The characteristics of participants
Among the total recruited participants, the prevalence of preterm LBW and FT-LBW was 3.4% and 6.1%, respectively. Table 1 shows the characteristics of participants in the full-term study. Of the total number, 25 were FT-LBW and 361 were FT-NBW. The pre-pregnancy BMI of FT-LBW mothers was significantly lower than that of FT-NBW mothers (P = 0.044). Birth weight of infants was also significantly different between FT-LBWs and FT-NBWs (P < 0.001). Statistical difference was not found in maternal age, GWG, socioeconomic status, household income or maternal educational level between groups. The gestational age (P < 0.001) and infant’s gender (P = 0.023) were different between FT-LBWs and FT-NBWs. After the adjustment of maternal age and GWG, pre-pregnancy BMI was significantly associated with FT-LBW (adjusted odds ratio: 0.77; 95% confidence interval: 0.63–0.93; P = 0.007).
The characteristics of participants for DNA methylation analysis
The pre-pregnancy BMI of FT-LBW mothers was significantly lower than that of FT-NBW mothers (FT-LBWs, 17.1 ± 1.6 kg/m2 vs. FT-NBWs, 22.2 ± 2.9 kg/m2; P = 0.013). Birth weight was also significantly different between FT-LBWs and FT-NBWs (2,344 ± 18 g vs. 3,011 ± 7 g; P < 0.001). Less significant difference was found in maternal age, GWG, or gestational age. (Data shown in Table S1)
Identification of DMGs
A total of 583 hypermethylated reference sequences were determined to correspond to 483 hyper-differentially methylated genes (DMGs) and 58 hypomethylated reference sequences were determined to correspond to 35 hypo-DMGs in FT-LBW infants, compared to the FT-NBW infants (Figure. 1).
Enrichment analysis and functional annotation
To investigate the effect of birth weight on fetal biological processes, we performed biological enrichment analysis of hyper-DMGs and hypo-DMGs in FT-LBW using the GOTERM_BP algorithm. The 483 hyper-DMGs identified were annotated to 11 biological processes in FT-LBW infants. The functions of hypomethylated genes could not be verified. Table 2 shows GO term and genes of each enrichment biological process. Furthermore, these 11 biological processes could be further categorized as “immune system,” “DNA metabolism and repair,” and “organism growth and organization”.
Of these, five enrichment biological processes were classified into the “immune system” category: clearance of foreign intracellular DNA by conversion of DNA cytidine to uridine (APOBEC3A_B and APOBEC3A), macrophage differentiation (CASP8, CASP10, and BMP4), apoptotic mitochondrial changes (BH3, IFIT2, and AIFM2), negative regulation of viral genome replication (IFI16, ADAR, et al.), and negative regulation of inflammatory response (NLRP12, SHARPIN, et al).
Two enrichment biological processes were identified into the “DNA metabolism and repair” category: purine nucleotide metabolic process (GMPR2, NME/NM23, and GUK1) and nucleotide-excision repair (HUS1, ERCC1, et al).
Another four enrichment biological processes were classified into the “organism growth and organization” category: ovarian cumulus expansion (EREG and BMPR1B), positive regulation of multicellular organism growth (GHR, GHRL, et al), osteoblast differentiation (CREB3L1, FBL, et al), and protein phosphorylation (STRADB, BMPR1B, et al).