Mendelian randomization supports causality between maternal hyperglycemia and epigenetic regulation of leptin gene in newborns

Abstract

Leptin is an adipokine that acts in the central nervous system and regulates energy balance. Animal models and human observational studies have suggested that leptin surge in the perinatal period has a critical role in programming long-term risk of obesity. In utero exposure to maternal hyperglycemia has been associated with increased risk of obesity later in life. Epigenetic mechanisms are suspected to be involved in fetal programming of long term metabolic diseases. We investigated whether DNA methylation levels near LEP locus mediate the relation between maternal glycemia and neonatal leptin levels using the 2-step epigenetic Mendelian randomization approach. We used data and samples from up to 485 mother-child dyads from Gen3G, a large prospective population-based cohort. First, we built a genetic risk score to capture maternal glycemia based on 10 known glycemic genetic variants (GRS10) and showed it was an adequate instrumental variable (β = 0.046 mmol/L of maternal fasting glucose per additional risk allele; SE = 0.007; P = 7.8 × 10(-11); N = 467). A higher GRS10 was associated with lower methylation levels at cg12083122 located near LEP (β = -0.072 unit per additional risk allele; SE = 0.04; P = 0.05; N = 166). Direction and effect size of association between the instrumental variable GRS10 and methylation at cg12083122 were consistent with the negative association we observed using measured maternal glycemia. Lower DNA methylation levels at cg12083122 were associated with higher cord blood leptin levels (β = -0.17 log of cord blood leptin per unit; SE = 0.07; P = 0.01; N = 170). Our study supports that maternal glycemia is part of causal pathways influencing offspring leptin epigenetic regulation.

Keywords: BMI, Body Mass Index; CDA, Canadian Diabetes Association; CHUS, Centre hospitalier universitaire de Sherbrooke; CpGs, CG dinucleotides; DNA methylation; DNAm, DNA methylation; DOHaD, Developmental Origins of Health and Disease; GCT, Glucose Challenge Test; GDM, Gestational Diabetes Mellitus; GRS, Genetic Risk Score; IADPSG, International Association of the Diabetes and Pregnancy Study Groups; IV, Instrumental Variable; MAGIC, Meta-Analyses of Glucose and Insulin-related traits Consortium; MDS, Multidimensional Scaling; MR, Mendelian Randomization; Mendelian randomization; OGTT, Oral Glucose Tolerance Test; SGA, Small for Gestational Age; SNPs, Single Nucleotide Polymorphisms; TSLS, Two-Stage Least Square; fetal programming; gestational diabetes; glycemia; leptin; mQTL, methylation Quantitative Trait Locus; obesity; pregnancy.

Figure 1.

Conceptual model of the 2-step epigenetic MR applied for maternal exposure and fetal outcome. E_m is a maternal exposure — environmental, nutritional, or maternal metabolic status. G_m represents one or more maternal SNPs acting as the instrumental variable representing E_m. DNAm levels are measured at one CpG site in offspring samples (CpG_o). G_o is one or more fetal SNPs acting as the IV to represent DNAm levels at CpG_o. P_o is the offspring phenotype of interest. β_OBS1 represents association between E_m and methylation levels at CpG_o (observational analysis). β_G1 represents association between G_m and E_m (should capture a fair proportion of the variance for a good instrument). β_IV1 represents the effect of G_m on CpG_o (IV analysis). β_OBS2, β_G2, and β_IV2 are analogous to β_OBS1, β_G1, and β_IV1 for step 2.

Figure 2.

Application of 2-step epigenetic MR to test the association between maternal glycemia and neonatal leptin levels using Gen3G data. For Step 1, the exposure (E_m) is measured fasting plasma glucose at 2nd trimester. GRS₁₀ is the IV of the exposure (G_m). DNAm was measured at cg12083122 site (LEP annotation) in cord blood cells (CpG_o). For step 2, CpG_o represent the same methylation site (cg12083122). We were unable to identify a SNP to place as G_o (among 5 SNPs tested) that would be an appropriate IV to represent DNAm at cg12083122 site in cord blood. Offspring outcome (P_o) is cord blood leptin levels (log-transformed).