Background In utero exposure to maternal hyperglycemia has been associated with an increased risk for the development of chronic diseases in later life. These predispositions may be programmed by fetal DNA methylation (DNAm) changes that persist postnatally. However, although some studies have associated fetal exposure to gestational hyperglycemia with DNAm variations at birth, and metabolic phenotypes in childhood, no study has yet examined how maternal hyperglycemia during pregnancy may be associated with offspring DNAm from birth to five years of age. Hypothesis Maternal hyperglycemia is associated with variation in offspring DNAm from birth to 5 years of age. Methods We estimated maternal hyperglycemia using the area under the curve for glucose (AUCglu) following an oral glucose tolerance test conducted at 24–30 weeks of pregnancy. We quantified DNAm levels in cord blood (n = 440) and peripheral blood at five years of age (n = 293) using the Infinium MethylationEPIC BeadChip (Illumina). Our total sample included 539 unique dyads (mother–child) with 194 dyads having DNAm at both time-points. We first regressed DNAm M-values against the cell types and child age for each time-point separately to account for the difference by time of measurement for these variables. We then used a random intercept model from the linear mixed model (LMM) framework to assess the longitudinal association between maternal AUCglu and the repeated measures of residuals of DNAm. We adjusted for the following covariates as fixed effects in the random intercept model: maternal age, gravidity, smoking status, child sex, maternal body mass index (BMI) (measured at first trimester of pregnancy), and a binary variable for time-point. Results In utero exposure to higher maternal AUCglu was associated with lower offspring blood DNAm levels at cg00967989 located in FSD1L gene (β = − 0.0267, P = 2.13 × 10–8) in adjusted linear regression mixed models. Our study also reports other CpG sites for which DNAm levels were suggestively associated (P < 1.0 × 10–5) with in utero exposure to gestational hyperglycemia. Two of these (cg12140144 and cg07946633) were found in the promotor region of PRDM16 gene (β: − 0.0251, P = 4.37 × 10–07 and β: − 0.0206, P = 2.24 × 10–06, respectively). Conclusion Maternal hyperglycemia is associated with offspring DNAm longitudinally assessed from birth to 5 years of age.

Plasminogen activator inhibitor (PAI-1) expression has been associated with a higher risk of development of obesity. DNA methylation (DNAm) is an epigenetic mechanism regulating gene transcription and likely involved in the fetal programming of childhood obesity. Our study aimed to assess the associations between PAI-1 gene (SERPINE1) DNAm, plasma PAI-1 levels, and adiposity at five years of age. We analyzed DNAm and anthropometric data from 146 girls and 177 boys from the Gen3G prospective birth cohort. We assessed adiposity using BMI z-scores, waist circumference, total skinfolds, and percentages of total, android, and trunk fat measured by dual-energy radiography (DXA). We estimated blood cell DNAm levels at 15 CpG sites within SERPINE1 using the methylationEPIC array. After correction for multiple testing, we found that lower DNAm in SERPINE1 intron 3 (cg11353706) was associated with greater adiposity levels in girls (waist circumference: r = −0.258, p = 0.002; skinfolds: r = −0.212, p = 0. 013; android fat: r = −0.215, p = 0.015; BMI z-score: r = −0.278, p < 0.001) and that lower DNAm in the SERPINE1 promoter (cg19722814) was associated with higher plasma PAI-1 levels in boys (r = −0.178, p = 0.021). Our study suggests that DNAm levels at the SERPINE1 gene locus are negatively correlated with adiposity, but not with plasma PAI-1 levels, in young girls only.