To our knowledge, this is the first study investigating the DNA methylation changes in different trimesters and after delivery. The results provide the supporting evidence that DNA methylation changes during pregnancy, which may be important for maternal adaptation to pregnancy. Meanwhile, the DNA methylation patterns during and after pregnancy were different, implying that puerperium repair may also act through DNA methylation mechanisms. In addition, different patterns of DNA methylation between trimesters were identified in the present study. Therefore, based on our findings, studies related to DNA methylation during pregnancy should specify the time that the samples were obtained, because gestational age is important to interpret the results.
According to the IPA analysis of the 14,018 identified CpG sites, the top canonical pathway involved is the “protein ubiquitination pathway”, which is a reversible process due to the presence of deubiquitinating enzymes that can cleave ubiquitin from modified proteins (Table 2) [20]. Protein ubiquitination is important for normal placental growth and development [21]. Besides, the epigenetic regulatory role of long non-coding RNAs in the ubiquitin proteasome system and collagen remodeling may be related to spontaneous preterm labor and preterm premature rupture of membranes [22–24]. TNFR2 signaling was figured out by IPA analysis as well. TNF-α has also been associated with inflammatory mechanisms related to implantation, placentation, and pregnancy outcome. TNF-α is secreted by immune cells and works by binding to TNFR1 and TNFR2 cell receptors. Elevation of TNF-α is associated with recurrent pregnancy loss, early and severe preeclampsia, and recurrent implantation failure, all “idiopathic” or related to aPL positivity, which implies its important role in maintaining normal pregnancy [25].
In this study, most of the CpG sites were in Group 9 (N = 8,639, 61.63%), which means although DNA methylation changes throughout the gestation period, most of them were subtle (differences of beta value between − 0.02 and 0.02). In addition, there were 4,168 (29.73%) CpG sites in Group 7 (995, 7.1%) and Group 8 (3,173, 22.64%), the two groups with significant DNA methylation changes between the first and second trimesters. This period is the stage for embryonic development and organogenesis [26]. Indeed, the results of the IPA analyses of Groups 7 and 8 in the present study provided the supporting evidence, showing that hematologic system development and function, tissue development, tissue morphology and hematopoiesis were the top systems and functions involved. Some of these systems and functions overlap with those in Groups 1, 2, 4 and 6. In contrast, the IPA analysis of Group 5 yielded very different results. The top five systems and functions involved were nervous system development and function, embryonic development, hair and skin development and function, humoral immune response, and urinary system development and function. Since Group 5 was the group with a significant increase in DNA methylation between the second and third trimesters, further studies are needed to investigate the functional implications between DNA methylation changes in this period and the development and functions of these systems. On the other hand, the results of the IPA analyses of the canonical pathways involved in different groups are quite diverse. Pathways involved in glucose homeostasis were found in Groups 1, 2, 4, 6 and 8, while those involved in the immune system were identified in Groups 5, 6, 7 and 8. These findings suggest that changes in glucose homeostasis and the immune system during pregnancy may be at least partly mediated by DNA methylation mechanisms. For example, our previous study showed that many genes had different methylation patterns in the GDM and non-GDM groups [16]. In addition, hypomethylation of the IL-10 gene in maternal blood and increased plasma concentrations of IL-10 before delivery were noted in women with gestational diabetes [27].
Pauwels et al. noted that the mean global DNA methylation percentage before pregnancy (6.89%) was significantly higher than the mean methylation percentage at 12 weeks of gestation (6.24%; p = .007), 30 weeks of gestation (6.36%; p = .04), and at delivery (6.35%; p = .04) [12]. In our study, the mean methylation degree of the top 1,000 CpG sites with significant changes was the highest in the postpartum period (mean M-value = -0.396), followed by the first trimester (mean GA at 10 weeks, mean M-value = -0.917, p < .001), the second trimester (mean GA at 25 weeks, mean M-value = -0.967, p < .001) and the third trimester (mean GA at 38 weeks, mean M-value = -0.971, p < .001). Both studies showed that the DNA methylation degree was much higher in the non-pregnant state (before pregnancy in Pauwels’ study and after delivery in our study) than that during pregnancy. These findings are in agreement with the concept that more genes are expressed during pregnancy for maternal changes and adaptation. The findings of the present study suggest that puerperium repair may act through DNA methylation mechanisms.
The study has several limitations. The first limitation is the small sample size, which makes it difficult to perform further detailed analysis. However, the present study did prove the concept of “DNA methylation changes during pregnancy” and provided a great picture of these changes. Further studies with more samples are needed to present more detailed or subgroup analyses. The second limitation is that all the pregnant women included were Han Chinese. Further studies of other ethnic groups are required to see if there are racial differences.