Heritable DNA Methylation in CD4+ Cells among Complex Families Displays Genetic and Non-Genetic Effects

Day, Kenneth; Waite, Lindsay L.; Alonso, Arnald; Irvin, Marguerite R.; Zhi, Degui; Thibeault, Krista S.; Aslibekyan, Stella; Hidalgo, Bertha; Borecki, Ingrid B.; Ordovás, José M.; Arnett, Donna K.; Tiwari, Hemant K.; Absher, Devin

doi:10.1371/journal.pone.0165488

Cited by 19 publications

(30 citation statements)

References 73 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Using the authors’ criteria for statistical significance, we overlapped these EWAS “hits” with mQTL data [49, 54, 62, 108, 110, 112, 113, 150, 155]. We selected the mQTL studies (all included in Table 2) to match the cell types or tissues studied in the EWAS.…”

Section: Relevance Of Mqtls and Hap-asm For Interpreting Ewas Datamentioning

confidence: 99%

“…~5% of the index SNPs were both mQTLs and eQTLs[108]Adipose tissue (648), replication set PBL (200)Illumina 450 K Methyl; multiple genotyping arrays, eQTLs: HT-12 V3 BeadChips; validations by WGBSmQTLs detected at ~28% the CpGs, with tissue-specificity; 22% of eQTLs were in LD with at least one mQTL; ~4% were in LD with a GWAS SNP; mQTLs associated with eQTLs and GWAS SNPs were enriched in enhancers[110]Cord blood (174), PBL (90), TC (125), FC (111), pons (106), cerebellum (105)Illumina: 27 K Methyl BeadChips; multiple Illumina and Affy genotyping arraysmQTLs detected at ~5% of the CpGs; overlap observed between ancestral groups, developmental stages, and tissue types; brain mQTL SNPs enriched in bipolar disorder GWAS peaks and miRNA-binding sites[155]TC (44), neurons (18), glia (22), T-cells (54), placenta (37)Illumina 450 K Methyl and 2.5 M SNP chips; validation by bis-seq and ox-bis-seq~3000 strong mQTLs identified; more than half tissue-restricted and ~900 located near GWAS signals; mQTLs enriched in polymorphic CTCF-binding sites and TFBS, and enriched in eQTLs located within 20 kb[49]Fetal brain (166), matched adult PFC, striatum and cerebellum (83)Illumina 450 K Methyl; 2.5 M SNP chipsMost fetal mQTLs also present in adult brain, but ~1/3 showed differential effects; mQTLs enriched in repressive and poised histone marks; mQTLs enriched in CTCF motifs, eQTLs, and schizophrenia-associated GWAS peaks[112]PBL (85)Illumina 27 K Methyl; OmniExpress SNP chips1287 smoking associated DM CpGs and 770 mQTLs identified. Among these, 43 CpGs were both smoking DM and mQTL[150]Adipose tissue (119)Illumina 450 K Methyl; Omni SNP chips; eQTL:Affymetrix Human Gene 1.0 ST arraymQTLs detected in ~3% of the CpGs; enriched in CGI shelves and shores and depleted in promoter regions and CGI; ~1% of mQTL SNPs (or proxy) were obesity-associated GWAS SNPs; 2% of the SNPs showed both mQTL and eQTL[113]CD4+ T cells (717)Illumina 450 K Methyl; Affy 6.0 SNP chipsOf ~20,000 heritable CpGs identified by modeling family structure, 15,133 were cis -mQTLs; 1329 trans-mQTLs and 4113 CpGs showing no evidence of cis or trans mQTL[54]Monocytes (197), neutrophils (197), and CD4+ T cells (132)Illumina 450 K Methyl; WGS; RNA-seq for ASE and ChIP-seq for hQTLsmQTLs affect 10% of CpGs, hQTLs found in 28 and 12% of H3K4me1 and H3K27ac peaks; 345 GWAS index SNPs (or SNPs in high LD with a GWAS index SNPs) colocalized with mQTLs and/or hQTLs[37]This list of studies is representative of the historical progression of the field and is not meant to be comprehensive. All experiments include internal statistical validations of the microarray and sequencing data; secondary validations refer to downstream assays by independent methods.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Genetic–epigenetic interactions in cis: a major focus in the post-GWAS era

et al. 2017

View full text Add to dashboard Cite

Studies on genetic–epigenetic interactions, including the mapping of methylation quantitative trait loci (mQTLs) and haplotype-dependent allele-specific DNA methylation (hap-ASM), have become a major focus in the post-genome-wide-association-study (GWAS) era. Such maps can nominate regulatory sequence variants that underlie GWAS signals for common diseases, ranging from neuropsychiatric disorders to cancers. Conversely, mQTLs need to be filtered out when searching for non-genetic effects in epigenome-wide association studies (EWAS). Sequence variants in CCCTC-binding factor (CTCF) and transcription factor binding sites have been mechanistically linked to mQTLs and hap-ASM. Identifying these sites can point to disease-associated transcriptional pathways, with implications for targeted treatment and prevention.

show abstract

Section: Relevance Of Mqtls and Hap-asm For Interpreting Ewas Datamentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Genetic–epigenetic interactions in cis: a major focus in the post-GWAS era

et al. 2017

View full text Add to dashboard Cite

show abstract

“…Most mQTL mapping studies thus far rely on DNA methylation data generated using array-based platforms [36–38]. However, the falling cost of sequencing and the development of high-throughput sequencing-based approaches to measure DNA methylation levels makes mQTL mapping using sequencing data increasingly feasible.…”

Section: Introductionmentioning

confidence: 99%

“…Early attempts to perform mQTL mapping with bisulfite sequencing data have yielded promising results [35, 49, 53]. However, existing mQTL mapping methods are designed with array data in mind [37, 38]. To maximize power, mQTL mapping using sequencing data requires new statistical methods development that can properly account for two of its distinctive features.…”

Section: Introductionmentioning

confidence: 99%

IMAGE: High-powered detection of genetic effects on DNA methylation using integrated methylation QTL mapping and allele-specific analysis

Fan

Vilgalys

Sun

et al. 2019

Preprint

View full text Add to dashboard Cite

Identifying genetic variants that are associated with methylation variation -an analysis commonly referred to as methylation quantitative trait locus (mQTL) mapping --is important for understanding the epigenetic mechanisms underlying genotype-trait associations. Here, we develop a statistical method, IMAGE, for mQTL mapping in sequencing-based methylation studies. IMAGE properly accounts for the count nature of bisulfite sequencing data and incorporates allele-specific methylation patterns from heterozygous individuals to enable more powerful mQTL discovery. We compare IMAGE with existing approaches through extensive simulation. We also apply IMAGE to analyze two bisulfite sequencing studies, in which IMAGE identifies more mQTL than existing approaches.Keywords: allelic specific methylation; ASM; methylation quantitative trait locus; mQTL; IMAGE; bisulfite sequencing; binomial mixed model; penalized quasilikelihood.mapping approaches (as allele-specific expression estimates have been shown to do for eQTL mapping: [51, 52]). Early attempts to perform mQTL mapping with bisulfite sequencing data have yielded promising results [35, 49, 53]. However, existing mQTL mapping methods are designed with array data in mind [37, 38]. To maximize power, mQTL mapping using sequencing data requires new statistical methods development that can properly account for two of its distinctive features.First, methylation data collected in sequencing studies are counts, not continuous representations like those produced by arrays. Specifically, methylation level estimates at a given cytosine base are based on both the total read count at the site and the subset of those reads that are unconverted by sodium bisulfite (or other processes [43]). Previous mQTL studies have dealt with these data by first computing a ratio between the methylated count and the total count, and then treating this ratio as an estimate of the true methylation level [35, 49]. However, the count nature of the raw data means that the mean and variance of the computed ratio are highly interdependent. This relationship is not captured by previously deployed linear regression methods, which likely leads to loss of power. Indeed, similar losses of power are well-documented for differential methylation analysis [40] and differential expression analysis of RNA-seq data [54][55][56][57]. To overcome this challenge, statistical methods for sequencing-based differential methylation analysis now adapt overdispersed count models, including beta-binomial models [58][59][60][61][62] and binomial mixed models [40, 63, 64], to properly model the mean-variance relationship and potential over-dispersion. In differential methylation analysis, these approaches can substantially improve power compared with normalization based approaches [30, 65, 66]. Because mQTL mapping is conceptually similar and can be effectively viewed as genotype-based differential methylation analysis, extending over-dispersed binomial models to mQTL mapping is a promising approach. ResultsMethod Overview and S...

show abstract

“…DNA methylation (DNAm) is heritable and plays a critical role in brain development and function through transcriptional regulation (1, 2). Family and twin studies have investigated the heritability of DNAm for sites across the genome in easily accessible tissues (3–7), but, to our knowledge, there have been limited studies of methylation heritability in brain tissue. Indeed, only one study estimated the heritability of DNAm levels of individual CpG dinucleotides attributable to local single nucleotide polymorphisms (SNPs) in postmortem brain from unrelated individuals, but the estimation was limited to ~21,000 CpG sites primarily within promoters (8).…”

Section: Introductionmentioning

confidence: 99%

Integrating brain methylome with GWAS for psychiatric risk gene discovery

Han¹,

Lin

Wang

et al. 2018

Preprint

View full text Add to dashboard Cite

DNA methylation (DNAm) is heritable and plays a role in brain development and function through transcriptional regulation. Aberrant DNAm in human brain has been linked to psychiatric disorders, potentially as mediators of common genetic risk variants. In this study, we hypothesize that common risk variants for psychiatric disorders may act through affecting DNAm level in human brain. We first aimed to investigate the heritability pattern of DNAm levels in the human prefrontal cortex. Secondly, through imputation-driven methylome-wide association study (MWAS), we aimed to identify CpG sites whose methylation levels are genetically associated and that show methylation-trait associations in the prefrontal cortex of patients with schizophrenia or bipolar disorder. Our heritability analysis showed that, of ~370,000 CpG sites measured with the Illumina HumanMethylation450 microarray, 17% were heritable (p < 0.05), with a mean heritability of 0.22. Heritable CpG sites were enriched in intergenic regions, CpG shore, and regulatory regions in prefrontal cortex. Our MWAS approach identified known and potentially novel risk genes harboring CpG sites of methylation-trait associations for schizophrenia or bipolar disorder, which were not detectable using three alternative strategies (blood-based methylome reference, transcriptome-wide association study, and two gene-based association tests). Gene set enrichment analysis for genes with methylation-trait association evidence revealed pathways clearly related to neuronal functions, but also highlighted additional biological mechanisms that may underlie psychiatric disorders, such as microRNA-related regulation. In conclusion, our results showed the power of integrating brain methylation data with GWAS for psychiatric risk gene discovery, with potential applications in brain-related disorders or traits. peripheral tissue (blood), which may not correlate with methylation signals in the brain (18, 19).Additionally, due to the challenges of accessing brain tissues, studies that examined brain samples were often limited by small sample size. Finally, it is hard to establish a causal role between DNAm change and disease as it is unclear whether aberrant DNAm causes disease or vice versa.Genome-wide association studies (GWAS) indicate that most common genetic risk variants underlying complex diseases are noncoding and enriched in regulatory genomic regions in cells or tissues related to disease (20), suggesting that risk variants may act through regulation of cell/tissue-specific molecular mechanisms, such as methylation or gene expression. For example, SCZ risk variants from GWAS were enriched at enhancers active in brain, but not in tissues unlikely to be relevant to SCZ (21). Furthermore, statistical genetic approaches like linkage disequilibrium (LD) score regression have shown enrichment of trait heritability in regions with functional annotations that are trait-related and cell-type-specific, including enrichment of heritability for SCZ and BP in functional regions specific to th...

show abstract

Heritable DNA Methylation in CD4+ Cells among Complex Families Displays Genetic and Non-Genetic Effects

Cited by 19 publications

References 73 publications

Genetic–epigenetic interactions in cis: a major focus in the post-GWAS era

Genetic–epigenetic interactions in cis: a major focus in the post-GWAS era

IMAGE: High-powered detection of genetic effects on DNA methylation using integrated methylation QTL mapping and allele-specific analysis

Integrating brain methylome with GWAS for psychiatric risk gene discovery

Contact Info

Product

Resources

About