gene and the disease in several large pedigrees (64). These findings appeared to rule out the APP gene as the disease locus. Subsequent reports appeared to exclude other areas of the long arm of Ch 21 and shed doubt on the original report (41, 51). Only one report confirmed a linkage to the proximal region of the long arm of Ch 21 in early-onset families (EOAD) (19).

The situation became clearer as a result of a large multicenter collaboration. In this study, 48 pedigrees were genotyped with five polymorphic DNA markers mapping to the proximal region of the long arm and two independent methods of statistical analysis were used: relative likelihood and affected pedigree member methods. Both methods indicated significant evidence for linkage on Ch 21 around four of the five markers. However, the majority of pedigrees did not contribute to the positive lod scores at these loci. When divided on the basis of age of onset, late-onset families showed no evidence for linkage to Ch 21, whereas the early-onset group gave significant evidence of linkage to this region. However, the observation of two peak lod scores on either side of the marker map even in early-onset families suggested the possibility of genetic heterogeneity even within this group. These results clearly indicated that despite the general uniformity of the AD phenotype, FAD was not a single homogeneous disorder. It now became clear that failure to detect and localize the disease locus on Ch 21 had been due, at least in part, to genetic heterogeneity. Indeed, each candidate gene would need to be excluded in each family (see Issues in long-Term treatment of Anxiety Disorders and Biological Markers in Alzhimer's Disease for related topics).

Although it was recognized that most cases of AD were not due to a gene defect on Ch 21, interest in the APP gene as a candidate disease locus was revived for some EOAD cases. Researchers were fortunate that a single large family existed with Ch 21-linked EOAD. This family (F23) provided a large part of the positive lod score in the collaborative study (60) and in our own study (19). Typing all individuals in the family for markers along the entire length of Ch 21 led to the identification of two recombination events which delineated a region between D21S1 and D21S17 in which the disease locus must lie. Approximately 20–30 megabases lying between these DNA markers, including the APP gene, was inherited in all affected family members (18).

At this time a mutation in exon 17 of the APP gene was reported in association with HCHWA-D. A G-to-C mutation resulting in a glutamic acid to glutamine substitution was discovered at position 693 of the full-length APP770 transcript (25) (see Fig. 2). These workers had found the mutation by sequencing the two exons which encode the deposited b-A4 peptide. This strategy was employed for F23 and led to the discovery of the first mutation that co-segregates with AD: a G-to-A transition at codon 717 in exon 17, causing a valine-to-isoleucine substitution (18). The mutation was detected in all affected family members but not in unaffected individuals over the age of onset nor in 100 unrelated controls; this provided evidence that the mutation could be pathogenic. Since then the mutation has been reported in an additional five Japanese families, one family from the United States, one Canadian family, two Italian families, and one British family (9, 16, 18). Further support for the probable pathogenicity of this mutation is the failure to reveal other mutations in the other 17 exons or a 330-bp regulatory region of the APP promoter in either of the two British families exhibiting this mutation (16). Soon after this, two other mutations were reported in codon 717: (i) a valine-to-glycine substitution in a British family (7) and (ii) a valine-to-phenylalanine substitution in a US family (35) both multiply affected by AD.

A double point mutation was found in two large Swedish families with EOAD (probably related) (33). Here a G-to-T and an A-to-C transversion result in amino acid changes: lysine to asparagine and methionine to leucine, respectively. These mutations were found in exon 16 at codons 670 and 671 of the APP770 transcript. A family was described with an unusual mixed phenotype that associates with a single base change in codon 692 of exon 17. One patient exhibited presenile dementia (onset 49.3 years), and four others had cerebral hemorrhage at a mean age of 39.5 years. The biopsies from one hemorrhaging patient showed extensive b-A4 immunological reactivity. The mutation is a C-to-G transversion resulting in an alanine-to-glycine substitution and has been detected in the histologically confirmed AD case in the absence of the HCHWA-D or other known AD mutations (21). This mutation was found in four other patients with cerebral hemorrhage and in seven who developed dementia, with a mean age of onset of 49 years and from the same family. The mutation did not occur in unaffected individuals ranging in age from 64 to 76; however, the pathogenicity of this mutation is in some doubt because of the absence of the mutation in a single elderly dementing patient (onset 61 years). The age of onset in this individual was significantly higher than in the individuals with mutations and may represent a nongenetic case of AD (a phenocopy). In addition, two other individuals of expected onset age had the 692 mutation and were healthy at the time of reporting. A C-to-T base change has also been reported in codon 713 of exon 17 (23). The resulting alanine-to-valine substitution was described in a single patient with schizophrenia who had a family history of the disorder. However, no other living relatives were available for study, and the mutation was not found in 86 unrelated schizophrenics or 156 controls (10). An alanine-to-threonine mutation also in codon 713 was described in a sporadic case of AD (onset 59 years) but not in five unaffected elderly relatives (6). Caution should be exercised with such findings because they may simply be rare polymorphisms. Alternatively, the observation of mutations in elderly unaffected relatives may reflect incomplete penetrance.

In summary, four mutations have been found in a total of 15 families with EOAD. This only represents a small proportion of familial EOAD [estimates range from 5% to 25% (65)] and an even smaller proportion of the total number of cases of AD (50).

Although mutations have been found in the APP gene in several AD families, it has not yet been demonstrated definitively that the mutations are sufficient to cause the disease. It is possible that they merely predispose to AD. However, a primary causal role is supported by the detection of mutations only in affected family members. Further supporting evidence comes from the fact that families with the 717 valine-to-isoleucine mutation are from different racial origins and that none of the Japanese or Caucasian families share APP haplotypes, indicating independent mutational events.

There are several possibilities as to how mutations in the APP gene might cause disease. The presence of an extra copy of a normal APP gene in trisomy 21 suggests that overexpression of normal APP may be sufficient to cause premature b-A4 deposition. These mutations could theoretically have their effect at one of several different levels: They could influence the rate of transcription or mRNA stability; they could affect the rate of translation of the mRNA or the post-translational modification of the protein; they could alter the proteolytic processing of APP; or for those mutations within the b-A4 sequence they may alter the physicochemical properties of the peptide. The mutations in the APP gene which are associated with the AD phenotype (i.e., mutations at codons 670/671 and 717) do not lie within the sequence encoding the b-A4 peptide, whereas mutations associated with HCHWA-D at codons 692 and 693 lie within this sequence (see Fig. 2). The 717 mutations lie 3 amino acid residues from the predicted carboxyl terminal end of b-A4. However, to-date, b-A4 has not been sequenced from the brain of an individual with an APP717 mutation, and these mutations may or may not be within the deposited sequence in such individuals. It has been suggested that these mutations affect proteolytic processing of APP because all four seem to occur in sequences flanking b-A4.

APP proteolytic processing is complicated and, as yet, not completely understood, although evidence suggests that the proportion of APP processed by different routes varies in different cell types. The detection of extracellular amino-terminal fragments in cultured mammalian cells transfected with full-length APP cDNA and a variety of hybrid and modified APPs indicated that specific cleavage occurs within the extracellular domain of b-A4 (57) (see Fig. 2). This was corroborated by results from nontransfected neuronal PC12 cells (1). The enzyme, termed "a-secretase," cleaves at lysine 16 of b-A4 (encoded by codon 687) but appears to have a broad sequence specificity (67). a-Secretase cleaves cell-surface-membrane-bound APP to produce a soluble "secreted" amino-terminal derivative extracellularly and a short C-terminal fragment that is degraded within the endosomal/ lysosomal compartment of the cell. The cleavage occurs within the b-A4 fragment and therefore cannot lead to deposition of intact fragments. The 692 mutations for mixed AD/HCHWA-D and 693 HCHWA-D mutation occur only about 6 amino acid residues away from the a-secretase cleavage site, but as yet there is no evidence that these mutations have any effect on cleavage.

Cellular studies have revealed that there are alternative pathways of APP processing. A lysosomal route of APP degradation is indicated by the finding of C-terminal fragments within lysosomes. The C-terminal fragment is probably derived from the a-secretase cleavage whereas the larger fragment which contains an intact b-A4 sequence is probably derived from the "b-secretase" cleavage. The "b-secretase" activity was detected by studying APP processing in mixed brain cell cultures. This activity cleaves APP between the methionine residue of APP770 at codon 671 and the aspartic acid residue at APP672 (55). Cleavage at this site produces fragments containing intact b-A4. The Swedish mutation (33) occurs in the two amino acids preceding this cleavage site. It is likely that this cleavage site is crucial to the production of soluble b-A4, a normal processing product detectable in body fluids (20, 56). Indeed, transient transfection of human kidney 293 cells with DNA constructs encoding APP containing the base substitutions at 670 and/or 671 of the Swedish double mutation leads to a six-to eightfold increase in soluble b-A4 production. The 670 mutation is largely responsible for this increase in soluble b-A4 detectable in the media compared to wild-type control (8). It would appear that the production of soluble b-A4 is distinct from the endosomal/lysosomal pathway of C-terminal degradation because truncation of the cytoplasmic tail of APP does not prevent enhanced generation of soluble b-A4 in these cells. Recently, similar increases in the production of b-A4 derivatives were produced in human neuroblastoma cells transfected with constructs containing the mutation at codon 670 (4). No effect on soluble b-A4 levels was observed upon transfection of constructs containing APP717 mutations. Similar transfection of human neuroglioma cells with constructs containing the HCHWA-D mutation do not cause any detectable changes (15).

Preliminary evidence suggests that the mutation causing HCHWA-D may cause premature b-A4 deposition by altering the secondary structure of the peptide rather than altering APP processing (14).

In summary, there are several possible processing pathways involving membrane-bound APP, all of which could be perturbed by the known mutations to accelerate b-A4 deposition. However, as already noted, the known mutations only account for a very small number of FAD cases. If b-A4 deposition is central to the disease process, then defects in any of the enzymes, precursors, cofactors, activators, and inhibitors that are involved in the processing of APP could lead indirectly to premature b-A4 deposition. Because potentially amyloidogenic fragments of APP occur in normal brain and cerebrospinal fluid (CSF), AD could result from abnormal polymerization of normal degradation products. It is possible that the APP mutations merely predispose to AD and other amyloidopathies, perhaps simply by bringing forward the age of onset. Other genetic and environmental factors may be involved, particularly because most cases of AD are sporadic.

The search for other disease loci gathered pace following the recognition of the genetic heterogeneity of AD. One of the early genetic investigations of FAD that failed to show linkage to Ch 21 (40) used a mixture of families, but the majority were of late onset (mean age: >60 years). When this group was reexamined using the affected pedigree member method of linkage analysis which excludes unaffected members, two chromosomal regions gave significant results: proximal 19q (see Fig. 3a) and the FAD-linked region of Ch 21 (40). Division of the families by age of onset revealed that the associations were between LOAD and Ch 19 and between EOAD and Ch 21. Indeed one of the EOAD families was later found to have a mutation in the APP gene (18). Independently, several groups have presented supporting evidence for linkage to Ch 19 (46). Two association studies have provided further support for this region of the genome.

Recently, several independent lines of evidence have provided support for the hypothesis that the disease locus on Ch 19 is the apolipoprotein E (ApoE) gene. ApoE is one of the many different proteins found to associate with b-A4 amyloid fibrils (5, 36, 58). Immunohistochemical studies show that ApoE accumulates extracellularly in senile plaques and intracellularly in NFT in both autopsy and biopsy samples (36, 70). This ApoE seems to bind tightly to b-A4 because ApoE immunoreactivity is enhanced when sections are pretreated with formic acid, which is thought to denature amyloid polymers (36). ApoE staining colocalizes with Congo red stain as evidenced by ultrastructural studies of autopsy tissue. However, local ApoE production, as well as leakage and neuronal uptake of serum protein occurring in the aged brain, could contribute to ApoE immunoreactivity in the brain (70). In vitro binding studies have also shown tight binding of ApoE to b-A4 (61, 62). ApoE and a related protein apolipoprotein CII (Apo CII) map to the region of Ch 19 originally implicated in the linkage studies. Association studies have recently been carried out using polymorphisms within both of these genes. The frequency of the "F" allele of Apo CII was found to be increased in family members affected by AD compared to unrelated controls in one study (53) but was not reproduced in a second study (49). However, association studies using polymorphism within the ApoE gene has demonstrated a robust and reproducible increase of the ApoE4 allele within familial and sporadic cases of late onset AD (11, 48, 49, 62). This is perhaps somewhat surprising when one considers that the ApoCII locus is less than 50 kb from the ApoE locus.

There are three common isoforms of ApoE encoded by the alleles E2, E3, and E4. The variants differ at residues 112 and 158: e3 cysteine 112, cysteine 158; e4 arginine 112, cysteine 158; e2 cysteine 112, arginine 158. A study of 2000 chromosomes estimates the frequency of the E3 and E4 alleles in the normal population to be 0.78 and 0.14, respectively (31). The e2 variant is defective in binding to the low-density-lipoprotein receptor (LDL-R). Individuals with abnormal lipoprotein metabolism and E2/E2 homozygosity are at increased risk for developing type III hyperlipoproteinemia. The e4 variant has a greater affinity for lipoprotein particles than does e3, resulting in more efficient lipid clearance, leading to down-regulation of hepatic LDL-R and enhanced plasma levels of cholesterol and triglycerides (28).

In an initial study of 30 AD families with mixed age of onset, the E4 allele frequency was 0.52 ± 0.06 compared to that of age-matched unrelated controls, 0.16 ± 0.03 (62). When restricted to affected and unaffected family members from 42 LOAD onset families, AD occurred in 91% of individuals homozygous for the E4 allele and in 47% of those with the E3/E4 allele, whereas only 20% had AD with allele combinations E3/E3, E3/E2, and E2/E4 (11). Similar trends have since been described (39). Enrichment for the E4 allele has also been observed in cases of sporadic AD (42, 44, 48). Immunocytochemistry of autopsy material from LOAD patients has revealed that the b-A4 burden (plaque numbers and amount of congophilic angiopathy) shows a positive correlation with the number of E4 alleles (44, 54).

ApoE is one of the several glycosylated proteins associated with plasma lipids (28). It plays a major role in regulation of lipid transport by acting as a ligand for binding ApoE-containing lipoproteins to the low-density lipoprotein receptor (LDL-R) and for binding chylomicron remnants to the LDL-R-related protein (LRP) (2). The ApoE gene is located on the proximal long arm of Ch 19q13.2 and consists of 4 exons and 3 introns covering 3597 bp. The gene encodes a 299-amino-acid protein, a third of which is rich in the basic residues arginine and lysine and is involved in receptor binding. The remaining 976-bp C-terminal region is the amphipathic a-helical lipid-binding domain (12, 38). ApoE is synthesized in many organs, and the highest levels are produced in the liver and the brain. It is the main lipoprotein in the CSF, where it is brain-derived from glia and astrocytes but not from neurons. LRP is the major receptor that binds ApoE in the brain and is particularly abundant in astrocytes (44). The lipid transport function of ApoE is thought to extend to the delivery of lipid to regenerating nerves and is postulated to be involved in response to neuronal injury. The levels of ApoE in the CSF rise significantly with several central nervous system disorders (5).

The genetic data reviewed here strongly support the hypothesis that possession of an Apo E4 allele is an important risk factor for the development of LOAD and sporadic AD but not for EOAD. It has been proposed that ApoE4 acts as a "pathological molecular chaperone" that binds to soluble b-A4 and enhances b-pleated sheet formation and amyloid fibril stability (70). Investigation of possible differences in the binding characteristics of ApoE alleles to synthetic b-A4 are underway (61).

The genetic evidence for the E4 allele as a risk factor above is compelling, yet many cases of LOAD sporadic AD (29% and 38%, respectively) have no E4 allele (11, 44). However, the precise risk associated with the ApoE4 allele requires use of an epidemiologically based sample. Unpublished data, presented at the 1993 Society for Neuroscience meeting by Schellenberg and colleagues, showed that in an epidemiologically based sample of 11,000 individuals there was a much smaller enrichment of the E4 allele with AD (0.26 compared to 0.19 nondementing individuals) than observed in previous studies. The presence of an E4 allele is therefore not diagnostic for AD. The allelic association between AD and ApoE could arise from linkage disequilibrium with a nearby predisposing locus on Ch 19. One possible candidate for such a role is an APP-like protein which has recently been mapped to the proximal portion of the long arm of human Ch 19 (68). The dosage effects of the ApoE4 allele (11), if confirmed by other groups, would strongly support the notion that ApoE is the disease locus rather than being linked to the real disease locus. Whatever the role of ApoE, the pronounced allelic association in many cases of LOAD and sporadics has served to bring the hitherto little supported disease locus on Ch 19 back into hot contention.

Convincing evidence of an alternative locus for FAD was reported by four independent groups in 1992. Linkage was reported to the marker D14S43 on Ch 14 at q24.3 in a collection of EOAD families (52) (see Fig. 3b). Schellenberg's findings were quickly followed by others reporting an AD locus on Ch 14 in unrelated families. The strongest evidence for linkage to Ch 14 was found in two large EOAD Belgian pedigrees. Linkage to D14S43 was found with a maximum lod score of +13.25 at zero recombination (63). St. George-Hyslop (59) reported linkage to Ch 14 near markers D14S43 and D14S53 with a maximum multipoint lod score of +23.4. The pedigrees were a combination of 21 families of mixed origin, with a mean family age of onset ranging from 42 to 84 years. These workers report that significant linkage to Ch 14 was found in the four EOAD families used in the original linkage report to Ch 21 (60). The lod scores for each family are much higher with the Ch 14 markers, suggesting that the small positive lod scores on Ch 21 were a chance event and did not represent true linkage. A peak lod score of +7.8 at no recombination was generated from nine EOAD families, indicating tight linkage to D14S43 (34). Subsequent to these original reports, genetic linkage maps for Ch 14 have been published using additional polymorphic markers in q24.3 (69). Use of these markers in the linked AD families has identified D14S63/S57 and D14S61 as the centromeric and telomeric flanking markers, respectively. Four markers show no recombination with the disease (D14S43, D14S76, D14S71, and D14S77). The distance between the flanking markers is approximately 12 cM, but the majority of this distance is between D14S63/S57 and D14S77. It is therefore a high priority to identify additional polymorphic markers in this region of the chromosome. Two candidate genes map to this portion of the long arm of Ch 14: c-fos and heat-shock protein 70. So far, no defects in these genes have been reported.

The Volga Germans, a group of families originating from the Volga region of Russia, do not show linkage to markers on Ch 21, 14, or 19, indicating further genetic heterogeneity in AD (52).

A mutation has been reported in codon 331 of mitochondrial NADH dehydrogenase subunit 2 (ND2) (26). The mutation occurred in 10 of 19 AD patients and in two of six patients with amyotrophic lateral sclerosis (ALS). Mutations within the mitochondrial genome are known to accumulate with aging although their pathological significance is unclear, particularly because the mitochondrial genome is highly polymorphic. The 331 mutation could be a common polymorphism or the discovery of a mutation that increases the risk of AD. If AD-causing mutations enhance susceptibility of neurons to damage, then defects in a host of metabolically sensitive cellular processes, which might otherwise be below a threshold for damage or be compensated for, may be sufficient to initiate a pathway leading to AD and AD-like pathology.

A small proportion of EOAD appears to be caused by mutations in the APP gene encoding b-A4, a major component in diffuse and aggregated plaques. One of the mutations appears to initiate a pathogenic process by overproduction of soluble b-A4. A second disease locus causing EOAD has been localized to the long arm of Ch 14 in band q24.3. Linkage and molecular genetic studies are underway to both narrow down the candidate region and to clone the region in yeast artificial chromosomes: a prelude to gene isolation and characterization. A third locus on the long arm of Ch 19 has been implicated in predisposition to late-onset AD. Evidence from association studies supports the notion that this locus is the apolipoprotein E gene or another gene very close by. It is clear that the number of genetic defects causing AD will not be confined to the loci on chromosomes 14, 19, and 21 because linkage cannot be demonstrated in families of Volga German origin. The determination of whether APP or b-A4 plays a major role in the pathogenesis of all cases of AD awaits the isolation of these other genes. As more gene defects are discovered, and a knowledge of the role and properties of their products are accumulated, the potential for drug design of many-fold greater phenotypic specificity and efficacy increases.

published 2000