Abstract
Background: Mutations in the presenilin-1 gene (PS1) account for a majority of patients with early-onset familial AD. However, the clinical indications and algorithms for genetic testing in dementia are still evolving.
Methods: The entire open reading frame of the PS1 gene was sequenced in a series of 414 consecutive patients referred for diagnostic testing, including 372 patients with AD and 42 asymptomatic persons with a strong family history of AD.
Results: Forty-eight independent patients screened had a PS1 mutation including 21 novel mutations. In addition, 3% of subjects (11/413) had a known polymorphism, the Glu318Gly substitution. The majority of the mutations were missense substitutions but there were three insertions and Δexon 10 mutation. With six exceptions (codons 35, 178, 352, 354, 358, and 365) most of the mutations occurred at residues conserved in the homologous PS2 gene or in PS1 of other species.
Conclusions: Eleven percent of a referral-based series of patients with AD can be explained by coding sequence mutations in the PS1 gene. The high frequency of PS1 mutations in this study indicates that screening for PS1 mutations in AD is likely to be successful, especially when directed at patients with a positive family history with onset before 60 years (90% of those with PS1 mutations were affected by age 60 years). This will also have significance for the secondary identification of at-risk relatives who might be candidates for future prophylactic therapies for AD.
In approximately 5% of patients with AD, the disease appears to be transmitted as an autosomal dominant mendelian trait with age-dependent penetrance.1 Mutations in the presenilin-1 gene (PS1)2 account for a majority of cases of early-onset familial AD (18 to 50%).3 More than 70 different mutations, mainly missense substitutions, have been reported in families of different ethnic origins.4 Most PS1 mutations are associated with classic presenile AD except the Glu318Gly substitution, which is a polymorphism in a nonconserved region of the gene unrelated to AD.5,6⇓
PS1 has also been suggested as a potential risk factor for late-onset AD based on an association between AD and noncoding polymorphisms within intron 97 or polymorphisms in the 5′ promoter element (-2154G→A and –2823I→D).8 However, these results have not yet been widely replicated and their functional significance is unclear.9,10⇓ Because these noncoding polymorphisms are not suitable for predictive or diagnostic testing, the current study was limited to searching for coding sequence variants.
We report the experience of mutation screening in a series of consecutive patients with AD referred for diagnostic testing at the University of Toronto (n = 21) or at Athena Diagnostics (n = 393). Although this is not a rigorously designed population-based study, it does reflect the experience of clinicians with cases in which there is a high index of suspicion for a diagnosis of familial AD.
Methods.
Subjects.
A total of 414 subjects from the United States, Germany, and Canada participated in the study (54% women and 46% men). A majority of the individuals tested had a clinical diagnosis of AD, but 10% of the patients (42 individuals) were either asymptomatic or had insufficient clinical features to support a definitive diagnosis. The exact information regarding the age at onset was not available for all subjects in this study. However, based on the age at the time of testing, at least 78% of the patients had presenile forms of AD (290 patients were <65 years of age). The mean age of patients at the time of testing was 58 ± 11 years (range, 27 to 92 years).
Sequencing.
Total RNA and genomic DNA were isolated from blood leukocytes.2 The entire open reading frame of the PS1 gene was recovered by reverse transcriptase PCR and then sequenced using four forward and four reverse primers for cycle sequencing.2 Exon 4 and in some cases exon 9 (exon numbering according to Rogaev et al.11), which undergo alternative splicing, were isolated from genomic DNA by PCR and were then sequenced separately. Mutations were detected by direct inspection of the fluorescent chromatographs and by analysis using the Factura and Sequence Navigator software (version 2.0) (Applied Biosystems, Foster City, CA).2
Mutation analysis.
Where there were sufficient informative family members, the segregation of novel mutations was assessed using assays based on PCR from genomic DNA to expedite screening for mutations in the relevant exons. For the Leu235Pro, Gln222Arg, and Ile213Leu mutations, the 144-basepair PCR product was amplified with primers 974 (5′-TGTGGTGGGAATGA-3′) and 892 (5′-TGAAATCACAGCCAAGATGAG-3′). A reaction volume of 20 μL containing 100 ng of DNA, 20 pmol of each primer, 2 μL of 10 × PCR reaction buffer (Qiagen, Missisauga, Canada), 1 U Taq polymerase, 250 mkM dNTPs, and 1 μCi α-32P-deoxycytidine triphosphate was thermocycled for 35 cycles of 94° for 30 seconds, 57° for 20 seconds, and 72° for 20 seconds. The Gln222Arg PCR products were digested with 1 U MspI for 1 hour at 37°, and the resulting restriction fragments (wild-type = 144 basepair; mutation = 96 basepair, 48 basepair) were resolved on a 6% nondenaturing polyacrylamide gel and visualized by autoradiography. The Ile213Leu PCR product was digested with PflmI (wild-type = 124 basepair and 22 basepair; mutation = 144 basepair). The Leu235Pro PCR product was analyzed by direct sequencing approach. The Met146Leu was assessed as previously described.2
Statistical analysis.
To determine whether there were particular age groups of AD-affected patients that have a higher frequency of PS1 mutations, we used Kaplan-Meier survival analysis12 and receiver operating characteristic curves13 to evaluate the probability of finding PS1 mutations by age. These analyses were restricted to symptomatic patients with AD with known age (n = 362). We used the presence or absence of any PS1 mutation as the “outcome measure” (all subjects had AD) and plotted the age at time of testing as the “time-to-event” variable (these statistics should not be confused with the sensitivity and specificity of our assays for detecting PS1 mutations per se, which are 100% accurate).
Results.
During the 2-year interval between May 1997 and January 2000, 414 patients were referred for diagnostic screening. Forty-eight independent cases involving a PS1 mutation were found, including 36 unique mutations, of which 21 have not been previously described (table). These novel mutations were not found in 126 normal chromosomes from unrelated subjects. Glu206Ala, Ala431Glu, and Ile143Thr were the most frequent mutations in our data set. Together they cover 28% of all independent mutations.
Mutations in the open reading frame of the PS1 gene found during screening of 414 consecutive referrals for genetic AD
Evidence for cosegregation with AD was available for only four families in which there were multiple living affected members (Ile213Leu; Gln222Arg; Leu235Pro and Met146Leu mutations). The Glu318Gly polymorphism was detected in 11 independent cases (3%), a frequency that is similar to the frequency previously published for normal populations.6 This result is therefore consistent with prior suggestions that Glu318Gly is a normal variant in a nonconserved residue.
With the exception of a Δexon 10 splicing mutation and three in-frame insertions, the mutations were missense substitutions. The insertion of TTATAT is predicted to result in a missense substitution at codon 156 (Tyr156Phe) plus the insertion of an Ile and Tyr immediately after codon 156. Codon 156 is in a highly conserved region of the gene, indicating that this mutation is likely to have a dramatic effect on the function of PS1 protein. Indeed this insertion is associated with a very early onset of AD (28 years) and spastic paraplegia.14 We have also identified three patients with double mutations in the PS1 gene. In the first case, a Met146Val was coinherited with a Ser365Tyr (whether these mutations are in cis or in trans is currently unknown). Codon 146 is the site of several previously known AD-related mutations and is fully conserved in most animal presenilins, except hop1 and spe4 of Caenorhabditis elegans. Codon 365 has not been previously reported as the site of an AD-related mutation; it is conserved in PS1 of all vertebrates but not in PS2. The two other patients with double PS1 mutations in trans were sisters who had inherited Ile143Thr from their father and the Ile439Val mutation from their mother (who was asymptomatic at age 55 years). The Ile143 residue is conserved in most animal presenilins except Sel12, and the Ile439 residue is conserved in most animal and plant presenilins, except hop1 and spe4 of C. elegans. All three patients with double PS1 mutations had a very early age at onset (≤35 years of age); however, insufficient samples are available to determine whether this is a statistically significant deviation from the expected age at onset for single PS1 mutations (many of these patients also have a very early age at disease onset).
As would be expected, patients with AD with positive PS1 findings were significantly younger (46 ± 11 years) than patients with AD with negative PS1 findings (60 ± 11 years) (p < 0.0001). To determine whether age might be used as a pretest predictor of the presence of PS1 mutations, we computed receiver operating characteristic curves (figure 1). If age ≤42 years is employed as a pretest cut-off age for PS1 mutation screening, then only 40% of AD-affected individuals in our series who have PS1 mutations would have been tested, although very few of the non-PS1 mutation patients with AD would be tested (1-specificity = 5.4%). However, if age ≤55 years is used as a pretest cut-off, then 83% of patients with AD who have PS1 mutations would have been tested. This would be accompanied by a modest increase in the number of PS1-normal patients with AD who would have been tested (1-sensitivity = 36.9%). Use of pretest cut-off ages >60 years results in only a small further gain in the proportion of patients with AD with PS1 mutations who would have been identified for testing and is accompanied by testing of increasing numbers of patients with AD without PS1 mutations. Similar results are obtained using survival analysis (figure 2). Among those with PS1 mutations, nearly 90% are affected by age 60 years compared with only 60% of those without a PS1 mutation.
Figure 1. Receiver operating characteristics curve, derived from 10 evenly distributed age cut-off points, which plots the age-based sensitivity against the age-based false-positives rate (1-specificity).
Figure 2. Survival curves, which compare the proportion of patients with AD with a certain age in the groups positively or negatively tested for PS1 mutations. Wild-type = thin line; mutations = thick line.
Discussion.
To date, mutations pathogenic for AD have been found in >22% of the coding region of PS1, affecting all exons except exon 4. The majority of the new mutations reported here involve residues in highly conserved transmembrane domains and are at or near putative membrane interfaces, both of which appear to be favored sites for pathogenic mutations (see the table).3 Three new mutations were also detected in the carboxyterminal region of PS1 (exon 13), including an Ile439Val substitution, which is the most distal carboxyterminal PS1 mutation identified to date.
Despite the fact that evidence of cosegregation was available for only four of the mutations described here, several observations contradict the idea that the other novel mutations are simply rare, innocent polymorphisms. First, all except six of the mutations affect residues that are conserved between the PS1 and PS2 genes and are also conserved in evolution. Second, other AD-linked PS1 mutations have been previously reported at codons that are not conserved in the PS2 gene (e.g., Glu123Lys15 and Leu282Arg16). However, like the nonconserved mutations described here, they are often associated with a relatively late age at disease onset (range, 40 to 62 years). Third, of the six mutations that are not conserved between PS1 and PS2 (Ser178Pro, Thr354Ile, Arg358Gln, Ser365Tyr, and the insertion of Arg after codon 352), are all conserved in PS1 of other vertebrates. Furthermore, codon 178 is affected in two independent AD families (one of which demonstrates cosegregation of Ser178Pro with AD in two siblings) and is in a highly conserved domain of PS1. The Arg35Gln mutation (the most distal N-terminal PS1 substitution identified to date) affects a residue that is not conserved but that is located adjacent to the evolutionarily conserved Glu at codon 34.
Four mutations (at codons 354, 358, 365, and 352) are clustered in a small region of the cytoplasmic loop that demonstrates substantial sequence conservation (in contrast to the nonconserved region of the loop containing the Glu318Gly polymorphism). These mutations are clustered near a functionally important region of the hydrophilic loop that contains both the caspase recognition site of PS1 and the consensus sequence for protein kinases A and C, which phosphorylate PS1.17,18⇓ Although deletion of the loop domain (residues 304 to 371) has been reported to have no effect on the ability of other mutations placed in cis elsewhere in PS1 gene to augment Aβ42 production in vitro,19 this does not mitigate against mutations in this loop causing AD. Indeed, a splicing mutation, which deletes exon 10 from the mRNA transcript, and results in the truncation of the loop domain combined with the missense substitution of Ser290, is clearly associated with AD.20 Cumulatively, these observations argue that these mutations are likely to be pathogenic rather than innocent polymorphisms. Analysis of Aβ42 levels in cultured cells would provide biochemical support for this notion.
We also identified one patient who presented with AD-type dementia and spastic paraplegia. This patient carried an unusual mutation (insertion of TTATAT) in a highly conserved transmembrane domain 3 interface region of the PS1 gene. A similar clinical phenotype, usually accompanied by abundant diffuse Aβ “cotton wool” plaques without congophilic cores and only minor neuritic dystrophy,21 has been described in five families with different PS1 mutations (including the exon 10 splicing mutant).22,23⇓ The mechanism by which this phenotype arises is currently unclear.
The frequency of PS1 mutations in this referral-based series indicates that screening for PS1 mutations in early-onset AD is indeed likely to be successful. Such screening is especially likely to be productive when directed toward persons with a positive family history24 and with age at onset before 60 years. The current absence of proven effective prophylactic treatments for AD currently limits such screening to roles in confirmation of clinical diagnoses and presymptomatic testing for genetic counseling. However, if any of the prevention or treatment strategies currently under investigation (e.g., nonsteroidal anti-inflammatory agents, estrogens, β- or γ- secretase inhibitors, or Aβ immunization25,26⇓ proves to be effective as prophylactics, presymptomatic detection of mutation carriers will become very important. This study suggests that screening for PS1 mutations among appropriately selected cases would then likely be highly cost effective.
- Received February 1, 2001.
- Accepted April 5, 2001.
References
Disputes & Debates: Rapid online correspondence
NOTE: All authors' disclosures must be entered and current in our database before comments can be posted. Enter and update disclosures at http://submit.neurology.org. Exception: replies to comments concerning an article you originally authored do not require updated disclosures.
- Stay timely. Submit only on articles published within the last 8 weeks.
- Do not be redundant. Read any comments already posted on the article prior to submission.
- 200 words maximum.
- 5 references maximum. Reference 1 must be the article on which you are commenting.
- 5 authors maximum. Exception: replies can include all original authors of the article.
- Submitted comments are subject to editing and editor review prior to posting.