WP leader: Vardhman Rakyan
Epigenomic studies applied to complex disease and complex phenotypes are generating insights into the aetiology and pathogenesis of type 2 diabetes. Epigenetic modifications, such as DNA methylation, can be studied across the genome and may vary according to genetic architecture and the environment. In early development, the epigenetic state of a developing embryo and fetus undergoes widespread changes (such as demethylation and remethylation) and as such, changes in the periconceptual and pregnancy environment may have significant changes in the epigenomic profile of a fetus at birth. We hypothesised that the epigenome of South Asian offspring could be modified by delivery of the targeted intervention to mothers in WP3 and/or vary according to maternal and fetal outcomes. Furthermore, we considered that epigenomic differences in these offspring could underlie the molecular mechanisms by which fetal programming of future type 2 diabetes might occur.
382 cord blood samples from the WP3 intervention trial were selected for the epigenetic discovery experiment. Extracted DNA was bisulphite-converted and hybridized to the Illumina Infinium HumanMethylation 450 Beadchip which allows genome-wide interrogation of approximately 450,000 pre-defined CpG sites, using fluorescent beads to detect methylated vs. unmethylated CpG sites. The data was subsequently quality control checked and 351 samples were found suitable for ongoing analysis.
Genome-wide analysis of methylation differences in cord blood was performed according to maternal booking BMI (underweight vs. normal weight vs. overweight/obese). Maternal and neonatal phenotypic characteristics and assays were used in quantitative trait analyses with methylation in cord blood. Maternal phenotypes studied included vitamin D status and vitamin B12 status at each antenatal visit as well as delta change of each. Other maternal phenotypes analysed were maternal gestational diabetes status and maternal BMI. Neonatal quantitative traits included cord blood assays of serum vitamin D, vitamin B12, homocysteine, folate (serum and red blood cell), leptin, ferritin, triglycerides, total cholesterol, HDL, blood glucose, serum insulin. Other measures analysed as quantitative variables included neonatal measurements: birth weight, birth length, ponderal index, mid-upper arm circumference, gestational age at delivery and placental weight.
Current work is investigating the potential confounding influence of maternal age, parity and gestational age on these methylation differences. Robust methylation variants, unconfounded by other influences, will be taken forward into replication experiments.
We have identified methylation variation associated with specific birth phenotypes relating to fetal growth and nutrient transport. Methylation analysis was put through stringent quality control checking and control for false discovery in the context of multiple testing. We are reassured by the commonality of methylation variants across related phenotypes (e.g. fetal size) as well as the preliminary replication of methylation variants in specific genes already identified in published research on birth weight and DNA methylation (e.g. HIF3A). Additional work is being performed to understand whether there are confounding influences on these DNA methylation differences, e.g. the effect of gestational age at delivery, or whether reference-based cell count correction identifies methylation differences according to cellular composition of the cord blood samples.