Show summary
Epigenome-Wide Association Study of objectively measured moderate physical activity in pregnancy
Nicolas Fragoso-Bargas1,2, Sindre Lee-Ødegård2, Gunn-Helen Moen2,3,4,5, Julia O. Opsahl2, Line Sletner2,6, Anne Karen Jenum2, Paul Franks7, Rashmi B. Prasad8, Leif C. Groop8, Elisabeth Qvigstad1,2, Kåre I. Birkeland2, Kåre R. Richardsen9, Christine Sommer1
1Oslo University Hospital, Oslo, Norway, 2University of Oslo, Oslo, Norway, 3The University of Queensland Diamantina Institute, Brisbane, Australia, 4Norwegian University of Science and Technology, Trondheim, Norway, 5University of Bristol, UK, 6Akershus University Hospital, Oslo, Norway, 7Lund University Genetic and Molecular Epidemiology, Malmö, Sweden, 8Lund University Diabetes Centre, Malmö, Sweden,9Oslo Metropolitan University, Oslo, Norway.
Physical activity (PA) is recommended during pregnancy to reduce the risk of excessive weight gain and gestational diabetes mellitus, and to improve general health. DNA methylation (DNAm) has been proposed as a potential molecular mechanism through which physical activity (PA) mediates the effects on the transcriptome. Here we performed an epigenome-wide association study (EWAS) of objectively measured moderate physical activity in pregnancy, which has not been performed previously.
In EPIPREG (a sub-sample of the population-based STORK Groruddalen study) we quantified DNAm at gestational week 28 (±2 SD) with Illumina’s Infinium Methylation EPIC BeadChip. With SenseWear™ Pro3 Armband, we measured hours per day of moderate PA (3.0 to 6.0 metabolic equivalent for task (MET)) and steps/day. DNAm outliers that were 4 times the interquartile range below or above the 25TH and 75TH percentiles respectively were removed, before the beta values were transformed to M-values. Of the 472 that passed the EWAS quality control, 239 European (EUR) and 109 South Asian women (SA) had two valid days of recorded PA (19.2h per day), and were non-outliers (±4 SD) in their respective ethnic group. We used two approaches to test the association between M-values and moderate PA (hours/day): 1) A trans-ethnic approach using mixed linear models adjusted for blood cell proportions, age, smoking, number of steps, number of valid days (between 2 to 8), and ethnicity as a random intercept, and 2) through separate analyses in EUR and SA using linear modelling adjusted with the before mentioned covariates. We accepted a false discovery rate of 10%.
In the trans-ethnic analysis, only cg08148261 located in the promoter area of LDHB was significantly associated with moderate PA. In Europeans, eight sites were significant: cg19031974 (BSN-AS2), cg19053287 (MCF2L) cg14930390 (MUTED), cg18517050 (CIAPIN1) and cg26191378 (TMUB1), located near promoter areas, cg27160968 (ESRRA) and cg15879426 (c17orf101) located in exon boundaries, and cg10421808 which is unannotated. In South Asians, only cg05999993 in the promoter area of AOAH was associated with moderate PA. Most of the sites showed decreased DNAm at increased moderate PA (hours/day), while in cg10421808, cg14930390 and cg15879426 the opposite was observed.
In conclusion, we found several significant associations between DNAm and moderate PA. Several CpGs were located near gene promoter areas, which could imply altered transcription of the respective genes. These findings suggest that PA may exert changes in DNAm with important transcriptional implications during pregnancy.