Upregulation of microRNA-3687 promotes gestational diabetes mellitus by inhibiting follistatin-like 3

Introduction

Gestational diabetes mellitus (GDM) presents considerable dangers throughout pregnancy, impacting both maternal and neonatal health and elevating the probability of long-term consequences, including type 2 diabetes, obesity, and cardiovascular disease in both the mother and child. GDM impacts over 14 % of pregnancies worldwide, demonstrating its widespread prevalence [1]. In China, the prevalence is notably high at 14.8 % (95 % confidence interval: 12.8–16.7 %) [2]. Typically diagnosed after 24 weeks of gestation, GDM can lead to complications earlier in pregnancy, highlighting the urgency for early identification and intervention to mitigate risks. GDM diagnosis relies on oral glucose tolerance tests administered after critical stages of fetal development have passed, limiting early intervention opportunities. The lack of definitive biomarkers for early GDM detection underscores the need for novel diagnostic tools with both diagnostic and prognostic capabilities. Early identification of GDM is crucial for initiating timely interventions to manage blood glucose concentrations and reduce the risk of adverse pregnancy outcomes and other complications.

Non-coding RNAs (ncRNAs) comprise approximately 60 % of the transcribed human genome and are involved in controlling pathways, developmental processes, and pathological states [3], [4], [5], [6]. Based on their length, they can be broadly divided into two types: long ncRNAs (lncRNAs), which are more than 200 nucleotides, and small ncRNAs, including microRNAs (miRNAs), which are usually<200 nucleotides (21–25 nucleotides long). miRNAs are naturally occurring tiny ncRNAs involved in the posttranscriptional modulation of gene expression by attaching themselves to target mRNAs and adjusting their translation [7]. Mature miRNAs bind to the 3′ untranslated regions of mRNAs, influencing their translation [8], [9], [10]. Due to their stability in bodily fluids and tissue-specific abundance, miRNAs hold promise as potential biomarkers for disease. They have been implicated in various aspects of disease pathology, including their diagnostic, prognostic, and predictive roles in conditions like GDM and its associated health consequences [11], [12], [13].

In this study, we developed a cell model of GDM using BeWo cells and identified 220 significantly differentially expressed genes (DEGs) through high throughput sequencing. Under hyperglycemic (HG) conditions (25 mM glucose), we observed a marked upregulation of miR-3687 and a significant downregulation of its target gene, FSTL3, both implicated in lipid metabolism. Real-time polymerase chain reaction (RT-PCR) analysis was employed for further evaluating these findings in placenta samples from GDM patients and controls. The observed findings indicated that follistatin-like 3 (FSTL3) was lower and miR-3687 was considerably increased in GDM placenta samples compared to controls. The dual luciferase reporter test findings established that miR-3687 directly targets FSTL3. The present study advanced our understanding of the pathophysiology of GDM and could provide critical novel insights for the clinical treatment of the condition.

Materials and methods

Results

DEGs in BeWo cells induced by glucose

The present study employed high throughput sequencing to identify 220 crucial DEGs in BeWo cells under high glucose (HG) conditions, comprising 158 upregulated and 62 downregulated DEGs (fold change ≥1.5 or ≤2/3, p<0.01). Principal component analysis (PCA) and volcano plots illustrating the distribution and significance of DEGs were depicted in Figures 1A and B, respectively. A heatmap displaying the top 10 DEGs based on their increment order by p-value was presented in Figure 1C.

Figure 1:

DEGs in BeWo cells after induction by HG conditions compared with LG conditions. PCA (A), volcano plot (B), and heat map (C) of all differential DEGs.

Eight miRNAs exhibited significant differential expression in BeWo cells, all consistently upregulated under HG conditions. Specifically, miR-3687-1 and miR-3687-2 showed significantly higher expression levels in the HG group, implicating their involvement in lipid metabolism (Figure 2A). A literature review established that FSTL3 was a target gene of miR-3687, had high expression in the placenta, and participates in lipid metabolism. Consistent with this, FSTL3 expression was significantly reduced in BeWo cells under HG conditions.

Figure 2:

miR-3687-1 and miR-3687-2 (A) shows significantly higher expression levels in BeWo cells in HG group compared with LG group. A PCA diagram of differential lncRNA (B), and the differential expressions of LINC01556, lncRNAGAS6-AS2, lncRNAMIR31HG, and lncTP53TG1 (C) in Bewo cells in HG group compared with LG group. A heatmap of the differential AS in BeWo cells (D) in HG group compared with LG group.

Furthermore, 87 lncRNA transcripts were significantly differentially expressed in BeWo cells under HG conditions compared to LG conditions. Among these, 54 lncRNAs were upregulated, while 33 were downregulated. Principal component analysis of the differentially expressed lncRNAs was shown in Figure 2B. Many of these lncRNAs have poorly understood functions, but literature searches revealed associations of LINC01556, lncRNAGAS6-AS2, lncRNAMIR31HG, and lncTP53TG1 with type 2 diabetes complications, as depicted in Figure 2C.

Additionally, a total of 1,597 alternative splicing (AS) events were significantly altered in BeWo cells under HG conditions compared to LG conditions, as illustrated in the heatmap in Figure 2D.

Functional analysis of DEGs

The functions of the DEGs and their potential associations with GDM were investigated with KEGG pathway analyses. Analysis of the 158 significantly upregulated DEGs in BeWo cells under HG conditions revealed their involvement in several pathways, including transcriptional misregulation in cancer, amoebiasis, taurine and hypotaurine metabolism, systemic lupus erythematosus, adrenergic signaling in cardiomyocytes, glycosaminoglycan biosynthesis-heparan sulfate/heparin, alcoholism, vasopressin-regulated water reabsorption, serine and threonine metabolism, glycine, and glyoxylate and dicarboxylate metabolism (Table 2, Figure 3A).

Figure 3:

The top 10 KEGG pathways of the significantly upregulated DEGs (A) and downregulated DEGs (B). Top 10 GO pathways of the significantly expressed AS.

Conversely, KEGG analysis of downregulated DEGs highlighted their significant roles in pathways such as small cell lung cancer, toxoplasmosis, mucin-type O-glycan biosynthesis, apoptosis, viral myocarditis, VEGF signaling pathway, arrhythmogenic right ventricular cardiomyopathy, pathways in cancer, ECM-receptor interaction, and hypertrophic cardiomyopathy (Table 3, Figure 3B).

Among the top 40 significantly upregulated DEGs in BeWo cells under HG conditions, ranked by their FPKM values, miR-3687-1 and miR-3687-2 exhibited the highest expression levels among these DEGs, emphasizing their involvement in lipid metabolism.

Table 3:

The top 10 enriched pathways in downregulated DEGs.

Term	Database	ID	Input number	Background	p-Value	Corrected p	Input
Small cell lung cancer	KEGG pathway	hsa05222	2	86	0.006199	0.139378	BIRC7|LAMA2
Toxoplasmosis	KEGG pathway	hsa05145	2	119	0.011615	0.139378	BIRC7|LAMA2
Mucin type-O-glycan biosynethesis	KEGG pathway	hsa00512	1	31	0.043097	0.222517	GALNT8
Apoptosis-multiple species	KEGG pathway	hsa04215	1	33	0.045819	0.222517	BIRC7
Viral myocarditis	KEGG pathway	hsa05416	1	60	0.081892	0.222517	LAMA2
VEGF signaling pathway	KEGG pathway	hsa04370	1	61	0.083204	0.222517	SH2D2A
Arrhythmogenic right ventricular cardiomyopathy (ARVC)	KEGG pathway	hsa05412	1	74	0.100109	0.222517	LAMA2
Pathways in cancer	KEGG pathway	hsa05200	2	397	0.105348	0.222517	BIRC7|LAMA2
ECM-receptor interaction	KEGG pathway	hsa04512	1	82	0.110372	0.222517	LAMA2
Hypertrophic cardiomyopathy , HCM	KEGG pathway	hsa05410	1	83	0.111648	0.222517	LAMA2

Furthermore, the roles of 1,598 differentially expressed AS events were predicted using GO analysis. The analysis indicated that these AS events were crucial in pathways including the regulation of small GTPase-mediated signal transduction, mitotic cell cycle, DNA-dependent transcription, gene expression, retrograde transport from endosome to Golgi, regulation of Rab GTPase activity, mitotic chromosome condensation, DNA repair, heme biosynthesis, and centrosome organization (Figure 3C).

Expression of miR-3687 and FSTL3 in the placenta samples of GDM and control groups

After identifying potential functions of differentially expressed miR-3687 and its target gene FSTL3, the focus shifted to investigating their roles in GDM development. The levels of miR-3687 and FSTL3 expression in placental tissues from GDM patients and controls were measured using RT-PCR analysis. The findings showed that GDM placenta samples had considerably higher expression of miR-3687 than control samples (Figure 4A). On the other hand, FSTL3 expression in GDM placental tissues was markedly less than in controls (Figure 4B). These findings suggested that miR-3687 and FSTL3 contributed to the pathogenesis of GDM.

Figure 4:

miR-3687 and FSTL3 relative expression level were verified in the placenta samples from the GDM group and the normal women by qPCR (A and B). The luciferase reporter assay of miR-3687 and FSTL3 (C).

Discussion

GDM represents a prevalent medical complication during pregnancy, characterized by normal glucose metabolism before conception followed by altered glucose metabolism, inadequate insulin secretion, and sustained production of anti-insulin-like hormones throughout pregnancy. Currently, screening and diagnosis of GDM typically occur in the mid to late stages of pregnancy, limiting the window for blood glucose intervention and management. Poor glycemic control can result in severe maternal and newborn problems, emphasizing the vital importance of early detection and diagnosis of GDM. Despite the importance of early detection, definitive biomarkers for early GDM identification are lacking. Consequently, researchers actively investigate molecular biomarkers to facilitate early diagnosis and intervention. Previous studies have demonstrated that BeWo trophoblast cells exhibit functional, metabolic, and morphological similarities to primary human trophoblast cultures [21], 22]. And varying glucose concentrations such as 10, 25, 30, or 35.5 mM have been utilized to mimic the in vitro GDM environment [23], [24], [25], [26]. In other reasearches, human trophoblastic cell line HTR8/SVneo were also cultured with different doses of glucose to simulate GDM [27], 28]. However, there was currently no standardized glucose concentration and cell line for GDM models, and different cell line and glucose conditions may influence the experimental outcomes. We selected 25 mM glucose (HG group) to treat BeWo cells to establish a GDM cell model and 5.5 mM glucose (LG group) as the control group according to U.Weiss et al. [14]. We identified genes potentially implicated in GDM-related glucose metabolism through high-throughput sequencing and bioinformatics analyses. This approach provided insights into the molecular mechanisms underlying GDM pathophysiology and offered potential targets for future therapeutic interventions and biomarker development.

Initially, we conducted comprehensive RNA-sequencing of BeWo cells, revealing significant upregulation of 158 genes and downregulation of 62 genes under HG conditions. Furthermore, eight miRNAs exhibited substantial differential expression in the HG group relative to the LG group, all continuously elevated. Notably, miR-3687-1 and miR-3687-2 exhibited substantial upregulation, as depicted in Figure 2. Prior research had associated miR-3687 with multiple cancers, including prostate cancer and bladder cancer, wherein it regulated genes pertinent to cell adhesion, angiogenesis, cell cycle control, JAK-STAT signaling, MAPK signaling, nitric oxide signaling, and VEGF signaling [29], [30], [31]. However, its role in GDM remains unexplored.

FSTL3 is a downstream target gene of miR-3687, highly expressed in the placenta and involved in lipid metabolism. FSTL3, a circulating glycoprotein, is pivotal in glucose metabolism and obesity [32]. FSTL3, an external regulatory protein, interacts with activin A, a member of the transforming growth factor β superfamily, consequently affecting glucose metabolism [33], 34]. Thadhani et al. reported reduced first-trimester circulating levels of FSTL3 in pregnant women who subsequently developed GDM [35]. Similarly, Hu et al. demonstrated significantly lower maternal and placental FSTL3 concentrations in GDM women during the third trimester, suggesting its potential involvement in GDM pathogenesis [36].

In our study, FSTL3 was notably downregulated in BeWo cells under HG conditions, contrasting with the upregulation of miR-3687. Database and literature evidence further supported that miR-3687 regulated genes involved in glucose and lipid metabolism by suppressing FSTL3 expression. Studies also suggested that miR-3687 expression in human adipocytes can be induced by metformin, underscoring their potential roles in GDM onset [37].

After full-term delivery in clinical practice, placental tissues were obtained from six GDM and six control patients to confirm the roles of miR-3687 and FSTL3 in the etiology of GDM. PCR analysis confirmed significant overexpression of miR-3687 and simultaneous underexpression of FSTL3 in the GDM full-term placenta compared to controls, consistent with our previous experiments. Furthermore, dual luciferase reporter gene assays verified the direct targeting of FSTL3 by miR-3687. Although there were no notable variations in age, gestational age, or medical history across the patient groups, individuals with GDM demonstrated substantially higher body mass index, blood triglyceride, and cholesterol levels than the control group. The data indicated that miR-3687 likely modulates glycolipid metabolism by suppressing FSTL3 expression, therefore playing a role in the etiology of GDM.

lncRNAs are emerging as potential new molecular biomarkers of GDM. Unlike microRNAs, lncRNAs are heterogeneous in size and function, localized in poorly conserved genomic regions [38]. Lu et al. examined lncRNA expression profiles throughout early gestation. Five lncRNAs (RP11-160H22.5, LOC149086, PCBP1-AS1, RP11-230G5.2, and XLOC_014172) correlated with macrosomia in pregnancy were discovered, with RP11-230G5.2 and XLOC_014172 demonstrating moderate predictive capability for macrosomia in GDM pregnancies within larger cohorts [39]. Several lncRNAs had drawn attention in various aspects of diabetes and related complications: LINC01556 in diabetic retinopathy, GAS6-AS2 in the PI3K/AKT signaling pathway, MIR31HG in lipid metabolism, and LINC00324 in diabetes wound repair [40], [41], [42], [43]. Our study identified 87 lncRNAs with significant expression changes in BeWo cells under HG conditions, including 54 upregulated and 33 downregulated lncRNAs. However, their expression levels were generally low, necessitating further research to elucidate their specific roles in GDM.

Event-related genes provide new insights for predicting and diagnosing GDM. Alongside significantly differentially expressed ncRNAs, our analysis highlighted 1597 AS events under HG conditions, a noteworthy finding in this study. For predicting the functional consequences of these differentially expressed AS events, GO analysis was used. Notably, four genes were identified as credible AS events based on their p values: Nuclear receptor coactivator 6 (NCOA6), associated with insulin secretion and glucose metabolism in pancreatic β-cells [44], RIPK3 and RNF216, associated with weight-loss regain in obese subjects, G protein-coupled receptor 78 (GPR78), a key target for anti-diabetic compounds, and PQBP1, involved in lipid metabolism [44], [45], [46], [47], [48]. These differentially expressed AS events likely contribute to the pathogenesis of GDM, warranting further investigation.

Other DEGs in BeWo cells under HG conditions might also play roles in GDM pathogenesis. ZnHIT2, a zinc finger protein critical in energy metabolism, was significantly upregulated in HG-cultured BeWo cells. SNORD3A, associated with ER stress, showed significant upregulation [49], 50]. However, in HG-cultured BeWo cells, RGS2 and IGFBP2 were substantially downregulated. IGFBP2 increases insulin sensitivity, whereas RGS2 increases insulin resistance [51], 52]. These findings suggested that gene regulation under HG conditions involved multiple genes with potentially conflicting roles in metabolic pathways. This complexity underscored the need for further cellular and in vivo experiments to elucidate each gene’s expression and regulatory changes in GDM research.

So far there were no relevant literature reports on miR-3687 in GDM. We detected and firstly reported the differential expression of miR-3687 as well as other key differential genes between GDM and healthy individuals, trying to provide a better understanding of the development and progression of GDM. Although some preliminary findings have been made in this study and the expression of differential miR-3687 and FSTL3 has been preliminarily verified by tissue verification, limitations for this study should be outlined. The main study limitation is the relatively small sample size. Notably, comprehensive management of GDM throughout pregnancy is often a key factor influencing maternal and fetal outcomes. However, in clinical practice, the adherence of most patients is poor, leading to fluctuations in blood sugar levels and the occurrence of fetal complications. Based on enough informed consent, we included six normal pregnant women as control and another six patients with GDM as experiment group. The inclusion and exclusion criteria of subject recruitment was very strict and the clinical parameters of the two groups, such as maternal age, gravida, parity, and gestational age at delivery were not significantly different. Moreover, the placental samples used in this study were obtained from women with well-controlled glucose levels. Although the small sample size may limit the generalizability of this study’s findings, it ensures the authenticity of the clinical data from this group of patients and their good adherence. We hope that in the future, more pregnant women will be willing to participate in such clinical research in the future, allowing us to further expand our sample size and make more discoveries that can aid in the prevention and treatment of diseases. Secondly, only one experimental validation method was used in this study, the lack of further functional verification and mechanism research may limit the accurate explanation of the biological significance of differential genes. Ideally, these results should be verified by multiple validation methods such as animal experiments. We hope that the discussion of these limitations will attract more researcher’ attention and promote the in-depth exploration of this field. We believe that through further investigations we can more comprehensively reveal the molecular mechanisms of GDM and provide more in-depth theoretical support for the diagnosis and prevention of GDM.

Our study revealed differential expression of various ncRNAs, especially miRNAs and lncRNAs, in placental and related cells under HG conditions. Notably, miR-3687 emerged as a significant regulator, influencing glucose and lipid metabolism in patients with GDM by targeting the downstream gene FSTL3. The results obtained imply that miR-3687 was essential to the pathophysiology of GDM. Therefore, despite the mentioned limitations, our study offered new insights into the molecular pathways causing GDM through ncRNAs.