Proximal and distal effects of genetic susceptibility to multiple sclerosis on the T cell epigenome

Abstract

Identifying the effects of genetic variation on the epigenome in disease-relevant cell types can help advance our understanding of the first molecular contributions of genetic susceptibility to disease onset. Here, we establish a genome-wide map of DNA methylation quantitative trait loci in CD4⁺ T-cells isolated from multiple sclerosis patients. Utilizing this map in a colocalization analysis, we identify 19 loci where the same haplotype drives both multiple sclerosis susceptibility and local DNA methylation. We also identify two distant methylation effects of multiple sclerosis susceptibility loci: a chromosome 16 locus affects PRDM8 methylation (a chromosome 4 region not previously associated with multiple sclerosis), and the aggregate effect of multiple sclerosis-associated variants in the major histocompatibility complex influences DNA methylation near PRKCA (chromosome 17). Overall, we present a new resource for a key cell type in inflammatory disease research and uncover new gene targets for the study of predisposition to multiple sclerosis.

Fig. 1. Characterization of genome-wide significant cis-mCpGs and cis-mQTLs.

a Enrichment of the identified cis-mCpGs in comparison to all tested CpGs in relation to UCSC CpG islands and gene functional regions (annotations from R IlluminaHumanMethylationEPICanno.ilm10b2.hg19 package) and chromatin states modeled for CD4⁺ T cells (annotations from Roadmap Epigenomics Project, sample #E043). Significant enrichment/depletion are shown using upwards/downwards arrows, respectively. b Percentage of our identified cis-mCpGs available in the BLUEPRINT study and the comparison between the two studies regarding the direction of effect and significance of association for the lead mSNPs are shown on the left. Comparison between the effect sizes found in our study and in BLUEPRINT are shown on the right.

Fig. 2. Colocalized cis-mQTL effects of MS susceptibility loci.

a The top 3 colocalized MS-cis-mQTL effects are illustrated (posterior probability >0.95 for colocalization between MS susceptibility and ≥ 1 cis-mQTL effect). The MS GWAS rows (blue dots) show -log(p-value) of association between SNPs and MS susceptibility in the discovery phase of the 2019 IMSGC GWAS. CpG rows (plum dots) show -log(p-value) of association between SNPs and the specified cis-CpGs methylation levels. Locations of the local CpGs measured with the Infinium MethylationEPIC kit are shown using light blue vertical lines. Gene exon/intron positions are based on Ensembl 93. Chromatin state annotations for CD4⁺ T cells are downloaded from the Roadmap Epigenomics Project (sample #E043). Green vertical lines represent the genomic location of the colocalized CpGs: Long vertical lines traversing all rows represent the top specified CpGs, while other vertical lines represent the additional colocalized cis-mCpGs with posterior probability >0.8. All genomic positions are in GRCh37 (hg19) coordinates. b Comparison between MS-cis-mQTL colocalization posterior probabilities using mQTL summary statistics from our study and BLUEPRINT for the available common CpGs. Vertical red line represents the threshold for high colocalization posterior probability in our study (i.e., 0.8). c The cis-mQTL, cis-eQTL, and CpG-mRNA association for the chromosome 5 top MS-cis-mQTL colocalized effect (n = 156, 36 and 36, respectively). Methylation levels are shown in M-values. Gene expression values are in TPM. Boxplot center lines represent median; the lower and upper hinges correspond to the 25th and 75th percentiles; the lower/upper whisker extends from the hinge to the smallest/largest value no further than 1.5× inter-quartile range. Error band represents 95% confidence interval.

Fig. 3. Colocalized trans-mQTL effect of MS susceptibility locus centered on rs3809627.

a The top colocalized MS-trans-mQTL effect observed between MS susceptibility locus centered on rs3809627 in chromosome 16 and cg27018912 located in a CpG island in chromosome 4. Left: Association (-log(p-value)) between SNPs located in the chromosome 16 locus and MS susceptibility (blue dots) and the trans-mCpG cg27018912 methylation levels (plum dots), accompanied by visualization of the CpGs, genes and modeled chromatin states in the chromosome 16 locus. Green vertical lines represent additional colocalized cis-mCpGs (posterior probability >0.8). Right: Genomic position of the trans-mCpGs (cg27018912, cg19409579 and cg04235768) on chromosome 4 (vertical green lines) in relation to nearby CpGs, genes, and modeled chromatin states. The top colocalized trans-mQTL effect is shown in the bottom using methylation M-values (n = 156). b Replication of the identified trans-mQTL effect between rs3809627 MS susceptibility variant (chromosome 16) and CpGs in the CpG island located on chromosome 4 in three independent studies. Methylation levels are shown in M-values. Data is shown for cg19409579 in the BLUEPRINT study as the top available trans-mCpG measured with the Infinium HumanMethylation450 array. Methylation data shown for the PhenoGenetic study are the average methylation levels of all the measured CpGs in the identified CpG island. Boxplot center lines represent median; box limits represent upper and lower quartiles. c Significant CpG-mRNA association was observed between cg19409579 trans-mCpG and PRDM8 gene expression in the BLUEPRINT study. Methylation levels are in M-values. Gene expression levels are ComBat normalized values available from the BLUEPRINT. Error band represents 95% confidence interval. Boxplots (a, b): center lines represent median; the lower and upper hinges correspond to the 25th and 75th percentiles; the lower/upper whisker extends from the hinge to the smallest/largest value no further than 1.5× inter-quartile range.

Fig. 4. cis and trans DNA methylation effects of MS MHC polygenic score.

a Manhattan plot shows association p-values between CpG methylation levels and MS-associated MHC polygenic score. Red horizontal line shows the significance threshold (FDR < 5%). Top CpGs in the two significant regions are specified. b t-statistics for positive and negative associations between the methylation levels of the CpGs located inside the extended MHC region and the MHC polygenic score are shown (red: positive, blue: negative). The inset illustrates the genomic location of the significant CpGs in the MHC class II region in relation to nearby CpGs, genes, and modeled chromatin states. Blue/red vertical lines represent the locations for CpGs with negative/positive associations with the MHC polygenic score, respectively. Association between the top identified CpG, cg09139047 located in the HLA-DRB1 gene, and MHC polygenic score (shown in Z-scores) is visualized in the top right corner. Colors represent the number of HLA-DRB1*1501 alleles. c The genomic location for cg03805787 (associated with the MHC polygenic score in chromosome 17, red vertical line) is shown in relation to surrounding CpGs, genes, and modeled chromatin states. Bottom panel illustrates the association between cg03805787 methylation levels (M-values) and MHC polygenic score (Z-scores) in our main study of MS subjects, as well as the replication in the healthy subjects of the PhenoGenetic study. Error bands (b, c) represent 95% confidence interval. MHC = Major Histocompatibility Complex.

Fig. 5. DNA methylation effects of MS total polygenic score.

a Manhattan plot shows association p-values between CpG methylation levels and MS total polygenic score. Red horizontal line shows the significance threshold (FDR < 5%). Top CpGs in the three significant regions are specified. The genomic location for cg16050799 (the top associated CpG outside the MHC region, red vertical line) in relation to nearby CpGs, genes, and modeled chromatin states, in addition to its association with the MS total polygenic score (Z-scores) is further shown. Error band represents 95% confidence interval. Methylation levels are shown in M values. b Genomic locations of MS susceptibility alleles nominally associated with cg16050799 methylation (p < 0.05) are shown. Gene sets enriched for genes associated with these loci are specified.