Dynamic alterations of DNA methylation profiles during pregnancy in maternal blood
In order to illustrate the DNA methylation alterations in different trimesters in maternal blood, we categorized DNA methylation changes in 1st, 2nd and 3rd trimesters to four groups according to the trend. We defined DNA methylation baseline level of 1st trimester as “0”, DNA methylation level increase between trimesters as “1” and DNA methylation level decrease between trimesters as “-1”. Thus, we detected four patterns of DNA methylation profiles (0. 1. -1; 0. 1. 1; 0. -1. 1; 0. -1. -1) in either control or exercise participants (Fig. 2A, Fig. 2B). Red lines indicate significant changes of differentially methylated positions (DMPs) (p < 0.05) and grey lines indicate non-significant changes (Fig. 2A, Fig. 2B).
There were 26573 DMPs in the control group and 15913 DMPs in the exercise group for 0. 1. -1 trend (Fig. 2C). DMPs were defined as both a significant change for 2nd vs. 1st trimester (p < 0.05) and also a significant change for 3rd vs. 2nd trimester (p < 0.05). Among them, 2562 DMPs in control group and 1345 DMPs in exercise group were significantly altered in 0. 1. -1 pattern (Fig. 2C). Similarly, for 0. -1. 1 trend, we found 35363 DMPs (2014 significant DMPs) in the control group and 14635 DMPs (1076 significant DMPs) in the exercise group respectively (Fig. 2C). Interestingly, there were fewer DMPs for 0. 1. 1 and 0. -1. -1 trends both in control and exercise groups (Fig. 2A, Fig. 2B). Only two significant DMPs (cg13769674, cg08530065) in the control and one significant DMP (cg04260608) in the exercise-trained group for 0. 1. 1 trend were identified. Similarly, one DMP (cg08586855) and two DMPs (cg12394567, cg02033213) were found for 0. -1. -1 trend in control and exercise groups, respectively (Fig. 2C).
Differentially methylated regions (DMRs) are genomic regions with different DNA methylation status across the biological samples and regarded as possible functional regions involved in gene transcriptional regulation. Compared to 1st trimester in maternal blood, 17781 DMRs were revealed in 2nd trimester in the control group, among which 1293 DMRs were located in promoters. 6565 DMRs were identified in the exercise group and 615 DMRs were located in promoters. 4530 DMRs and 3145 DMRs were found in 3rd trimester compared to 2nd trimester in control and exercise groups, respectively. 477 DMRs and 280 DMRs were located in promoters accordingly (Fig. 3A).
DNA methylation levels of specific genes were altered by exercise in maternal blood
We next examined the DMPs with opposite trends in control and exercise groups in order to determine whether exercise training affects the DNA methylation trends in maternal blood. We compared the significant DMPs between control and exercise groups with four opposite trend combinations (trends 0. -1. 1 and 0. 1. -1; trends 0. 1. -1 and 0. -1. 1; trends 0. 1. 1 and 0. -1. -1; trends 0. -1. -1 and 0. 1. 1, respectively). There were no significant CpG sites identified in the latter two opposite trend combinations (trends 0. 1. 1 and 0. -1. -1; trends 0. -1. -1 and 0. 1. 1). Three CpG sites and corresponding genes UMAD1 (UBAP1-MVB12-associated (UMA) domain containing 1, cg12309238), RPA3 (replication protein A3, cg12309238), PLAGL2 (PLAG1 like zinc finger 2, cg25811820) and POFUT1 (protein O-fucosyltransferase 1, cg25811820) showed significantly differential trends in control (0. -1. 1) and exercise (0. 1. -1) groups in maternal blood (Fig. 4). Six CpG sites and corresponding genes including SPATA17 (spermatogenesis associated 17, cg17026642), GPATCH2 (G-patch domain containing 2, cg17026642), CEP170 (centrosomal protein 170, cg08258520), MPHOSPH10 (M-phase phosphoprotein 10, cg20854010), MCEE (methylmalonyl-CoA epimerase, cg20854010) as well as MRGPRD (MAS related GPR family member D, cg11903239) demonstrated opposite trends 0. 1. -1 (control) and 0. -1. 1 (exercise) during pregnancy (Fig. 4).
The global DNA methylation profile of cord blood is different from that of maternal blood
Next, we compared the DNA methylation profiles in maternal blood and cord blood. We studied the epigenome-wide variation in DNA methylation with principal component analysis (PCA), which shows that the leading axis of epigenomic variation reflects the tissue type (Fig. 5A). Dots represent blood samples and are colored accordingly. The heatmaps show DNA methylation comparisons between cord blood and maternal blood in different trimesters (Fig. 5B, Fig. 5C, Fig. 5D). The differential genes of maternal and cord blood were then grouped into three categories using Gene Ontology (GO) system: biological process (describing the larger cellular or physiological role carried out by the gene, coordinated with other genes) and cellular component (describing the location in the cell where the gene product executes its function) and molecular function (describing the molecular activity of a gene) (Fig. 5E). Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway demonstrated that “metabolic pathways” was the most significant pathway enrichment in differential genes comparing maternal and cord blood (Fig. 5F). The corresponding proteins of differential genes of maternal and cord blood were analyzed using protein-protein interactions (PPIs) network, whereby the physical contacts of high specificity is established between potential protein molecules (Fig. 5G).
Compared to 3rd trimester in maternal blood, 82676 DMRs were demonstrated in cord blood in control, among which 9464 DMRs were located in promoters. 78318 DMRs were identified in the exercise group and 9130 DMRs were located in promoters (Fig. 3B).
Maternal exercise alters gene-specific DNA methylation levels in cord blood
The methylation microarray data showed that five specific CpG sites cg02878244, cg02819231, cg02505749, cg11660360, cg03084276 were significantly differentially methylated in the exercise group compared the control group in cord blood (Fig. 6A). These CpG sites were located within the CpG islands of Transcription Start Site (TSS) of four specific genes including developing brain homeobox 1 (DBX1), F-box and leucine rich repeat protein 2 (FBXL2), potassium two pore domain channel subfamily K member 9 (KCNK9) and prostaglandin reductase 1 (PTGR1) (Fig. 6C).
KEGG pathway analysis revealed that “aldosterone synthesis and secretion” was the most significant pathway enrichment in cord blood from the exercise versus the control groups (Fig. 6B).