Skip Navigation


Human Molecular Genetics Advance Access originally published online on September 22, 2004
Human Molecular Genetics 2004 13(22):2885-2892; doi:10.1093/hmg/ddh299
This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
13/22/2885    most recent
ddh299v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (11)
Right arrowRequest Permissions
Google Scholar
Right arrow Articles by Bray, N. J.
Right arrow Articles by O'Donovan, M. C.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Bray, N. J.
Right arrow Articles by O'Donovan, M. C.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?

Human Molecular Genetics, Vol. 13, No. 22 © Oxford University Press 2004; all rights reserved

Allelic expression of APOE in human brain: effects of epsilon status and promoter haplotypes

Nicholas J. Bray1, Luke Jehu1, Valentina Moskvina2, Joseph D. Buxbaum3, Stella Dracheva3, Vahram Haroutunian3,4, Julie Williams1,2, Paul R. Buckland1, Michael J. Owen1 and Michael C. O'Donovan1,*

1Department of Psychological Medicine and 2Biostatistics Bioinformatics Unit, School of Medicine, Cardiff University, Cardiff, CF14 4XN, UK, 3Department of Psychiatry, Mount Sinai School of Medicine, New York, NY 10021, USA and 4Mental Illness Research, Education and Clinical Centres (MIRECC), Bronx Veterans Affairs Medical Centre, Bronx, New York, NY 10468, USA

Received August 9, 2004; Accepted September 11, 2004


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 RESULTS
 DISCUSSION
 MATERIALS AND METHODS
 REFERENCES
 
The {varepsilon}4 haplotype of APOE is the only undisputed genetic risk factor for late-onset Alzheimer's disease (LOAD). It has been proposed that at least two other polymorphisms in the promoter of the APOE gene (–219G>T and –491A>T) might also contribute to disease susceptibility, and modulate the impact of structural changes in the ApoE protein, by altering its expression. In order to assess the extent of cis-acting influences on APOE expression in human brain, highly quantitative measures of allele discrimination were applied to cortical RNA from individuals heterozygous for the epsilon alleles. A small, but significant, increase in the expression of {varepsilon}4 allele was observed relative to that of the {varepsilon}3 and {varepsilon}2 alleles (P<0.0001). Similar differences were observed in brain tissue from confirmed LOAD subjects, and between cortical regions BA10 (frontopolar) and BA20 (inferior temporal). Stratification of {varepsilon}4/{varepsilon}3 allelic expression ratios according to heterozygosity for the –219G>T promoter polymorphism revealed significantly lower relative expression of haplotypes containing the –219T allele (P=0.02). Our data indicate that, in human brain, most of the cis-acting variance in APOE expression is accounted for by the {varepsilon}4 haplotype, but there are additional, small, cis-acting influences associated with promoter genotype.


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
RESULTS
 DISCUSSION
 MATERIALS AND METHODS
 REFERENCES
 
Apolipoprotein E (ApoE) serves a central role in lipid metabolism and is the major apolipoprotein synthesized in brain (1). The ApoE protein exists as three common isoforms (designated ApoE2, ApoE3 and ApoE4), resulting from two non-synonymous single nucleotide polymorphisms in the APOE gene. These isoforms differentially impact on a variety of biological parameters, including plasma cholesterol levels (2), neurite outgrowth (3) and amyloid deposition (4).

Possession of the {varepsilon}4 allele of the APOE gene (encoding ApoE4) is the only undisputed genetic risk factor for the common, late-onset, form of Alzheimer's disease (LOAD), increasing risk in a dose-dependent manner (5,6). However, the {varepsilon}4 allele is neither necessary nor sufficient for expression of the disease, and a large proportion of {varepsilon}4 homozygotes surviving to 80 years do not show cognitive impairment (7). The {varepsilon}2 allele (encoding ApoE2) confers a protective effect (8).

It has been postulated that, in addition to the structural changes in ApoE, cis-acting variation within APOE regulatory sequence might also contribute to disease susceptibility, by influencing gene expression. Transgenic mouse models indicate that APOE dosage can significantly influence pathological hallmarks of human Alzheimer's disease, including amyloid deposition and neuritic degeneration (9,10). Furthermore, that cis-acting regulatory variants in human APOE have the potential to influence LOAD susceptibilty is suggested by a study in which variation in quantitative measures of peripheral lipid metabolism were better explained by considering genotype data from additional polymorphic sites in the vicinity of APOE, rather than {varepsilon}2–4 status alone (11). It should, however, be noted that the implications of that study for Alzheimer's disease are unknown, as it is unclear if those aspects of APOE function as indexed by peripheral lipid metabolism are relevant to LOAD pathogenesis. Moreover, the extent of cis-acting influences on APOE expression in human brain is also currently unclear.

A powerful method for investigating cis-acting influences on gene expression involves comparing the relative level of each mRNA transcript in individuals who are heterozygous for an expressed polymorphism. This ‘within-subjects’ approach can allow detection of genuine cis-acting effects, whilst controlling for the trans-acting factors than can confound measures of total expression between samples. One such study, using an RFLP-based assay, reported evidence for intrinsic effects of epsilon status on APOE mRNA expression in brain, with {varepsilon}3 expression being greatly increased relative to {varepsilon}4 in {varepsilon}3{varepsilon}4 heterozygotes (12). Elevated {varepsilon}3 expression was observed in both LOAD cases and controls, although, in the former, the magnitude of the difference was smaller. However, these findings were not replicated in a subsequent study, in which the potential methodological shortcomings of RFLP-based assays were clearly demonstrated (13).

Molecular genetic studies provide some support for association between LOAD and two polymorphisms in the promoter of APOE, denoted –491A>T and –219G>T (also known as Th1/E47cs) (14–16). Both polymorphisms have been shown to influence APOE expression in vitro (17), raising the possibility that the association is a direct result of cis-acting effects on gene expression, rather than by virtue of linkage disequilibrium (LD).

Given the importance of APOE as a risk factor for LOAD, and the potential importance of variation in APOE expression, we have undertaken a thorough investigation of the allelic expression of APOE in human brain using highly quantitative measures of allele discrimination (18,19) in a large series of heterozygous samples. Our objectives were to establish whether epsilon status influences APOE expression in brain, and to determine the extent to which APOE expression is subject to other cis-acting influences, in particular the effects of promoter polymorphisms of current interest in Alzheimer's disease research.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
DISCUSSION
 MATERIALS AND METHODS
 REFERENCES
 
We analysed three sets of samples comprising a total of 202 brain samples from 142 unrelated individuals. Of an initial set of 60 non-AD samples, 15 were {varepsilon}3{varepsilon}4 heterozygotes, 11 were {varepsilon}2{varepsilon}3 heterozygotes and two were {varepsilon}2{varepsilon}4 heterozygotes. Eleven {varepsilon}3{varepsilon}4 heterozygotes were identified in the second sample of 22 LOAD subjects, with no {varepsilon}2 alleles observed. Thirteen of the third set of 60 samples with a mixture of psychiatric diagnosis used for within-subject comparisons between brain regions were {varepsilon}3{varepsilon}4 heterozygotes and four were {varepsilon}2{varepsilon}3 heterozygotes. Observed allele frequencies for the three groups are shown in Table 1.


View this table:
[in this window]
[in a new window]
  		Table 1. Observed allele frequencies for genotyped polymorphisms in sample groups
 
Measures of APOE allele-specific expression were performed using expressed SNPs 334T>C and 472C>T. The {varepsilon}3 allele is determined by the haplotype 334T–472C, the {varepsilon}2 allele by 334T–472T and the {varepsilon}4 allele by 334C–472C. The use of polymorphisms within mRNA sequence permits discrimination between alleles transcribed from each chromosome.

Corrected cDNA ratios for all {varepsilon}3{varepsilon}4 and {varepsilon}2{varepsilon}3 heterozygotes are shown in Table 2. Analysis of corrected cDNA ratios from the initial sample of 15 {varepsilon}3{varepsilon}4 heterozygotes showed a small, but significant, elevation of the C-allele at SNP 334T>C ({varepsilon}4) relative to the T-allele ({varepsilon}3), in each of three replication experiments. Repeat assays showed good reproducibility of individual cDNA allele ratios, with an average co-efficient of variation (SD/mean) of 0.04. Combining data from all 15 individuals across all experiments, expression of the {varepsilon}4 allele was increased in cDNA by an average of 14% relative to the {varepsilon}3 allele (mean corrected ratio=1.14, P<0.0001), with individual ratios ranging from a 6 to a 25% relative increase (Fig. 1).


View this table:
[in this window]
[in a new window]
  		Table 2. Corrected cDNA ratios in {varepsilon}3{varepsilon}4 and {varepsilon}2{varepsilon}3 heterozygotes
 


View larger version (14K):
[in this window]
[in a new window]
  		Figure 1. Comparison between corrected genomic and cDNA ratios assayed at SNP 334T>C in {varepsilon}3{varepsilon}4 subjects. For all individuals, the C-allele determines {varepsilon}4, and the T-allele determines {varepsilon}3. The ratio is therefore expressed as the level of {varepsilon}4/{varepsilon}3. Column 2 shows cDNA data points from non-AD cases (n=15), and column 3 shows cDNA data points from confirmed LOAD cases (n=11). The data for genomic DNA are the averages of two measurements for each individual sample. The data for cDNA are the averages of six estimates for each individual sample. For both non-AD and LOAD cases, the level of {varepsilon}4 is significantly greater than {varepsilon}3 in cDNA (P<0.0001). No differences are observed in the extent of allelic distortion between non-AD and LOAD cases (P=0.76).

 
Expression ratios derived from the 11 {varepsilon}3{varepsilon}4 heterozygotes from the LOAD sample yielded a similar pattern, with an average 15% relative increase in {varepsilon}4 compared with {varepsilon}3 expression (P<0.0001). There was again good reproducibility of individual cDNA ratios across assays (SD/mean=0.04), with averaged increases ranging from 9 to 19% (Fig. 1). No significant differences were observed between ratios derived from the initial sample and those from LOAD cases (P=0.76).

Ratios obtained for the additional 13 {varepsilon}3{varepsilon}4 heterozygotes for which tissue from both BA10 and BA20 were available again showed a significant increase in the expression of {varepsilon}4 allele (BA10 mean increase=11%, P<0.0001, BA20 mean increase=10%, P<0.0001). Within-subject comparisons between the two brain regions indicated no significant differences in the extent of allelic imbalance (paired t-test, P=0.28).

A smaller difference was observed in the relative expression of {varepsilon}2 and {varepsilon}3 alleles (SNP 472C>T) in the initial sample of 11 {varepsilon}2{varepsilon}3 heterozygotes (Fig. 2). The average increase in the 472C ({varepsilon}3) allele over the 472T ({varepsilon}2) allele across experiments was only 4% (ratio=1.04), ranging from equal expression (ratio=1) to a 10% relative increase in the {varepsilon}3 allele (ratio=1.10). Although this just meets conventional criteria for significance (P=0.04), an increase in relative {varepsilon}3 expression was not observed in the four {varepsilon}2{varepsilon}3 heterozygotes from the USA sample, in tissue derived from either BA10 or BA20 (mean ratios=1.00 and 0.99, respectively). Combining data from all 15 samples of frontal cortex yielded a mean {varepsilon}3/{varepsilon}2 ratio of 1.03 (P=0.06).



View larger version (12K):
[in this window]
[in a new window]
  		Figure 2. Comparison between corrected genomic and cDNA ratios assayed at SNP 472C>T in (non-AD) {varepsilon}2{varepsilon}3 subjects (n=11). For all individuals, the C-allele determines {varepsilon}3 and the T-allele determines {varepsilon}2. The ratio is therefore expressed as the level of {varepsilon}3/{varepsilon}2. The data for genomic DNA are the averages of two measurements for each individual sample. The data for cDNA are the averages of six estimates for each individual sample. The level of the {varepsilon}3 allele is slightly higher than {varepsilon}2 in cDNA (P=0.04).

 
Two of the initial 60 samples were {varepsilon}2{varepsilon}4 heterozygotes, and therefore heterozygous at both SNPs 334T>C and 472C>T. This allowed comparison between allele ratios derived from the two separate assays, which in the absence of alternative splicing, should be comparable, despite being amplified and assayed using different primer sets. The cDNA ratios, corrected by the average genomic for each assay, were 1.14 and 1.10 for one sample, and 1.12 and 1.08 for the other sample, at SNPs 334T>C and 472C>T, respectively. Thus, in both cases, calculated ratios differed by only 4% between the two independent assays. This sample is too small for statistical analysis, but, as expected from our observations that the relative expression of {varepsilon}4>{varepsilon}3, while {varepsilon}3≥{varepsilon}2, the {varepsilon}4 allele was expressed at a higher level than {varepsilon}2 in each case.

To investigate the influence of known promoter polymorphisms on allelic expression, 60 unrelated Caucasians and 22 Caucasian LOAD cases were genotyped for promoter SNPs –491A>T and –219G>T (observed allele frequencies are presented in Table 1). Although other variants of potential regulatory significance exist (e.g. –427T>C, +113G>C), these particular polymorphisms were selected on the basis of having the strongest prior evidence for association with LOAD (16). The sample used for within-subject comparisons between brain regions were not included in this analysis as it is ethnically more heterogeneous. The use of samples with different genetic backgrounds is potentially problematic for calculating the conditional diplotype probabilities as the diplotypes are less likely to follow Hardy–Weinberg expectations.

To determine if any of the variance in relative {varepsilon}4 expression is attributable to heterozygosity at the promoter sites, the corrected cDNA ratios obtained from analysis of {varepsilon}3{varepsilon}4 heterozygotes at SNP334T>C were stratified according to whether individuals were heterozygous or homozygous for the two promoter loci.

Ratios derived from the 13–219G>T heterozygotes and 13–219G>T homozygotes were compared by using two-factorial ANOVA, where LOAD diagnosis was included as a factor. Of the 13 –219G>T homozygotes, nine were homozygous for the –219T allele and four were homozygous for the –219G allele. The average increase of the {varepsilon}4 allele, relative to {varepsilon}3, was 12% in –219G>T heterozygotes (range 6–23%) compared with 16% in –219G>T homozygotes (range 9–25%) (Fig. 3). A significant main effect of heterozygosity versus homozygosity at the –219G>T polymorphism was observed (P=0.02), with no significant effect of LOAD diagnosis (P=0.43). Knowledge of phase is required to allow inference about the direction of effect of the 219G and T alleles. Diplotypes are unambiguous with respect to individuals who are homozygous at the promoter polymorphism, but not for those who are heterozygous at both this and the assayed site. However, given the observed haplotype frequencies, for –219G>T/334T>C double heterozygotes, the probability that the 219G nucleotide is in phase with 334T ({varepsilon}3) and the 219T nucleotide is in phase with 334C ({varepsilon}4) is 0.85. As the –219G>T heterozygotes showed a reduced relative increase in {varepsilon}4 expression compared with –219G>T homozygotes, and because in the case of the heterozygotes the –219T allele is predicted to be in phase with the {varepsilon}4 allele in 85% of cases, we can infer that the –219T allele is associated with reduced APOE expression compared with the –219G allele.



View larger version (13K):
[in this window]
[in a new window]
  		Figure 3. Comparison between corrected {varepsilon}4/{varepsilon}3 cDNA ratios from –219G>T homozygotes and –219G>T heterozygotes, as assayed at SNP 334T>C. Ratios derived from {varepsilon}3{varepsilon}4 heterozygotes that are also heterozygous for SNP –219G>T show lower ratios of {varepsilon}4/{varepsilon}3 expression than do {varepsilon}3{varepsilon}4 heterozygotes that are homozygous for SNP –219G>T (P=0.02).

 
Of the 26 {varepsilon}3{varepsilon}4 heterozygotes included in the analysis, six were heterozygous for the –491A>T polymorphism, one was homozygous for the –491T allele and 19 were homozygous for the –491A allele. The average increase of the {varepsilon}4 allele relative to {varepsilon}3 was 16% in –491A>T heterozygotes (range 9–25%) compared with 14% in –491A>T homozygotes (range 6–23%) (Fig. 4). Differences in allele ratio did not approach statistical significance (P=0.57). Again, there was no significant effect of LOAD diagnosis (P=0.73). However, given individual genotypes at SNPs 334T>C and –219G>T, for four of the six –491A>T heterozygotes, the probability that the –491T nucleotide is in phase with the 334T ({varepsilon}3) allele and the –491A nucleotide is in phase with the 334C ({varepsilon}4) allele is only 0.61. As the phase probability is not much greater than chance, it is likely that phase will be reversed in a proportion of cases, distorting any directional effect on {varepsilon}4 expression. The –491A>T heterozygote with the greatest phase certainty (P=0.99) was predicted to possess the –491A nucleotide in phase with –219G and 334C ({varepsilon}4) and the –491T nucleotide in phase with –219G and 334T ({varepsilon}3). This sample, the only individual that was homozygous for the ‘high expression’ 219G allele, whereas heterozygous for the –491A>T polymorphism, showed the greatest elevation in relative 334C ({varepsilon}4) expression of all assayed samples (25%).



View larger version (13K):
[in this window]
[in a new window]
  		Figure 4. Comparison between corrected {varepsilon}4/{varepsilon}3 cDNA ratios from –491A>T homozygotes and –491A>T heterozygotes, as assayed at SNP 334T>C. No significant differences were observed between the two groups (P=0.57).

 

   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
 DISCUSSION
MATERIALS AND METHODS
 REFERENCES
 
Using a highly quantitative method of allele discrimination (18,19), we have compared the relative expression of the APOE {varepsilon}2, {varepsilon}3 and {varepsilon}4 alleles in brain RNA from individual heterozygotes. By comparing the relative expression of alleles within individuals, the assay has a perfect internal control for variables such as RNA quality, disease status (with the exception of multiplicative effects between other risk factors and gene expression), cell loss as a result of degenerative changes and the trans-acting effects of other loci or environmental factors (e.g. cause of death, drug treatment, nutritional status, etc.). All of these variables can be expected to influence each allelic variant of mRNA equally. The assay therefore permits detection of small cis-acting influences on gene expression, even with a background of large inter-individual variation in total gene expression, a feature that has been previously observed in studies of total APOE mRNA expression (20). A further advantage is that the assay is robust to secondary effects on gene expression. For example, possession of an allele that influences disease as a result of impaired protein function might, by trans-acting homeostatic mechanisms, result in a compensatory increase in gene expression.

We found increased expression of the {varepsilon}4 allele over that of the {varepsilon}3 allele in {varepsilon}3{varepsilon}4 heterozygotes, a finding that replicated in two additional samples: LOAD subjects and a mixed population of subjects with neuropsychiatric disorders. We also observed a small increase in relative {varepsilon}3 expression in {varepsilon}2{varepsilon}3 individuals, but this finding did not replicate in the small number of neuropsychiatric patients in which this genotype was observed. Although not a comprehensive survey of different brain regions, analysis of the frontal and temporal cortex yielded no evidence for differences by neuroanatomical region.

The relatively small difference in allele ratio observed in {varepsilon}3{varepsilon}4 heterozygotes in this study is in stark contrast with the findings of a previous study (12), where an RFLP-based design was used. In that study, a large increase was observed in the level of {varepsilon}3 relative to {varepsilon}4 mRNA in both control and LOAD samples, although smaller differences were seen in the LOAD group. Our data are more compatible with those of a subsequent study (13), in which no major differences in the expression of the two alleles were found in either control or LOAD {varepsilon}3{varepsilon}4 heterozygotes. It is also notable that, in the subsequent study (13), an average 4% relative increase in {varepsilon}4 expression was found, with a maximum of 22%. Using an RFLP-based design similar to the previous study, they also showed that estimation of allele ratio using this method is critically dependent on the number of PCR cycles (due to digestion-resistant heteroduplex formation), suggesting that the former results were due to artefact. The primer extension method used in the present study is unaffected by this problem and thus the present data are likely to provide a more accurate estimate of relative allelic expression of brain APOE.

The small increase in {varepsilon}4 expression observed in this study could possibly reflect an intrinsic effect of the alleles on transcription efficiency or RNA stability. Alternatively, the finding may reflect strong LD between the {varepsilon}2–4 conferring haplotype and an untested regulatory variant, of which there may be many (11). Increased representation of the {varepsilon}4 allele is unlikely to have resulted from differences in PCR or primer extension efficiency, as this is controlled (and corrected for) by concurrent assay of genomic DNA under the same conditions. Although a genomic control cannot be applied to limit potential confounding from differential RT efficiency, neither is this phenomenon likely to explain the observed differences for {varepsilon}3{varepsilon}4 heterozygotes, as inter-individual differences in ratios were largely preserved between separate RT reactions.

Our finding of increased expression of the {varepsilon}4 allele at the RNA level is consistent with a recent study in which the encoded ApoE4 isoform was found to account for an elevated proportion of total ApoE protein in the CSF of {varepsilon}3{varepsilon}4 heterozygotes (21). This finding is in contrast to measures of ApoE4 in plasma (2,21), where it is associated with lower protein expression, suggesting important allele-specific differences in the CNS and periphery in the control of synthesis and/or degradation of ApoE.

The magnitude of {varepsilon}4 over-expression was found to be virtually identical between the initial sample and the sample of confirmed LOAD cases. The initial sample was not selected to be a specific control group for the LOAD sample, and although it did not contain any individual with a diagnosis of AD, it is possible that some individuals might have developed the disease had they survived to old age. However, that no differences were observed between this sample and confirmed LOAD cases is consistent with previous comparisons using age-matched control groups in terms of both relative {varepsilon}4 RNA expression (13) and the relative level of ApoE4 protein (21), suggesting that, if there are common risk factors for LOAD that influence risk of disease by altering APOE expression, they do not selectively interact to markedly increase the expression of specific APOE mRNA species.

Though significant, the influence of cis-acting polymorphisms on steady-state levels of APOE mRNA in cerebral cortex is small, with a maximum relative difference in expression observed in this study of 25%. The 56 APOE heterozygotes assayed in this study provided >99% power to detect the effect of a distinct regulatory variant occurring in the general population at a frequency of 0.05, and almost 90% power to detect the effect of one present at a frequency of 0.02. This degree of power effectively excludes the existence of even fairly rare polymorphisms in and around APOE, exerting large direct effects on net APOE mRNA expression.

The third objective of this study was to determine if APOE mRNA levels in vivo are influenced by polymorphisms that affect transcription activity in vitro, and which have, in several studies, been associated with LOAD. Stratification of observed {varepsilon}4/{varepsilon}3 allele ratios according to heterozygosity for the promoter polymorphisms indicated a very small but statistically significant effect of the –219G>T polymorphism on allelic expression, consistent with it having a functional effect in brain. The –219T allele has previously been associated with reduced transcriptional activity in vitro (17) and with reduced plasma ApoE in vivo (22). The present finding of a reduced ratio of {varepsilon}4/{varepsilon}3 expression when the probability is that the {varepsilon}4 allele is in phase with the –219T allele, and the {varepsilon}3 allele in phase with the –219G allele, suggests that the –219T allele also decreases APOE expression in brain. As with measures of total expression, our data do not allow us to distinguish between direct and indirect association, but the consistency of our finding with in vitro studies suggests that this is a direct effect.

The conditional probabilities for diplotypes suggest that, for ~15% of –219G>T heterozygotes (i.e. two of the 13), the higher expression –219G allele is in phase with {varepsilon}4 in {varepsilon}3{varepsilon}4 individuals. As we do not know which, if any, individual diplotypes in this analysis are misclassified, we cannot accurately assess the impact of this on our estimate of the magnitude of effect. However, if we assume that the two heterozygotes displaying the least reduction in {varepsilon}4/{varepsilon}3 ratio are misclassified, we obtain an estimate of the maximum possible size of the effect. Excluding these from analysis reveals a mean net effect of the 219T allele of a 6% reduction in relative {varepsilon}4 expression. We stress that this analysis provides a guide to the maximum effect; we cannot conclude that the effect is of this size because we have no compelling case to exclude any single individual from the analysis.

The –491A allele of the –491A>G promoter polymorphism has been reported to show higher transcriptional activity in vitro (17), and to be associated with increased measures of total APOE expression in plasma (23) and brain (24). In the present study, heterozygosity at this site did not significantly influence relative {varepsilon}4 expression in any one direction. It should be noted, however, that power to detect a small, directional effect would have been diminished by the small sample size and low predicted consistency of phase. Interestingly, although only a single observation of all assayed samples, the one showing the greatest elevation in relative 334C ({varepsilon}4) expression (25%) was also the only sample in which, with high probability (0.99), the –491A (high expression predicted in vitro) nucleotide was in phase with both –219G (high expression in vitro and in this study) and 334C (high expression in this study).

Our study illustrates the importance of complementing in vivo and in vitro work. In vitro work requires analysis of the effects of variants in alien genomic and cellular contexts, and therefore the findings must be examined for compatibility in vivo. Nevertheless, without in vitro work, while associations can be observed, it is impossible to attribute in vivo effects on expression to specific variants. This would require that the effect of all other cis-acting variants on the same haplotype be controlled for—effectively an impossible task, as the relevant variants could be legion, and distributed across large genomic sequences.

To summarize, using a relatively large series of brain samples, we have demonstrated that any common influences on APOE expression attributable to cis-acting polymorphism are small. We have also effectively excluded the existence of even fairly rare cis-acting haplotypes of large effect. Of the variance that does occur, most is accounted for by the {varepsilon}4 haplotype, with the {varepsilon}4 allele of APOE being expressed at a significantly higher level than {varepsilon}3. This is highly reproducible and appears to be independent of AD status. We have also shown that the –219G promoter allele is associated with higher expression of APOE mRNA in vivo, a finding compatible with previous in vitro studies. Our study provides important functional in vivo plausibility to the existing body of association data, suggesting that the –219T promoter polymorphism may modulate risk of AD independent of APOE epsilon status (16). However, though demonstration of functionality enhances the case, strong genetic support from samples that are larger than are currently available to any single group is now required to force the conclusion that the extremely small effects we have observed are relevant to LOAD susceptibility.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
 DISCUSSION
 MATERIALS AND METHODS
REFERENCES
 
Samples
Initial experiments were performed using post-mortem frontal or temporal cortex from 60 unrelated anonymized Caucasians, of which 50 were from the UK (the MRC London Neurodegenerative Diseases Brain Bank), and 10 were from Sweden (Department of Clinical Neuroscience, Karolinska Institute, Stockholm). All were free from a diagnosis of psychiatric or neurological disorder at the time of death. Of these, 15 were {varepsilon}3{varepsilon}4 heterozygotes (12M, 3F, mean age=55, SD=20), 11 were {varepsilon}2{varepsilon}3 heterozygotes (9M, 4F, mean age=61, SD=18) and two were {varepsilon}2{varepsilon}4 heterozygotes (2F, mean age=65, SD=35). The LOAD sample consisted of post-mortem temporal cortex from 22 unrelated anonymized UK Caucasians with a neuropathological diagnosis of late-onset LOAD, obtained from the MRC London Neurodegenerative Diseases Brain Bank. Of these, 11 were {varepsilon}3{varepsilon}4 heterozygotes (3M, 8F, mean age=77, SD=12), with no {varepsilon}2 heterozygotes observed. Within-subject comparisons of expression between brain regions were made using a sample of 13 {varepsilon}3{varepsilon}4 heterozygotes (10M, 3F), drawn from an additional 60 unrelated anonymized individuals of mixed ethnicity and mixed neuropsychiatric diagnoses from the USA (The Mount Sinai School of Medicine, Department of Psychiatry Alzheimer's disease and Schizophrenia Brain Bank), of which there were also four {varepsilon}2{varepsilon}3 heterozygotes (2M, 2F). Tissue was available from cortical regions BA10 (frontopolar) and BA20 (inferior temporal) for each of these 60 individuals. For all samples, genomic DNA was extracted using standard phenol–chloroform procedures, and total RNA was extracted using the RNAwizTM isolation reagent (Ambion). Total RNA was treated with DNase prior to reverse transcription using random decamers and the RETROscriptTM kit (Ambion).

Genotyping
Genotyping for APOE {varepsilon}2/{varepsilon}3/{varepsilon}4 was performed using an RFLP-based assay described previously (25). Genotyping of SNP –491A>T was performed using SNaPshot primer extension, with amplification primers 5'-GCCTAGCCCCACTTTCTTTT-3' and 5'-CACAGTGGGCGAATCACTTA-3', and the extension primer 5'-CGAATCACTTAAGGTCAGGAG-3'. SNP –219G>T was genotyped using primers 5'-AGAATGGAGGAGGGTGTCCG-3' and 5'-ACTCAAGGATCCCAGACTTG-3', followed by restriction digestion with HpaII and EcoN1.

Allele expression assay
Genomic DNA from all subjects was initially genotyped, as described, in order to identify heterozygotes for SNPs 334T>C ({varepsilon}4 heterozygotes) and 472C>T ({varepsilon}2 heterozygotes). In each individual allelic expression experiment, cDNA from heterozygotes was assayed as two separate RT reactions, alongside duplicates of the corresponding genomic DNA samples. Each experiment was repeated on two separate occasions (i.e. a total of three experiments). DNase-treated RNA samples did not yield detectable product in the absence of an RT step. For assays of relative {varepsilon}4 expression, a 182 bp fragment containing only SNP 334T>C was amplified from cDNA or genomic DNA using primers: 5'-GCCTACAAATCGGAACTGGA-3' and 5'-AGCTCCTCGGTGCTCTGG-3'. For assays of relative {varepsilon}2 expression, a 67 bp fragment containing only SNP 472C>T was amplified from cDNA or genomic DNA using primers: 5'-CTGCGTAAGCGGCTCCTC-3' and 5'-CCCCGGCCTGGTACACTG-3'. PCRs were carried out in a total reaction volume of 12 µl, containing either cDNA or genomic DNA template, 1xPCR buffer, dNTPs at 0.1 mM, primers at 0.24 mM, 0.3 U ‘Hot Star’ taq polymerase (Qiagen) and 6% DMSO, with 35 cycles and an annealing temperature of 60°C. Amplified samples were incubated with 1 U shrimp alkaline phosphatase (Amersham) and 2 U exonuclease I (Amersham) for 45 min at 37°C and then for 15 min at 85°C prior to primer extension reactions. Primer extension was carried out using the SNaPshot Multiplex Kit (Applied Biosystems). The extension primer for SNP 334T>C was: 5'-GCGCGGACATGGAGGACGTG-3'. The extension primer for SNP 472C>T was: 5'-CCGATGACCTGCAGAAG-3'. Primer extension reactions were performed in a total volume of 10 µl, containing 2 µl treated PCR product, 4.5 µl SNaPshot kit, 2.5 µl water and 1 pM extension primer. Primer extension thermocycling conditions consisted of an initial step of 95°C for 2 min, followed by 25 cycles of 95°C for 5 s, 43°C for 5 s and 60°C for 5 s. Following primer extension, reaction products were treated with 0.5 U shrimp alkaline phosphatase (Amersham) for 45 min at 37°C and then for 15 min at 85°C. Aliquots of 1 µl SNaPshot reaction product were combined with 9 µl Hi-Di formamide and loaded onto a 3100 DNA sequencer (Applied Biosystems). Products were electrophoresed on a 36 cm capillary array at 60°C and data processed using Genescan analysis version 3.7 software (Applied Biosystems). Peak heights of allele-specific extended primers were determined using Genotyper version 2.5 software (Applied Biosystems), and the ratios used as an index of the relative expression of the two alleles in each individual sample. The same analytic conditions were used for genomic DNA and cDNA so that the average of all of the ratios observed in genomic DNA in each experiment (representing a 1 : 1 ratio of the two alleles) could be used to correct allelic ratios obtained from cDNA analyses for any inequalities in allelic representation specific to the assay (26).

Statistical analysis
Differences in allelic expression were tested by comparing genomic ratios with cDNA ratios from the same heterozygous samples. All group comparisons were analyzed by t-test (two-tailed) or, where diagnosis was included as a factor, by two-factorial ANOVA.

Predicted haplotype frequencies were calculated by using EH plus (27), and these formed the basis for calculation of individual diplotype probabilities. We calculate the probability that an individual carries a specific diplotype by first reconstructing all possible combinations of diplotypes for an individual, given the observed genotypes at each locus. We then use the expected distribution of diplotype frequencies, given the observed haplotypes frequencies within the specific sample to identify the most probable diplotype for that individual. The probability for the diplotype within an individual is then the frequency of that diplotype divided by the sum of the frequencies for all possible diplotypes.

Calculation of power to detect the effects of unknown regulatory variants is based on the binomial distribution, Hardy–Weinberg equilibrium at the regulatory SNP, and no LD with the marker SNP. The probability of an individual being homozygous at a putative regulatory locus with alleles in Hardy–Weinberg equilibrium is p2+q2, where p and q are the two allele frequencies. The probability that, of n individuals, all are homozygous (and therefore undetected by our assay) for the regulatory polymorphism is then (p2+q2)n. This also applies for n individuals selected for heterozygosity at the marker locus if there is no relationship (i.e. LD) between the genotypes at each locus. The power to detect at least one heterozygote is then 1–(p2+q2)n. If the marker and regulatory SNP are in LD, then a higher proportion of people selected for heterozygosity at the marker will also be heterozygous for the regulatory SNP, and the power will be increased.


   ACKNOWLEDGEMENTS
 
We are grateful to the MRC London Neurodegenerative Diseases Brain Bank (UK) and the Department of Clinical Neuroscience at the Karolinska Institute (Stockholm, Sweden) for donating brain tissue. This work was funded by the Medical Research Council (UK).


   FOOTNOTES
 
* To whom correspondence should be addressed. Tel: +44 2920743242; Fax: +44 2920746554; Email: odonovanmc{at}cardiff.ac.uk


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
 DISCUSSION
 MATERIALS AND METHODS
 REFERENCES
 

  1. Koch, S., Donarski, N., Goetze, K., Kreckel, M. Stuerenburg, H.J., Buhmann, C. and Beisiegel, U. (2001) Characterization of four lipoprotein classes in human cerebrospinal fluid. J. Lipid Res., 42, 1143–1151.[Abstract/Free Full Text]

  2. Boerwinkle, E. and Utermann, G. (1988) Simultaneous effects of the apolipoprotein E polymorphism on apolipoprotein E, apolipoprotein B, and cholesterol metabolism. Am. J. Hum. Genet., 42, 104–112.[Web of Science][Medline]

  3. Nathan, B.P., Bellosta, S., Sanan, D.A., Weisgraber, K.H., Mahley, R.W. and Pitas, R.E. (1994) Differential effects of apolipoproteins E3 and E4 on neuronal growth in vitro. Science, 264, 850–852.[Abstract/Free Full Text]

  4. Rebeck, G.W., Reiter, J.S., Strickland, D.K. and Hyman, B.T. (1993) Apolipoprotein E in sporadic Alzheimer's disease: allelic variation and receptor interactions. Neuron, 11, 575–580.[CrossRef][Web of Science][Medline]

  5. Strittmatter, W.J., Saunders, A.M., Schmechel, D., Pericak-Vance, M., Enghild, J., Salvesen, G.S. and Roses, A.D. (1993) Apolipoprotein E: high-avidity binding to beta-amyloid and increased frequency of type 4 allele in late-onset familial Alzheimer disease. Proc. Natl Acad. Sci. USA, 90, 1977–1981.[Abstract/Free Full Text]

  6. Corder, E.H., Saunders, A.M., Strittmatter, W.J., Schmechel, D.E., Gaskell, P.C., Small, G.W., Roses, A.D., Haines, J.L. and Pericak-Vance, M.A. (1993) Gene dose of apolipoprotein E type 4 allele and the risk of Alzheimer's disease in late onset families. Science, 261, 921–930.[Abstract/Free Full Text]

  7. Hyman, B.T., Gomez-Isla, T., Briggs, M., Chung, H., Nichols, S., Kohout, F. and Wallace, R. (1996) Apolipoprotein E and cognitive change in an elderly population. Ann. Neurol., 40, 55–66.[CrossRef][Web of Science][Medline]

  8. Talbot, C., Lendon, C., Craddock, N., Shears, S., Morris, J.C. and Goate, A. (1994) Protection against Alzheimer's disease with apoE epsilon 2. Lancet, 343, 1432–1433.[Web of Science][Medline]

  9. Bales, K.R., Verina, T., Dodel, R.C., Du, Y., Altstiel, L., Bender, M., Hyslop, P., Johnstone, E.M., Little, S.P., Cummins, D.J. et al. (1997) Lack of apolipoprotein E dramatically reduces amyloid beta-peptide deposition. Nat. Genet., 17, 263–264.[Web of Science][Medline]

  10. Holtzman, D.M., Bales, K.R., Tenkova, T., Fagan, A.M., Parsadanian, M., Sartorius, L.J., Mackey, B., Olney, J., McKeel, D., Wozniak, D. and Paul, S.M. (2000) Apolipoprotein E isoform-dependent amyloid deposition and neuritic degeneration in a mouse model of Alzheimer's disease. Proc. Natl Acad. Sci. USA, 97, 2892–2897.[Abstract/Free Full Text]

  11. Stengard, J.H., Clark, A.G., Weiss, K.M., Kardia, S., Nickerson, D.A., Salomaa, V., Ehnholm, C., Boerwinkle, E. and Sing, C.F. (2002) Contributions of 18 additional DNA sequence variations in the gene encoding apolipoprotein E to explaining variation in quantitative measures of lipid metabolism. Am. J. Hum. Genet., 71, 501–517.[CrossRef][Medline]

  12. Lambert, J.C., Perez-Tur, J., Dupire, M.J., Galasko, D., Mann, D., Amouyel, P., Hardy, J., Delacourte, A. and Chartier-Harlin, M.C. (1997) Distortion of allelic expression of apolipoprotein E in Alzheimer's disease. Hum. Mol. Genet., 6, 2151–2154.[Abstract/Free Full Text]

  13. Growdon, W.B., Cheung, B.S., Hyman, B.T. and Rebeck, G.W. (1999) Lack of allelic imbalance in APOE epsilon3/4 brain mRNA expression in Alzheimer's disease. Neurosci. Lett., 272, 83–86.[CrossRef][Web of Science][Medline]

  14. Bullido, M.J., Artiga, M.J., Recuero, M., Sastre, I., Garcia, M.A., Aldudo, J., Lendon, C., Han, S.W., Morris, J.C., Frank, A. et al. (1998) Polymorphism in the regulatory region of APOE associated with risk for Alzheimer's dementia. Nat. Genet., 18, 69–71.[CrossRef][Web of Science][Medline]

  15. Lambert, J.C., Pasquier, F., Cottel, D., Frigard, B., Amouyel, P. and Chartier-Harlin, M.C. (1998) A new polymorphism in the APOE promoter associated with risk of developing Alzheimer's disease. Hum. Mol. Genet., 7, 533–540.[Abstract/Free Full Text]

  16. Lambert, J.C., Araria-Goumidi, L., Myllykangas, L., Ellis, C., Wang, J.C., Bullido, M.J., Harris, J.M., Artiga, M.J., Hernandez, D., Kwon, J.M. et al. (2002) Contribution of APOE promoter polymorphisms to Alzheimer's disease risk. Neurology, 59, 59–66.[Abstract/Free Full Text]

  17. Artiga, M.J., Bullido, M.J., Sastre, I., Recuero, M., Garcia, M.A., Aldudo, J., Vazquez, J. and Valdivieso, F. (1998) Allelic polymorphisms in the transcriptional regulatory region of apolipoprotein E gene. FEBS Lett., 421, 105–108.[CrossRef][Web of Science][Medline]

  18. Yan, H., Yuan, W., Velculescu, V.E., Vogelstein, B. and Kinzler, K.W. (2002) Allelic variation in human gene expression. Science, 297, 1143.[Free Full Text]

  19. Bray, N.J., Buckland, P.R., Williams, N.M., Williams, H.J., Norton, N., Owen, M.J. and O'Donovan, M.C. (2003) A haplotype implicated in schizophrenia susceptibility is associated with reduced COMT expression in human brain. Am. J. Hum. Genet., 73, 152–161.[CrossRef][Web of Science][Medline]

  20. Yamagata, K., Urakami, K., Ikeda, K., Ji, Y., Adachi, Y., Arai, H., Sasaki, H., Sato, K. and Nakashima, K. (2001) High expression of apolipoprotein E mRNA in the brains with sporadic Alzheimer's disease. Dement. Geriatr. Cogn. Disord., 12, 57–62.[CrossRef][Web of Science][Medline]

  21. Fukumoto, H., Ingelsson, M., Garevik, N., Wahlund, L.O., Nukina, N., Yaguchi, Y., Shibata, M., Hyman, B.T., Rebeck, G.W. and Irizarry, M.C. (2003) APOE epsilon 3/epsilon 4 heterozygotes have an elevated proportion of apolipoprotein E4 in cerebrospinal fluid relative to plasma, independent of Alzheimer's disease diagnosis. Exp. Neurol., 183, 249–253.[CrossRef][Web of Science][Medline]

  22. Lambert, J.C., Brousseau, T., Defosse, V., Evans, A., Arveiler, D., Ruidavets, J.B., Haas, B., Cambou, J.P., Luc, G., Ducimetiere, P. et al. (2000) Independent association of an APOE gene promoter polymorphism with increased risk of myocardial infarction and decreased APOE plasma concentrations—the ECTIM study. Hum. Mol. Genet., 9, 57–61.[Abstract/Free Full Text]

  23. Laws, S.M., Taddei, K., Martins, G., Paton, A., Fisher, C., Clarnette, R., Hallmayer, J., Brooks, W.S., Gandy, S.E. and Martins, R.N. (1999) The –491AA polymorphism in the APOE gene is associated with increased plasma apoE levels in Alzheimer's disease. Neuroreport, 10, 879–882.[Web of Science][Medline]

  24. Laws, S.M., Hone, E., Taddei, K., Harper, C., Dean, B., McClean, C., Masters, C., Lautenschlager, N., Gandy, S.E. and Martins, R.N. (2002) Variation at the APOE –491 promoter locus is associated with altered brain levels of apolipoprotein E. Mol. Psychiatr., 7, 886–890.[CrossRef][Web of Science][Medline]

  25. Hixon, J.E. and Vernier, D.T. (1990) Restriction isotyping of human apolipoprotein E by gene amplification and cleavage with Hha1. J. Lipid Res., 31, 1432–1433.

  26. Hoogendoorn, B., Norton, N., Kirov, G., Williams, N., Hamshere, M.L., Spurlock, G., Austin, J., Stephens, M.K., Buckland, P.R., Owen, M.J. and O'Donovan, M.C. (2000) Cheap, accurate and rapid allele frequency estimation of single nucleotide polymorphisms by primer extension and DHPLC in DNA pools. Hum. Genet., 107, 488–493.[CrossRef][Web of Science][Medline]

  27. Zhao, H., Curtis, D. and Sham, P.C. (2000) Model-free analysis and permutation tests for allelic associations. Hum. Hered., 50, 133–139.[Web of Science][Medline]


Add to CiteULike CiteULike   Add to Connotea Connotea   Add to Del.icio.us Del.icio.us    What's this?


This article has been cited by other articles:


Home page
 Hum Mol GenetHome page
T. M. Caffrey, C. Joachim, S. Paracchini, M. M. Esiri, and R. Wade-Martins
Haplotype-specific expression of exon 10 at the human MAPT locus
Hum. Mol. Genet., December 15, 2006; 15(24): 3529 - 3537.
[Abstract] [Full Text] [PDF]

Home page
 Hum Mol GenetHome page
N. J. Bray, A. Preece, N. M. Williams, V. Moskvina, P. R. Buckland, M. J. Owen, and M. C. O'Donovan
Haplotypes at the dystrobrevin binding protein 1 (DTNBP1) gene locus mediate risk for schizophrenia through reduced DTNBP1 expression
Hum. Mol. Genet., July 15, 2005; 14(14): 1947 - 1954.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
13/22/2885    most recent
ddh299v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (11)
Right arrowRequest Permissions
Google Scholar
Right arrow Articles by Bray, N. J.
Right arrow Articles by O'Donovan, M. C.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Bray, N. J.
Right arrow Articles by O'Donovan, M. C.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?