DNA methylation of insulin signalling pathways in human primary muscle stem cells is associated with HOMA2-IR in older adults

Context: While ageing is associated with increased insulin resistance (IR), the molecular mechanisms underlying increased IR in muscle, the primary organ for glucose clearance, have yet to be elucidated in older individuals. As epigenetic processes are suggested to contribute to development of ageing-associated diseases, in human primary muscle stem cells (myoblasts) from community dwelling older individuals we investigated whether differential DNA methylation was associated with IR. Methods: We measured DNA methylation (Infinium HumanMethylationEPIC BeadChip) in myoblast cultures from vastus lateralis biopsies (119 male/females, mean age 78.24 years) from the Hertfordshire Sarcopenia Study extension (HSSe) and examined differentially methylated Cytosine phosphate Guanine (CpG) sites (dmCpG), regions (DMRs) and gene pathways associated with HOMA2-IR (derived from fasting insulin/glucose) and HbA1c. Results: 38 dmCpGs (false discovery rate (FDR)<0.05), and 8 DMRs (Stouffer<0.05) were associated with HOMA2-IR; 6 dmCpGs and 3 DMRs were robust to adjustment for BMI, including DMRs within T-box transcription factor (TBX1) and nuclear receptor subfamily-2 group F member-2 (NR2F2). HOMA2-IR associated dmCpGs were enriched in genes linked with skeletal muscle development and insulin signalling. 49 dmCpGs were associated with HbA1c, with 48 robust to adjustment for BMI. The only dmCpG associated with both HOMA2-IR and HbA1c was cg13451048, located within Exoribonuclease Family Member 3 (ERI3). HOMA2-IR and HbA1c dmCpGs were not associated with accelerated epigenetic ageing. Conclusions: These findings suggest differential DNA methylation in muscle stem cells may contribute to age-related decline in glycaemic control. The identification of dmCpGs, associated genes and pathways may provide novel targets for pharmacological intervention. A total of 145 samples had sufficient DNA for methylation analysis. This included 7 technical inter-chip and intra-chip replicates. Replicates were removed after normalization but before inference (the replicate with the lowest MAD score was removed). Nineteen samples showed aberrant methylation densities and MAD scores lower than -5 were removed from subsequent analysis. This resulted in 119 samples which were taken forward for further analysis. A total of 7 inter-chip and intra-chip replicates were included in the analysis.