Background: Maternal food intake is an important environmental determinant that could plausibly act via epigenetic factors to affect foetal adaptation to the prevailing environmental conditions. As suggested by the "Barker" hypothesis this mechanism might prepare the foetus for optimal functioning in the specific environment. Maternal nutrition is particularly relevant in affecting methylation. Animal studies have highlighted epigenetic changes following under-nutrition or altered dietary intake. Dietary manipulation of pregnant female rodents alters coat colour phenotype and is associated with epigenetic modifications in the agouti gene . Folate and choline deficiencies in maternal diet in rodents have also been shown to be associated with altered methylation patterns , . Maternal exposure to a high-fat diet in mice has been associated with increased body length and higher insulin sensitivity and it has been suggested this effect might too be mediated via epigenetic changes .In humans, studies on the effects of the Dutch Famine have found specific methylation (either hypo- or hyper-) patterns of different genes following peri-conceptional exposure to famine (but not exposure later in gestation). Maternal stress in utero can also affect methylation patterns, and epigenetic mechanisms have been suggested as mediators of the transgenerational effect of stress in animal models .

Maternal ED have a major influence on nutrition and therefore there is a strong biological plausibility that ED will have an effect on methylation patterns , however methylation patterns in the offspring of women with ED have not been previously studied.

Aim: To determine if maternal ED in pregnancy affect DNA methylation in the offspring.

Methodology:

Outcomes: In ALSPAC genome-wide DNA methylation data generated using the Illumina HumanMethylation450 BeadChip array is currently being generated on 1,000 mother-child pairs at multiple time-points (birth, age 7, age 17, mother antenatal and mother 17 years after delivery). These data will provide a reference "unexposed" population with whom we will compare cord blood methylation in offspring of women who experienced ED. The 'case' group (n=100) will be drawn from ALSPAC and a comparison of cases and controls will be undertaken to identify differentially methylated regions (DMRs)of interest. Arrays will be run at the ALSPAC lab facility using an identical protocol to that used to generate the unselected population data and incorporating controls to account for any potential batch effects.

DMRs of interest will then be selected and analysis of these target loci will be undertaken in additional ALSPAC samples. This replication will be undertaken using IlluminaVeraCode technology which allows custom multiplex analysis of 96 CpG sites in a single assay. It provides a cost effective way of exploring epigenetic variation in large sample sets. In addition to investigating the influence of maternal ED on offspring methylation, the persistence of epigenetic variation observed throughout childhood will be appraised by analysing the same DMRs in samples from children at age 7 and 15-17 years in ALSPAC and comparing these to the data measured at birth.

Statistical analyses: Linear regression modelling including relevant covariates and random batch effects will be applied to identify DMRs in genome-wide IlluminaHumanMethylationBeadChip data by modelling the methylation level at each individual CpG site as quantified by the 'beta value'. Data will also be analysed using logistic regression (or a Kruskal Wallis test) when data can be categorised (e.g. ED vs control). Replication data generated from VeraCode will be analysed using the same statistical approaches.