Methylation quantitative trait locus analysis of chronic postsurgical pain uncovers epigenetic mediators of genetic risk

Abstract

Background: Overlap of pathways enriched by single nucleotide polymorphisms and DNA-methylation underlying chronic postsurgical pain (CPSP), prompted pilot study of CPSP-associated methylation quantitative trait loci (meQTL). Materials & methods: Children undergoing spine-fusion were recruited prospectively. Logistic-regression for genome- and epigenome-wide CPSP association and DNA-methylation-single nucleotide polymorphism association/mediation analyses to identify meQTLs were followed by functional genomics analyses. Results: CPSP (n = 20/58) and non-CPSP groups differed in pain-measures. Of 2753 meQTLs, DNA-methylation at 127 cytosine-guanine dinucleotides mediated association of 470 meQTLs with CPSP (p < 0.05). At PARK16 locus, CPSP risk meQTLs were associated with decreased DNA-methylation at RAB7L1 and increased DNA-methylation at PM20D1. Corresponding RAB7L1/PM20D1 blood eQTLs (GTEx) and cytosine-guanine dinucleotide-loci enrichment for histone marks, transcription factor binding sites and ATAC-seq peaks suggest altered transcription factor-binding. Conclusion: CPSP-associated meQTLs indicate epigenetic mechanisms mediate genetic risk. Clinical trial registration: NCT01839461, NCT01731873 (ClinicalTrials.gov).

Keywords: CPSP; DNA methylation; PARK16; epigenetics; genetics; meQTL; mechanisms; methylation quantitative trait; postoperative pain.

Conflict of interest statement

Financial & competing interests disclosure

The project was supported by the 5K23HD082782 through the Eunice Kennedy Shriver National Institute of Child Health and Human Development, National Institutes of Health (PI: Chidambaran), Center for Pediatric Genomics, Shared Facility Discovery Award, from Cincinnati Children’s Hospital Medical Center (PI: Chidambaran) and 1R01AR075857-01 through the National Institute of Arthtritis and Muscoskeletal and Skin Diseases. HJ was supported by NIH grant (R01AI141569-01) and NIH/NIEHS P30ES023513 (EHSC scholar fund). The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH. A Barski is a co-founder of Datirium, LLC. Datirium developed SciDAP bioinformatics platform used for ATAC data analysis in this study. V Chidambaran is a co-inventor of issued patent # US 10,662,476 B2 (24 May 2020) (Personalized pain management and anesthesia: preemptive risk identification and therapeutic decision support). The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

No writing assistance was utilized in the production of this manuscript.

Figures

Figure 1.. Analysis workflow.

Red text indicates tests; blue text indicates results.

Figure 2.. Causal inference test steps.

Omnibus test: intersection/union test for p-value of causal interference test.

Figure 3.. Enriched gene regulatory mechanisms within the 127 CpGs that causally mediate associations between SNPs and chronic postsurgical pain.

(A) Epigenetic markers enriched within these regions – histone marks, expression trait quantitative loci, and chromatin states. The X-axis indicates the significance (-log p-value), using the RELI software package (see Methods). The size of each circle indicates the fold-enrichment relative to background. Chromatin states are based on combinations of histone marks using the ChromHMM tool and data from RoadMap Epigenomics: 12_EnhBiv, bivalent enhancer; 2_TssAFlnk, region flanking an active transcription start site; 3_TxFlnk, transcribed regions; 7_Enh, enhancer. (B) Transcription factor binding site motifs enriched within the DNA sequences of these regions. The Y-axis indicates regulatory proteins (e.g., transcription factors), in decreasing order of significance.

Figure 4.. Differentially methylated CpG sites - meQTL pairs associated with chronic post-surgical pain in the PARK16 locus.

Figure 5.. DNA methylation–meQTL associations.

(A) and (B) show association of rs708723 (RAB29) genotypes with differential DNAm at RAB7L1 (cg16031515) and PM20D1 (rs16334093) CpG sites respectively, (C & D) show association of rs960603 (PM20D1) genotypes with differential DNA methylation at RAB7L1 (cg16031515) and PM20D1 (rs16334093) CpG sites respectively. In all panels, DNAm beta values for CpG sites are represented on y-axis and genotypes (AA, AG and GG) on the x-axis. (A & C), red dots represent DNAm values in CPSP phenotypes while blue dots represent controls. CPSP phenotypes are more represented in AG and AA genotypes for both meQTLs compared to wild type GG. This is consistent with odds ratios for CPSP being 3.194 (95% CI 1.345-7.586; p = 0.009) for rs708723 (risk allele A), and 4.866 (95%CI 1.382 – 17.310); p = 0.014 for rs960603 (risk allele A). DNAm values are represented as box and whisker plots in (B & D), with colors denoting different CpG sites. Red plots in (B & D) show that DNAm at cg16031515 (RAB7L1) was significantly lower for higher risk genotypes (AG and AA) compared to GG genotypes for both rs708723 and rs960603 (cg16031515 vs rs708723 genotypes: beta=-0.105, t-stat=-5.584, p < 0.001 and cg16031515 vs rs960603: beta= -0.145, t-stat=5.923, p < 0.001). Blue plots in (B & D) show DNAm at cg16334093 (PM20D1) was significantly higher for risk genotypes (AG and AA) compared to GG for both variants (cg16334093 vs rs708723: beta=0.154; t-stat=6.820, p < 0.001; cg16334093 vs rs960603 genotypes: beta=0.212, t-stat=7.232, p < 0.001). The direction of associations with DNAm at RAB7L1 and PM20D1 with meQTLs associated with higher CPSP risk was consistent for all meQTLs on PARK16. This suggests that association of RAB29/RAB7L1 and PM20D1 meQTLs with CPSP risk is mediated by DNAm at CpG sites on these genes. *p

Figure 6.. Mediation model showing mediation of methylation quantitaive trait loci association with chronic postsurgical pain by DNA methylation.

Mediation model denoting mediation by DNAm at CpG sites on two genes (RAB7L1 and PM20D1) of association of meQTLs rs823114 G_A (NUCKS1) (A), rs960603 G_A (PM20D1) and rs708723 (RAB7L1) (B) with CPSP. Odds ratios for CPSP (all meQTLs are associated with increased Odds for CPSP) are presented. Causal inference test was significant for all these interactions. The direction of association of meQTL on DNAm and DNAm on CPSP are denoted by + (positive) and - (negative) signs. In panel B, the red font indicates specific additional CpG sites affected by meQTL rs708723. Net effect of all meQTLs was to decrease DNAm at RAB7L1 sites and thus decreased the protective effect of RAB7L1 CpG DNAm on CPSP, and to increase DNAm at PM20D1 associated with increased risk for CPSP. CPSP: Chronic post-surgery pain; DNAm: DNA methylation; meQTL: Methylation quantitative trait loci; OR: Odds ratio.