Context-dependent features of transcriptomic landscapes in pregnant mother-neonate dyads of preeclampsia

Article information

Clin Exp Pediatr. 2026;.cep.2025.02565

Publication date (electronic) : 2026 February 19

doi : https://doi.org/10.3345/cep.2025.02565

Yu-Chun Cheng, MD^1,*, Yun-Ju Lai, MD^2,*, Wei-Shiung Lian, PhD^3,4, Ching-Chang Tsai, MD², Hsin-Hsin Cheng, MD², Hong-Ren Yu, MD, PhD¹, Mao-Meng Tiao, MD¹, Jiunn-Ming Sheen, MD, PhD¹, Ying-Lun Hsu, MD¹, Feng-Sheng Wang, PhD^,3,4, I-Chun Lin, MD, PhD^,1

¹Department of Pediatrics, Kaohsiung Chang Gung Memorial Hospital and Chang Gung University, College of Medicine, Kaohsiung, Taiwan

²Department of Obstetrics and Gynecology, Kaohsiung Chang Gung Memorial Hospital and Chang Gung University, College of Medicine, Kaohsiung, Taiwan

³Core Laboratory for Phenomics and Diagnostics, Department of Medical Research, Kaohsiung Chang Gung Memorial Hospital and Chang Gung University, College of Medicine, Kaohsiung, Taiwan

⁴Center for Mitochondrial Research and Medicine, Kaohsiung Chang Gung Memorial Hospital and Chang Gung University, College of Medicine, Kaohsiung, Taiwan

Corresponding author: I-Chun Lin, MD, PhD. Department of Pediatrics, Chang Gung Memorial Hospital-Kaohsiung Medical Center; Graduate Institute of Clinical Medical Sciences, Chang Gung University, College of Medicine, 123, Ta-Pei Rd., Niao Sung Hsiang, Kaohsiung County 83301, Taiwan Email: uc22@cgmh.org.tw

Co-corresponding Author: Feng-Sheng Wang, PhD. Department of Medical Research, Kaohsiung Chang Gung Memorial Hospital, Taiwan; No. 123, Dapi Rd., Niaosong Dist., Kaohsiung City 83301, Taiwan Email: wangfs@ms33.hinet.net

^*

These authors contributed equally to this study as co-first authors.

Received 2025 October 26; Revised 2025 December 17; Accepted 2025 December 23.

Abstract

Introduction

Preeclampsia (PE) is a serious gestational disorder, with high morbidity and mortality for the pregnant mothers and their fetuses, affecting 2% to 8% of all pregnancies [1]. PE impacts the long-term cardiometabolic health of the pregnant mother and child after pregnancy [2-6]. Those with a history of PE have a 2- to 4-fold-higher risk of early coronary artery calcification [7], premature atherosclerosis [8], hypertension [9], and heart failure [10], independent of traditional cardiovascular factors such as age and body mass index. Prenatal exposure to PE is also a risk factor for cardiometabolic diseases (CMDs) [3,11]. However, the pathogenesis underlying the impacts of PE on CMD in the affected children and mothers remains unknown.

In the past decades, extensive evidences highlight the importance of developmental origins of health and disease, suggesting that the environment in the uterus and the mothers’ illness during the fetal development may make these grownup children more susceptible to certain diseases in the future [12,13]. Endothelial dysfunction occurs starting during late gestation in gestational hypertension [14]. We have found that PE may have neonatal cardiovascular manifestations at birth and that coronary dilatation is associated with PE-related endothelial inflammation as well as perinatal mortality [15,16]. Vascular dysfunction in pulmonary and systemic circulation occurs after birth in infants of maternal PE [17]. Children and young adults born to mothers with PE tend to have higher blood pressure than those born to normotensive mothers [18].

PE and CMDs are multifactorial diseases, and PE plays an independent role on CMD development. Gestation is the critical period shared by the pregnant women with their fetuses; both the women and infants are at risk of developing CMDs after PE. Thus, we tested our hypothesis that certain genes are commonly affected in pregnant mother-neonate dyads of PE after gestation. We used RNA sequencing and bioinformatics to analyze the transcriptomic profiles of blood leukocytes from the pregnant women and neonates and those of umbilical endothelial cells. The results were validated, and the clinical correlations were explored.

Methods

Results

The study population included 44 pregnant mother-neonate dyads in each group (Table 1). Those with PE had significantly higher maternal body mass index and antepartum systolic, diastolic, and mean blood pressure (MBP) than the normotensive individuals. Neonates born to mothers with PE had significantly lower GA, birth body weight (BBW), birth length, birth head circumference, as well as a higher percentage of small-for-GA and Cesarean section deliveries than neonates born to those with normotension. Eight of the PE pregnant women (18.18%) had taken aspirin prophylactics since a median GA of 12.50 (IQR, 11.25–26.50) weeks. Approximately 18% of the PE pregnant women had either type I and II diabetic mellitus or gestational diabetic mellitus.

Table 1.

Basic profiles of study subjects and maternal and perinatal factors of normotension versus preeclampsia groups

1. Transcriptomic profile of MBLs

The DEGs between the groups were assessed using t tests with criteria of P<0.05 and |log2FC|≥2. The results are depicted as a volcano map and heatmaps with fold changes in the selective genes of MBLs as shown in the Fig. 1A and B, respectively. The 224 DEGs of the MBLs were enriched in the GO pathways, including those related to positive regulation of cell proliferation, immune response, inflammatory response, negative and positive regulation of smooth muscle cell proliferation, extracellular exosome, extracellular region, serine-type endopeptidase activity, and cytokine activity (Fig. 1C–E). The KEGG enrichment analysis showed that the DEGs were enriched in pathways including those associated with cell adhesion molecules, complement and coagulation cascades, and tumor necrosis factor signaling (Fig. 1F).

Fig. 1.

Identification of differentially expressed genes (DEGs) in the maternal blood leukocytes (MBLs). (A) The volcano maps of DEGs indicated 116 upregulated genes and 108 downregulated genes of MBLs in preeclampsia (PE). (B) The heatmaps with fold changes of selective DEGs of MBLs. (C–E) Gene Ontology enrichment analysis of DEGs in biological process (BP), cellular component (CC), and molecular function (MF). (F) Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis of DEGs. SMC, smooth muscle cell; TNF, tumor necrosis factor.

2. Transcriptomic profile of CBLs

The DEGs of the CBLs were assessed. The results are depicted as a volcano map and heatmaps with fold changes in the selective genes in Fig. 2A and B, respectively. The 177 DEGs of the CBLs were enriched in GO pathways, including those related to angiogenesis, platelet degranulation, blood coagulation, extracellular exosomes, guanyl-nucleotide exchange factor activity, and serine-type endopeptidase inhibitor activity (Fig. 2C–E). The DEGs of the CBLs in the KEGG enrichment analysis were enriched in pathways associated with the hematopoietic cell lineage as well as complement and coagulation cascades (Fig. 2F).

Fig. 2.

Identification of differentially expressed genes (DEGs) in cord blood leukocytes (CBLs). (A) Volcano maps of DEGs showing 66 upregulated genes and 111 downregulated genes of CBLs in preeclampsia. (B) Heatmaps with fold changes of selective DEGs of CBLs. (C–E) Gene Ontology enrichment analysis and (F) Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis of the DEGs. MF, molecular function.

3. Overlapping DEGs and mRNA expression levels between MBLs and CBLs

Venn diagram analysis revealed that the DEGs overlapping between MBLs and CBLs included ADAM metallopeptidase with thrombospondin type 1 motif 2 (ADAMTS2), arachidonate 15-lipoxygenase type B (ALOX15B), beta-1,3-glucuronyltransferase 1 (B3GAT1), gamma- aminobutyric acid A receptor epsilon (GABRE), kinesin family member 21A (KIF21A), lamin A/C (LMNA), microtubule-associated serine/threonine kinase family member 4 (MAST4), matrix metallopeptidase 8 (MMP8), oncostatin M (OSM), and solute carrier family 12 member 1 (SLC12A1) (Fig. 3A). These genes were enriched in KEGG pathways related to the extracellular matrix, chloride transmembrane transport, negative regulation of cell proliferation, multicellular organism catabolic processes, and collagen catabolic processes (Fig. 3B).

Fig. 3.

Overlapping differentially expressed genes (DEGs) of maternal blood leukocytes (MBLs) and cord blood leukocytes (CBLs) and the mRNA expression levels of the selective genes in preeclampsia (PE). (A) Venn diagram and (B) Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis of the overlapping DEGs of the MBLs and CBLs. (C and D) mRNA levels of overlapping genes were validated using real-time reverse transcription polymerase chain reaction and compared between the normotension (Normo) and PE groups in the MBLs and CBLs, respectively. Horizontal lines on violin plots indicate median and interquartile range. *P<0.05, **P<0.01, and ***P<0.001 compared by the Mann-Whitney U test.

Further mRNA validation showed that the PE group had significantly higher MBL levels of ADAMTS2 and LMNA; higher CBL levels of ADAMTS2, KIF21A, GABRE, LMNA, MAST4, and SLC12A1; lower MBL levels of GABRE; and lower CBL levels of MMP8 and OSM than the normotension group (Fig. 3C and D, Supplementary Fig. 1). The mRNA levels of ADAMTS2 in the MBLs and CBLs, and of ALOX15B, GABRE, and LMNA in the CBLs in the PE group negatively correlated with GA and BBW (Table 2). The CBL levels of ADAMTS2, ALOX15B, GABRE, and MAST4 positively correlated with MBP (Table 2). The CBL level of MMP8 positively correlated with GA and BBW and negatively correlated with maternal MBP (Table 2). The mRNA levels of LMNA in the MBLs and MAST4 in the CBLs significantly determined the risk of PE in the adjustment model for GA (Table 3). Multivariate analysis in the PE-adjustment model revealed that the MBL level of ADAMTS2 and the CBL levels of ADAMTS2, GABRE, and MMP8 independently determined the GA and BBW, and that the MMP8 level in CBLs independently determined maternal MBP (Table 4).

Table 2.

Spearman correlations (r) of mRNA expression levels in MBLs and CBLs with GA, BBW, and maternal MBP individually in the normotension and preeclampsia groups

Table 3.

Binary logistic regression in adjustment model of gestational age for independent factors of the mRNA expression levels to predict preeclampsia versus normotension group

Table 4.

Multivariable analysis of independent factors of mRNA expression levels of leukocytes to determine neonatal BBW, GA, and maternal MBP by stepwise linear regression in adjustment of PE

4. Transcriptomic profiles and mRNA expression levels in the UAECs and UVECs

We used the same method described in the prior section to assess the DEGs in the UAECs and UVECs. The results are depicted as volcano and heatmaps of selective genes for UAECs (Fig. 4A and B, respectively) and UVECs (Fig. 5A and B, respectively). The 168 UAEC DEGs were enriched in GO pathways, including those related to cardiocyte differentiation, heart morphogenesis, cell-cell adhesion via plasma membrane adhesion molecules, cardiac muscle cell development, basolateral and cytoplasmic side of plasma membrane, guanosine triphosphate (GTP) binding, GTPase and metallopeptidase activity, and transmembrane transporter activity (Fig. 4C–E). The UAEC DEGs were enriched in the KEGG pathways including those associated with axon guidance, hematopoietic cell lineage, gap junctions, and the nuclear factor kappa B signaling pathway (Fig. 4F). The 138 DEGs of the UVECs were enriched in GO pathways, including those related to leukocyte cell-cell adhesion, regulation of histone modification and chromatin organization, clathrin-coated vesicle, transport vesicle membrane, SNARE binding, and 3',5'-cyclic-GMP phosphodiesterase activity (Fig. 5C–E). The DEGs of UVECs were enriched in the KEGG pathways associated with cell adhesion molecules, calcium signaling, intestinal immune network for IgA production, renin secretion, and Th1 and Th2 cell differentiation (Fig. 5F). Only 6 DEGs overlapped between UAECs and UVECs, such as DEAD/H-box helicase 11 like 2 (DDX11L2, a pseudogene). Almost no DEGs overlapped between UACEs and blood leukocytes and between UVECs and MBLs (Supplementary Table 2).

Fig. 4.

Identification of differentially expressed genes (DEGs) in the umbilical arterial endothelial cells (UAECs). (A) Volcano maps of DEGs indicative of 77 upregulated genes and 91 downregulated genes of UAECs in preeclampsia (PE). (B) Heatmaps with fold changes in the selective DEGs of UAECs (n=3 in the normotensive [Normo] group; n=4 in the PE group). (C–E) Gene Ontology enrichment analysis, and (F) Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis of the DEGs. BP, biological process; CC, cellular component; MF, molecular function; GTP, guanosine triphosphate; cGMP-PKG, cyclic guanosine monophosphate-protein kinase G.

Fig. 5.

Identification of differentially expressed genes (DEGs) in the umbilical venous endothelial cells (UVECs). (A) Volcano maps of DEGs showing 80 upregulated genes and 58 downregulated genes of UVECs in preeclampsia (PE). (B) Heatmaps with fold changes of selective genes of UVECs (n=3 each group). (C–E) Go enrichment analysis, and (F) Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis of the DEGs. BP, biological process; CC, cellular component; MF, molecular function.

Further validation of some DEGs from UAECs revealed that the PE group had significantly higher mRNA levels of FAT Atypical Cadherin 3 (FAT3) and solute carrier family 25 member 18 (SLC25A18) in the UAECs, and significantly lower DDX11L2 levels in their UVECs than the control group (Fig. 6A and B). Arterial endothelial levels of SLC25A18 negatively correlated with maternal DBP in both groups (Fig. 6C). The arterial endothelial levels of SLC6A15 negatively and positively correlated with maternal DBP in the normotension and PE group, respectively (Fig. 6D).

Fig. 6.

Messenger RNA expression levels of umbilical arterial and venous endothelial cells (UAECs and UVECs, respectively). (A and B) mRNA levels of some genes, including DDX11L2, FAT3, PIEZO2, SLC25A18, and SLC6A15, were assessed in UAECs (n=8 each group) and UVECs (n=9 in normotension [Normo] group; n=13 in preeclampsia [PE] group) using real-time reverse transcription polymerase chain reaction. The horizontal lines on the violin plots indicate medians and interquartile ranges. *P<0.05 between the 2 groups by Mann-Whitney U test. (C and D) Spearman correlations between the cord blood leukocyte (CBL) mRNA levels of SLC25A18 and SLC6A15 and maternal diastolic blood pressure (DBP) in the Normo and PE groups.

Discussion

In this observational study, we demonstrated the characteristic patterns of transcriptomic profiles in the pregnant mother-neonate dyads after PE. Our results suggest that PE context-dependently affects pregnant mother-neonate dyads through multiple signaling pathways. Furthermore, ADAMTS2, GABRE, and LMNA may be the DEGs of peripheral leukocytes that are most commonly affected in pregnant mother-neonate dyads. The levels of ADAMTS2 in the pregnant mother as well as neonatal ADAMTS2, GABRE, and MMP8 independently determined neonatal BBW and GA. Neonatal MMP8 levels independently determined pregnant mother MBP. Additionally, pregnant mother LMNA and neonatal MAST4 levels independently determined the development of PE. In contrast, the transcriptomic profiles of UAECs and UVECs differed from each other and from those of MBLs and CBLs.

Until now, PE transcriptomic profiling has been conducted, mostly using placenta tissue [28], and maternal blood to early predict PE risk [29]. Suvakov et al. [30] reported 29 commonly enriched pathways among PE, hypertension, and heart failure: inflammation, metabolism, angiogenesis, and the renin-angiotensin system. Similarly, our maternal transcriptomic profile was enriched in inflammation, immune, and smooth muscle cell proliferationrelated signaling pathways, and some pathways related to extracellular matrix and collagen catabolic processes overlapped between mothers and neonates [30]. While comparing our data with those from another study result that reported 250 DEGs in PE placentas from 3 combined datasets [28]; we identified several common pathways, but no DEGs, between the studies.

Ten DEGs overlapped between maternal and neonatal leukocytes, of which LMNA and MMP8 [31], as well as the ADAMTS [32] and GABA signaling pathway [33,34], involved in PE. However, little is known about the relationships between MAST4, KIF21A, or B3GAT1 and PE or CMDs. LMNA, encoding a part of proteins making up of nuclear lamina, is associated with cardiomyopathy, hypertension [35], premature atherosclerosis [36], and involved in PE progression and risk [37,38], and serum lamin A levels are lower in pregnant women with PE or with fetal congenital heart disease [38]. In contrast, we found that the LMNA mRNA levels were substantially higher in pregnant mothers and neonates affected by PE than in the controls. The higher the neonatal LMNA level, the lower the neonatal GA and BBW in the PE group but the LMNA level in PE mothers was independent of GA. Thus, further studies are warranted to determine how LMNA influences the cardiometabolic phenotype in PE, considering the severity of PE and the potential GA-dependent expression of LMNA. MMP8 is involved in breakdown of extracellular matrix, and is associated with blood pressure (BP) [39] and the severity of metabolic syndromes [40]. DNA hypomethylation of MMP8 was observed in the omental arteries of PE [31]. Our results also showed that PE mothers tended to have higher MMP8 mRNA levels and significantly higher plasma MMP8 levels, partly similar to the results of a previous study (Supplementary Fig. 2) [41]. In contrast, the MMP8 mRNA and plasma levels were lower in the PE neonates in our study. The lower the neonatal MMP8 level, the younger the GA, the lower the BBW, and the higher the maternal blood pressure, implying more severe PE. ADAMTS2, encoding preproprotein to generate mature procollagen N-proteinase, is involved in cardiovascular physiology and disease (e.g., antiangiogenic properties, capability to reduce endothelial proliferation and cardiac hypertrophy) [42]. ADAM and ADAMTS family play an important role in normal pregnancy [43]. Our result displayed that PE mother-neonate dyads had higher ADAMTS2 levels that were negatively correlated with GA. GABRE is a gene encoding the epsilon subunit of the gamma-aminobutyric acid (GABA) A receptor, modifying GABA inhibitory network. The previous studies indicated that the α3 subunit of GABA receptor is upregulated and the pi subunit is downregulated in PE placenta [33,34], but the relationship between GABRE and PE is little known. For a long time, GABA is believed to lower BP and has a cardiovascular protection [44,45], and able to protect and regenerate islet beta cells [46]. In our study, the trends of GABRE levels in mother-neonate dyads were opposite, which may reflect a potential cause-effect after PE exposure. Taken together, much evidence shows the association of aforementioned DEGs with PE, hypertension, and the severity of metabolic syndromes. Therefore, the precise roles of these DEGs in the grownup PE offspring warrant further exploration.

PE strongly affects the structure of umbilical vessels with apoptotic and disrupted arterial endothelium, especially in severe PE [47,48]. We found that the transcriptomic profiles of UAECs and UVECs were mostly different. One reason for these differences may be that the UAECs experience higher blood pressure in PE than UVECs, and such shear stress contributes to additional endothelial dysfunction [49]. Conversely, UVECs may be more affected by PE-related oxidative and proinflammatory mediators as well as all placenta-releasing substances. Heart-related signaling pathways (e.g., cardiocyte differentiation, heart morphogenesis, and cardiac muscle cell development) were more involved in the transcriptomic profile of the UAECs than in those of CBLs and UVECs. SLC25A18 showed a negative correlation with DBP in both groups, while SLC6A15 exhibited an opposite trend—negative in normotensive and positive in preeclamptic pregnancies—suggesting a possible alteration in metabolic regulation under hypertensive conditions. However, the mRNA level of endothelial SLC6A15 expression was not statistically validated by quantitative polymerase chain reaction, this observation should be interpreted with caution. PE context-dependently increases the risk of congenital heart disease in neonates [50], highlighting the need to investigate the cell- and tissue-specific effects of PE.

The main strength of our study is identifying the context-dependent transcriptomic profiles in PE as well as some correlations and independent determinants of altered gene expression in perinatal and peripartum features. Our results have implications for clinical practice and research: First, the altered expression of various genes highlights the importance of long-term surveillance of PE to explore the relationships between individual gene expression and cardiometabolic phenotypes for people with a pregnancy history of PE and their adult offspring. A multigene panel based on our hub genes could be developed to investigate the sustained or imprinting effects of PE on CMD development. Second, maternal LMNA levels may be used as early predictors of PE development and the clinical implications. Finally, the connection of ADAMTS2, GABRE, and MMP8 with low BBW may be valuable in CMD risk prediction because low BBW increase the risk of adult CMDs [51].

The limitations of our study are mainly attributable to the prematurity and low BBW of neonates born to mothers with PE. Recruiting healthy GA- or BBW-matched neonates is difficult in clinical practice. We performed logistic regression analyses with adjusted models of GA or PE to eliminate the confounding effects of the strong correlations between GA and the expression levels of several genes. We used data from MBLs and CBLs obtained in 2017; as such, more DEGs may be obtained using updated methodologies as RNA sequencing has advanced since 2017. The generalizability of this study may be limited because all the participants were Taiwanese mothers and neonates of East Asian origin. The development and outcomes of PE and CMDs vary with ethnicity.

In conclusion, gene expression is context-dependent in multiple signaling pathways in the pregnant mother-neonate dyads in PE. Our findings can be used for developing precision medicine for managing future CMD risks in these populations and warrant studies of the long-term pathophysiological influence of these genes on those with PE and their children.

Supplementary materials

Article information Continued