Abundant quantitative trait loci exist for DNA methylation and gene expression in human brain.

Analysis of CpG methylation and mRNA measures across four human brain regions.(A,B), unsupervised cluster analysis of CpG methylation levels at autosomal loci and mRNA expression levels. Data arising from each brain region are colored accordingly and demonstrate consistent separation of cerebellum, pons and cortical samples, with separation of frontal and temporal cortex samples using mRNA transcript levels. (C,D) Tissue based pairwise comparisons of CpG methylation and mRNA expression. The analyses in these figures used cleaned data. Histograms show the distribution of CpG methylation levels and mRNA expression levels for each tissue. Scatter plots are direct comparison of the level of each detected transcript or methylation level in each tissue pair. Notably frontal cortex and temporal cortex show the most similar patterns of CpG methylation and expression; conversely, all comparisons against cerebellar tissue show this tissue to have the most distinct patterns for all measures.

Quantitative trait loci (QTL) for CpG methylation (methQTL), mRNA expression (eQTL) in four brain regions.(A,B), methQTL and eQTL respectively, where the Y-axis represents the physical location of the QTL SNP and the X-axis the physical location of the QTL probe (CpG site or mRNA). The size of each point reflects the relative strength of the association (R2). (A) The excess of points along the diagonal illustrates that detected methQTL are predominantly cis acting; but that trans methQTLs with high R2 values exist across multiple tissues. (B) Shows the majority of eQTL loci are in cis and that the in cis effect sizes are larger that those observed in trans. (C–G) cis methQTL results showing a symmetric association between methylation level and variants on both sides of the CpG site in all four brain regions. (H–L) cis eQTL results showing a symmetric association between expression level and variants on both sides of the transcription start site (TSS, corrected for strand) in all four brain regions. G and L show the average p value for significant methQTL or eQTL across regions. For both methQTL and eQTL the more significant cis QTLs tended to be both closer to the transcriptions start site or CpG methylation site and common across tissue regions tested.

Peak enrichment for QTLs.Plots showing the large cis enrichment of significantly identified QTLs over trans. For the paper and result tables cis has been defined as SNPs and probes (CpG and mRNA) that are ±1 Mb of each other. At this annotated definition of cis the enrichment of results are on the right of these two plots x-axis, where the plots are all four brain tissue regions together. The left plot is for CpGs and the right for mRNA. At 1 Mb the enrichment of cis to trans for CpG QTLs is 4,427-fold and for mRNA is 7,255-fold. These plots show how this enrichment changes when considering other distances as a threshold for cis, the X-axis are these other distances (non-linear) at 50 distances that are four/fifths the size of the next larger distance. The plots are based on the proportions of significant cis and trans SNP/Probe pairings from the possible pairings at different distances.

Comparison of QTLs for CpG methylation and mRNA expression across tissue regions.Any QTL that passed our threshold for significance in at least one tissue was included in the Ternary plots. The color of the points in the ternary plots reflects the cumulative R2 value for all tissues tested within each plot. Points toward the center indicate an equal R2 value across the three regions under investigation. Points toward the corner of a plot indicate a high R2 in one of the three tissues; points toward the edges of the plot indicate a high R2 in two of the three tissues. (A–H) Comparing methQTL in every three-way combination of the the four tissues for cis (A–D) and trans (E–H). (I–P) eQTL comparisons across each group of three tissues for cis and trans effects. Notably the cumulative R2 is generally higher for cis compared to trans loci across both methQTL and eQTL. Green circles highlight a cluster of relatively high cumulative R2 values driven primarily by the observed R2 within cerebellar tissue. These points were revealed to be a cis eQTL involving 20 SNPs and two neighboring transcripts, PPAPDC1A and C10orf85. (Q–T) Boxplots show expression level plotted against genotype for one of these eQTL SNP-transcript pairs (SNP rs2182513 and PPAPDC1A) and illustrate that this is a tissue specific QTL limited to the cerebellum.

Intersection of QTLs for CpG and mRNA traits.Plots shown for each tissue region; cerebellum, frontal cortex, pons and temporal cortex. Per tissue for every pairing of CpG site and mRNA transcript where the CpG was within a CpG island and within 1Mb of the mRNA TSS and both the CpG and mRNA have a significant cis QTL; i.e. a triplet is formed between SNP, mRNA and CpG, where the SNP was significantly correlated in cis with either the CpG or mRNA of the CpG-mRNA cis pairs under consideration. For these triplets we plotted the eQTL R2 (X axis) and the methQTL R2 (Y axis). The R2 values are shifted into one of four quadrants (without changing the effect size) based on the positive and negative combinations of correlations for the eQTLs and methQTLs. A positive correlation with mRNA means that the level of expression is increased with the minor allele of the SNP and a negative correlation means that the mRNA expression level is decreased with the minor allele of the SNP. For CpGs a positive correlation implies that the level of methylation is increased at a CpG site with the minor allele and a negative correlation implies that the level of methylation is decreased with the minor allele. The top-left quadrant contains negative eQTL and positive methQTL correlations. The top-right quadrant contains eQTLs and methQTLs where the correlation was positive in both. The bottom-left quadrant contains eQTLs and methQTLs where the correlation was negative in both. The bottom-right quadrant contains positive eQTL and negative methQTL correlations. The triplets are plotted as circles where the radius of the circle represent the R2 value between the CpG site and mRNA transcript. The color of circle represents the distance between the CpG site and the mRNA TSS, where red indicates that the CpG site is close and upstream to the mRNA TSS; blue indicates that the CpG site is close to the mRNA TSS but downstream; and black indicates that the CpG site is farther from the mRNA TSS both up or downstream. Triplets that fall along the main diagonals are those where a significant methQTL and eQTL are present; the CpG and mRNA pairs for these triplets are closer together and have a larger R2 between the CpG and mRNA. Triplets that are on or close to the X axis are those where there is a significant eQTL but a lack of signal for the methQTL, likewise the items that are on or close to the Y axis are those with a significant methQTL and a lack of signal for the eQTL.