Targeting Adipose Tissue Functional Methylomes by Next-Generation Capture Sequencing Reveals Novel Metabolic Trait Associated Variants

Fiona Allum1,2, Xiaojian Shao1,2, Frédéric Guénard3, Marie-Michelle Simon1,2, Stephan Busche1,2, Maxime Caron1,2, Tony Kwan1,2, Todd Richmond4, Daniel Burgess4, André Tchernof5, Simon Marceau5, Mark Lathrop1,2, Marie-Claude Vohl3, Tomi Pastinen1,2, Elin Grundberg1,2

1. Department of Human Genetics, McGill University, Montreal, QC, Canada; 2. McGill University and Genome Quebec Innovation Centre, Montreal, QC, Canada; 3. Institute of Nutrition and Functional Foods (INAF), Université Laval, Québec, QC, Canada; 4. Roche NimbleGen, Madison, WI, US; 5. Québec Heart and Lung Institute, Université Laval, Québec, QC, Canada.

DNA methylation (DNAme) is an epigenomic modification with important roles in gene regulation and disease susceptibility. Most genome-wide DNAme studies (EWAS) of multifactorial disease traits use targeted arrays or enrichment methodologies preferentially covering CpG-dense promoter regions to characterise sufficiently large samples. However, we and others have shown that disease-linked DNAme sites are depleted in these regulatory elements (i.e. promoters) due to their static nature. In order to assess DNAme profiles more efficiently at dynamic sites, we recently developed a customizable, cost-effective approach, MethylC-Capture Sequencing (MCC-Seq), for sequencing functional methylomes while simultaneously providing genetic variation information. Using a custom adipose tissue (AT) panel design covering ~2.5M CpGs and ~1.3M SNPs, we previously established that MCC-Seq is as accurate as alternative approaches for DNAme profiling by systematic comparisons with Whole-Genome Bisulfite Sequencing (R=0.97), Illumina 450K array (R=0.96) and Agilent SureSelect (R=0.99). We also showed that MCC-Seq accurately calls 96% of heterozygous SNPs. Here, we apply MCC-Seq in an EWAS of metabolic disease-related traits in 105 AT samples from deeply phenotyped obese individuals (BMI>40kg/m2) undergoing gastric surgery. Focusing on triglyceride (TG) levels, the most variable trait in our cohort, we identified 3224 and 649 nominally significant CpGs with permutation p-values of ≤0.001 and ≤0.0001, respectively. Overlapping these CpGs with 18 AT-specific chromatin states (NIH Roadmap Consortium) that were recently published, we noted significant enrichment of TG-associated CpGs within active enhancer states (EhnA1, EhnA2) at p≤0.001 (1.7-fold; p< 2.2x10-16) that was strengthened at p≤0.0001 (2.1-fold; p=4.8x10-11). In contrast, depletion in promoter regions (TssA, TssFlnk, TssFlnkU, TssFlnkD) was observed at both p-value cutoffs. Next, we looked at TG-associated (p≤0.0001) CpGs that also demonstrated significant-association (p≤0.001) with at least one other metabolic trait in an effort to pinpoint sites highly informative of metabolic disease states. Of the 649 TG-associated sites evaluated, 30 showed significant-associations to another metabolic trait at p≤0.001, mainly with HDL-cholesterol (28/30 CpGs). The localization of these sites within the 18 chromatin states revealed improved enrichment within active enhancer states (5.2-fold; p=9.2x10-6). Our study approach enables efficient and cost-effective cataloguing of functional and disease-relevant epigenetic and genetic variants, representing a practical approach for large-scale EWAS. Our results further highlight that such variants are enriched in active enhancers of disease-relevant cell types, which have previously been overlooked due to their underrepresentation on commonly used targeted arrays.