Discovery of placental microRNAs associated with maternal insulin sensitivity during pregnancy

Abstract

Context:

During pregnancy, maternal insulin sensitivity decreases, supporting transfer of nutrients to the fetus; when excessive, this can lead to gestational diabetes mellitus. The physiological decline in insulin sensitivity is likely caused by placental factors; however, the identity of these placental factors remains unclear.

Objective:

To identify placental microRNAs (miRNAs) associated with maternal insulin sensitivity during pregnancy.

Design:

A prospective pregnancy cohort study called Genetics of Glucose regulation in Gestation and Growth. We assessed insulin sensitivity using the Matsuda index during the second trimester of pregnancy, and microtranscriptome expression in placental samples collected at delivery. We accounted for confounders including maternal age, fetal sex, gestational age at delivery, gravidity and maternal body mass index, and surrogate variables, capturing sampling and technical variability.

Setting:

Centre Hospitalier Universitaire de Sherbrooke, Canada.

Participants:

A total of 434 pregnancies were included. The mean (SD) maternal age was 28.6 (4.4) years; Matsuda index, 7.7 (4.9); and gestational age at delivery, 39.4 (1.4) weeks.

Main outcome measure(s):

Placental miRNAs expression (n = 952 miRNAs).

Results:

We identified 18 placental miRNAs negatively and 1 miRNA positively correlated with Matsuda index (FDR p < 0.05). Gene ontology and tissue expression analysis of the genes targeted by the identified placental miRNAs suggest they may influence metabolic regulation, potentially acting as endocrine factors in skeletal muscle and adipose tissue and as paracrine factors within the placenta.

Conclusions:

We identified placental miRNAs that may act as endocrine and paracrine factors to modulate maternal insulin sensitivity during pregnancy.

1. Introduction

During pregnancy, insulin sensitivity decreases by 50–60 %, a physiological shift to direct more nutrients to the growing fetus [1]. However, when insulin sensitivity is excessively impaired in peripheral tissues and/or there is insufficient insulin secretion during pregnancy, this leads to the development of gestational diabetes mellitus (GDM) [2] contributing to maternal and fetal complications [3]. Notably, GDM cases characterized by impaired insulin sensitivity are more likely to develop complications compared to other GDM subtypes, such as those with predominant insulin-secretory defects [4].

The placenta is likely a key contributor to the decline in maternal insulin sensitivity, as it produces biological factors (proteins, hormones, and microRNAs) that alter maternal metabolism to increase nutrient availability and placental transfer, thereby supporting fetal growth and development [5,6]. MiRNAs are small non-coding RNAs that play a critical role in the post-transcriptional regulation of mRNA transcripts. They can function as autocrine, paracrine, or even endocrine factors circulating in blood and exerting effects beyond their cell of origin [7]. Additionally, miRNAs are suspected to contribute to a variety of diseases including obesity and diabetes. For instance, in obesity-associated cardiovascular risks, many miRNAs regulate peroxisome proliferator-activated receptors (PPARs) in adipose tissue, which regulate insulin sensitivity and can contribute to the development of type 2 diabetes [8]. It has been hypothesized that placental miRNAs can be released into maternal circulation and modulate gestational insulin sensitivity and glucose metabolism pathways involved in the pathophysiology of GDM [9].

Previous investigations have used a candidate approach to identify placental miRNAs involved in glucose regulation or GDM. However, this approach yielded findings that are inconsistent and were limited to a few selected miRNAs [10]. To overcome these limitations, we employed a genome-wide agnostic approach to identify novel placental miRNAs that regulate insulin sensitivity during pregnancy, using data from a large population-based pregnancy cohort and placenta samples collected at birth. We hypothesized that specific placental miRNAs are associated with insulin sensitivity measured in late second trimester, which is a key pathophysiologic feature of GDM.

2. Materials and methods

2.1. Gen3G cohort

The Genetics of Glucose regulation in Gestation and Growth (Gen3G) Cohort is a prospective cohort for the study of pregnant women based in Sherbrooke, Quebec, Canada [11]. The study population was predominantly white (self-identified); in the current analysis, we included only white individuals (9 non-white individuals were excluded based on identification as outliers on the microRNA dataset principal components plots). Briefly, we invited women aged at least 18 years who were pregnant in their first trimester and receiving prenatal care with plans to deliver at the Centre Hospitalier Universitaire de Sherbrooke (CHUS; Sherbrooke, Quebec, Canada) to participate. We enrolled 1024 participants between January 2010 and June 2013 with informed consent from each participant. Exclusion criteria included a history of preexisting diabetes or laboratory evidence of overt diabetes (first trimester A1c ≥ 6.5 % or 1h-glucose ≥10.3 mmol/L post-50g glucose challenge test), use of medications that influence glucose tolerance, multiple pregnancies, and drug and/or alcohol abuse. The institutional review board at CHUS approved all study protocols.

2.2. Measures at enrollment in first trimester (V1)

We collected data and samples at three time points during pregnancy [11]. The first visit occurred between weeks 5–16 of gestation during which we collected demographic data, medical history and anthropometric measurements. Research staff assessed weight using a calibrated electronic scale, and height using a wall stadiometer; we calculated body mass index (BMI) in kg/m².

2.3. Measures at late second trimester visit (V2)

Visit 2 coincided with the end of second trimester and occurred between weeks 24 and 30 of gestation. At this visit, we updated medical history, repeated anthropometry measures, and participants performed a 75g oral glucose tolerance test (OGTT), during which we collected blood at fasting, 1h and 2h time points.

2.4. Measures at delivery (V3)

At delivery (V3), we obtained information on mode of delivery, gestational age, and infant sex from medical records. We also collected data on pregnancy complications, including preeclampsia and gestational hypertension, which we classified according to standard clinical criteria.

2.5. Maternal Matsuda Index

We selected Matsuda index because it is the surrogate marker that is most strongly correlated with insulin sensitivity measured by clamps and was validated in pregnancy [12,13]. In our Gen3G cohort, we have previously used Matsuda index to subtype GDM [4] and to discover new placental factors (proteins) linked with insulin sensitivity and risk of GDM [14]. We measured plasma glucose and insulin values during three time-points (fasting, 1h, and 2h) of the 75g-OGTT using the glucose hexokinase method (Roche Diagnostics) and the Human Milliplex map kits (EMD Millipore), respectively. We assessed insulin sensitivity with the Matsuda Index using the original clamp-validated formula as validated in pregnancy [12,13] using the following calculation: 10, 000/√ ((fasting glucose (mg/dL) × fasting insulin (mIU/L) × (mean glucose (mg/dL) × mean insulin during OGTT (mIU/L))).

2.6. Placental tissue collection

Trained study staff at CHUS collected placenta within 30 min of delivery using a standard protocol [11]. Briefly, 1-cm³ tissue was sliced from the maternal facing side (including some decidual tissue) of the placenta from the participants. Each collected sample was immediately placed in RNA Later (Invitrogen) and conserved at 4 °C for at least 24 h before being stored in −80 °C until RNA extraction.

2.7. RNA extraction, sequencing and quality control

Laboratory personnel extracted total RNA (Average = 19.7 ± 7.1 μg) using miRNeasy kit (Qiagen) following the manufacturer’s recommended protocol and checked the quality of each sample using an Aligent Bioanalyzer to determine the RNA Integrity Number (RIN; average RIN = 6.7 ± 0.8). For samples with a RIN value ≥ 5, 3ug of each sample was shipped to Novogene (Wayne, New Jersey, USA). Unique Molecular Index (UMI) were assigned during library preparation with the QIAseq miRNA Library Kit (Qiagen). Small RNA sequencing was performed on the Illumina NovaSeq 6000 platform, generating 75-bp single-end reads.

We used the exceRpt pipeline (V4.6.3) for small RNA sequencing data analysis [15]. In brief, exceRpt employs fastx (V3.0) for adaptor trimming of FASTQ reads, STAR (V2.5.1b) for alignment and miRBase v21 annotations. We quantified 2244 miRNAs among the 467 sequenced placenta samples. We removed miRNAs with low abundance by including only those with ≥6 reads in at least 20 % of samples, resulting in 952 miRNAs for downstream analysis. Nine participants of non-European ancestry were identified as outliers via principal component analysis and excluded, leaving 434 samples with available Matsuda Index data.

2.8. Statistical analyses

Prior to analysis, we computed normalizing factors using the R statistical software package edgeR [16] then normalized and transformed gene counts to log₂ counts per million reads (CPM) using Voom from the Limma R package [17]. We adjusted models for maternal age, fetal sex, gestational age at delivery, gravidity and maternal BMI (measured at first trimester of pregnancy) as biological covariates, in addition to surrogate variables (SVs). We derived SVs from the CPM with the R package SmartSVA to account for technical and unmeasured sources of variability, including batch effects and cell-type composition [18,19]. In-depth analysis of the resulting SVs led us to include only the first 4 SVs, as many of the additional SVs appeared to capture biological variance of interest. We used Limma to identify placental miRNAs associated with log₂ transformed maternal Matsuda Index as an independent continuous variable in our models. We considered statistically significant q-value <0.05 after adjustment for multiple testing using the false discovery rate (FDR) by Benjamini-Hochberg [20]. We conducted sensitivity analyses by further adjusting our models for delivery mode (vaginal, C-section preceded by labor, and elective C-section). We performed secondary analyses of the miRNAs identified in our primary analyses to compare the placental microRNA expression levels between GDM and non-GDM pregnancies. We performed all analyses using R statistical software (version 4.0.3).

2.9. miRNA gene targets

From the list of miRNAs significantly correlated with the Matsuda index (FDR q < 0.05), we investigated their gene targets through miRNet [21] using default settings and miRTarBase v9.0 [22] with the minimal network option enabled.

2.10. Pathway and tissue enrichment of identified miRNA-targeted genes

We conducted pathway and tissue enrichment analyses to better understand the potential roles of the genes targeted by Matsuda-associated miRNAs. Pathway analysis was performed with ShinyGO (v0.82) [23], while tissue enrichment was assessed with EnrichR [24–26], focusing on GTEx version 8 datasets (GTEx Aging Signatures 2021 and GTEx Tissues V8 2023). We also selected pathway enrichment output from Jensen tissue 2.0 [27], ARCHS4 database [28], and MGI phenotype level 4 2024 [29].

3. Results

3.1. Clinical characteristics

Participant characteristics are summarized in Table 1. Women in the study had a mean ± SD age of 28.6 ± 4.4 years and an early pregnancy BMI of 25.4 ± 5.6 kg/m². The mean ± SD Matsuda Index as measured during the second trimester (visit 2) was 7.7 ± 4.9 arbitrary units (AU). A total of 36 women (8 %) developed GDM. The mean ± SD Matsuda index in non-GDM participants was 8.00 ± 4.92, while in GDM was 4.30 ± 2.25.

Table 1.

Clinical Characteristics of the Gen3G participants with Placental microRNA data (n = 434).

Maternal Characteristics	Mean ± SD or N (%)
Gestational age first trimester visit (weeks at V1)	9.8 ± 2.2
Maternal Age (years)	28.6 ± 4.4
Primigravid BMI @ V1 (kg/m2)	158 (36 %) 25.4 ± 5.6
Systolic blood pressure (mmHg)	110.6 ± 9.9
Diastolic blood pressure (mmHg)	69.5 ± 6.9
Gestational age second trimester visit (weeks at V2)	26.5 ± 1.0
Gestational Diabetes	36 (8 %)
Matsuda Index (AU)	7.7 ± 4.9
Fasting glucose @ V2 (mmol/L)	4.2 ± 0.4
1-hr glucose during OGTT @ V2 (mmol/L)	7.1 ± 1.6
2-hr glucose during OGTT @ V2 (mmol/L)	5.9 ± 1.4
Mode of Delivery
Vaginal	361 (83 %)
Labor preceding C-section	20 (5 %)
Elective C-section	53 (12 %)
Child Characteristics
Sex	230 M/204 F
Gestational age at delivery (weeks)	39.4 ± 1.4
Newborn weight (g)	3392 ± 473.3
Placental weight (g)	541.8 ± 128.8

3.2. Differential placenta miRNA expression in relation to insulin sensitivity

Of the 2244 placental miRNAs detected, 952 met the threshold of ≥6 counts in at least 20 % of participants and were included in downstream analyses. Maternal Matsuda Index was associated with placental expression of 19 miRNAs (FDR-adjusted q < 0.05; Fig. 1) in models adjusted for maternal age, fetal sex, gestational age at delivery, maternal BMI, gravidity and four SVs. Among these, one miRNA (miR-6870–3p) was positively correlated with the Matsuda index, while 18 were negatively correlated, indicating higher placental miRNA was linked to lower insulin sensitivity (Table 2). Many of these identified miRNAs had reported potential roles in pregnancy complications, or in inflammation/metabolism based on prior literature (Supplementary Table 1). In our primary analyses, miR-141–3p and miR-29c-3p, members of the miRNA miR-200 and miR-29 families respectively, showed the strongest associations (Table 2). Further adjusting our models for mode of delivery showed very similar associations (Supplementary Table 2).

Fig. 1.

Differentially expressed placental microRNAs associated with maternal insulin sensitivity, estimated by Matsuda Index derived from the 26-week oral glucose tolerance test. Maternal Matsuda Index (log₂ transformed) was used as a continuous variable; associations with placental microRNAs adjusted for maternal age, gestational age, fetal sex, gravidity, maternal BMI in the first trimester, and the first four surrogate variables (from the placental microRNA dataset). The volcano plot demonstrates microRNAs with FDR <5 % with blue dots and microRNA annotation.

Table 2.

Placental microRNAs associated with insulin sensitivity estimated by the Matsuda Index at the second trimester of pregnancy.

microRNA	Average expression (TMM)	Slope	P-value	Adjusted P-value
hsa-miR-141-3p	14.140	−0.07	5.2E-06	4.9E-03
hsa-miR-29c-3p	11.813	−0.07	2.8E-05	1.3E-02
hsa-miR-9-3p	4.282	−0.16	7.5E-05	2.4E-02
hsa-miR-141-5p	7.352	−0.06	1.5E-04	2.8E-02
hsa-miR-376c-5p	7.066	−0.07	1.6E-04	2.8E-02
hsa-miR-376b-5p	7.061	−0.07	1.8E-04	2.8E-02
hsa-miR-3611	1.802	−0.10	2.7E-04	3.6E-02
hsa-miR-590-5p	5.330	−0.08	3.1E-04	3.6E-02
hsa-miR-30e-5p	12.075	− 0.04	3.4E-04	3.6E-02
hsa-miR-3664-3p	3.513	−0.07	4.0E-04	3.7E-02
hsa-miR-29a-3p	13.234	−0.10	4.6E-04	3.7E-02
hsa-miR-449a	2.455	−0.09	4.7E-04	3.7E-02
hsa-miR-376a-5p	5.538	−0.07	5.4E-04	3.7E-02
hsa-miR-1307-5p	5.533	−0.17	5.7E-04	3.7E-02
hsa-miR-188-3p	1.219	−0.09	5.8E-04	3.7E-02
hsa-miR-561-5p	5.654	−0.08	8.4E-04	4.5E-02
hsa-miR-590-3p	8.148	−0.07	8.5E-04	4.5E-02
hsa-miR-6870-3p	−1.456	0.18	8.6E-04	4.5E-02
hsa-miR-9-5p	7.368	−0.14	9.4E-04	4.7E-02

We conducted a secondary analysis for the 19 identified Matsuda-associated miRNAs to compare their placental expression in pregnancies complicated with GDM (n = 36) to non-GDM pregnancies (n = 398). We found 7 miRNAs nominally associated (P < 0.05) with GDM status in the expected direction of association (Supplementary Table 3); for example, miR-141–3p placental levels were higher in GDM pregnancies (FC = 0.11; P = 8.6× 10–3) in line with the higher placental expression levels in relation to lower insulin sensitivity in our primary analyses. However, after accounting for multiple testing (using Bonferroni for 19 miRNAs tested), only miR-561–5p was statistically associated with GDM status.

3.3. Potential roles and functions – enrichment pathways

To further explore the potential biological roles of Matsuda-associated miRNAs, we analyzed their predicted target genes with miRNet [21]. Each miRNA may target tens to hundreds of genes, and genes can be targeted by multiple miRNAs [7]; therefore, we focused on targets shared by more than one of the 18 miRNAs negatively correlated with Matsuda (Fig. 2). We next performed gene ontology of the target genes. Many were enriched in biological processes such as cellular metabolism and proliferation (Fig. 3). Nuclear Factor of Activated T-cells (NFAT) binding emerged as a significantly enriched molecular function, and protein phosphatase type 1 complex was identified as an enriched cellular component (Fig. 3).

Fig. 2.

MiRNAs negatively correlated with insulin sensitivity target multiple genes. A total of 73 genes were identified as gene targets of more than one of the 18 miRNAs negatively correlated with insulin sensitivity. Blue boxes represent miRNAs while purple boxes represent gene targets.

Fig. 3.

Gene ontology analysis of gene targets reveals involvement in proliferation and metabolism pathways. Gene ontology analysis was performed on the 73 genes targeted by the 18 miRNAs with placental expression levels negatively correlated with insulin sensitivity. The analysis categorized ontology pathways into biological processes, molecular functions, and cellular components.

Using the list of 73 genes targeted by more than one Matsuda-associated miRNA (Fig. 2), we examined enrichment in expression patterns in publicly available datasets of human tissues. These genes showed enrichment in skeletal muscle and adipose tissues (white and brown adipose) across multiple datasets (Fig. 4A–C), suggesting a potential endocrine role influencing maternal physiology in peripheral tissues sensitive to insulin actions. Furthermore, the ARCHS4 tissue dataset and MGI phenotype annotations indicated enrichment in placental expression and potential roles in organogenesis (Fig. 4D and E) suggesting roles of paracrine actions within the placenta as well as influences on fetal development.

Fig. 4.

Tissue expression analysis identifies muscles and adipocyte tissue (white and brown adipose) among those expressing genes targeted by miRNA that are negatively correlated with insulin sensitivity. Significant adjusted P values are listed.

4. Discussion

In the present study, we investigated the association of maternal insulin sensitivity with genome-wide placental miRNA profiles in a large, population-based pregnancy cohort. Using the validated Matsuda Index estimated at 24–30 weeks of pregnancy, we identified 19 placental miRNAs associated with maternal insulin sensitivity—one positively and 18 negatively. Downstream analyses in publicly available datasets suggest the identified placental miRNAs may exert both endocrine and paracrine actions.

Target gene analysis of the Matsuda-associated miRNAs indicated 73 predicted genes. Functional enrichment analysis revealed pathways central to cell proliferation and cellular metabolism, both critical regulators of insulin action and glucose tolerance [30–32]. NFAT protein binding was the most significantly enriched molecular function. Alteration in NFAT has been implicated in diabetes pathophysiology [33] and in vitro studies demonstrated reduced NFAT levels in a type 2 diabetes β-cell model [34]. The protein phosphatase type 1 complex (cellular component enriched feature) acts downstream of the AKT pathway and is essential for insulin signaling and glucose deposition in hepatocytes [35].

Previous studies have shown that placental miRNAs can be exported into the maternal circulation, where they may impact glucose uptake in peripheral tissues [36]. In support of this endocrine hypothesis, our enrichment analysis revealed that target genes of the Matsuda-associated miRNAs were preferentially expressed in skeletal muscles and adipose tissues (white and brown adipose). Adipocytes are key players in insulin resistance contributing to development of type 2 diabetes [37]. Among the identified target genes enriched in adipose tissues, we noted Forkhead box O 3 (FOXO3), and Vascular endothelial growth factor A (VEGFA). FOXO3, a member of the FOXO family, regulates proliferation, cellular homeostasis, and metabolism [38]. Although its role in adipocytes is not fully defined, evidence suggests that reduced FOXO3 levels in adipocytes improves liver insulin sensitivity through tissue crosstalk [39]. VEGFA encodes a growth factor critical for angiogenesis, which can influence glucose metabolism in a tissue-specific manner [40,41]; for example, adipocyte-specific overexpression of VEGFA increased vascularization within adipose tissue, leading to overall greater energy expenditure and insulin sensitivity [42].

Skeletal muscle is the primary site of insulin-mediated glucose uptake and a key determinant of whole-body insulin sensitivity [43,44]. Identified target genes enriched in muscle included Kelch-like ECH-associated protein 1 (KEAP1), and E2F Transcription Factor 3 (E2F3). KEAP1 is highly expressed in muscle and placenta [45] and functions as a repressor protein that binds nuclear factor erythroid 2-related factor 2 (NRF2), regulating oxidative stress [46,47]. E2F3 is a transcription factor that is required for the proliferation of most cells [48]. Similar to KEAP1, E2F3 is enriched in muscle cells [45] and plays an important role in insulin sensitivity. Elevated E2F3 levels enhance muscle cell proliferation and increase Peroxisome Proliferator-Activated Receptor Gamma Coactivator 1-Alpha (PPARGC1A) expression levels in skeletal muscles, leading to greater insulin sensitivity [49]. Both KEAP1 and E2F3 are targets of miR-141–3p, the placental miRNA most strongly negatively associated with maternal insulin sensitivity in our analyses.

Many of the identified Matsuda-associated placental miRNAs have been reported in prior studies investigating pregnancy complications (pre-eclampsia, spontaneous abortion, preterm labor, etc.) [50–52]. Of particular interest among the list, we found multiple members of the miR-200 family (miR-141–3p and miR-141–5p) and of the miR-29 family (miR-29c-3p and hsa-miR-29a-3p) which have been implicated in insulin signaling during pregnancy. Their potential roles are discussed below.

4.1. MiR-200 family of miRNA

The miR-200 family has been extensively studied for their role in cancer and tumorigenesis [53]. Notably, these miRNAs may participate in the epithelial-mesenchymal transition, a process important for both tumorigenesis and placental development, where trophoblasts become more mesenchymal and invade the maternal decidua [54]. In mice, members of this family also regulate early pregnancy endometrial development [55]. Our analysis revealed two family members-miR-141–5p and miR-141–3p exhibiting higher placental expression levels in relation to lower maternal insulin sensitivity. While both may contribute to placental development and healthy gestation [51,56], only miR-141–3p has been studied for its role in insulin sensitivity. Conversely, miR-141–5p has been linked to preeclampsia through the regulation of the MAPK pathway [57].

4.1.1. MiR-141–3p

MiR-141–3p emerged as the miRNA most strongly associated with insulin sensitivity. As pregnancy progresses and the placenta develops, insulin sensitivity physiologically declines [1], and in parallel, miR-141–3p placental expression increases [58]. This pattern suggests that the expression of miR-141–3p throughout pregnancy plays an important physiological role. In support of this, miR-141–3p has been shown to repress insulin like growth factor 2 (IGF2) levels in mouse trophoblast stem cells, leading to the downregulation of AKT signaling pathway [59], which is central to insulin sensitivity in various tissues [60]. Additionally, insulin like growth factor 2 receptor (IGF2R), the receptor responsible for binding IGF2 and initiating the downstream signaling, is also predicted to be a target of miR-141–3p. Using the miRTarBase v9.0 database, we observed that miR-141–3p regulates additional genes implicated in insulin sensitivity and diabetes, including KEAP1, and E2F3. Together, these data support our hypothesis that higher placental miR-141–3p contributes to reducing peripheral insulin sensitivity.

4.2. MiR-29 family of miRNA

The miR-29 family of miRNAs are expressed in most tissues and are implicated in the regulation of glucose metabolism in multiple cell-types, including β-cells and hepatocytes. This family of has been hypothesized to contribute to metabolic conditions such as obesity, insulin sensitivity, and diabetes [61]. We identified two members of the miR-29 family, miR-29c-3p, and miR-29a-3p, where higher levels were associated with lower insulin sensitivity. Both are predicted to play important roles in diabetes and insulin sensitivity. While some studies have linked miR-29a-3p to lower insulin-like growth factor 1 (IGF1) levels in pineal gland tumors, we focus our discussion on miR-29c-3p based on substantial prior evidence supporting its role in glycemic regulation [62].

4.2.1. miR-29c-3p

MiR-29c-3p has been studied for its role in insulin and glucose regulation. Notably, miR-29c-3p expression in β-cells increases in response to elevated glucose levels. When miR-29–3p levels were overexpressed in β-cells in vitro, insulin secretion decreased [63]. Furthermore, a study in rat muscle found that miR-29c-3p expression negatively correlated with the levels of the membrane bound glucose transporter GLUT4 and intracellular hexokinase HK2, the first enzyme in glycolysis [64]. Reduced GLUT4 and HK2 protein levels in murine skeletal myocytes and adipocytes have been linked with lower insulin sensitivity and diabetes [65,66].

Another potential mechanism by which miR-29c-3p might influence insulin sensitivity could be via Cyclin-dependent kinase 6 (CDK6), identified as one of its predicted gene targets. CDK6 is a G1 cell-cycle kinase that regulates the transcription of pro-proliferation genes, although its role in glucose regulation is debated [67]. On one hand, CDK6 activity in pancreatic cells promotes β-cells proliferation, enhancing glucose homeostasis and insulin secretion [68]. On the other hand, CDK6 regulates beige adipocyte formation, which play a role in modulating insulin sensitivity [69]; for example, in mice harboring an inactivation mutation of CDK6 in adipocytes, blood glucose clearance was more rapid and resulted in lower blood glucose concentration compared to wildtype controls [70]. While the effects of CDK6 appear to be tissue-specific and context-dependent, one possibility is that placental miR-29c-3p levels inhibit CDK6 in adipocytes during pregnancy and contribute to lower insulin sensitivity.

4.3. Strengths and limitations

Our study has several strengths. First, we had a relatively large sample size, consisting of 434 placenta samples, collected from a population-based pregnancy cohort with detailed measures allowing us to account for some potential confounding variables. We used an agnostic approach to measure and analyze differential expression levels of placental miRNAs in relation to insulin sensitivity in pregnancy. Another strength is the use of the Matsuda index which was validated against euglycemic-hyperinsulinemic clamps, including in pregnant individuals [12,13]. However, Matsuda index remains a surrogate measure of insulin sensitivity that can vary based on the population studied and the period of gestation. Our study has additional limitations. This was an observational study, limiting causal inference. Furthermore, placental samples were collected at delivery, raising the possibility that observed associations reflect consequences of maternal physiology rather than contributing factors. As many placental miRNAs expression varies across gestation [71], it would be of high interest to measure insulin sensitivity and placental miRNAs in early gestation to assess potential roles of placental miRNA in early pregnancy, while individuals are physiologically more insulin sensitive, but such studies are technically and ethically challenging. We did not demonstrate that the identified placental miRNAs are released in maternal circulation in concomitant maternal samples or in placental perfusion experiments; yet many placental miRNAs are detectable in maternal circulation [9](e. g. miR-141 has been detected in maternal plasma and correlated to placental expression) [72]. Finally, the study population was white and based in a single geographic region in Canada, which may limit generalizability.

5. Conclusions

Several miRNAs of placental origin are released into maternal circulation throughout pregnancy and their roles and impact on maternal and fetal health are still being elucidated. In the present study, we identified 19 placental miRNAs associated with the Matsuda Index-estimated insulin sensitivity measured in the late second trimester. Placental miRNAs that are released in maternal circulation have potential as biomarkers and therapeutic targets [73], but future studies assessing their absolute abundance are needed to evaluate this possibility. Our findings are hypotheses generating towards potential endocrine and paracrine roles of the identified placental miRNAs associated with maternal insulin sensitivity. Ongoing work includes experimentally validating these miRNA gene targets to support our hypothesis that these miRNAs act as endocrine factors in insulin responsive tissues, such as skeletal muscle and adipose tissues. If proven, this would open a novel therapeutic field to improve insulin sensitivity inside and outside of pregnancy.

Abbreviations:

GDM

Gestational Diabetes Mellitus

miRNA

microRNA

PPARs

Peroxisome Proliferator-Activated Receptors

Gen3G

Genetics of Glucose regulation in Gestation and Growth

A1c

Hemoglobin A1c

BMI

Body Mass Index

OGTT

Oral Glucose Tolerance Test

PCA

Principal Component Analysis

SVs

Surrogate Variables

FDR

False Discovery Rate

AU

Arbitrary Units

NFAT

Nuclear Factor of Activated T-cells

KEAP1

Kelch-like ECH-associated protein 1

NRF2

Nuclear Factor Erythroid 2–Related Factor 2

E2F3

E2F Transcription Factor 3

PPARGC1A

Peroxisome Proliferator-Activated Receptor Gamma Coactivator 1-Alpha

FOXO3

Forkhead Box O3

VEGFA

Vascular Endothelial Growth Factor A

IGF2

Insulin-Like Growth Factor 2

IGF2R

Insulin-Like Growth Factor 2 Receptor

MAPK

Mitogen-Activated Protein Kinase

CDK6

Cyclin-Dependent Kinase 6

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.placenta.2025.10.021.

Data availability

The Gen3G placental microRNA data and pregnancy phenotypes are available on dbGAP (https://www.ncbi.nlm.nih.gov/projects/gap/cgi-bin/study.cgi?study_id=phs003151.v1.p1).