Increased TGF-β/Activin-Smad2 signaling is associated with pancreatic β-cell dysfunction and glucose intolerance in gestational diabetes mellitus

Abstract

Background

Gestational diabetes mellitus (GDM) is the most common metabolic disease during pregnancy and increases the prevalence of type 2 diabetes in both mothers and children. GDM management provides an opportunity to prevent and lower the global burden of diabetes across life. Molecular mechanisms underlying GDM are not completely understood. In this study, we explore the role of transforming growth factor beta (TGF-β) signaling in GDM, as this pathway reportedly affects pancreatic β-cell development, function, and proliferation.

Methods

We developed a GDM animal model. Serum circulating levels of TGF-β family ligands were measured in mice and human GDM. Pancreatic TGF-β signaling was investigated via gene and protein expression.

Results

Our GDM animal model recapitulates the main pathophysiological features of human GDM, including glucose intolerance, decreased insulin sensitivity and pancreatic β-cell malfunction. Islets from GDM mice showed impaired insulin secretion and content, altered ion channel activity, and decreased β-cell replication rate. This was accompanied by increased Smad2 signaling activation. Elevated serum activin-A and inhibin levels were found in mice and human GDM, suggesting their role as upstream signaling transducers of pancreatic Smad2 activation. Pharmacological inhibition of TGF-β/Activin-Smad2 signaling in mouse pancreatic islets resulted in improved pancreatic β-cell function and regeneration capacity.

Conclusions

Our data suggest that disruption of the pancreatic Smad2 pathway plays a critical role in the pathogenesis of GDM, contributing to abnormal glucose homeostasis and inadequate insulin secretion. Attenuation of this signaling pathway may represent a putative therapeutic target for GDM.

Highlights

• •High fat diet just before and during pregnancy leads to gestational diabetes in mice.
• •Activin and inhibin serum levels are increased in human and mice gestational diabetes.
• •Enhanced pancreatic Smad2 signaling contributes to inadequate insulin secretion in gestational diabetes.
• •Inhibition of Smad2 signaling improves pancreatic β-cell function and proliferation.

1. Introduction

Gestational diabetes mellitus (GDM) is defined as any degree of glucose intolerance with onset or first recognition during pregnancy [1]. It is considered one of the most common medical complications of the gestational period [2]. The global standardized prevalence of GDM has increased considerably over the past decades and currently stands at 14.2% [3], although GDM incidence has been reported as high as 27.0% in certain regions, such as South East Asia [3]. In most cases, normal glucose tolerance is restored after delivery; however, women with GDM are at increased risk of developing type 2 diabetes mellitus (T2DM) in the postpartum period, highlighting that GDM diagnosis also carries long-term metabolic implications [4]. Accordingly, long-term follow up studies show that up to 20–60% of women with GDM will develop diabetes within 5–10 years after giving birth, and approximately 10% will develop the disease shortly after delivery [5]. Besides, women who develop GDM have a 10-fold higher risk of progressing to T2DM compared to their peers [6]. From this life course perspective, GDM can be considered a chronic maternal metabolic disorder. Evidence to date confirms that GDM also places the offspring at a higher risk of developing metabolic disorders, including T2DM and obesity [7]. Therefore, the road from GDM to T2DM represents a well-recognized continuum, which provides a unique opportunity to prevent and lower the long-term burden of metabolic disorders in affected individuals.

Despite the importance of GDM, the molecular mechanisms underlying the disease are not completely understood. During normal pregnancy, there is a progressive decline in maternal insulin sensitivity. This is a physiological response that serves as a protective mechanism to ensure an adequate nutrient supply to the growing fetus. Maternal euglycemia is maintained as insulin resistance is counterbalanced by a compensatory response of pancreatic β-cell function and mass, the failure of which manifests as GDM [8]. These adaptive responses are thought to be driven, at least in part, by increased production of maternal hormones, such as 17β-estradiol, human placental lactogen, prolactin, and progesterone [[8], [9], [10]]. Potential causes of inadequate β-cell function and failure in GDM remain elusive [11].

The transforming growth factor-beta (TGF-β) family of cytokines comprises a number of highly conserved and structurally related proteins, which regulate various important biological processes, including cell growth, differentiation, apoptosis, extracellular matrix remodeling, and fibrosis [12]. Based on sequence similarity, TGF-β, bone morphogenic proteins (BMPs), and activin/inhibin are considered the three major subgroups of this family, while Nodal and related factors form a separate subgroup mainly implicated in mesoderm and endoderm specification [13]. All members of the TGF-β superfamily signal through a serine/threonine kinase receptor system which involves type-I and type-II receptors. In brief, ligand binding induces activation of the serine/threonine kinase of RII domain, which then phosphorylates RI on specific serine and threonine residues. Thereafter, the activated type I receptor phosphorylates receptor-regulated Smad proteins (R-Smads), Smad2 and 3 for TGF-β/Activin signaling, and Smad1, 5, and 8 for BMP signaling. Activated R-Smad binds to the co-Smad, Smad4, forming heteromeric complexes that translocate into the nucleus. In the nucleus the Smad complex regulates the transcription of target genes [14]. TGF-β ligands, their receptors, and Smad proteins are present in the endocrine pancreas [15,16].

TGF-β/Smad signaling is critically involved in pancreas development, β-cell differentiation and function. In adult pancreatic islets, TGF-β signaling modulates β-cell proliferation, insulin secretion, and survival through downstream effectors such as Smad2 and Smad3 [[16], [17], [18]]. Increasing evidence suggests that dysregulation of this pathway may contribute to the pathogenesis of both type 1 diabetes mellitus (T1DM) and T2DM, as well as associated complications including diabetic nephropathy [16,17,[19], [20], [21]]. Notably, attenuation of TGF-β signaling has demonstrated protective effects against the development of both diabetes and obesity [17,[22], [23], [24]].

While Smad3 has been extensively characterized for its role in repressing β-cell proliferation and insulin gene transcription [[23], [24], [25], [26], [27]], the specific function of Smad2 within the pancreas remains less well defined. Emerging evidence indicates that pancreatic β-cell–specific deletion of Smad2 enhances glucose tolerance, increases insulin secretion, and augments β-cell mass and proliferation. This is observed under both basal conditions and in the context of high-fat diet–induced hyperglycemia [22]. Collectively, these findings suggest a negative regulatory role for Smad2 in β-cell function and underscore its potential as a therapeutic target for improving β-cell performance in T2DM.

In the context of GDM, current understanding of the role of TGF-β signaling remains limited. Its dysregulation has been associated with impaired placental function and adverse pregnancy outcomes such as pre-eclampsia [28,29]. In particular, increased Smad2 activity may contribute to the development of this latter condition [30,31]. Despite these advances, the specific contribution of Smad2 and the TGF-β/Smad signaling axis to the pathophysiology of GDM, particularly within the pancreas, remains unexplored. This study was designed to fill a critical gap in understanding the role of TGF-β/Activin-Smad signaling pathway in GDM, by integrating complementary clinical and preclinical data. To support this end, we developed a relevant animal model that closely recapitulates key features of human GDM, providing a controlled system for mechanistic studies that help to overcome certain inherent limitations of clinical research. Our work highlights how dysregulation of this pathway contributes to pancreatic β-cell dysfunction and impaired regenerative capacity, offering novel insights into GDM pathophysiology.

2. Material and methods

2.1. Animals

C57BL/6 female mice (9–10 weeks of age) were used in this study. C57BL/6 mice were obtained from Envigo (C57BL/6JOlaHsd; Barcelona, Spain) for breeding by the animal experimentation service of our institution, which supplied the animals for the study. Animals were kept under controlled and standardized conditions with ad libitum access to food and water. The experimental procedures were evaluated and approved by the corresponding local Animal Ethics Committee (Miguel Hernandez University) and the competent regional authority (Generalitat Valenciana) (approvals ID: 2019/VSC/PEA/0243 and 2024-VSC-PEA-0194). All procedures were conducted in accordance with current European legislation conforming to the guidelines from Directive 2010/63/EU of the European Parliament on the protection of animals used for scientific purposes.

The mice were randomly assigned to different experimental groups based on body weight. For the GDM model, mice were fed a standard control diet (10 % energy from fat; D12450J, Research Diet, USA) or high fat diet (HFD) (60% energy from fat; D12492, Research Diet, USA) for three days prior to mating and throughout pregnancy. Pregnancy was determined by the presence of a vaginal plug the morning after mating, which was identified as gestational day (GD) 0. Experiments were performed at specified time points. Non-pregnant mice were fed a control diet or HFD for the equivalent length of time. For ex vivo experiments, adult female C57BL/6 mice (8–10 weeks old) were used.

2.2. Intraperitoneal glucose and insulin tolerance tests

Female mice were fasted for 6 h (8 am–2 pm) during the light phase to induce a post-absorptive state comparable to an overnight fast in humans, minimizing metabolic stress and aligning with their natural circadian feeding behavior. Then, they were intraperitoneally injected with glucose (1.5 g/kg body weight for the intraperitoneal glucose tolerance test (IPGTT) and in vivo glucose stimulated insulin secretion (GSIS)). For the intraperitoneal insulin tolerance test (IPITT)) insulin (0.5 UI/kg body weight for non-pregnant and 0.9 UI/kg body weight for pregnant mice) was administered. Blood was sampled from the tail vein for glucose measurements during the IPGTT at 0, 10, 20, 30, 60, and 120 minutes (min) and during the ITT at 0, 15, 30, 45, and 60 min. In the IPGTT, blood was collected from the saphenous vein for insulin measurement at 0, 10, and 30 min. Glycemia was monitored using an automatic glucometer (Accu-Chek Compact plus; Roche, Madrid, Spain). Plasma insulin levels were determined using an ELISA (Crystal Chem, Downers Grove, IL, USA). IPGTT and IPITT were performed on gestational day 15 (GD15).

2.3. Mouse serum measurements

Blood samples for serum analysis were collected at decapitation in a non-fasting state at GD16. Hormone concentrations were determined using mouse ELISAs: Insulin (Mercodia, Uppsala, Sweden), leptin, adiponectin, C-peptide, and progesterone (Crystal Chem, Downers Grove, IL, USA), placental lactogen (Biomatik, Ontario, Canada), 17β-estradiol (Cayman Chemical, Ann Arbor, USA), and prolactin (Thermo Fisher Scientific, Carlsbad, CA, USA). For serum TGF-β family ligand quantification, the following mouse ELISAs were used: TGF-β1, activin-A (Thermo Fisher Scientific), and inhibin (Abbexa, Cambridge, UK).

For each ELISA assay (both in mouse and human samples, as detailed later), commercially available sandwich ELISA kits were used in strict accordance with the manufacturers' instructions. Sample volumes were chosen according to the expected concentration range and specific kit instructions. Plates coated with capture antibodies were incubated with samples and standards for 1–2 h at room temperature or overnight at 4 °C, depending on the manufacturer's recommendations. Washing steps were performed using an automated plate washer, with each well washed a minimum of 3–5 times between steps as specified in the respective protocols. Detection antibodies, either enzyme-conjugated or biotinylated, were then applied, followed by incubation with the appropriate chromogenic substrate, typically tetramethylbenzidine (TMB). The enzymatic reaction was terminated by the addition of sulfuric acid, and absorbance was measured using a microplate reader (Bio-Rad, Madrid, Spain). Quantification was achieved by generating standard curves through the recommended curve-fitting method, and sample concentrations were interpolated using GraphPad Prism software. Assay performance metrics, including the lower limit of detection, as well as intra- and inter-assay variability, were consistent with the manufacturers' specifications and validated prior to sample analysis.

2.4. Islet isolation and cell culture

Pancreatic islets of Langerhans were isolated using collagenase (Sigma, Madrid, Spain) digestion, as previously described [32]. The isolation medium contained (in mM): 115 NaCl, 10 NaHCO₃, 5 KCl, 1.1 MgCl₂, 1.2 NaH₂PO₄, 2.5 CaCl₂, 25 HEPES, and 5 d-glucose, pH 7.4, as well as 0.25% bovine serum albumin (BSA) or 0.1% BSA for western blotting experiments. For primary cell culture, isolated islets were dispersed into single cells via trypsin enzymatic digestion. Cells were centrifuged and resuspended in RPMI 1640 without phenol red (Gibco, Thermo Fisher Scientific) and with 10% charcoal dextran treated serum (Gibco), 2 mM glutamine, 100 U/mL penicillin, and 0.1 mg/mL streptomycin (Thermo Fisher Scientific). Cells were then planted on glass covers and cultured at 37 °C in 95% humidified air and 5% CO₂, for 24 h for electrophysiological recordings.

For in vitro experiments, isolated islets or single cells were cultured in RPMI 1640 without phenol red supplemented with 10% charcoal dextran treated serum, 2 mM glutamine, 100 U/mL penicillin, and 0.1 mg/mL streptomycin at 37 °C in 95% humidified air and 5% CO₂, for 48 h in the presence of vehicle, 5 nM activin-A (Bio-Techne R&D Systems, Abingdon, UK) and/or 5 nM inhibin (Bio-Techne R&D Systems) and/or 10 μM galunisertib (MedChemExpress, New Jersey, USA).

2.5. Ex vivo insulin secretion and content measurements

Freshly isolated pancreatic islets were transferred to isolation medium (described in 2.4) and incubated for 2 h at 37 °C in a humidified atmosphere containing 95% air and 5% CO₂ to allow recovery. This recovery period is critical to allow the islets to stabilize metabolically and physiologically following the isolation procedure. Then, islets were handpicked and size-matched under a stereomicroscope to ensure uniformity in size across experimental groups. Groups of five size-matched islets were placed into individual wells of a 24-well culture plate containing 400 μL of secretion buffer. The secretion buffer contained (in mM): 140 NaCl, 4.5 KCl, 2.5 CaCl₂, 1 MgCl₂, 20 HEPES and 2.8 mM glucose, pH 7.35. This basal glucose concentration serves as a resting condition to equilibrate islets prior to stimulation. Islets were incubated for 1 h at 37 °C with 5% CO₂/95% air. Subsequently, islets were transferred to new wells containing 400 μL of secretion buffer with varying glucose concentrations (2.8, 8.3, or 16.7 mM) and incubated for an additional 1 h under the same conditions. After incubation, the supernatants were collected and stored at −80 °C for subsequent insulin secretion analysis.

For insulin content determination, islets were handpicked and transferred to 20 μL of ethanol/HCl buffer, then incubated overnight at 4 °C with gentle agitation. The supernatant was collected the following day for insulin content measurement.

Insulin secretion and content were quantified using a mouse insulin ELISA kit (Mercodia). To account for variability in islet size and cellular content, insulin secretion and content values were normalized to total islet protein content. Protein concentration was measured using the Bradford protein assay.

2.6. Patch clamp recordings

K_ATP channel activity was recorded using standard patch-clamp recording procedures in isolated pancreatic β-cells as previously described [33]. Approximately 80–90% of the cells were identified as β-cells by their response to high glucose, consisting of action currents in the cell-attached configuration. For the patch-clamp recordings of voltage-gated K⁺ and Ca²⁺ currents, the standard whole-cell configuration was used. For K_ATP channel activity and voltage-gated K⁺ currents, bath solution contained (in mM) 5 KCl, 135 NaCl, 2.5 CaCl₂, 10 HEPES, and 1.1 MgCl₂ and was supplemented with glucose as indicated (pH 7.4 with NaOH). For voltage-gated Ca²⁺ currents, bath solution contained (in mM): 118 NaCl, 20 TEA-Cl, 5.6 CaCl₂, 1.2 MgCl₂, 5 HEPES, and 5 glucose (pH: 7.4 with NaOH). For recording K_ATP channel activity, the pipette solution contained (in mM): 140 KCl, 1 MgCl₂, 10 HEPES, and 1 EGTA (pH 7.2). For recordings of voltage-gated K⁺ currents, the pipette was filled with 120 mM KCl, 1 mM MgCl₂, 1 mM CaCl₂, 3 mM MgATP, 10 mM EGTA, and 10 mM HEPES (pH: 7.15 with KOH). A similar medium was used for the Ca²⁺ current measurements, except that KCl was equimolarly replaced by CsCl and pH adjusted with CsOH (pH: 7.20 with CsOH). K⁺ currents were recorded in response to depolarizing voltage pulses of −60 to +80 mV from a holding potential of −70 mV. Ca²⁺ currents were recorded in response to depolarizing voltage pulses of −60 mV to +60 mV from a holding potential of −70 mV. For K⁺ and Ca²⁺ currents density quantification, K⁺ and Ca²⁺ currents (in pA) were normalized to the cell capacitance (in pF).

2.7. RNA isolation and gene expression analysis using real-time quantitative PCR

Total RNA was extracted from pancreatic islets isolated from mice using the RNeasy Micro Kit (Qiagen, Madrid, Spain) according to the manufacturer's instructions and quantified with a Nanodrop 2000 (Thermo Fisher Scientific, Waltham, MA, USA), followed by cDNA synthesis using a high-capacity cDNA reverse transcription kit (Applied Biosystems, Foster City, CA, USA). Quantitative real-time polymerase chain reaction (qRT-PCR) assays were performed at a final volume of 10 μL, containing 200 nM of each primer, 1 μL of cDNA, and 1 × IQ SYBR Green Supermix (Bio-Rad, Hercules, CA, USA), using a CFX96 Real Time System (Bio-Rad). The resulting values were analyzed using the CFX Maestro 2.3 (Bio-Rad) and expressed as the relative expression normalized against control levels using the comparative 2^−ΔΔCT method. HPRT (Hypoxanthine-guanine phosphoribosyl transferase) was used as the housekeeping gene. Further information on primer sequences is described in Supplemental Table 1.

2.8. Western blotting

Pancreatic islets of Langerhans were employed as the sample material for Western blotting analysis. Proteins were separated on precast gels (8–16% Mini-Protean TGX, Bio-Rad) and subsequently transferred to polyvinylidene disulfide (PVDF) membranes (Amersham-Cytiva, Germany). Membranes were blocked in Tris-buffered saline/Tween (TBST) buffer containing 5% low-fat milk protein for 1 h at room temperature. Membranes were then incubated overnight at 4 °C with primary antibodies anti-pSmad2 (ab188334, Abcam, Cambridge, UK, 1:500), anti-Smad2 (D43B4, Cell Signaling, Danvers, MA, USA, 1:1000), anti-TGF-βRI (ab235578, Abcam, 1:500), anti-β-Actin (A5316, Sigma, 1:5000). The following antibodies were also used anti-p-Smad3 (sc-517575, Santa Cruz Biotechnology, Heidelberg, Germany, 1:500), anti-smad3 (sc-101154, Santa Cruz, 1:200), anti-TGF-βRII (ab61213, Abcam, 1:500), and ALK4 (ARG40270, Arigo Biolaboratories, Zhubei, Taiwan, 1:1000), as shown in supplemental figures. Thereafter, the membranes were washed with TBST buffer and incubated for 1 h at room temperature with appropriate secondary peroxidase-conjugated antibodies (Bio-Rad, Richmond, CA, USA; 1:5000). Membranes were developed using SuperSignal West Femto chemiluminescent reagent (Thermo Scientific, Rockford, IL, USA), and visualized using the BioRad ChemiDoc XRS + System (Bio-Rad). The intensity of the bands was quantified using Image Lab software (version 4.1, Bio-Rad).

2.9. β-cell mass and proliferation analysis in pancreatic sections

Pancreas were removed at GD15 or equivalent length of time for non-pregnant mice, weighed and fixed in ice-cold 4% paraformaldehyde overnight at 4 °C. Then pancreata were embedded in paraffin, and tissue sections (5 μm) were prepared. Antigen retrieval was performed by heating the samples for 20 min in citrate buffer (10 mM, pH 6.0), and blocking was achieved through incubation in phosphate-buffered saline (PBS) containing 1% BSA, 0.1% triton-X100, and 10% goat serum at room temperature for 2 h. After washing, samples were incubated overnight at 4 °C with anti-insulin antibody (8138, Cell Signaling, 1:400), washed, and, then incubated with goat anti-mouse Alexa Fluor 594 secondary antibody (A32742, Life Technologies, Carlsbad, CA, USA, 1:500) for 1 h at room temperature. Nuclei were stained with 10 μg/mL Hoechst (Invitrogen, Barcelona, Spain). Samples were mounted with Fluoromont-G (Life Technologies). Images were acquired using the IN Cell Analyzer 6000 system (GE Healthcare, Little Chalfont, UK). Each slide was completely scanned by capturing images from different fields that did not overlap. The total pancreatic and insulin areas were quantified using ImageJ software (National Institutes of Health, Bethesda, MD, USA). Quantification was performed on at least three sections per pancreas, separated by 200 μm, from each animal. The β-cell area was calculated by measuring the total insulin-stained area normalized against the total pancreatic area. The β-cell mass was calculated by multiplying the β-cell area by the pancreas weight [34].

For the analysis of β-cell proliferation, pancreas sections were incubated overnight at 4 °C in the presence of primary antibodies (anti-insulin antibody (8138, Cell Signaling Technology, 1:400) and anti-Ki67 antibody (12202, Cell Signaling Technology, 1:225)) and, subsequently, with goat anti-mouse Alexa Fluor 594 (A32742, Life Technologies) and goat anti-rabbit 488 (A32731, Life Technologies) secondary antibodies (1:500) for 1 h at room temperature. Nuclei were stained with 10 μg/mL Hoechst (Invitrogen). Slides were mounted with Fluoromont-G (Life Technologies). Images were acquired using a Zeiss LSM900 with Airyscan 2 confocal microscope (Carl Zeiss Microscopy GmbH, Germany) at 40× magnification. Ki-67-positive nuclei were scored only in cells that were also positive for insulin. Quantification analysis was performed using ImageJ software (National Institutes of Health, Bethesda, MD, USA).

2.10. β-cell proliferation analysis in primary islet cell culture

Cells were fixed with 4% paraformaldehyde for 10 min at room temperature and washed with PBS. Cells were permeabilized with 0.5% Triton X-100 for 3 min. Non-specific interactions were blocked with 1% BSA, 0.1% Triton-X100, and 10% goat serum in PBS for 1 h at room temperature. Cells were then incubated with anti-insulin antibody (8138, Cell Signaling Technology, 1:400) and anti-Ki67 antibody (12202, Cell Signaling Technology, 1:100) overnight at 4 °C. After washing, cells were incubated with secondary antibodies goat anti-mouse Alexa Fluor 594 and goat anti-rabbit 488 both for 1 h. The nuclei were stained with 10 μg/mL Hoechst (Invitrogen) for 10 min. Samples were mounted using Fluoromont-G (Life Technologies). Images were acquired using a Zeiss LSM900 with Airyscan 2 confocal microscope 20× magnification. The proliferation rate was expressed as the percentage of β-cells with Ki67-positive nuclei with respect to the total number of β-cells. Quantification analysis was performed using ImageJ software (National Institutes of Health, Bethesda, MD).

2.11. Clinical patient data, sample collection, and measurements

Patients were recruited at the Hospital Vinalopó (Elche, Spain). Women were screened for GDM at 24–28 weeks following the Spanish Diabetes and Pregnancy Group recommendations. Subjects were challenged with a 1 h 50 g glucose challenge test; if values were ≥140 mg/dL, participants underwent a 3 h 100 g oral glucose tolerance test. GDM was diagnosed if two or more plasma glucose levels met or exceeded the threshold according to the National Diabetes Data Group and 3rd Workshop-Conference on Gestational Diabetes [35]. Women were excluded if they had pre-existing diabetes, prior gestational diabetes, or multifetal gestation; or if patients were on medication known to affect carbohydrate metabolism during pregnancy. The experimental protocol was reviewed and approved by the Ethics Committee “Comité de Ética de la Investigación con Medicamentos de los Hospitales Universitarios Torrevieja y Elche-Vinalopó” (approval number AP11-BIODIAGES). All participants gave their written, informed consent. The research was conducted in accordance with the ethical principles of the Declaration of Helsinki.

Blood samples were obtained between 27 and 29 weeks of pregnancy for biochemistry analysis. Plasma insulin and C-peptide levels were determined using chemiluminescence on an Atellica IM analyzer (Siemens Healthineers). Triglycerides were determined using a spectrophotometric method (Atellica IM analyzer, Siemens Healthineers). Human ELISA kits were used to quantify Adiponectin (Crystal Chem), Tumor necrosis factor-α (TNF-α), Interleukin-6 (IL-6), TGF-β1, and BMP-2 (Invitrogen), activin-A (R&D Systems), and inhibin (Raybiotech Life Inc, Georgia, USA).

2.12. Statistical analysis

After testing for normality, we performed the Student's t-test, one-way analysis of variance (ANOVA), or two-way ANOVA when variables were normally distributed, and the Kruskal–Wallis test when variables were not normally distributed. Statistical analysis was performed using GraphPad Prism Software 10.0 (GraphPad Software, USA). Data are shown as mean ± SEM, and statistical significance was set at P < 0.05 for all analyses. The statistical tests used in each experiment are indicated in each corresponding figure legend.

3. Results

3.1. Glucose homeostasis in a GDM animal model and human GDM patients

Female mice were fed a control diet or HFD three days before mating and throughout pregnancy. These experimental groups will be referred to throughout the text as control pregnant mice (CT-P) or HFD-pregnant mice (HFD-P), respectively. In parallel, non-pregnant (NP) mice were fed either a control diet or HFD during the equivalent length of time to that of pregnant mice. These groups will be named CT-NP and HFD-NP, respectively.

We first studied the impact of high fat feeding in whole-body glucose homeostasis. To this end, an IPGTT was performed at GD15 in pregnant mice or after the equivalent time in non-pregnant mice. We found that HFD administration resulted in marked glucose intolerance in HFD-P mice compared to CT-P, CT-NP, and HFD-NP mice (Figure 1A). This impaired glucose tolerance manifested in a significantly higher glucose area under the curve (AUC) over the course of the IPGTT (Figure 1B). HFD-NP mice exhibited some degree of glucose intolerance compared to CT-NP mice, although the effect was more modest than that observed in HFD-P mice (Figure 1A). As expected, CT-P mice showed slightly higher blood glucose levels in response to glucose load than CT-NP mice [36]. We also observed that fasting glycemic levels were higher in HFD mice than in the control groups (Figure 1C). No changes in glucose tolerance were observed prior to mating (Figure S1A–B). Next, we measured insulin release in response to glucose administration as a surrogate of pancreatic β-cell function. For this purpose, the stimulation index was determined as the ratio of insulin levels at 10 or 30 min after glucose administration to levels at 0 min. We found that insulin secretion was markedly diminished in HFD-P mice compared to CT-P mice, as well as to CT-NP and HFD-NP mice (Figure 1D) at 10 min; the same trend was observed at 30 min (Figure 1D). Fasting serum insulin levels were higher in HFD-P mice than in the other groups which is an indicator of insulin resistance (Figure 1E). No significant differences were found in fed insulin levels apart from an increase associated with pregnancy (Figure 1F). Similar results were found in C-peptide levels (Figure S1C). To further explore insulin sensitivity, an IPITT was performed. Insulin sensitivity decreased in both HFD-NP and HFD-P mice (Figure 1G–H). We found that leptin serum levels were upregulated in pregnant mice although the effect was more pronounced in HFD-P mice (Figure 1I). Adiponectin, which serves as an insulin sensitizer, was clearly decreased in HFD-P mice compared to that in CT-P mice (Figure 1J).

Figure 1.

Study of glucose homeostasis and insulin sensitivity in control and gestational diabetes (GDM) mice. (A) IPGTT in CT-NP (n = 24), HFD-NP (n = 23), CT-P (n = 23), and HFD-P (n = 23) mice at GD15 or equivalent length of time. (B) Area under the curve (AUC) from the IPGTT. (C) Fasting blood glucose levels in CT-NP (n = 24), HFD-NP (n = 23), CT-P (n = 23), and HFD-P (n = 23) mice. (D) In vivo insulin secretion in response to glucose determined as the ratio of insulin levels at 10 or 30 min to levels at 0 min for each mouse (CT-NP, n = 18; HFD-NP, n = 17; CT-P, n = 14; HFD-P, n = 26). (E) Fasting and (F) fed serum insulin levels (fasting insulin: CT-NP, n = 18; HFD-NP, n = 16; CT-P, n = 13; HFD-P, n = 26; fed insulin: CT-NP, n = 24; HFD-NP, n = 21; CT-P, n = 21; HFD-P, n = 25). IPITT in (G) non-pregnant mice (CT-NP, n = 6; HFD-NP, n = 6) and (H) pregnant-mice (CT-P, n = 7; HFD-P, n = 12) at GD15 or equivalent period. (I) Leptin and (J) adiponectin serum levels (leptin: CT-NP, n = 16; HFD-NP, n = 17; CT-P, n = 18; HFD-P, n = 18; adiponectin: CT-NP, n = 17; HFD-NP, n = 17; CT-P, n = 17; HFD-P, n = 17). Data are presented as means ± SEM. Statistical comparisons were performed using Two-way ANOVA followed by Tukey's (A, B, C, D, E, I, J), Fisher's LSD (F), or Bonferroni's (G, H) post hoc tests. In (A) $$$ P < 0.001, $$$$ P < 0.0001 (CT-NP vs. HFD-NP); &&&& P < 0.0001 (CT-NP vs. CT-P); ++++ P < 0.0001 (CT-NP vs. HFD-P); ∗P < 0.05, ∗∗∗P < 0.001, ∗∗∗∗P < 0.0001 (CT-P vs. HFD-P); #P < 0.05, ##P < 0.01 (CT-P vs. HFD-NP); ˆˆˆˆ P < 0.0001 (HFD-NP vs. HFD-P). In the rest of panels ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001 and ∗∗∗∗P < 0.0001.

No changes in the weights of liver (Figure S1D), mesenteric adipose (Figure S1I), or pancreas (Figure S1K) tissue were found in HFD-P mice. In contrast, perigonadal tissue weight increased (Figure S1G), while perirenal adipose tissue weight decreased (Figure S1J) in HFD-P compared to CT-P mice. We observed decreased muscle weight (Figure 1SE-F) and increased retroperitoneal tissue weight in HFD-P vs. non-pregnant mice (Figure S1H).

Similarly, we found that pregnant women diagnosed with GDM showed higher fasting insulin and C-peptide levels than healthy pregnant controls, which is a strong indicator of insulin resistance (Figure 2A–B). Accordingly, we found that the Homeostatic Model Assessment of Insulin Resistance (HOMA-IR), a validated index for assessing insulin resistance, was greater in GDM vs. the control group (Figure 2C). Triglyceride levels were upregulated in GDM (Figure 2D). Adiponectin was significantly decreased in individuals with GDM (Figure 2E). Moreover, TNF-α and IL-6, pro-inflammatory cytokines associated with insulin resistance and diabetes, appeared increased in the GDM group compared to controls (Figure 2F–G).

Figure 2.

Serum measurements in Control and GDM patients. Serum fasting levels of (A) insulin and (B) C-peptide in control and GDM patients. (C) Quantification of HOMA-IR index. Serum levels of (D) TG, (E) adiponectin, (F) TNFα and (G) IL-6. Control (n = 10–11), GDM (n = 13–14). Data are presented as means ± SEM. Statistical comparisons were performed using the Student's t-test. ∗P < 0.05, ∗∗P < 0.01.

These findings indicate that HFD-P mice developed glucose intolerance, aggravated insulin resistance, altered cytokine levels and decreased pancreatic β-cell function, consistent with our findings in GDM patients and what is described in the literature [[37], [38], [39]]. Our results indicate that this animal model is suitable for studying GDM.

3.2. Study of pancreatic β-cell function and mass

Gene expression analysis was conducted in islets isolated from CT-NP, HFD-NP, CT-P, and HFD-P mice, focusing on genes associated with the function and identity of pancreatic β-cells. The most remarkable changes were decreased expression of insulin and Mafa genes in HFD-P compared to CT-P mice (Figure 3A–B). Mafa expression was clearly upregulated in CT-P islets versus islets from non-pregnant mice (Figure 3B). Furthermore, we observed increased expression levels of Pdx1 and Hnf4α (Figure 3C–D) in pregnancy compared to non-pregnancy, the difference was even greater in HFD-P vs CT-P mice for Hnf4α. Gck gene expression was increased in HFD-P compared to HFD-NP mice (Figure 3E), while no differences were found in Glut2 gene levels (Figure 3F). Kir6.2 and Sur1 gene expression was upregulated in islets isolated from HFD-P mice compared to those from non-pregnant mice (Figure 3G–H).

Figure 3.

Study of pancreatic β-cell function in control and GDM mice. Islet mRNA levels of (A)Insulin, (B)Mafa, (C)Pdx1, (D)Hnf4α, (E)Gck, (F)Glut2, (G)Kir6.2, and (H)Sur1 from CT-NP (n = 10), HFD-NP (n = 8), CT-P (n = 11), and HFD-P (n = 12–13) mice at GD16 or equivalent length of time. (I) Ex vivo glucose-stimulated insulin secretion at 2.8 mM, 8.3 mM, and 16.7 mM of glucose in batches of isolated size-matched islets from non-pregnant mice (CT-NP, n = 9–15; HFD-NP, n = 11–15). (J) Islet insulin content measurement in batches of islets from non-pregnant mice (CT-NP, n = 42; HFD-NP, n = 48). (K)Ex vivo glucose-stimulated insulin secretion in batches of isolated size-matched islets from pregnant mice (CT-P, n = 12–15; HFD-P, n = 12–13). (L) Islet insulin content measurement in batches of islets from pregnant mice (CT-P, n = 43; HFD-P, n = 44). Insulin secretion and content were normalized against total islet protein levels. (M) Representative recordings of K_ATP channel activity in pancreatic β-cells isolated from CT-NP, HFD-NP, CT-P, and HFD-P mice. Channel openings are represented by downward deflections, reflecting inward currents due to high K⁺ content of the pipette. The abolition of K_ATP channel activity and generations of action currents at 8 mM glucose was used as a positive control for pancreatic β-cell identity. The graph shows the quantification of K_ATP channel activity in the presence of 8 mM glucose. Glucose effects were measured after 10 min of acute application in each condition. Data are represented as a percentage of activity with respect to resting conditions (0 mM glucose). Experiments were conducted at 32–34 °C (CT-NP, n = 7; HFD-NP, n = 7; CT-P, n = 8; HFD-P, n = 8 cells). (N) Voltage-gated Ca²⁺ currents. Representative recordings of voltage-gated Ca²⁺ currents in response to 500 ms depolarizing pulses (−60 mV to +80 mV from a holding potential of −70 mV [inset]). Average relationship between voltage-gated Ca²⁺ current density (currents in pA normalized to cell size in pF) and voltage of pulses (CT-NP, n = 22; HFD-NP, n = 31; CT-P, n = 15; HFD-P, n = 17 cells). Data are presented as means ± SEM. Statistical comparisons were performed using Two-way ANOVA followed by Tukey's (A-H, M, and N) or Sidak's (I, K) post hoc tests. Student's t-test was used in L. ∗P < 0.05, ∗∗P < 0.01, and ∗∗∗∗P < 0.001. In (N) $ P < 0.05, $$$ P < 0.001, $$$$ P < 0.0001 (CT-NP vs. HFD-NP); & P < 0.05 (CT-NP vs. CT-P); + P < 0.05, ++ P < 0.01, +++ P < 0.001 (CT-NP vs. HFD-P).

We then analyzed pancreatic β-cell function. We measured glucose-stimulated insulin secretion in isolated islets exposed to different glucose concentrations. Islets from HFD-NP mice displayed reduced insulin secretion in response to glucose compared to islets from CT-NP mice; although the effect was only significant at the highest glucose concentration (16.7 mM) (Figure 3I). The decline of insulin secretion capacity was more evident in islets from HFD-P mice, which exhibited decreased insulin release in response to both 8.3 and 16.7 mM glucose concentrations compared to islets from CT-P mice (Figure 3K). In addition, basal insulin secretion was also found to be downregulated (Figure 3K). Insulin content was reduced in islets from HFD-P mice (Figure 3L), while no effect was observed in HFD-NP mice (Figure 3J).

It is well established that the K_ATP channel serves as a metabolic sensor playing a central role in insulin secretion. Here, we evaluated K_ATP channel activity using the patch clamp technique. We found that pancreatic β-cells from HFD-P mice showed increased activity of the K_ATP channel and decreased channel closure in response to glucose compared to those from CT-P, CT-NP, and HFD-NP. This was accompanied with a lower occurrence of action currents (Figure 3M), indicating reduced action potential firing, which is instrumental in activating voltage-gated Ca²⁺ channels and insulin release. No differences were observed between CT-P and CT-NP nor HFD-NP mice, suggesting that the effect was specific for HFD-P β-cells (Figure 3M). Additionally, voltage-gated Ca²⁺ currents were analyzed as they are also a key part of the molecular pathway triggering insulin release. As shown in Figure 3N, Ca²⁺ currents in response to depolarizing voltages pulses were decreased in HFD-NP β-cells compared to CT-NP cells. A similar effect was observed in HFD-P β-cells (Figure 3N). In contrast, voltage-gated K⁺ currents were similar in all experimental groups (Figure S2A–B). The observed alterations in ion channel expression and activity provide an explanation for the diminished insulin secretion observed in mice with GDM.

Next, we examined potential pancreatic β-cell mass changes among the different experimental groups. Morphometric analysis revealed that both pancreatic β-cell area and mass were increased in CT-P and HFD-P vs. CT-NP mice (Figure 4A–B). We observed decreased β-cell area in HFD-NP vs. CT-NP mice (Figure 4A), but no differences in β-cell mass (Figure 4B). The number of proliferating pancreatic β-cells, identified by double staining of insulin and Ki67, were increased in both CT-P and HFD-P mice (Figure 4C). Of note, despite the increment compared to non-pregnant conditions, HFD-P pancreatic β-cells showed a marked decrease in proliferation when compared to those from CT-P mice (Figure 4C).

Figure 4.

Pancreatic β-cell mass and proliferation in control and GDM mice. (A) Quantification of pancreatic β-cell area and (B) pancreatic β-cell mass in CT-NP (n = 7), HFD-NP (n = 7), CT-P (n = 7) and HFD-P (n = 7) at GD15 or equivalent length of time. (C) Representative images of pancreas sections stained for insulin (red), Ki67 (green), and nuclei with Hoechst (blue). Scale bar, 20 μm. The graph shows the quantification of pancreatic β-cell proliferation (CT-NP, n = 8; HFD-NP, n = 8; CT-P, n = 7; HFD-P, n = 7). Data are presented as means ± SEM. Statistical comparisons were performed using Two-way ANOVA followed by Tukey's or Sidak's post hoc tests. ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001, and ∗∗∗∗P < 0.0001.

3.3. Maternal hormonal changes in GDM

Next, we evaluated possible alterations in the plasma levels of maternal hormones, given their important role in adaptive changes and glucose control in pregnancy. We observed that placental lactogen and 17β-estradiol (Figure 5A–B) were decreased in HFD-P mice compared to those in controls. In contrast, progesterone levels were increased in HFD-P mice (Figure 5C), while no significant changes in prolactin levels were found (Figure 5D).

Figure 5.

Maternal hormonal changes in GDM mice. Serum levels of (A) placental lactogen (CT-P, n = 18; HFD-P, n = 18), (B) 17 β-estradiol (CT-P, n = 12; HFD-P, n = 12), (C) progesterone (CT-P, n = 19; HFD-P, n = 19), and (D) prolactin (CT-P, n = 17; HFD-P, n = 17) measured at GD16. Data are presented as means ± SEM. Statistical comparisons were performed using the Student's t-test (A and B), or Mann–Whitney test (C). ∗P < 0.05, ∗∗P < 0.01.

3.4. TGF-β signaling in GDM

We quantified the circulating levels of key TGF-β family ligands, TGF-β1, activin-A, and inhibin, and found no differences in TGF-β1 serum levels among mice groups (Figure 6A). However, a marked increase in activin-A levels was quantified in HFD-P vs. CT-P mice. In contrast, in non-pregnant mice, HFD produced the opposite effect with a clear decrease in activin-A in HFD-NP compared to CT-NP mice (Figure 6B). A similar result was observed for inhibin levels, a marked increase in the GDM group (HFD-P) compared to the control group (CT-P), and a decline in HFD-NP vs. CT-NP mice. Furthermore, a marked decrease associated with pregnancy was also observed in CT-P compared to CT-NP mice for inhibin (Figure 6C). To compare with humans, we measured serum levels of the same TGF-β family ligands in GDM patients. We observed that TGF-β1 levels were significantly decreased in GDM women (Figure 6D). In addition, activin-A and inhibin concentrations were upregulated in GDM patients (Figure 6E–F). We found that BMP-2 levels were enhanced in GDM women; however, no statistical significance was reached (Figure 6G).

Figure 6.

Serum TGF-β ligands in mice and human GDM patients. Mice serum levels of (A) TGF-β1 (CT-NP, n = 20; HFD-NP, n = 13; CT-P, n = 21; HFD-P n = 19), (B) activin-A (CT-NP, n = 22; HFD-NP, n = 19; CT-P, n = 22; HFD-P n = 28), and (C) inhibin (CT-NP, n = 18; HFD-NP, n = 17; CT-P, n = 12; HFD-P n = 17). Serum levels of (D) TGF-β1, (E) activin-A, (F) inhibin, and (G) BMP-2 in control and GDM patients (Control (n = 10–11); GDM (n = 13–14)). Data are presented as means ± SEM. Statistical comparisons were performed using Two-way ANOVA followed by Sidak's post hoc test (B, C), the Student's t-test (D, E), or Mann–Whitney test (F). ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001.

We next evaluated potential changes in the expression and activity of the main components of the TGF-β/Smad pathway at the endocrine pancreas level in mice. The islets of Langerhans from the different experimental groups (CT-NP, HFD-NP, CT-P, and HFD-P) were isolated, and gene and protein expression assays were conducted.

The most remarkable changes were found in the islets from the GDM group (HFD-P). Increased mRNA expression levels of Tgf-βRI (also known as Activin receptor-like kinase 5 or Alk5) and Smad2 were observed in HFD-P mice islets compared to CT-P mice (Figure 7A–B). Augmented levels of Smad3 were also found, although the difference was only significant compared to CT-NP mice (Figure 7C). Smad4 was found to be slightly decreased in pregnant compared to that in HFD-NP mice (Figure 7D). Furthermore, a marked trend of increased Smad7 expression was found in HFD-P compared to CT-P mice (Figure 7E). No significant changes in Tgf-β1, Tgf-βRII, Alk4, activin-A (inhba), and inhibin (inha) gene expression were found in HFD-P vs. CT-P mice (Figure S3A–E).

Figure 7.

Pancreatic TGF-β/Smad signaling in control and GDM mice. Pancreatic islet mRNA expression levels of (A)Tgf-βRI, (B)Smad2, (C)Smad3, (D)Smad4, and (E)Smad7 from CT-NP (n = 10), HFD-NP (n = 8), CT-P (n = 11), and HFD-P (n = 12–13) mice. Representative western blot images and corresponding quantification showing (F) pSmad2, Smad2, and pSmad2/Smad2 protein levels in islets isolated from the different experimental mice groups (CT-NP n = 11; HFD-NP, n = 9; CT-P, n = 8; HFD-P, n = 11), (G) TGF-βRI protein levels in islets isolated from the different experimental mice groups (CT-NP, n = 13; HFD-NP, n = 13; CT-P, n = 12; HFD-P, n = 13). All measurements were performed at GD16 or an equivalent length of time. Data are presented as means ± SEM. Statistical comparisons were performed using Two-way ANOVA, followed by Sidak's (A, B, F, and G) or Fisher's LSD (C, D, and E) post hoc tests. ∗P < 0.05.

We assessed whether TGF-β/Smad signaling could be differently regulated in GDM (HFD-P group) and/or pregnancy by evaluating changes in the signaling pathway activation at the protein level. Increased phosphorylation of Smad2, as well as elevated pSmad2/Smad2 levels, were observed in islets from HDF-P mice compared to those from CT-P mice (Figure 7F). This agrees with the upregulated expression of TGF-βRI protein (Figure 7G). No differences were observed in pSmad3 or the pSmad3/Smad3 ratio (Figure S4A), nor in TGF-βRII (Figure S4B) or ALK4 (Figure S4C) protein expression.

3.5. Pharmacological inhibition of TGF-β/Activin-Smad2 signaling in pancreatic islets

As activin-A and inhibin serum levels were found to be elevated in both mice and pregnant women with GDM, it is important to consider the potential implications of these findings. We hypothesized that they may be upstream signaling transducers of pancreatic Smad2 activation. Thus, we incubated pancreatic islets in the presence of activin-A and inhibin and observed that pSmad2 and the pSmad2/Smad2 ratio at protein level were dramatically increased compared to controls (Figure S5A). Next, we determined the impact of Galunisertib (GL; LY2157299 monohydrate), a small molecule inhibitor (SMI) of TGF-βRI kinase that specifically downregulates the phosphorylation of Smad2 [40]. We found that GL drastically inhibited the phosphorylation of Smad2 in the presence of the ligands as well as the ratio of pSmad2/Smad2 (Figure 8A). GL also decreased TGF-βRI expression under the same experimental conditions (Figure 8B).

Figure 8.

Inhibition of TGF-β/Activin-Smad2 signaling in pancreatic islets of Langerhans. Representative western blot images and corresponding quantification showing (A) pSmad2, Smad2, and pSmad2/Smad2 or (B) TGF-βRI protein levels in islets cultivated for 48 h in the presence of vehicle (CT), galunisertib (GL) (10 μM), activin-A (5 nM) + inhibin (5 nM) (A + I) or activin-A + inhibin + galunisertib (A + I + GL) (n = 6–8). (C) Ex vivo glucose-stimulated insulin secretion in response to 2.8 mM, 8.3 mM, and 16.7 mM of glucose in CT, GL, A + I, and A + I + GL batches of treated islets (n = 12–18). (D) Insulin content measured in CT, GL, A + I, and A + I + GL batches of treated islets (n = 42–49). (E) Representative images of islet cells stained for insulin (red), Ki67 (green), and nuclei with Hoechst (blue). Scale bar, 20 μm. The graph shows the quantification of pancreatic β-cell proliferation (CT, n = 6712; GL, n = 6123; A + I, n = 5969 and A + I + GL, n = 6559 cells). Data are presented as means ± SEM. Statistical comparisons were performed using One-way ANOVA with Tukey's post hoc test (A, B, and D), Kruskal–Wallis with Dunn's post hoc test (E), or Two-way ANOVA with Tukey's post hoc test (C). ∗P < 0.05, ∗∗P < 0.01, ∗∗∗∗P < 0.0001; #P < 0.05 Student's t-test (A).

Accordingly, we investigated whether the TGF-β family ligands may impact pancreatic β-cell function and whether GL may act as a modulator of the ligand-induced response. Thus, we quantified glucose-stimulated insulin secretion and found that islets treated with both activin-A and inhibin showed a marked decrease in insulin release in response to the highest glucose concentration tested (16.7 mM; Figure 8C) and a slight decrease in insulin content levels (Figure 8D). Treatment with GL blunted this response, suggesting a protective effect (Figure 8C). We also observed that treatment with both ligands led to diminished pancreatic β-cell proliferation and that incubation with GL prevented this effect. Treatment with the inhibitor alone increased pancreatic β-cell replication (Figure 8E).

We next explored the individual effects of activin-A and inhibin on TGF-β pathway activation and subsequent Smad2 phosphorylation in pancreatic islets. Islets treated with activin-A for 48 h showed increases in pSmad2 levels and the pSmad2/Smad2 ratio. In contrast, inhibin treatment did not have any effect (Figure S6A). No changes were observed in TGF-βRI expression neither in response to activin-A nor inhibin (Figure S6B). In addition, we found that GL inhibited the increase in pSmad2 levels and the pSmad2/Smad2 ratio induced by activin-A (Figure S6C). Decreased expression of TGF-βRI was observed when islets were incubated in the presence of activin-A and GL (Figure S6D). Furthermore, activin-A promoted a marked decrease in pancreatic β-cell insulin secretion (Figure S6E) and proliferation rate (Figure S6F), an effect which was abolished by GL (Figure S6E–F).

4. Discussion

The current study first focused on developing a preclinical animal model of GDM that closely resembles the pathophysiology of the disease in women. Our research revealed that HFD treatment three days before and during pregnancy results in a GDM-like phenotype, including elevated fasting blood glucose levels, marked glucose intolerance, aggravated insulin resistance, impaired insulin secretion, and compromised β-cell remodeling.

Different animal models have been described in the literature for GDM study; however, they all include certain limitations hindering the accurate pathophysiology of the disease. Consequently, GDM modeling remained a challenge when we approached this work. The main induction strategies described so far for GDM development include surgical, chemical, genetic and diet-induced models [[41], [42], [43]]. Surgical and chemical methods display significant disadvantages as both evoke severe and irreversible hyperglycemia associated with acute and extensive pancreatic β-cell death; more similar to T1DM than GDM, since, unlike T1DM, GDM is not caused by a lack of insulin but by the inability of pancreatic β-cells to adapt to the insulin resistance inherent to pregnancy [[41], [42], [43]]. Genetic manipulation normally leads to severe impairment of glucose tolerance even in the non-pregnant state, and cannot fully reflect the complex interaction between polygenic and environmental factors, thus limiting their use for GDM studies [42]. Nutritional manipulation, including HFD, and high fat and high-sucrose diet (HFHS), has been shown to better mimics the core characteristics of human GDM and the metabolic context of the disease, considering that obesity and diet quality are important risk factors for GDM. This is true for rodents but also for larger animal models, including sheep and dogs, which may facilitate transfer of findings to clinical management [43]. Furthermore, diet-induced models have also been successful for the study of GDM metabolic consequences for the offspring in the short and long-term [42].

To date, the most common experimental procedures involving diet include administration of HFD (45% or 60%) or HFHS for four- or six-weeks prior to mating and during pregnancy. This dietary intervention typically leads to the development of glucose intolerance of varying severity in the mid-to late stages of pregnancy [[44], [45], [46], [47], [48], [49]] and increased fasting insulin levels [46,49]. However, elevated blood glucose levels and glucose intolerance are observed prior to mating [47,49], a finding that aligns with our observations following a four-week treatment with HFD in female mice prior to conception (unpublished data). This is important to consider since perturbed glucose metabolism before the onset of pregnancy does not accurately represent most cases of human GDM, but rather reflects a pre-diabetes condition which may be unveiled during pregnancy. By contrast, in our GDM animal model, mice were challenged with HFD (60%) for three days prior to mating and throughout pregnancy, with no glucose tolerance impairment before pregnancy. In line with our results, a study conducted by Pennington and collaborators have shown that acute exposure to HFHS, one week prior to mating and during pregnancy, promoted glucose intolerance and decreased insulin response to glucose at GD17.5 [50]. It is also important to remark that GDM is a form of glucose intolerance caused by insulin resistance and pancreatic β-cell malfunction during pregnancy [37,38]. Our GDM mice developed glucose intolerance associated with increased insulin resistance and manifested reduced in vivo insulin release in response to glucose, reflecting the basis of GDM. Furthermore, the observed adverse metabolic changes were dependent on pregnancy, as is seen during GDM. Proof of this, is that glucose intolerance and pancreatic β-cell malfunction were much more evident in pregnant mice fed with HFD compared to non-pregnant mice fed the same diet for an equivalent period.

We isolated pancreatic islets from GDM mice and found that the capacity of the β-cells to secrete insulin in response to glucose was dramatically diminished. Basal insulin secretion as well as insulin content was also disrupted. Decreased insulin synthesis and secretion may be attributable to various factors. First, we observed that the gene expression of both insulin and Mafa, a key regulator of insulin transcription and a master regulator of genes important for β-cell function, were markedly decreased in β-cells from GDM mice. The INS gene has been considered as a candidate for the study of GDM in humans. The INS-VNTR class III, a polymorphic sequence associated with less efficient transcription of the INS gene, has been shown to be more frequent in women who develop GDM [51], although discrepancies can be found depending on the genetic background. Second, we reported that electrical ion channel activity was altered in pancreatic β-cells from GDM mice. Glucose-induced inhibition of K_ATP channel activity was impaired, leading to reduced β-cell excitability in response to glucose, which is consistent with the reduced insulin release observed. This finding is of great relevance given the central role of K_ATP channel in coupling metabolism and insulin secretion, and that its impaired regulation is a cause of diabetes [52]. Furthermore, voltage-gated Ca²⁺ currents were decreased, which may contribute to reduced insulin secretion capacity in GDM β-cells. These changes, described here for the first time, may help to explain the inability of β-cells to adapt to the increased insulin demand imposed by pregnancy leading to GDM.

We also found that pancreatic β-cell proliferation rate was clearly increased at GD15 in pregnant compared to non-pregnant mice. It is well described in the literature that proliferation of pancreatic β-cells starts around GD9 with a maximum peak around days GD12-15 [8]. However, we noted that β-cell proliferation in GDM mice was markedly lower than that in CT-P mice, suggesting that the plasticity of pancreatic β-cells in GDM was reduced.

Another important observation of our study was that maternal hormone levels, which play a central role in β-cell adaptation in pregnancy, were deregulated in GDM. We observed decreased placental lactogen and 17β-estradiol circulating levels while progesterone was increased. Placental lactogen concentration rises during normal gestation and correlates with enhanced β-cell proliferation and β-cell insulin secretion. The role of estradiol and progesterone in pregnancy is less understood [8], but it is proposed that estradiol increases insulin content and secretion, and exerts protective effects against oxidative stress and apoptosis [8,53]. Likewise, progesterone displays pro-apoptotic effects in β-cells which may contribute to reducing β-cell mass in the peri-partum period [8], and counterbalance the stimulatory effects of lactogens on β-cells during pregnancy [54].

Next, we investigated the role of TGF-β signaling in GDM. To our knowledge, this is the first study to demonstrate that TGF-βRI/Smad2 signaling was upregulated in GDM pancreatic islets compared to controls. This was evidenced by enhanced Smad2 and TGF-βRI mRNA expression levels as well as increased Smad2 activation as indicated by elevated pSmad2, pSmad2/Smad2 and TGF-βRI protein levels. Notably, these changes concurred with a decline in pancreatic β-cell function, including decreased insulin secretion and content, altered ion channel activity, and diminished β-cell proliferation. The fact that inhibiting Smad2 signaling abrogated most of these β-cell alterations suggests that Smad2 may represent an important target for GDM. In agreement with this notion, previous studies have revealed that mice with specific Smad2 ablation in pancreatic β-cells presented a higher β-cell proliferation rate [27,55], particularly after pancreatectomy [27]. More recently, it has been demonstrated that pancreatic β-cell-specific Smad2 knockout mice manifested improved glucose tolerance, increased in vivo and ex vivo insulin secretion, and enhanced β-cell mass and proliferation [22]. Smad2 inactivation also promoted differential expression of certain transcription factors, which are essential for pancreatic β-cell function and identity, such as increased expression of the Mafa gene [22]. In line with these findings, we observed that Smad2 upregulation in GDM was associated with a significant decline in Mafa and insulin gene expression. This may be relevant at the mechanistic level, as it has been demonstrated that Smad2 physically interacts with Mafa, thereby inhibiting transcriptional activation of the insulin gene promoter [56]. In a similar manner, TGF-βRI knockdown in human pancreatic β-cells upregulated Mafa expression through the AKT/FOXO1 pathway and inhibited dedifferentiation [57].

Importantly, other studies have shown that Smad3 plays a role in regulating pancreatic β-cell proliferation, apoptosis and insulin secretion [[25], [26], [27]] and that attenuation of Smad3 signaling may have beneficial effects in glucose homeostasis in db/db and HFD-fed mice [23,24]. Here we reported, for the first time, elevated pSmad2 activation in GDM pancreatic islets without changes in pSmad3, suggesting a specific role for Smad2 in the impairment of β-cell function and proliferation in GDM, and therefore an important target for the disease.

We also observed elevated serum levels of activin-A and inhibin, both in our GDM mice model and in human GDM patients, suggesting their role as essential upstream signaling transducers of pancreatic Smad2 activation in GDM. This finding is significant for several reasons. First, these molecules may serve as biomarkers reflecting the pathophysiological mechanisms of GDM leading to glucose intolerance and impaired pancreatic β-cell function. Second, since these are readily obtained from blood samples, they may improve identification of women at higher risk of developing GDM. Current information on TGF-β signaling in GDM is very scarce, with some evidence supporting increased serum maternal levels of activin-A in GDM, which were normalized after insulin therapy and restoration of normal blood glucose levels [58].

Previous studies claim that activin-A behaves as a negative regulator of pancreatic β-cell differentiation and function in adult islets. Activin-A treatment significantly decreased mature β-cell gene expression and insulin secretion in MIN6 cells, and mouse and human islets [59]. Others have reported a stimulatory effect of activin-A on insulin secretion in rats [60] and human islets [61,62]. Contradictory results may be due to treatment duration, as reduced insulin secretion was reported after a prolonged activin-A treatment of 72 h [59], while acute exposure displayed the opposite effect [60,61]. In human islets, acute exposure to activin-A decreased insulin secretion, an effect which was reversed by pre-treatment with the inhibitor follistatin [62]. Our findings support the notion that the addition of exogenous activin-A, alone or in combination with inhibin, results in reduced insulin secretory capacity of β-cells and decreased proliferation. No previous work has explored the effects of inhibin on pancreatic β-cell function.

Finally, our findings indicate that pancreatic Smad2 activation is upregulated in GDM and that targeting TGF-β/Activin-Smad2 signaling may offer novel therapeutic avenues for this pathology. Different approaches have been addressed to block TGF-β signaling, including neutralizing antibodies, antisense oligonucleotides and small molecular inhibitors (SMI) [63]. Among SMIs, galunisertib (LY2157299 monohydrate) is an inhibitor of TGF-βRI kinase, which specifically downregulates the phosphorylation of Smad2, and is under clinical investigation given its tolerability and pharmacodynamic profile with encouraging results [40,64]. Our in vitro studies demonstrate that treatment with this inhibitor increases pancreatic β-cell proliferation and prevents the decrease in β-cell number that occurs in response to TGF-β ligands. This may be clinically relevant if we consider that Smad2 activation restrains β-cell mass expansion in GDM. Growing evidence indicates that SMIs of TGF-β signaling promote increased β-cell replication in both mouse and human β-cells [65,66] and help to restore mature β-cell identity by inducing expression of β-cell transcription factors [67]. A recent study demonstrated that galunisertib can rescue apoptosis induced by the loss of GLIS3, a gene which has been associated with type 2, type 1, and neonatal diabetes, in β-like cells [68]. Importantly, our results reflect additional beneficial effects on blocking TGF-βRI/Smad2 pathway activation, as we found that galunisertib also promoted a marked improvement in pancreatic β-cell function with augmented insulin secretion in response to glucose.

5. Conclusions

Our findings reveal, for the first time, a concurrent elevation of circulating activin-A and inhibin serum levels in both women with GDM and a corresponding preclinical GDM model. This dysregulation is linked to enhanced Smad2 activation within pancreatic islets, which correlates with glucose intolerance, impaired pancreatic β-cell function and reduced β-cell regenerative capacity. Importantly, we show that pharmacological inhibition of the TGF-β/Activin-Smad2 signaling pathway improves β-cell function and promotes plasticity, highlighting its promise as a novel therapeutic target for GDM.

CRediT authorship contribution statement

Talía Boronat-Belda: Writing – review & editing, Visualization, Validation, Methodology, Investigation, Formal analysis. Hilda Ferrero: Writing – review & editing, Visualization, Validation, Methodology, Investigation, Formal analysis. Sergi Soriano: Writing – review & editing, Visualization, Validation, Methodology, Investigation, Formal analysis. Elena Ribes-García: Writing – review & editing, Resources, Methodology. Rubén Betoret-Gustems: Writing – review & editing, Resources, Methodology, Funding acquisition. Daniel Martínez-Bañón: Writing – review & editing, Resources, Methodology. Mónica Serrano-Selva: Writing – review & editing, Resources, Methodology. Juan Martínez-Pinna: Writing – review & editing, Investigation, Formal analysis. Ángel Nadal: Writing – review & editing, Resources, Funding acquisition. Iván Quesada: Writing – review & editing, Visualization, Resources, Methodology, Investigation, Funding acquisition, Formal analysis. Paloma Alonso-Magdalena: Writing – review & editing, Writing – original draft, Visualization, Validation, Supervision, Project administration, Methodology, Investigation, Funding acquisition, Formal analysis, Conceptualization.

Funding sources

This work was supported by grants PID2020-113112RB-I00 (PAM) and PID2023-146795OB-I00 (PAM and IQ) funded by Ministerio de Ciencia, Innovación y Universidades-Agencia Estatal de Investigación MICIU/AEI/10.13039/501100011033; ILISABIO22_AP11 Acción preparatoriaUniversidad Miguel Hernández (UMH) y Fundación para el Fomento de la Investigación Sanitaria y Biomédica de la Comunidad Valenciana (Fisabio)_ UNISALUT 2022 (PAM and RBB); CIPROM/2023/27 Generalitat Valenciana programa PROMETEO (PAM and AN).

CIBER is an initiative of Instituto de Salud Carlos III (Ministerio de Sanidad, Spain).

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following are the Supplementary data to this article:

Data availability

Data will be made available on request.