- Email: [email protected]
Genetic association between intronic polymorphism in presenilin-1 gene and late-onset Alzheimer's disease
Genetic association between intronic polymorphism in presenilin-1 gene and late-onset Alzheimer’s disease
Michelle
Wragg, Mike Hutton, Chris Talbot, and the Alzheimer’s Disease Collaborative Group* Introduction
Summary Mutations in the presenilin-1 (PS-1) gene are associated with early-onset Alzheimer’s disease. 40-50% of the risk for late-onset disease has been attributed to alleles at the apolipoprotein E (ApoE) locus. We have looked for an association between PS-1 and late-onset disease.
Background
Methods We collected blood samples from 208 white cases of dementia of the Alzheimer type and from 185 agematched controls (mean ages 76·9 and 76·2 years, respectively; 58% female in each series). Clinical diagnostic accuracy for Alzheimer’s disease in our patients is 96%. We also studied 29 African-American patients with dementia of the Alzheimer type and 50 age-matched controls (cases vs controls, 77·2 vs 72·0 years; 72 vs 77% female). We used PCR to test for an association between Alzheimer’s disease and a polymorphism within the intron 3’ to exon 8 of the PS-1 gene. The ApoE genotype of most of the cases and controls was known from previous
investigations. Findings Homozygosity of the 1 allele in the PS-1 gene
was
associated with a doubling of the risk for late-onset Alzheimer’s disease compared with the [12]/[22] genotype (odds ratio 1·97, 95% Cl 1·29-3·00). The proportion of Alzheimer’s disease cases in the white population that could be attributed to homozygosity at this locus, as estimated by the attributable fraction, was 0·22. This compares with 0·35 for a single copy of ApoE4 and 0·15 for two copies. The smaller African-American series showed similar distribution of PS-1 genotype between cases and controls.
Interpretation In our white series of cases, PS-1 accounted for about half as much of the risk for late-onset Alzheimer’s disease as did ApoE4.
*Listed at end of report
Department of Psychiatry, Washington University School of Medicine, St Louis, Missouri, (M Wragg MSc, C Talbot BSc); and Department of Psychiatry, Suncoast Alzheimer’s Disease Laboratories and Gerontology Center, University of South Florida, Tampa, Florida, USA (M Hutton PhD)
Correspondence to: Dr Alison Goate, Department of Psychiatry, Washington University School of Medicine, St Louis, MO 63100, USA
Mutations in the presenilin-1 (PS-1) gene on chromosome 14 cause a significant proportion of earlyonset, autosomal dominant Alzheimer’s disease.’ Many mutations that cause disease between the ages of about 30 and 50 years have been described.H Whilst the PS-1 gene is generally thought to be involved only in the rare, familial early-onset form of the disease, allele sharing between affected family members with late-onset disease has been observed.5 This allele sharing was not found with standard maximum-likelihood methods but with the affected-pedigree-member method of genetic analysis. These results suggest that the chromosome 14 locus was not behaving as an autosomal dominant in late-onset disease.5 Genetic analysis of late-onset Alzheimer’s disease has implicated two loci: the apolipoprotein E (ApoE) gene and the al-antichymotrypsin (AACT) gene. As much as 40-50% of the risk for late-onset disease is attributable to alleles at the ApoE locust ApoE4 is thought to increase risk of Alzheimer’s disease in a dose-dependent manner whilst the ApoE e2 allele may be protective.7-8 It has been reported that homozygosity at a polymorphism within the AACT gene modified the ApoE risk,9 although we could not repeat this observation. 10 We have now tested a polymorphism in the PS-1 gene for association with lateonset disease.
Patients, controls, Screening protocol
for
and methods
polymorphism
extracted from blood with standard procedures. During DNA sequence analysis in cases of early-onset Alzheimer’s disease, we and others have identified a common polymorphism within the intron 3’ to exon 8.3 The most common allele has an A at nucleotide 16 (allele 1) in this intron whilst the variant allele has a C at this position (allele 2). The polymorphism does not create or destroy any restriction enzyme sites. A mismatch primer was designed to contain two mismatched basepairs four and five basepairs from the 3’ end of the primer and five and six basepairs from the polymorphism. When incorporated into a PCR product this primer produces a BamHI cut-site when a C is present and no cut-site when an A is present at the polymorphic site. This enables the two alleles to be distinguished by digestion of a PCR product with BamHI. Forward primer’ was 951 5’ CACCCATTTACAAGTTTAGC 3’, reverse (mismatched) primer was 5’ CACTGATTACTAATTCAGGATC 3’. Use of these primers gives a PCR product of 199 basepairs which is cleaved by BamHI to produce fragments of 181 and 18 basepairs. 50-100 ng DNA was used as template in 25 p.L reactions. The reaction mix consisted of (final concentrations): 0-2 mmol/L dNTPs, 30 pmol/L each primer, 1XTNK50 buffer, and 0-5 U Taq DNA polymerase. PCR was done at: 94°C for 5 min (94°C 30 s, 45°C 30 s, 72°C 30 s) for 35 cycles, and 72°C for 3 min. 5 U BamHI was added and DNA
was
509
First figure=number of cases with genotype or, for alleles, number of alleles. Figure in parentheses number of cases screened in each category. Cases vs controls, X2, *p=0.007. tap=0.0017 (odds ratio 1.97, 95% CI 129-300), and tp=0’006.
Table 1: PS-1
carried out at 37°C for 3 h. The digested product was 3% agarose gel, which is sufficient to separate the digested product so that 200 and 182 basepair bands can be distinguished. Multiple samples were rechecked by sequencing to on
a
ensure our
allele-scoring system
was accurate.
genotyped as described." The coding region of PS-1 is encoded by exons 3-12. Each exon was amplified and sequenced as described. ApoE
was
12
Patients and controls Four different series of patients and controls were analysed. First, white cases and controls were recruited from the Greater St Louis metropolitan community through the Washington University Alzheimer’s Disease Research Center. Second, a similar, smaller population was collected from the Tampa metropolitan area through the Suncoast Gerontology Center. Third, the African-American case-control sample was recruited at St Louis as above. Recruitment for these three series was not based on a family history of Alzheimer’s disease; however, about 40-50% of participants have a first-degree relative with a diagnosis of dementia of the Alzheimer type (DAT). Diagnostic criteria for DAT are equivalent to or more stringent than those for probable and possible Alzheimer’s disease.13,14 Clinical diagnostic accuracy for histological Alzheimer’s disease in our patients is 96%." Controls for each group met all exclusion criteria for DAT but were not demented. Blood samples were collected from 208 white DAT patients (mean age 76-9 years, SD 8-8, 58% female) and 185 age-matched controls (76-2, 11-0, 58%). The mean age of onset of dementia in the DAT patients was 71-3 (8-9). Blood samples were collected from 29 African-American DAT patients (77-2, 9-6, 72% female) and 50 age-matched controls (72-0, 7-8, 77% female). The mean age of onset of dementia in the DAT patients was 71-8 (5-2). The fourth group was collected in the UK as part of a study on the genetics of Alzheimer’s disease." The 32 unrelated patients are white and come from families multiply affected by late-onset disease (mean age 81-1, SD 7-3, 78% female). The mean age of onset of dementia was 73-4 (6-6).
Statistical methods
Genotypic and allelic distributions were analysed with X2 tests and, for 2X2 contingency tables, with odds ratios. The effect of
Figure: Agarose gel showing PS-1 polymorphism genotypes Gel shows band pattern for PS-1 polymorphism m intron 3’ of exon 8. TT corresponds to 11 homozygotes and GG corresponds to 22 homozygotes. TG represents the 12 heterozygote bp=basepairs.
510
frequency.
Addition of numbers for 11, 12, and 22
genotypes gives total
genotype and allele distributions
digestion run
is
examined with multiple linear regression. Logistic regression analyses were used to examine the effect of gender, ApoE4 genotype, PS-1 genotype, and interactions between these variables to predict disease status. Two-tailed tests of significance were used throughout. All analyses were done on SPSSv6 for Windows. Attributable fraction (AF) was calculated from AF=f(OR-1)/[l+f(OR-1)], where f is the frequency of the risk factor in the population and OR is the odds ratio for developing Alzheimer’s disease in individuals with the risk factor compared with those without the risk factor.
genotype
on
age
at onset was
Results Because there were no significant differences in genotype or allele distributions for the PS-1 polymorphism between white controls and patients from St Louis and Tampa, we pooled data for these control and for these patients’ populations (table 1). The genotypic distribution in the controls was close to that expected under HardyWeinberg equilibrium (p=0-635). There were, however, significant differences in genotype (p=0’007) and allele (p=0’006) distributions between controls and patients in the white population. The effect was due to a higher proportion of DAT patients, who were homozygous for the absence of the artificial BamHI site (ie, individuals who were homozygous for allele 1 in figure 1). With the [22] genotype as a reference, the [12] genotype was not associated with increased risk of Alzheimer’s disease (odds ratio 0-98, 95% CI 0-57-1-68), whereas the [11] genotype was associated with an approximate doubling in risk (1-96, 1-11-3-48). In the analyses that follow, we combined the [12] and [22] genotypes into a single category. Analysis of probands from the UK families with latesimilar genotype onset Alzheimer’s disease gave to the American series (table 1), suggesting frequencies that the same effect might be operating, although the lack of an adequate control group for the UK cases precluded formal
analysis. ApoE genotypes of most of these groups have previously been determined and showed the expected increase in ApoE4 allele frequency and decrease in age of onset of Alzheimer’s disease associated with ApoE4 dosage.s,11 When the sample was split on the basis of ApoE genotype, the PS-1 [11] genotype was associated with a doubling of risk of Alzheimer’s disease in carriers of zero or one ApoE4 allele, but not in ApoE4 homozygotes (table 2). The small cell sizes for ApoE c4 homozygotes makes it difficult to interpret the significance of this observation. However, for the common ApoE4 genotypes, there was no interaction between ApoE and PS-1 genotypes. The association between the PS-1 [11] genotype and Alzheimer’s disease was present and of similar magnitude The
population stratification confounded
our
analysis because:
the genotype distributions in cases and controls were similar at each site (pO-11) and these distributions are similar to those in an independent British sample (MJ Owen, Department of Psychological Medicine, University of Wales College of Medicine, Cardiff, UK); the controls were close to Hardy-Weinberg equilibrium; and independent analysis of the results from the two centres gave similar effects (St Louis, odds ratio [95% CI] 1-82 These data are derived from table 1, except that available for 1 control and 4 cases.
ApoE genotype results were
Table 2: Distribution of controls and
cases
by PS-1.
or
not
ApoE4
genotypes
aged 65 years or less (odds ratio 184, 95% CI 0-97-3-50) and for cases with onset over the age of 65 years (2-10, 1-32-3-33).Further, there was no evidence, by linear regression analyses (including or excluding ApoE4 genotype as a variable) that PS-1 genotype for
cases
of onset of dementia in our series. We used logistic regression to model our data. Consistent with the results presented above, ApoE4 genotype (p=0’000003) and presence or absence of PS-1 [ll] genotype (p=0°0027) were significant predictors of disease status. There was no significant interaction When we between ApoE and PS-1 genotypes (p>0-86). did logistic regression analyses separately for men and women, the risk-increasing effect of the PS-1 [11] genotype was greater in women (odds ratio 2-24, 95% CI 1-21-4-15) than in men (1-71, 0-86-3-14). However, this difference was not as statistically significant, demonstrated by the absence of significant interaction between PS-1 genotype and sex (p=058) in the regression model on the total sample. A measure of the overall contribution of a risk factor to disease in a population is provided by the attributable fraction, which refers to the excess fraction of disease in a population that would not have occurred if the risk factor had been absent. For the PS-1 [11]genotype in our white series the attributable fraction was 0-22. This compares with 0-35 for possession of a single copy of ApoE4 and 0’15 for two copies of ApoE4. Thus, in our series, PS-1 accounted for about half as much of the morbidity as did
predicted age
ApoE4. Sequencing of the entire open-reading frame of the PS-1 gene in ten late-onset cases of Alzheimer’s disease who were homozygous for the 1 allele of the PS-1 gene and the 3 allele of the ApoE gene failed to reveal any differences from the previously published sequence. Analysis of the small sample of African-American cases and controls revealed that the frequency of this polymorphism was different in the African-American population compared with the whites, but that the frequencies were similar between Alzheimer’s disease cases and controls (table 1).
Discussion We found that
homozygosity for allele
1 of the PS-1 gene associated with increased risk for polymorphism development of Alzheimer’s disease in the white population. Because our finding was statistically significant at p<0002, a type 1 error is unlikely. Although tight ethnic matching of controls in North American populations can be difficult, we do not believe that was
[1,13-2’92]; Tampa,
2-70
[1-08-6-77]).
There are three possible explanations for the association we found. First, the polymorphism is biologically relevant to the disease. Second, the polymorphism is in linkage disequilibrium with biologically relevant variability elsewhere in the PS-1 gene. Third, the polymorphism is in disequilibrium with genetic variability in another adjacent gene. The third
explanation seems unlikely. The position of the polymorphism
in the intron between exons 8 and 9 suggests that if the polymorphism is biologically relevant, the most likely mode of action would be through the modulation of alternate splicing, particularly of exons 8 and 9 of the PS-1 gene. Exon 8 is alternately spliced in some tissues1s and a mutation in the acceptor site in the same intron led to early-onset disease in a UK family through the loss of exon 9.16 In addition, exon 8 is the site of the most prominent cluster of mutations leading to early-onset disease.1,2 Thus, a subtle alteration of alternate splicing caused by this genetic variability remains a possibility. However, if the polymorphism were a relevant cause of the disease, the association with disease would occur in all populations. Our data from the African-American series indicate that this is not the case. For this reason, we favour the explanation that this polymorphism is in linkage disequilibrium with relevant variability elsewhere in the PS-1 gene. Sequence analysis of the open-reading frame of ten cases with a [11] genotype failed to reveal coding changes. Thus, any relevant variability in the PS-1 gene must be in the promoter or other non-coding regions. Our data suggest several experiments. The strength of the association needs to be tested in other ethnic groups. The effect of the polymorphism on alternate splicing of the PS-1 gene should be examined. Other genetic variability within the .P.S*-7 gene should be sought, particularly in recognised promoter and control elements, and then this variability tested for association with disease, both in case-control series and in pedigree datasets. PS-1 has been found in plaques from "typical" Alzheimer’s disease cases.17 That observation and our data are consistent with the view that all cases of Alzheimer’s disease share common pathogenic mechanisms and that PS-1 is part of this biochemical pathway. Other members of the Alzheimer’s Disease Collaborative Group are: Frances Busfield, Sang Woo Han, Corinne Lendon, Robert F Clark, John C Morris, Dorothy Edwards, and Alison Goate (Departments of Psychiatry, Neurology and Occupational Therapy, Washington University School of Medicine, St Louis, Missouri); Eric Pfeiffer, Richard Crook, Guy Prihar, Helen Phillips, Matt Baker, and John Hardy (Departments of Psychiatry, Pharmacology, Biochemistry, and Neurology, Suncoast Alzheimer’s Disease Laboratories and Gerontology Center, University of South Florida, Tampa, Florida, USA); Martin Rossor and Henry Houlden (Department of Neurology, Imperial College of Science, Technology & Medicine, London W2); Eric Karran and Gareth Roberts (Department of Moleuclar Neuropathology, SmithKline Beecham, Harlow, Essex); and Nick Craddock (Department of Psychological Medicine, University of Wales College of Medicine, Cardiff, UK).
511
Work in AG’s laboratory was supported by a career development award from the NIH (AG00634), a Zenith Award from the Alzheimer’s Association, the Metropolitan Life Foundation, and the McDonnell Center for Cellular and Molecular Neurobiology. Work in JH’s laboratory was supported by NIH project grants (AG11871 and AG12028), the Metropolitan Life Foundation, and the American Health Assistance Foundation. SmithKline Beecham Advanced Technologies in Genetics provided support to JH and AG. NC is a Wellcome Trust Senior Clinical Research Fellow. The David and Frederick Barclay Foundation, the MRC and the Alzheimer’s Disease Society supported MR and the collection of British families has been supported by MRC grants to MR, JH, and AG. Collection of the St Louis samples was supported by NIA grants to JCM (AG03991 and AG05681) and pilot grants from the Alzheimer’s Disease and Related Disorders Program, University of Missouri Alzheimer’s Research Center, and the Alzheimer’s Association to CL. We thank the physicians and staff of the clinical core of the Alzheimer’s Disease Research Center at Washington University for subjects’ evaluation. Private donations to all academic centres are also acknowledged. We thank the families for their continuous enthusiasm for this project.
2
3 4
6
Schellenberg G, Payami H, Wijsman E, et al. Chromosome 14 and late onset familial Alzheimer’s disease. Am J Hum Genet 1993; 53: 619-28. Corder E, Saunders A, Strittmatter W,
512
7
8
1432-33. Kamboh MI, Narambir D, Sanghera K, Ferrell R, DeKosky S. APOE4-associated Alzheimer’s disease risk is modified by &agr;1-antichymotrypsin polymorphism. Nature Genet 1995; 10: 486-88. 10 Talbot CJ, Houlden H, Craddock N, et al. Polymorphism in AACT gene may lower age of onset of Alzheimer’s disease. NeuroReport (in
9
press). et al. Confirmation that the E4 allele is associated with late onset familial Alzheimer’s disease. Neurodegeneration 1993; 2: 283-86. 12 Hutton M, Busfield F, Wragg M, et al. Complete analysis of the presenilin 1 gene in early onset Alzheimer’s disease. NeuroReport
11 Houlden
H, Crook R, Duff K,
apolipoprotein
(in
13 McKhann
Sherrington R, Rogaev E, Liang Y, et al. Cloning of a gene bearing mis-sense mutations in early-onset familial Alzheimer’s disease. Nature 1995; 375: 754-60. Clark RF, Hutton M, Fuldner R, et al. The structure of the chromosome 14 Alzheimer’s disease gene, presenilin 1 (PS-1) and identification of six novel mutations in early onset AD families. Nature Genet 1995; 11: 219-22. Wasco W, Pettingell W, Jondro P, et al. Familial Alzheimer’s chromosome 14 mutations. Nature Med 1995; 1: 848. Cruts M, Backhovens H, Wang S-Y, et al. Molecular genetic analysis of familial early onset Alzheimer’s disease linked to chromosome 14q24.3. Hum Mol Genet 1995; 4: 2363-72.
5
onset
press).
References 1
E type 4 allele and the risk of Alzheimer’s disease in late families. Science 1993; 261: 921-23. Corder E, Saunders A, Risch N, et al. Protective effect of apolipoprotein E type 2 allele for late onset Alzheimer disease. Nature Genet 1994; 7: 180-83. Talbot C, Lendon C, Craddock N, Shears S, Morris JC, Goate A. Protection against Alzheimer’s disease with apoE ∈2. Lancet 1994; 343:
apolipoprotein
et
al. Gene dose of
G, Drachman D, Folstein M, Katzman R, Price D, Stadlan EM. Clinical diagnosis of Alzheimer’s disease: report of the NINCDS-ADRDA Work Group under the auspices of Department of Health and Human Services Task Force on Alzheimer’s Disease.
14 15
Neurology 1984; 34: 939-44. Berg L, Morris JC. Alzheimer’s disease. In: Terry RD, Katzman R, Bick K, eds. Alzheimer’s disease. New York: Raven Press, 1994: 9-21. Rogaev EI, Sherrington R, Rogaeva E, et al. Familial Alzheimer’s
disease in kindreds with missense mutations in a gene on chromosome 1 related to the Alzheimer’s disease type 3 gene. Nature 1995; 376: 775-78. 16 Perez-Tur J, Froelich S, Prihar G, et al. A mutation in Alzheimer’s disease destroying a splice acceptor site in the presenilin-1 gene.
NeuroReport (in press). 17 Wisniewski T, Palha JA, Ghiso J, Frangione B. S182 protein in Alzheimer’s disease neuritic plaques. Lancet 1995; 346: 1366.
Michelle
Wragg, Mike Hutton, Chris Talbot, and the Alzheimer’s Disease Collaborative Group* Introduction
Summary Mutations in the presenilin-1 (PS-1) gene are associated with early-onset Alzheimer’s disease. 40-50% of the risk for late-onset disease has been attributed to alleles at the apolipoprotein E (ApoE) locus. We have looked for an association between PS-1 and late-onset disease.
Background
Methods We collected blood samples from 208 white cases of dementia of the Alzheimer type and from 185 agematched controls (mean ages 76·9 and 76·2 years, respectively; 58% female in each series). Clinical diagnostic accuracy for Alzheimer’s disease in our patients is 96%. We also studied 29 African-American patients with dementia of the Alzheimer type and 50 age-matched controls (cases vs controls, 77·2 vs 72·0 years; 72 vs 77% female). We used PCR to test for an association between Alzheimer’s disease and a polymorphism within the intron 3’ to exon 8 of the PS-1 gene. The ApoE genotype of most of the cases and controls was known from previous
investigations. Findings Homozygosity of the 1 allele in the PS-1 gene
was
associated with a doubling of the risk for late-onset Alzheimer’s disease compared with the [12]/[22] genotype (odds ratio 1·97, 95% Cl 1·29-3·00). The proportion of Alzheimer’s disease cases in the white population that could be attributed to homozygosity at this locus, as estimated by the attributable fraction, was 0·22. This compares with 0·35 for a single copy of ApoE4 and 0·15 for two copies. The smaller African-American series showed similar distribution of PS-1 genotype between cases and controls.
Interpretation In our white series of cases, PS-1 accounted for about half as much of the risk for late-onset Alzheimer’s disease as did ApoE4.
*Listed at end of report
Department of Psychiatry, Washington University School of Medicine, St Louis, Missouri, (M Wragg MSc, C Talbot BSc); and Department of Psychiatry, Suncoast Alzheimer’s Disease Laboratories and Gerontology Center, University of South Florida, Tampa, Florida, USA (M Hutton PhD)
Correspondence to: Dr Alison Goate, Department of Psychiatry, Washington University School of Medicine, St Louis, MO 63100, USA
Mutations in the presenilin-1 (PS-1) gene on chromosome 14 cause a significant proportion of earlyonset, autosomal dominant Alzheimer’s disease.’ Many mutations that cause disease between the ages of about 30 and 50 years have been described.H Whilst the PS-1 gene is generally thought to be involved only in the rare, familial early-onset form of the disease, allele sharing between affected family members with late-onset disease has been observed.5 This allele sharing was not found with standard maximum-likelihood methods but with the affected-pedigree-member method of genetic analysis. These results suggest that the chromosome 14 locus was not behaving as an autosomal dominant in late-onset disease.5 Genetic analysis of late-onset Alzheimer’s disease has implicated two loci: the apolipoprotein E (ApoE) gene and the al-antichymotrypsin (AACT) gene. As much as 40-50% of the risk for late-onset disease is attributable to alleles at the ApoE locust ApoE4 is thought to increase risk of Alzheimer’s disease in a dose-dependent manner whilst the ApoE e2 allele may be protective.7-8 It has been reported that homozygosity at a polymorphism within the AACT gene modified the ApoE risk,9 although we could not repeat this observation. 10 We have now tested a polymorphism in the PS-1 gene for association with lateonset disease.
Patients, controls, Screening protocol
for
and methods
polymorphism
extracted from blood with standard procedures. During DNA sequence analysis in cases of early-onset Alzheimer’s disease, we and others have identified a common polymorphism within the intron 3’ to exon 8.3 The most common allele has an A at nucleotide 16 (allele 1) in this intron whilst the variant allele has a C at this position (allele 2). The polymorphism does not create or destroy any restriction enzyme sites. A mismatch primer was designed to contain two mismatched basepairs four and five basepairs from the 3’ end of the primer and five and six basepairs from the polymorphism. When incorporated into a PCR product this primer produces a BamHI cut-site when a C is present and no cut-site when an A is present at the polymorphic site. This enables the two alleles to be distinguished by digestion of a PCR product with BamHI. Forward primer’ was 951 5’ CACCCATTTACAAGTTTAGC 3’, reverse (mismatched) primer was 5’ CACTGATTACTAATTCAGGATC 3’. Use of these primers gives a PCR product of 199 basepairs which is cleaved by BamHI to produce fragments of 181 and 18 basepairs. 50-100 ng DNA was used as template in 25 p.L reactions. The reaction mix consisted of (final concentrations): 0-2 mmol/L dNTPs, 30 pmol/L each primer, 1XTNK50 buffer, and 0-5 U Taq DNA polymerase. PCR was done at: 94°C for 5 min (94°C 30 s, 45°C 30 s, 72°C 30 s) for 35 cycles, and 72°C for 3 min. 5 U BamHI was added and DNA
was
509
First figure=number of cases with genotype or, for alleles, number of alleles. Figure in parentheses number of cases screened in each category. Cases vs controls, X2, *p=0.007. tap=0.0017 (odds ratio 1.97, 95% CI 129-300), and tp=0’006.
Table 1: PS-1
carried out at 37°C for 3 h. The digested product was 3% agarose gel, which is sufficient to separate the digested product so that 200 and 182 basepair bands can be distinguished. Multiple samples were rechecked by sequencing to on
a
ensure our
allele-scoring system
was accurate.
genotyped as described." The coding region of PS-1 is encoded by exons 3-12. Each exon was amplified and sequenced as described. ApoE
was
12
Patients and controls Four different series of patients and controls were analysed. First, white cases and controls were recruited from the Greater St Louis metropolitan community through the Washington University Alzheimer’s Disease Research Center. Second, a similar, smaller population was collected from the Tampa metropolitan area through the Suncoast Gerontology Center. Third, the African-American case-control sample was recruited at St Louis as above. Recruitment for these three series was not based on a family history of Alzheimer’s disease; however, about 40-50% of participants have a first-degree relative with a diagnosis of dementia of the Alzheimer type (DAT). Diagnostic criteria for DAT are equivalent to or more stringent than those for probable and possible Alzheimer’s disease.13,14 Clinical diagnostic accuracy for histological Alzheimer’s disease in our patients is 96%." Controls for each group met all exclusion criteria for DAT but were not demented. Blood samples were collected from 208 white DAT patients (mean age 76-9 years, SD 8-8, 58% female) and 185 age-matched controls (76-2, 11-0, 58%). The mean age of onset of dementia in the DAT patients was 71-3 (8-9). Blood samples were collected from 29 African-American DAT patients (77-2, 9-6, 72% female) and 50 age-matched controls (72-0, 7-8, 77% female). The mean age of onset of dementia in the DAT patients was 71-8 (5-2). The fourth group was collected in the UK as part of a study on the genetics of Alzheimer’s disease." The 32 unrelated patients are white and come from families multiply affected by late-onset disease (mean age 81-1, SD 7-3, 78% female). The mean age of onset of dementia was 73-4 (6-6).
Statistical methods
Genotypic and allelic distributions were analysed with X2 tests and, for 2X2 contingency tables, with odds ratios. The effect of
Figure: Agarose gel showing PS-1 polymorphism genotypes Gel shows band pattern for PS-1 polymorphism m intron 3’ of exon 8. TT corresponds to 11 homozygotes and GG corresponds to 22 homozygotes. TG represents the 12 heterozygote bp=basepairs.
510
frequency.
Addition of numbers for 11, 12, and 22
genotypes gives total
genotype and allele distributions
digestion run
is
examined with multiple linear regression. Logistic regression analyses were used to examine the effect of gender, ApoE4 genotype, PS-1 genotype, and interactions between these variables to predict disease status. Two-tailed tests of significance were used throughout. All analyses were done on SPSSv6 for Windows. Attributable fraction (AF) was calculated from AF=f(OR-1)/[l+f(OR-1)], where f is the frequency of the risk factor in the population and OR is the odds ratio for developing Alzheimer’s disease in individuals with the risk factor compared with those without the risk factor.
genotype
on
age
at onset was
Results Because there were no significant differences in genotype or allele distributions for the PS-1 polymorphism between white controls and patients from St Louis and Tampa, we pooled data for these control and for these patients’ populations (table 1). The genotypic distribution in the controls was close to that expected under HardyWeinberg equilibrium (p=0-635). There were, however, significant differences in genotype (p=0’007) and allele (p=0’006) distributions between controls and patients in the white population. The effect was due to a higher proportion of DAT patients, who were homozygous for the absence of the artificial BamHI site (ie, individuals who were homozygous for allele 1 in figure 1). With the [22] genotype as a reference, the [12] genotype was not associated with increased risk of Alzheimer’s disease (odds ratio 0-98, 95% CI 0-57-1-68), whereas the [11] genotype was associated with an approximate doubling in risk (1-96, 1-11-3-48). In the analyses that follow, we combined the [12] and [22] genotypes into a single category. Analysis of probands from the UK families with latesimilar genotype onset Alzheimer’s disease gave to the American series (table 1), suggesting frequencies that the same effect might be operating, although the lack of an adequate control group for the UK cases precluded formal
analysis. ApoE genotypes of most of these groups have previously been determined and showed the expected increase in ApoE4 allele frequency and decrease in age of onset of Alzheimer’s disease associated with ApoE4 dosage.s,11 When the sample was split on the basis of ApoE genotype, the PS-1 [11] genotype was associated with a doubling of risk of Alzheimer’s disease in carriers of zero or one ApoE4 allele, but not in ApoE4 homozygotes (table 2). The small cell sizes for ApoE c4 homozygotes makes it difficult to interpret the significance of this observation. However, for the common ApoE4 genotypes, there was no interaction between ApoE and PS-1 genotypes. The association between the PS-1 [11] genotype and Alzheimer’s disease was present and of similar magnitude The
population stratification confounded
our
analysis because:
the genotype distributions in cases and controls were similar at each site (pO-11) and these distributions are similar to those in an independent British sample (MJ Owen, Department of Psychological Medicine, University of Wales College of Medicine, Cardiff, UK); the controls were close to Hardy-Weinberg equilibrium; and independent analysis of the results from the two centres gave similar effects (St Louis, odds ratio [95% CI] 1-82 These data are derived from table 1, except that available for 1 control and 4 cases.
ApoE genotype results were
Table 2: Distribution of controls and
cases
by PS-1.
or
not
ApoE4
genotypes
aged 65 years or less (odds ratio 184, 95% CI 0-97-3-50) and for cases with onset over the age of 65 years (2-10, 1-32-3-33).Further, there was no evidence, by linear regression analyses (including or excluding ApoE4 genotype as a variable) that PS-1 genotype for
cases
of onset of dementia in our series. We used logistic regression to model our data. Consistent with the results presented above, ApoE4 genotype (p=0’000003) and presence or absence of PS-1 [ll] genotype (p=0°0027) were significant predictors of disease status. There was no significant interaction When we between ApoE and PS-1 genotypes (p>0-86). did logistic regression analyses separately for men and women, the risk-increasing effect of the PS-1 [11] genotype was greater in women (odds ratio 2-24, 95% CI 1-21-4-15) than in men (1-71, 0-86-3-14). However, this difference was not as statistically significant, demonstrated by the absence of significant interaction between PS-1 genotype and sex (p=058) in the regression model on the total sample. A measure of the overall contribution of a risk factor to disease in a population is provided by the attributable fraction, which refers to the excess fraction of disease in a population that would not have occurred if the risk factor had been absent. For the PS-1 [11]genotype in our white series the attributable fraction was 0-22. This compares with 0-35 for possession of a single copy of ApoE4 and 0’15 for two copies of ApoE4. Thus, in our series, PS-1 accounted for about half as much of the morbidity as did
predicted age
ApoE4. Sequencing of the entire open-reading frame of the PS-1 gene in ten late-onset cases of Alzheimer’s disease who were homozygous for the 1 allele of the PS-1 gene and the 3 allele of the ApoE gene failed to reveal any differences from the previously published sequence. Analysis of the small sample of African-American cases and controls revealed that the frequency of this polymorphism was different in the African-American population compared with the whites, but that the frequencies were similar between Alzheimer’s disease cases and controls (table 1).
Discussion We found that
homozygosity for allele
1 of the PS-1 gene associated with increased risk for polymorphism development of Alzheimer’s disease in the white population. Because our finding was statistically significant at p<0002, a type 1 error is unlikely. Although tight ethnic matching of controls in North American populations can be difficult, we do not believe that was
[1,13-2’92]; Tampa,
2-70
[1-08-6-77]).
There are three possible explanations for the association we found. First, the polymorphism is biologically relevant to the disease. Second, the polymorphism is in linkage disequilibrium with biologically relevant variability elsewhere in the PS-1 gene. Third, the polymorphism is in disequilibrium with genetic variability in another adjacent gene. The third
explanation seems unlikely. The position of the polymorphism
in the intron between exons 8 and 9 suggests that if the polymorphism is biologically relevant, the most likely mode of action would be through the modulation of alternate splicing, particularly of exons 8 and 9 of the PS-1 gene. Exon 8 is alternately spliced in some tissues1s and a mutation in the acceptor site in the same intron led to early-onset disease in a UK family through the loss of exon 9.16 In addition, exon 8 is the site of the most prominent cluster of mutations leading to early-onset disease.1,2 Thus, a subtle alteration of alternate splicing caused by this genetic variability remains a possibility. However, if the polymorphism were a relevant cause of the disease, the association with disease would occur in all populations. Our data from the African-American series indicate that this is not the case. For this reason, we favour the explanation that this polymorphism is in linkage disequilibrium with relevant variability elsewhere in the PS-1 gene. Sequence analysis of the open-reading frame of ten cases with a [11] genotype failed to reveal coding changes. Thus, any relevant variability in the PS-1 gene must be in the promoter or other non-coding regions. Our data suggest several experiments. The strength of the association needs to be tested in other ethnic groups. The effect of the polymorphism on alternate splicing of the PS-1 gene should be examined. Other genetic variability within the .P.S*-7 gene should be sought, particularly in recognised promoter and control elements, and then this variability tested for association with disease, both in case-control series and in pedigree datasets. PS-1 has been found in plaques from "typical" Alzheimer’s disease cases.17 That observation and our data are consistent with the view that all cases of Alzheimer’s disease share common pathogenic mechanisms and that PS-1 is part of this biochemical pathway. Other members of the Alzheimer’s Disease Collaborative Group are: Frances Busfield, Sang Woo Han, Corinne Lendon, Robert F Clark, John C Morris, Dorothy Edwards, and Alison Goate (Departments of Psychiatry, Neurology and Occupational Therapy, Washington University School of Medicine, St Louis, Missouri); Eric Pfeiffer, Richard Crook, Guy Prihar, Helen Phillips, Matt Baker, and John Hardy (Departments of Psychiatry, Pharmacology, Biochemistry, and Neurology, Suncoast Alzheimer’s Disease Laboratories and Gerontology Center, University of South Florida, Tampa, Florida, USA); Martin Rossor and Henry Houlden (Department of Neurology, Imperial College of Science, Technology & Medicine, London W2); Eric Karran and Gareth Roberts (Department of Moleuclar Neuropathology, SmithKline Beecham, Harlow, Essex); and Nick Craddock (Department of Psychological Medicine, University of Wales College of Medicine, Cardiff, UK).
511
Work in AG’s laboratory was supported by a career development award from the NIH (AG00634), a Zenith Award from the Alzheimer’s Association, the Metropolitan Life Foundation, and the McDonnell Center for Cellular and Molecular Neurobiology. Work in JH’s laboratory was supported by NIH project grants (AG11871 and AG12028), the Metropolitan Life Foundation, and the American Health Assistance Foundation. SmithKline Beecham Advanced Technologies in Genetics provided support to JH and AG. NC is a Wellcome Trust Senior Clinical Research Fellow. The David and Frederick Barclay Foundation, the MRC and the Alzheimer’s Disease Society supported MR and the collection of British families has been supported by MRC grants to MR, JH, and AG. Collection of the St Louis samples was supported by NIA grants to JCM (AG03991 and AG05681) and pilot grants from the Alzheimer’s Disease and Related Disorders Program, University of Missouri Alzheimer’s Research Center, and the Alzheimer’s Association to CL. We thank the physicians and staff of the clinical core of the Alzheimer’s Disease Research Center at Washington University for subjects’ evaluation. Private donations to all academic centres are also acknowledged. We thank the families for their continuous enthusiasm for this project.
2
3 4
6
Schellenberg G, Payami H, Wijsman E, et al. Chromosome 14 and late onset familial Alzheimer’s disease. Am J Hum Genet 1993; 53: 619-28. Corder E, Saunders A, Strittmatter W,
512
7
8
1432-33. Kamboh MI, Narambir D, Sanghera K, Ferrell R, DeKosky S. APOE4-associated Alzheimer’s disease risk is modified by &agr;1-antichymotrypsin polymorphism. Nature Genet 1995; 10: 486-88. 10 Talbot CJ, Houlden H, Craddock N, et al. Polymorphism in AACT gene may lower age of onset of Alzheimer’s disease. NeuroReport (in
9
press). et al. Confirmation that the E4 allele is associated with late onset familial Alzheimer’s disease. Neurodegeneration 1993; 2: 283-86. 12 Hutton M, Busfield F, Wragg M, et al. Complete analysis of the presenilin 1 gene in early onset Alzheimer’s disease. NeuroReport
11 Houlden
H, Crook R, Duff K,
apolipoprotein
(in
13 McKhann
Sherrington R, Rogaev E, Liang Y, et al. Cloning of a gene bearing mis-sense mutations in early-onset familial Alzheimer’s disease. Nature 1995; 375: 754-60. Clark RF, Hutton M, Fuldner R, et al. The structure of the chromosome 14 Alzheimer’s disease gene, presenilin 1 (PS-1) and identification of six novel mutations in early onset AD families. Nature Genet 1995; 11: 219-22. Wasco W, Pettingell W, Jondro P, et al. Familial Alzheimer’s chromosome 14 mutations. Nature Med 1995; 1: 848. Cruts M, Backhovens H, Wang S-Y, et al. Molecular genetic analysis of familial early onset Alzheimer’s disease linked to chromosome 14q24.3. Hum Mol Genet 1995; 4: 2363-72.
5
onset
press).
References 1
E type 4 allele and the risk of Alzheimer’s disease in late families. Science 1993; 261: 921-23. Corder E, Saunders A, Risch N, et al. Protective effect of apolipoprotein E type 2 allele for late onset Alzheimer disease. Nature Genet 1994; 7: 180-83. Talbot C, Lendon C, Craddock N, Shears S, Morris JC, Goate A. Protection against Alzheimer’s disease with apoE ∈2. Lancet 1994; 343:
apolipoprotein
et
al. Gene dose of
G, Drachman D, Folstein M, Katzman R, Price D, Stadlan EM. Clinical diagnosis of Alzheimer’s disease: report of the NINCDS-ADRDA Work Group under the auspices of Department of Health and Human Services Task Force on Alzheimer’s Disease.
14 15
Neurology 1984; 34: 939-44. Berg L, Morris JC. Alzheimer’s disease. In: Terry RD, Katzman R, Bick K, eds. Alzheimer’s disease. New York: Raven Press, 1994: 9-21. Rogaev EI, Sherrington R, Rogaeva E, et al. Familial Alzheimer’s
disease in kindreds with missense mutations in a gene on chromosome 1 related to the Alzheimer’s disease type 3 gene. Nature 1995; 376: 775-78. 16 Perez-Tur J, Froelich S, Prihar G, et al. A mutation in Alzheimer’s disease destroying a splice acceptor site in the presenilin-1 gene.
NeuroReport (in press). 17 Wisniewski T, Palha JA, Ghiso J, Frangione B. S182 protein in Alzheimer’s disease neuritic plaques. Lancet 1995; 346: 1366.