Integrative genomic analyses in adipocytes implicate DNA methylation in human obesity and diabetes

**Background:** Obesity affects over 650 million people worldwide and is a major risk factor for type 2 diabetes (T2D). DNA methylation, a key epigenetic mechanism, has been implicated in obesity and T2D through pathways including developmental programming, diet, and genetic variation. However, previous studies were limited by tissue selection, cellular heterogeneity, and difficulty assigning causation. Adipocytes are the major cell type in adipose tissue and play critical roles in energy homeostasis and metabolic disease. This study aimed to overcome prior limitations by performing epigenome-wide association studies (EWAS) in isolated human subcutaneous and visceral adipocytes from individuals with extreme obesity and healthy controls.

**Methods:** The authors collected subcutaneous and visceral adipose tissue intraoperatively from people with extreme obesity (mean BMI 44.8 kg/m²) and healthy controls (mean BMI 24.9 kg/m²), matched for age, sex, and ethnicity. Adipocytes were isolated by collagenase digestion. Genome-wide DNA methylation was profiled in 190 samples using Illumina HumanMethylation450 and EPIC Beadchips across discovery and replication cohorts. RNA sequencing was performed in 89 samples from the replication cohort. Integrative analyses included: (i) identification of obesity-associated 5mC sites at FDR < 1% with replication at epigenome-wide significance (P < 1 × 10⁻⁷); (ii) assignment of methylation sites to target genes using promoter overlap, adipocyte capture Hi-C, GeneHancer, and topologically associated domains (TADs); (iii) transcription factor binding motif enrichment; (iv) two-sample Mendelian randomization (MR) using cis-SNPs as instrumental variables in 588 whole adipose tissue samples (TwinsUK) and large-scale GWAS (BMI, WHR, T2D, glycemic and lipid traits); and (v) functional validation via siRNA knockdown and CRISPR-activation in adipocyte models.

**Key Results:** In subcutaneous adipocytes, 4,485 5mC sites were associated with extreme obesity at FDR < 1%, of which 905 replicated in an independent sample (691 sentinel loci). In visceral adipocytes, 445 sites were identified at FDR < 1%, with 220 replicating (173 sentinel loci). Median methylation differences between obese and control were 5.8% (range 1.1–17.9%) in subcutaneous and 7.9% (range 2.9–21.5%) in visceral adipocytes, with systematic hypomethylation in obesity. Only 23 subcutaneous sentinels replicated in visceral adipocytes, indicating marked depot specificity. Sentinels were enriched in enhancer regions and transcription factor binding motifs (e.g., AP-1, KLF, ETS families). Methylation at 121 subcutaneous and 29 visceral sentinels was associated with expression of cis-genes (FDR < 0.01), including genes involved in insulin signaling (IRS2, ADIPOR2), adipogenesis (PRDM16, EBF2), and browning/beigeing. MR analyses provided causal evidence for methylation effects on obesity at 10 subcutaneous and 4 visceral loci, on central adiposity at 12 loci, on T2D at 4 loci, and on glycemic traits at 3 loci. Functional studies showed that silencing Prrc2a or Limd2 in 3T3-L1 adipocytes significantly reduced lipid accumulation (Prrc2a P = 3.5 × 10⁻⁵; Limd2 P = 0.0047), with Prrc2a knockdown reducing Pparg expression. CRISPR-activation confirmed that the differentially methylated region at LIMD2 regulates LIMD2 transcription (>2-fold increase, P < 0.0001).

**Clinical Implications:** This study provides a comprehensive resource of obesity-associated DNA methylation changes in human adipocytes and identifies novel causal links between epigenetic variation and metabolic disease. The depot-specific nature of these changes suggests that subcutaneous and visceral adipose tissues have distinct epigenetic regulatory mechanisms. The identification of PRRC2A and LIMD2 as functional targets of obesity-associated methylation, with roles in adipogenesis and lipid handling, offers new avenues for therapeutic development. The findings underscore the importance of studying cell-type-specific epigenomic variation and extreme trait sampling to uncover disease mechanisms. Limitations include the modest sample size for MR analyses (only 191 subcutaneous and 34 visceral sentinels had suitable cis-SNPs), the use of whole adipose tissue for mQTL discovery, and the need for further functional characterization of the many novel loci identified.