Methylation quantitative trait loci (meQTLs) are consistently detected across ancestry, developmental stage, and tissue type
- Alicia K Smith†1, 2Email author,
- Varun Kilaru†1,
- Mehmet Kocak3,
- Lynn M Almli1,
- Kristina B Mercer2,
- Kerry J Ressler1, 4,
- Frances A Tylavsky3 and
- Karen N Conneely5
© Smith et al.; licensee BioMed Central Ltd. 2014
Received: 14 May 2013
Accepted: 6 February 2014
Published: 21 February 2014
Individual genotypes at specific loci can result in different patterns of DNA methylation. These methylation quantitative trait loci (meQTLs) influence methylation across extended genomic regions and may underlie direct SNP associations or gene-environment interactions. We hypothesized that the detection of meQTLs varies with ancestral population, developmental stage, and tissue type. We explored this by analyzing seven datasets that varied by ancestry (African American vs. Caucasian), developmental stage (neonate vs. adult), and tissue type (blood vs. four regions of postmortem brain) with genome-wide DNA methylation and SNP data. We tested for meQTLs by constructing linear regression models of methylation levels at each CpG site on SNP genotypes within 50 kb under an additive model controlling for multiple tests.
Most meQTLs mapped to intronic regions, although a limited number appeared to occur in synonymous or nonsynonymous coding SNPs. We saw significant overlap of meQTLs between ancestral groups, developmental stages, and tissue types, with the highest rates of overlap within the four brain regions. Compared with a random group of SNPs with comparable frequencies, meQTLs were more likely to be 1) represented among the most associated SNPs in the WTCCC bipolar disorder results and 2) located in microRNA binding sites.
These data give us insight into how SNPs impact gene regulation and support the notion that peripheral blood may be a reliable correlate of physiological processes in other tissues.
KeywordsDNA methylation meQTL mQTL Developmental stage Ancestry Race Gene regulation Inter-individual variation Biomarker Brain
DNA methylation patterns vary widely across the genome by developmental stage and by tissue type , making study design a challenge when the most relevant tissue is not accessible from living humans. This is particularly true for studies of neurologic or psychiatric traits. Although the most relevant tissues, like brain, may be obtained from non-human subjects, animal models often fail to adequately reflect the range of symptoms seen in humans. As a result, peripheral tissues are commonly used to identify biomarkers due to their ready availability.
A recent report characterized methylation patterns extensively in DNA derived from brain and blood samples taken from the same individuals . The study confirmed that genome-wide DNA methylation is highly tissue-specific; however, the authors also noted a striking similarity between DNA methylation differences in the blood and brain (cerebellum and cortex) from the same individual, prompting the hypothesis that between-individual DNA methylation differences may be mediated by DNA sequence variation, which does not vary across tissues. Nevertheless, the study had only a limited number of subjects and was not designed to address this hypothesis directly.
Over the past few years, multiple studies [3–9] have revealed that DNA methylation at specific loci can be influenced by sequence variations, such that individual genotypes at a given locus may result in different patterns of DNA methylation due to allele-specific methylation. These sites are called methylation quantitative trait loci (meQTLs) and can influence the methylation pattern across an extended genomic region [3, 8]. Numerous studies have found meQTLs in blood leukocytes, buccal cells, brain tissue, and lymphoblast cell lines [3–7], though there is only modest overlap between the SNP-CpG relationships identified in each study.
The ability to detect an meQTL is influenced by statistical power, which may depend on SNP allele frequency and linkage disequilibrium (LD) structure, which vary based on ancestry, or the variance of methylation at a specific CpG site, which may differ across stages of development or tissue type. These SNP-CpG relationships have wide-reaching biological relevance and likely contribute to inter-individual variation in gene regulation. The relationship between sequence variants and DNA methylation patterns in different sets of individuals may also provide insight into complex traits in which a particular environmental exposure associates with a trait only if incurred at a particular developmental window. Delineating sets of potential regulatory SNPs across different ancestries, developmental stages, and tissue types will help us assess the extent to which meQTLs can be used for large-scale studies of a wide range of human traits. This study marks an initial step in that process.
Number of meQTLs detected in each dataset and the number and percent overlapping between datasets, in the form number (percent)
Comparison of meQTLs by cohort
Of the meQTLs identified in each cohort, 10.6-49.5% were unique to that cohort. Note that the number of meQTLs identified varied substantially across cohorts, in part due to differences in power across cohorts. For example, 2055 meQTLs were found in umbilical cord blood from Caucasian neonates (CB B) compared to 629 in umbilical cord blood from African American neonates (CB A). Although these two groups had the same sample size, linkage disequilibrium between markers is substantially lower in African Americans than in Caucasians , so SNPs in CB A are less likely to tag untyped meQTLs. Despite the between-cohort differences in power, a number of SNP-CpG associations were common across cohorts. To determine whether the observed overlap between any two cohorts was significantly different from that expected by chance, we used Fisher’s exact test as described in Methods. There was significant overlap in meQTL detection between CB A and CB B (Table 1; overlap of 21.3-69.5%, depending on which group was treated as the baseline). Further, comparisons of meQTLs detected in African American samples at birth (CB A) and adult peripheral blood (PB) had similar agreement (44.1-50.7%). Finally, comparisons between Caucasian blood samples (CB B) and brain tissues (FCTX, TCTX, CRBLM, and PONS) also demonstrated greater overlap in detected meQTLs than would be expected by chance (18.5-31.6%). However, the highest proportions of overlap occurred between brain regions (35.8-71.7%). There is 6.6-35.1% overlap between PB and the four brain regions, even though the cohorts differ by both ancestral group and tissue type. All of the overlap between pairs of cohorts mentioned above was highly significant (Fisher’s exact p < 1×10-308); however, a caveat of this analysis is that the four brain regions were sampled from the same set of individuals. To investigate whether the higher levels of overlap between brain tissues are due to the sampling of brain tissue from a common set of subjects, we also calculated overlap from pairs of brain samples after randomly partitioning the subjects into two non-overlapping groups. Additional file 2 shows that for each pairwise comparison, the overlap between detected meQTLs is only slightly smaller in cross-tissue comparisons when comparing tissue sampled from non-overlapping groups (i.e., group A vs. B) vs. sampled from the same group (A vs. A, or B. vs. B). For A vs. B comparisons, we again observed a significantly larger number of overlapping meQTLs than expected by chance (Fisher exact p-values < 10-284 for all comparisons) despite having much lower power in these partitioned analyses, suggesting that the observed overlap between brain tissues is not an artifact caused by the sampling of brain tissue from a common set of subjects.
To verify that the overlap observed in Table 1 is robust to differences in linkage disequilibrium (LD) patterns among SNPs, we also performed secondary analyses where we examined only a set of roughly independent SNPs. Although fewer meQTLs were detected overall because fewer comparisons were made, we observed similarly large proportions of overlap in meQTL detection across cohorts when working with this pruned set of SNPs (Additional file 3). To verify that our observed associations are robust to how methylation was modeled, we re-performed select analyses where we modeled methylation using M-values (the logit transform of β-values ) rather than modeling β-values directly. Resulting test statistics for β-values vs. M-values were extremely similar; a representative example is shown in Additional file 4.
Differences in meQTLs by tissue, developmental stage, and ancestral group
To compare our findings with earlier results demonstrating that meQTLs in cerebellum tissue were enriched in GWAS results for complex traits , we investigated whether our brain meQTLs were more likely to be represented among the top SNP associations from the WTCCC bipolar disorder study (defined here as p < 0.001, to be consistent with ). We found that the Holm-significant meQTLs identified in PONS were enriched (empirical p-value = .019) for SNP associations from the GWAS. When we performed a similar analysis using a looser definition of meQTLs (p < .001), we saw enrichment in FCTX (p < .0001), TCTX (p = .048), and PONS (p = .005), though not CRBLM, as observed in .
The meQTLs identified in this study were more likely to occur in miRNA binding sites (3.7-5.2%) than SNPs that were not meQTLs (2.9-3.1%) in multiple cohorts: FCTX (empirical p-value = .002), TCTX (p = .001), CRBLM (p < 1×10-4), PONS (p = .014), PB (p = .001), and CB A (p = .015). For example, rs3814309 associates with the proportion of DNA methylation at cg04987894 (GSTM5) in the FCTX (p = 1.22×10-16), TCTX (p = 2.68×10-12), CRBLM (p = 2.95×10-11), PONS (p = 9.88×10-14), CB A (7.76×10-6), CB B (p = 3.40×10-4), and PB (p = 2.63×10-4). Interestingly, rs3814309 also lies on a binding site for hsa-miR-1237 in all brain tissues examined.
This study compared meQTLs in seven cohorts of similar size and coverage. Although the number of meQTLs identified varied, those detected were often specific to a single cohort. Despite this, we saw a highly significant degree of overlap between cohorts made up of subjects from different developmental stages and ancestral groups, as well as between samples obtained from different tissues. Exploring the nature of the similarities and differences from this study will give us insight into the use of meQTLs for the interpretation of large-scale studies.
DNA methylation patterns vary substantially between different tissues , but the sequence variants influencing methylation patterns may be consistent across tissues, as exemplified in Figure 2 and Table 1. This study uncovered a correlation between meQTLs identified in umbilical cord blood and those identified in several brain regions in an independent set of adult samples. Some of these meQTLs are located in genes with common functions in the blood and brain. For example, in CB B and each of the brain cohorts, we saw multiple meQTLs across a region of up to 20.3-kb that associated with a CpG site near ATP-binding cassette, sub-family C (ABCC4). This gene encodes a drug transporter that functions at the blood–brain barrier . Still other genes have functions that are exclusive in the brain. For example, in CB B and each of the brain cohorts, meQTLs associated with a CpG site near 5-hydroxytryptamine (serotonin) receptor 6 (HTR6; Additional file 1), which is expressed on in many brain regions and plays a role in cognition [14, 15]. Recently, Lambe and colleagues characterized HTR6 expression levels in postmortem tissue over time . HTR6 expression varies substantially over the lifespan suggesting that there may be particular ages at which one is more susceptible to psychiatric illnesses. Thus, examination of genotype-dependent DNA methylation of HTR6 in peripheral blood may yield new information about the dynamic regulatory processes occurring in the brain. In contrast, genetic variation in and near APOE associates with methylation patterns of a CpG site in the first intron of that gene in FCTX, TCTX, and PONS, but not in CRBLM or PB. APOE is linked to Alzheimer’s disease , and this study suggests that there may be limited insight to be gained from studies of APOE regulation that are not conducted in specific brain regions.
When comparing meQTLs detected in the blood of neonates (CB A) versus an adult population (PB), we again noted substantial (44.1-50.7%) though not complete overlap. This is not particularly surprising since DNA methylation patterns differentiate by developmental stage as well as by tissue type . Thus, if a gene involved in growth and development is less active in adulthood, meQTLs could be less detectable in adults based on the variability in DNA methylation levels of that gene. For example, in this study, we found associations between several SNPs and a CpG site in the promoter of the apelin receptor (AGTRL1) that were observable in blood from neonates but not adults. Apelin and its receptor help to stimulate hematopoiesis in embryonic stem cells, and dysregulation of these genes is linked to numerous disorders related to blood flow problems and hypertension [18–21]. Because fetal development is a period of extensive cellular replication and growth, epigenetic changes may establish long-term patterns among exposed individuals [22–24]. Extrapolating from this, genetic variation may be more influential during specific periods across the lifespan. Indeed, a recent study of meQTLs suggests that, in some cases, DNA methylation may be the mechanism by which genetic and environmental risk factors contribute to age-related phenotypes, including LDL, lung function, and longevity .
The power to detect meQTLs may be influenced by variations in allele frequencies and LD structure between cohorts, meaning we would expect to see differences in cohorts in which the length of expected LD blocks vary. Consistent with this, there were 3.3 times more meQTLs detected in the Caucasian cord blood cohort (CB B) compared to the African American cord blood cohort (CB A). We saw significant, albeit incomplete, overlap between meQTLs identified in both cohorts. In fact, 14.0-45.1% were specific to either CB A or CB B. For example, SNPs near CFTR (cystic fibrosis transmembrane conductance regulator) associate with methylation of a CpG site in the promoter region of CFTR in umbilical cord blood samples from Caucasian but not African American neonates. Genetic variants in CFTR cause cystic fibrosis, a Mendelian condition that is more prevalent in Caucasians [26, 27]. Methylation-associated SNPs may be in LD with functional mutations that are more common in Caucasians than in African Americans, but the difference in genotype-dependent methylation patterns cannot be attributed to differences in the minor allele frequencies between the two cohorts. DNA methylation patterns vary significantly by ancestry [28–30], prompting several to suggest that epigenetic differences may contribute to the increased risk of African Americans for a variety of health conditions [31–33]. Studies of epigenetic differences may provide insight into health disparities , and meQTLs can help distinguish between genetic and epigenetic effects in such studies.
Because of their regulatory potential, identification of meQTLs may also be useful to harness the power of GWAS more efficiently . The majority of meQTLs from this study localized to intronic regions. Despite this, we still saw significant enrichment of the meQTLs from FCTX, TCTX, and PONS in a GWAS study of bipolar disorder (BPD) . A recent study suggests that the top associations from a GWAS of BPD are enriched for meQTLs in a cerebellum dataset; when only the SNPs that associated with DNA methylation patterns were examined for association with BPD, associations that met Bonferroni criteria were identified in multiple cohorts . In addition, the results of this study suggest enrichment of miRNA binding sites among meQTLs, indicating these SNPs may promote a more complex degree of epigenetic regulation.
Though we attempted to present results from similarly powered cohorts, this study has several limitations. We would expect to identify more meQTLs if the cohorts were larger or if DNA methylation were assessed more densely. Due to study design and cohort availability, between-ancestry comparisons were limited to neonates. While DNA methylation was assessed in distinct tissues, those tissues contain multiple cell types. Thus, PB included the full range of leukocyte subpopulations, and the brain regions contained both a variety of neuronal phenotypes and glial cells. The corrections for multiple comparisons were conservative, and we have every confidence that the sets of meQTLs we identified are robust. We would expect, however, to have greater power to detect meQTLs in a study of homogenous cell types. In addition, bisulfite conversion does not distinguish between methylcytosine and hydroxymethylcytosine, which may be present in varying proportions in the seven cohorts used for this study.
Our findings reveal similarities in genotype-dependent DNA methylation across a diverse range of subject characteristics and tissues. Though DNA methylation patterns can be highly tissue specific, the contribution of inter-individual variation in gene regulation remains understudied. We need to further explore and characterize such regions of the genome to facilitate population-based studies of disease and for biomarker development. More specifically, these data support the utility of studying peripheral blood when the ideal tissue is unavailable and provide scope for interpreting those results.
Characteristics of the study cohorts
Age, yrs (mean ± SD)
Sex (% Male)
42.5 ± 14.0
Cord blood A
0.0 ± 0.0
Cord blood B
0.0 ± 0.0
47.6 ± 23.8
48.2 ± 24.5
46.6 ± 23.0
46.5 ± 24.0
DNA methylation data
The methylation datasets for this analysis have been described previously [8, 30, 40]. For each cohort, DNA methylation was first assessed at 27,578 CpG sites using the HumanMethylation27 BeadChip (Illumina). Samples with probe detection call rates < 90% were excluded, as were those with an average intensity value of either < 50% of the experiment-wide sample mean or < 2,000 arbitrary units. Data points with detection p-values > 0.001 were set to missing. Estimated DNA methylation proportions or β-values were then computed for each CpG site and sampled as the ratio of methylated signal to total (methylated + unmethylated) signal. Principal components analysis (PCA) of the β-values was then used to identify and eliminate 9 outliers from PONS, 21 from FCTX, and 6 from CRBLM, where outliers were determined as samples that fell more than 2.5 standard deviations away from the origin in plots of the first two principal components (Additional file 7). Additional file 7 also indicates batch effects in the brain samples, so we adjusted for these in our analysis as described below. No other cohort had evidence of outliers, and samples were hybridized in a single batch for each of the other cohorts. CpG sites with 1000 Genomes Project (Pilot 1 Data Release 2010_03) variants physically contained within the Illumina probe were excluded from all analyses, since CpG-SNP associations involving sites with variants in the probe are likely to be technical artifacts. After this step, 20,093 CpG sites passing QC in all cohorts remained eligible for analysis.
Genotyping for each cohort has been described previously [8, 30]. Genotyping was performed using multiple platforms: Illumina Omni-Quad 1 M, Omni-Express BeadChips (PB), HumanHap550 (FCTX, TCTX, CRBLM, PONS), and Affymetrix Genome-Wide Human SNP Array 5.0 and 6.0 (CB A and CB B). PLINK was used to perform quality control analyses such that, for all datasets, SNPs that had a call rate < 95%, a minor allele frequency (MAF) < .05, or significant deviation from Hardy-Weinberg proportions (p < .00001) were excluded, as were samples with > 5% missing data. So that annotation information was consistent between all datasets, the positional information for all the genotype datasets was converted to build 36 of the Human genome using the UCSC liftover tool when necessary . All allelic designations are oriented to the forward strand.
MaCH 1.0  was used to impute missing genotypes for SNPs that were not common to each array. Caucasian datasets were imputed using the HapMap CEU + TSI phase 3 reference samples, and the African American datasets were imputed using unrelated individuals from HapMap ASW, CEU, LWK, MKK, TSI, and YRI Phase 3 reference samples. Imputed SNPs with an estimated r2 < .3 between imputed and true genotypes and those with posterior probabilities < .9 for the most likely genotype were excluded from subsequent analysis. After imputation, 222,888 SNPs with MAF > .05 passed quality control based on genotyping or imputation in all cohorts and were eligible for analysis.
All statistical analyses were conducted in R. The relationship between the β-value or proportion of methylation at each CpG site and each SNP within 50 kb of that site was examined via linear regression, where β-values were modeled as a linear function of the number of reference alleles (0, 1, or 2). As appropriate, we included covariates to account for Affymetrix array 5.0 vs. 6.0 (CB A and CB B), or hybridization batch (FCTX, TCTX, CRBLM, PONS; other cohorts were hybridized in a single batch). To eliminate the possibility of inflated test statistics due to the lack of power at particular sites, analysis was restricted to SNP-CpG combinations with complete data for 32 or more subjects in each cohort (i.e., residual degrees of freedom ≥ 30). A Holm (step-down Bonferroni) correction  was applied to adjust for the total number of tests performed in each dataset; this type of approach will maintain the experiment-wide type I error rate at 5% for independent tests and will be even more conservative for the correlated tests performed here . Gene ontology (GO) terms for biological processes were assigned using GeneCodis 2.0 [45, 46]. We used SNAP to map the location of each meQTL relative to its location within a gene .
To test whether the overlap of meQTLs identified between each pair of cohorts was significantly greater than that expected by chance, we used Fisher’s exact test to test for independence between meQTL status in one cohort vs. meQTL status in another cohort. To test for enrichment in other datasets, we used an empirical sampling strategy. We first used this strategy to test for significant overlap between the set of meQTLs identified in each cohort and the set of variants listed in the Wellcome Trust GWAS for bipolar disorder . To generate an empirical p-value for the enrichment test, for each cohort we randomly sampled sets of X SNPs from the full set of SNPs analyzed, where X is the number of SNPs identified as significant meQTLs in that cohort. To ensure that the distribution of MAF in the randomly drawn datasets was similar to the original, SNPs in each cohort were binned by MAF intervals of .025. For each random sample we then sampled the same number of SNPs from each bin as in the original set of significant meQTLs. For each random set, we computed the number of overlapping SNPs as the number of GWAS SNPs included in the random set of X SNPs, and compared this number to the number of overlapping SNPs in the original analysis. The empirical p-value was then the proportion of random SNP sets demonstrating at least as much overlap with the GWAS SNPs as observed in the original analysis.
We also used the strategy described above to compare enrichment of meQTLs from this study to published meQTL and eQTL results. For blood cohorts (PB, CB A, and CB B), we compared to eQTLs detected by Stranger and colleagues , and for brain cohorts (FCTX, TCTX, CRBLM, and PONS), we compared to meQTLs and eQTLs detected by Gibbs and colleagues . Each enrichment test was conducted with 10,000 randomly sampled SNP sets, similar to those described above.
To identify meQTLs that could influence miRNA binding, we examined the list of SNPs in predicted miRNA-mRNA binding sites from the MirSNP database (http://cmbi.bjmu.edu.cn/mirsnp). Similarly, we tested for enrichment of the miRNA SNPs among the set of significant meQTLs and generated an empirical p-value using 1000 randomly sampled sets of SNPs for each cohort.
For secondary analyses involving an independent set of SNPs, we used PLINK to prune the genotype data in windows of 50 bp (base pairs), removing one SNP from each pair of SNPs with r2 > 0.05. We then performed select analyses on this reduced set of roughly independent SNPs (summarized in Additional file 3).
The additional material contains three figures and four tables that provide supporting information for specific points in the paper.
Methylation quantitative trait loci
Linkage disequilibrium (LD)
- CB A:
Umbilical cord blood from African American neonates
- CB B:
Umbilical cord blood from Caucasian neonates
Adult peripheral blood
ATP-binding cassette, sub-family C
Cystic fibrosis transmembrane conductance regulator
Serotonin receptor 6
Minor allele frequency
We gratefully acknowledge the study participants and staff of CANDLE and the Grady Trauma Project. This research was supported by grants from the National Institute of Mental Health (MH085806 to A.K.S., MH071537 and MH096764 to K.J.R.), the National Institute of Child Health and Human Development (HD060713 and HD055462 to Julia Krushkal and Ronald M. Adkins), and the Howard Hughes Medical Institute (K.J.R.). Additional support came from a grant from the University of Tennessee Health Science Centers Clinical and Translational Science Institute to RMA and a grant from The Urban Child Institute to FT. The authors also acknowledge the support of the Center for Integrative and Translational Genomics and Dr. Robert Williams at the University of Tennessee Health Science Center. Finally, we thank Cheryl Strauss for her editorial assistance.
- Liang P, Song F, Ghosh S, Morien E, Qin M, Mahmood S, Fujiwara K, Igarashi J, Nagase H, Held WA: Genome-wide survey reveals dynamic widespread tissue-specific changes in DNA methylation during development. BMC Genomics. 2011, 12: 231-10.1186/1471-2164-12-231.PubMed CentralPubMedView ArticleGoogle Scholar
- Davies MN, Volta M, Pidsley R, Lunnon K, Dixit A, Lovestone S, Coarfa C, Harris RA, Milosavljevic A, Troakes C, et al: Functional annotation of the human brain methylome identifies tissue-specific epigenetic variation across brain and blood. Genome Biol. 2012, 13: R43-10.1186/gb-2012-13-6-r43.PubMed CentralPubMedView ArticleGoogle Scholar
- Bell JT, Pai AA, Pickrell JK, Gaffney DJ, Pique-Regi R, Degner JF, Gilad Y, Pritchard JK: DNA methylation patterns associate with genetic and gene expression variation in HapMap cell lines. Genome Biol. 2011, 12: R10-10.1186/gb-2011-12-1-r10.PubMed CentralPubMedView ArticleGoogle Scholar
- Shoemaker R, Deng J, Wang W, Zhang K: Allele-specific methylation is prevalent and is contributed by CpG-SNPs in the human genome. Genome Res. 2010, 20: 883-889. 10.1101/gr.104695.109.PubMed CentralPubMedView ArticleGoogle Scholar
- Zhang D, Cheng L, Badner JA, Chen C, Chen Q, Luo W, Craig DW, Redman M, Gershon ES, Liu C: Genetic control of individual differences in gene-specific methylation in human brain. Am J Hum Genet. 2010, 86: 411-419. 10.1016/j.ajhg.2010.02.005.PubMed CentralPubMedView ArticleGoogle Scholar
- Schalkwyk LC, Meaburn EL, Smith R, Dempster EL, Jeffries AR, Davies MN, Plomin R, Mill J: Allelic skewing of DNA methylation is widespread across the genome. Am J Hum Genet. 2010, 86: 196-212. 10.1016/j.ajhg.2010.01.014.PubMed CentralPubMedView ArticleGoogle Scholar
- Li Y, Zhu J, Tian G, Li N, Li Q, Ye M, Zheng H, Yu J, Wu H, Sun J, et al: The DNA methylome of human peripheral blood mononuclear cells. PLoS Biol. 2010, 8: e1000533-10.1371/journal.pbio.1000533.PubMed CentralPubMedView ArticleGoogle Scholar
- Gibbs JR, van der Brug MP, Hernandez DG, Traynor BJ, Nalls MA, Lai SL, Arepalli S, Dillman A, Rafferty IP, Troncoso J, et al: Abundant quantitative trait loci exist for DNA methylation and gene expression in human brain. PLoS Genet. 2010, 6: e1000952-10.1371/journal.pgen.1000952.PubMed CentralPubMedView ArticleGoogle Scholar
- Gamazon ER, Badner JA, Cheng L, Zhang C, Zhang D, Cox NJ, Gershon ES, Kelsoe JR, Greenwood TA, Nievergelt CM, et al: Enrichment of cis-regulatory gene expression SNPs and methylation quantitative trait loci among bipolar disorder susceptibility variants. Mol Psychiatry. 2013, 18 (3): 340-6. 10.1038/mp.2011.174.PubMed CentralPubMedView ArticleGoogle Scholar
- Stranger BE, Montgomery SB, Dimas AS, Parts L, Stegle O, Ingle CE, Sekowska M, Smith GD, Evans D, Gutierrez-Arcelus M, et al: Patterns of cis regulatory variation in diverse human populations. PLoS Genet. 2012, 8: e1002639-10.1371/journal.pgen.1002639.PubMed CentralPubMedView ArticleGoogle Scholar
- Shifman S, Kuypers J, Kokoris M, Yakir B, Darvasi A: Linkage disequilibrium patterns of the human genome across populations. Hum Mol Genet. 2003, 12: 771-776. 10.1093/hmg/ddg088.PubMedView ArticleGoogle Scholar
- Du P, Zhang X, Huang CC, Jafari N, Kibbe WA, Hou L, Lin SM: Comparison of Beta-value and M-value methods for quantifying methylation levels by microarray analysis. BMC Bioinform. 2010, 11: 587-10.1186/1471-2105-11-587.View ArticleGoogle Scholar
- Nagasaka Y, Oda K, Iwatsubo T, Kawamura A, Usui T: Effects of aripiprazole and its active metabolite dehydroaripiprazole on the activities of drug efflux transporters expressed both in the intestine and at the blood–brain barrier. Biopharm Drug Dispos. 2012, 33: 304-315. 10.1002/bdd.1801.PubMedView ArticleGoogle Scholar
- Fone KC: An update on the role of the 5-hydroxytryptamine6 receptor in cognitive function. Neuropharmacology. 2008, 55: 1015-1022. 10.1016/j.neuropharm.2008.06.061.PubMedView ArticleGoogle Scholar
- King MV, Marsden CA, Fone KC: A role for the 5-HT(1A), 5-HT4 and 5-HT6 receptors in learning and memory. Trends Pharmacol Sci. 2008, 29: 482-492. 10.1016/j.tips.2008.07.001.PubMedView ArticleGoogle Scholar
- Lambe EK, Fillman SG, Webster MJ, Shannon Weickert C: Serotonin receptor expression in human prefrontal cortex: balancing excitation and inhibition across postnatal development. PLoS One. 2011, 6: e22799-10.1371/journal.pone.0022799.PubMed CentralPubMedView ArticleGoogle Scholar
- Verghese PB, Castellano JM, Holtzman DM: Apolipoprotein E in Alzheimer’s disease and other neurological disorders. Lancet Neurol. 2011, 10: 241-252. 10.1016/S1474-4422(10)70325-2.PubMed CentralPubMedView ArticleGoogle Scholar
- Cobellis L, De Falco M, Mastrogiacomo A, Giraldi D, Dattilo D, Scaffa C, Colacurci N, De Luca A: Modulation of apelin and APJ receptor in normal and preeclampsia-complicated placentas. Histol Histopathol. 2007, 22: 1-8.PubMedGoogle Scholar
- Yu QC, Hirst CE, Costa M, Ng ES, Schiesser JV, Gertow K, Stanley EG, Elefanty AG: APELIN promotes hematopoiesis from human embryonic stem cells. Blood. 2012, 119: 6243-6254. 10.1182/blood-2011-12-396093.PubMedView ArticleGoogle Scholar
- Carpene C, Dray C, Attane C, Valet P, Portillo MP, Churruca I, Milagro FI, Castan-Laurell I: Expanding role for the apelin/APJ system in physiopathology. J Physiol Biochem. 2007, 63: 359-373.PubMedView ArticleGoogle Scholar
- O’Carroll AM, Lolait SJ, Harris LE, Pope GR: The apelin receptor APJ: journey from an orphan to a multifaceted regulator of homeostasis. J Endocrinol. 2013, 219: R13-R35. 10.1530/JOE-13-0227.PubMedView ArticleGoogle Scholar
- Waterland RA, Michels KB: Epigenetic epidemiology of the developmental origins hypothesis. Annu Rev Nutr. 2007, 27: 363-388. 10.1146/annurev.nutr.27.061406.093705.PubMedView ArticleGoogle Scholar
- Waterland RA, Jirtle RL: Early nutrition, epigenetic changes at transposons and imprinted genes, and enhanced susceptibility to adult chronic diseases. Nutrition. 2004, 20: 63-68. 10.1016/j.nut.2003.09.011.PubMedView ArticleGoogle Scholar
- Dolinoy DC, Weidman JR, Jirtle RL: Epigenetic gene regulation: linking early developmental environment to adult disease. Reprod Toxicol. 2007, 23: 297-307. 10.1016/j.reprotox.2006.08.012.PubMedView ArticleGoogle Scholar
- Bell JT, Tsai PC, Yang TP, Pidsley R, Nisbet J, Glass D, Mangino M, Zhai G, Zhang F, Valdes A, et al: Epigenome-wide scans identify differentially methylated regions for age and age-related phenotypes in a healthy ageing population. PLoS Genet. 2012, 8: e1002629-10.1371/journal.pgen.1002629.PubMed CentralPubMedView ArticleGoogle Scholar
- Watson MS, Cutting GR, Desnick RJ, Driscoll DA, Klinger K, Mennuti M, Palomaki GE, Popovich BW, Pratt VM, Rohlfs EM, et al: Cystic fibrosis population carrier screening: 2004 revision of American College of Medical Genetics mutation panel. Genet Med. 2004, 6: 387-391. 10.1097/01.GIM.0000139506.11694.7C.PubMed CentralPubMedView ArticleGoogle Scholar
- Strom CM, Crossley B, Buller-Buerkle A, Jarvis M, Quan F, Peng M, Muralidharan K, Pratt V, Redman JB, Sun W: Cystic fibrosis testing 8 years on: lessons learned from carrier screening and sequencing analysis. Genet Med. 2011, 13: 166-172. 10.1097/GIM.0b013e3181fa24c4.PubMedView ArticleGoogle Scholar
- Terry MB, Delgado-Cruzata L, Vin-Raviv N, Wu HC, Santella RM: DNA methylation in white blood cells: association with risk factors in epidemiologic studies. Epigenetics. 2011, 6: 828-837. 10.4161/epi.6.7.16500.PubMed CentralPubMedView ArticleGoogle Scholar
- Lam LL, Emberly E, Fraser HB, Neumann SM, Chen E, Miller GE, Kobor MS: Factors underlying variable DNA methylation in a human community cohort. Proc Natl Acad Sci U S A. 2012, 109 (Suppl 2): 17253-17260.PubMed CentralPubMedView ArticleGoogle Scholar
- Adkins RM, Krushkal J, Tylavsky FA, Thomas F: Racial differences in gene-specific DNA methylation levels are present at birth. Birth Defects Res A Clin Mol Teratol. 2011, 91 (8): 728-36. 10.1002/bdra.20770.PubMed CentralPubMedView ArticleGoogle Scholar
- Kuzawa CW, Sweet E: Epigenetics and the embodiment of race: developmental origins of US racial disparities in cardiovascular health. Am J Hum Biol. 2009, 21: 2-15. 10.1002/ajhb.20822.PubMedView ArticleGoogle Scholar
- Burris HH, Collins JW: Race and preterm birth–the case for epigenetic inquiry. Ethn Dis. 2010, 20: 296-299.PubMedGoogle Scholar
- Menon R, Conneely KN, Smith AK: DNA methylation: an epigenetic risk factor in preterm birth. Reprod Sci. 2012, 19: 6-13. 10.1177/1933719111424446.PubMedView ArticleGoogle Scholar
- Thayer ZM, Kuzawa CW: Biological memories of past environments: epigenetic pathways to health disparities. Epigenetics. 2011, 6: 798-803. 10.4161/epi.6.7.16222.PubMedView ArticleGoogle Scholar
- Tycko B: Mapping allele-specific DNA methylation: a New tool for maximizing information from GWAS. Am J Hum Genet. 2010, 86: 109-112. 10.1016/j.ajhg.2010.01.021.PubMed CentralPubMedView ArticleGoogle Scholar
- The Wellcome Trust Case Control Consortium: Genome-wide association study of 14,000 cases of seven common diseases and 3,000 shared controls. Nature. 2007, 447: 661-678. 10.1038/nature05911.PubMed CentralView ArticleGoogle Scholar
- Bradley RG, Binder EB, Epstein MP, Tang Y, Nair HP, Liu W, Gillespie CF, Berg T, Evces M, Newport DJ, et al: Influence of child abuse on adult depression: moderation by the corticotropin-releasing hormone receptor gene. Arch Gen Psychiatry. 2008, 65: 190-200. 10.1001/archgenpsychiatry.2007.26.PubMed CentralPubMedView ArticleGoogle Scholar
- Gillespie CF, Bradley B, Mercer K, Smith AK, Conneely K, Gapen M, Weiss T, Schwartz AC, Cubells JF, Ressler KJ: Trauma exposure and stress-related disorders in inner city primary care patients. Gen Hosp Psychiatry. 2009, 31: 505-514. 10.1016/j.genhosppsych.2009.05.003.PubMed CentralPubMedView ArticleGoogle Scholar
- Adkins RM, Tylavsky FA, Krushkal J: Newborn umbilical cord blood DNA methylation and gene expression levels exhibit limited association with birth weight. Chem Biodivers. 2012, 9: 888-899. 10.1002/cbdv.201100395.PubMedView ArticleGoogle Scholar
- Smith AK, Conneely KN, Kilaru V, Mercer KB, Weiss TE, Bradley B, Tang Y, Gillespie CF, Cubells JF, Ressler KJ: Differential immune system DNA methylation and cytokine regulation in post-traumatic stress disorder. Am J Med Genet B Neuropsychiatr Genet. 2011, 156B (6): 700-8.PubMed CentralPubMedView ArticleGoogle Scholar
- Hinrichs AS, Karolchik D, Baertsch R, Barber GP, Bejerano G, Clawson H, Diekhans M, Furey TS, Harte RA, Hsu F, et al: The UCSC genome browser database: update 2006. Nucleic Acids Res. 2006, 34: D590-D598. 10.1093/nar/gkj144.PubMed CentralPubMedView ArticleGoogle Scholar
- Li Y, Willer CJ, Ding J, Scheet P, Abecasis GR: MaCH: using sequence and genotype data to estimate haplotypes and unobserved genotypes. Genet Epidemiol. 2010, 34: 816-834. 10.1002/gepi.20533.PubMed CentralPubMedView ArticleGoogle Scholar
- Holm S: A simple sequentially rejective multiple test procedure. Scand J Stat. 1979, 6: 65-70.Google Scholar
- Sidak Z: Rectangular confidence regions for the means of multivariate normal distributions. J Am Stat Assoc. 1967, 62: 626-633.Google Scholar
- Nogales-Cadenas R, Carmona-Saez P, Vazquez M, Vicente C, Yang X, Tirado F, Carazo JM, Pascual-Montano A: GeneCodis: interpreting gene lists through enrichment analysis and integration of diverse biological information. Nucleic Acids Res. 2009, 37: W317-W322. 10.1093/nar/gkp416.PubMed CentralPubMedView ArticleGoogle Scholar
- Carmona-Saez P, Chagoyen M, Tirado F, Carazo JM, Pascual-Montano A: GENECODIS: a web-based tool for finding significant concurrent annotations in gene lists. Genome Biol. 2007, 8: R3-10.1186/gb-2007-8-1-r3.PubMed CentralPubMedView ArticleGoogle Scholar
- Johnson AD, Handsaker RE, Pulit SL, Nizzari MM, O’Donnell CJ, de Bakker PI: SNAP: a web-based tool for identification and annotation of proxy SNPs using HapMap. Bioinformatics. 2008, 24: 2938-2939. 10.1093/bioinformatics/btn564.PubMed CentralPubMedView ArticleGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited.