Deferoxamine addresses metabolic dysregulation and urinary tract infections in weight-associated gestational diabetes mellitus

Abstract

Background

The increasing prevalence of gestational diabetes mellitus (GDM), particularly among overweight or obese individuals, poses significant health risks. Excess iron contributes to oxidative stress, inflammation, and disruptions in immune and metabolic functions in GDM. Deferoxamine (DFO), an iron chelator, may offer a therapeutic solution by restoring immune and metabolic balance.

Methods

We conducted a comprehensive multi-omics analysis using GEO transcriptomic data, applying Weighted Gene Co-expression Network Analysis (WGCNA) and gene set enrichment analysis to identify key immune-metabolic genes. Molecular docking experiments with DFO were performed using AutoDock Vina, and interactions were visualized in PyMOL. Various in vitro assays—CCK-8, qRT-PCR, Western blot, immunofluorescence, ELISA, and colony formation tests—were conducted under high glucose conditions to assess the effects of DFO, focusing on LAMA3 and the PI3K/AKT signaling pathway.

Results

Key genes such as CDR2L, LIMCH1, LDLR, and LAMA3 were identified as being improperly regulated in cases of gestational diabetes mellitus (GDM). DFO demonstrated a significant affinity for these targets, especially LDLR. Functionally, DFO was found to improve cell survival during hyperglycemic stress, mitigate oxidative stress, and lower the concentrations of inflammatory cytokines like IL-6, IL-17, and IL-23. Notably, silencing or blocking LAMA3 reversed these effects, inhibiting the PI3K/AKT pathway, increasing apoptosis markers, and decreasing cell proliferation.

Conclusions

DFO holds potential as a targeted treatment for GDM associated with obesity by addressing iron excess and immune-metabolic dysregulation. LAMA3 plays a crucial role in mediating DFO’s anti-inflammatory and survival-promoting effects via the PI3K/AKT pathway. Further clinical studies are needed to explore DFO’s therapeutic potential in GDM.

Graphical Abstract

Supplementary Information

The online version contains supplementary material available at 10.1186/s40001-025-03013-4.

Highlights

• The research uses a multi-omics strategy to identify important immune genes (CDR2L, LIMCH1, LDLR, LAMA3) and immune pathways associated with the weight-related GDM

• As per the study, DFO not only regulates iron, reduces oxidative stress, and restores PI3K/AKT pathways but also affects immune responses, thus providing a new immunotherapy for GDM.

• Studies that reviewed both metabolic effects and immune modulating effects of DFO may pave the way for new personalized immunotherapy options for GDM.

• We developed a robust framework that integrates multi-omics approaches with experimentation, providing the means to advance precision medicine and build immune-targeted therapies for pregnancy-associated metabolic disorders.

Supplementary Information

The online version contains supplementary material available at 10.1186/s40001-025-03013-4.

Background

A growing number of women in low-income countries are developing gestational diabetes. Such concerns arise out of the increasing obesity rates, inactive lifestyles, and advanced maternal age [1–4]. Gestational diabetes mellitus (GDM) refers to a form of glucose intolerance that is diagnosed during pregnancy through an oral glucose tolerance test (OGTT) [5–8]. The short-term perils for the mother and baby are well acknowledged. However, the long-term risks are ignored. Mothers who have gestational diabetes are more likely to have gestational hypertension, preterm birth, and cesarean section delivery [9]. Notably, the risk of obesity and type 2 diabetes in later life is increased in children whose mothers had GDM in pregnancy [10, 11].

Development of GDM is influenced by the interaction of genetic, environmental, and immunometabolic factors [12]. Factors like obesity, physical inactivity, and older maternal age are common contributors to macrocosmic effect [13]. The varying rates of GDM between ethnic groups and places indicate the significance of genetic and environmental factors [14, 15]. Iron helps cells make energy and transport oxygen. However, too much at one time can damage tissues and cause chronic inflammation [16, 17]. The buildup of iron is being more frequently linked to metabolic disorders like type 2 diabetes [18]. The iron accumulation is increasingly associated with the occurrence of metabolic disorders, including type 2 diabetes [19]. Deferoxamine (DFO), an FDA-approved iron chelator, binds to ferric ions to lessen oxidative damage caused by iron [20]. Research studies have shown that it can lower oxidative stress and inflammation in a range of disease models, including diabetes [21]. DFO in GDM may reduce oxidative damage and replenish immune and metabolic functions.

GDM is primarily marked by insulin resistance, which is exacerbated by systemic inflammation, oxidative stress, and lipid metabolism disruptions at the molecular level [22, 23]. Research shows that iron metabolism plays a crucial role in GDM's immune response and oxidative stress. Disrupting iron balance leads to excess iron accumulation, producing reactive oxygen species (ROS) that impact lipid peroxidation. This process also activates immune cells, triggering inflammation and insulin resistance [24, 25]. Targeting iron-related oxidative and immune pathways may offer therapeutic potential [26–30]. Due to GDM's complex mechanisms, omics data integration (transcriptomics, proteomics, metabolomics) has been used to explore its underlying processes and identify potential biomarkers for treatment [31–35]. Multi-omics approaches, combined with advanced bioinformatics, help classify immune profiles and identify therapeutic targets, enabling personalized immunotherapy development [36–38]. Data-driven strategies are crucial for personalized medicine, ensuring interventions are tailored to individual immune-metabolic profiles [39–42]. GDM affects metabolism and increases the risk of urinary tract infections (UTIs). DFO is considered a potential treatment for GDM-related UTIs due to its anti-inflammatory and possibly anti-infectious properties. Studies show DFO, combined with diabetes medications, helps treat drug-resistant infections like MRSA and CRAB [43]. DFO also prevents bacterial biofilm formation, which contributes to chronic UTI persistence, highlighting its potential for treating both metabolic and infectious issues in GDM. However, it should be used cautiously during pregnancy due to risks like ototoxicity, ocular toxicity, and growth inhibition at high doses. Animal studies suggest teratogenic effects, and DFO should only be used when the benefits outweigh the risks, with fetal monitoring in future trials. Repurposing drugs like DFO saves time and money, and it holds promise as an immunometabolic agent. Customized GDM treatment plans could improve medication effectiveness, reduce infections, and improve outcomes for both mother and baby.

Materials and methods

Identification of disease targets in gestational diabetes mellitus associated with body weight

The transcriptomic microarray data of gestational diabetes mellitus and body weight were downloaded from Gene Expression Omnibus database (GEO) [44–48]. A systematic review was done using keywords: “obesity” and “gestational diabetes. Humans.” In R, using the Bioconductor package, background correction, normalization, and expression values were calculated during data processing. Using the limma package, we evaluated mRNA differential expression across sample groups with a significance threshold of P-value < 0.05 and fold change of 1.50 (|log2FC|≥ 0.58) [49, 50]. Genes were considered upregulated and downregulated with log2FC (Fold Change) ≥ 0.58, and log2FC ≤ -0.58, respectively. The package heatmap was used for clustering and visualizing differentially expressed genes. We took the log of the p-value and used log p-values and log2FCs to find differently expressed genes which were upregulated, downregulated, and unchanged. We implement a gene set enrichment analysis with the clusterProfiler package in R to study differences in biological functionality of GDM associated with body weight (BW) as compared with normal (N), utilizing “hallmark gene sets.”

Construction of weighted gene co-expression network

This study explored the relationship between gene expression and body mass in gestational diabetes mellitus. The genes were chosen based on having expression levels above the upper quartile of variance. The authors used “WGCNA” package of R software for the development of weighted gene co-expression network (WGCN) of GDM and body weight [51, 52]. To cluster the samples we did the clustering analysis, and to find the appropriate soft-threshold power (β) for constructing the adjacency matrix, a scale-free network model was employed. A topological overlap matrix (TOM) of genes was computed to find the similarity of genes. This made it possible to construct a dendrogram of genes with hierarchical clustering and modules’ identification. The tree dynamic cut technique was used to detect and merge gene module. For every gene module, ME was calculated. A correlation analysis was carried out to assess how modules' MEs correlate with clinical features of GDM patients. The relationship between the MEs and the clinical features was assessed using Pearson correlation coefficients. The researchers examined the genes in this module because the relationship of one key module with GDM with respect to body weight was strong.

Relative expression of core targets and ROC evaluation

We performed an intersection analysis using R and Perl programming to assess the relationship between WGCNA hub targets linked to BW-related GDM genes and genes with varying expressions correlated with GDM and BW [53]. A Venn diagram was illustrated using Venny 2.1 to display the findings and associations of these genes [44, 54]. This research can help understand how GDM develops in connection with weight and also help identify a possible marker for therapeutically targeting the disease. Next, weight-associated GDM expression matrix data were acquired from the GEO database for validation. We studied genes already linked to weight-related GDM. The ggpubr package (in R) was used to evaluate the expression of major targets in GDM samples. Graphs were created depicting their relative expression levels. In addition, we combined the normalized expression data of essential GDM targets with clinical data. ROC curves were constructed using R packages such as survival, caret, glmnet, survminer, and survivalROC in order to assess the diagnostic performance and accuracy of core genes for weight-related GDM.

Immune infiltration and immune function analysis of core targets in weight-related gestational diabetes mellitus

Researchers used advanced bioinformatics approaches to analyze the immune cell composition of weight-related GDM. Using the deconvolution algorithm-based CIBERSORT package, standardized gene expression data were used for the calculation of the relative proportions of various immune cells in the tissues [55, 56]. A matrix of genes deregulated in GDM, notably those involved in immune function. I used Perl scripts for identifying immune cells, while the microarray data were managed through background correction, normalization and computation of expression values using limma package in R. Using CIBERSORT we have evaluated the immune cell profiles and generated bar plots. Heatmaps of the immune cell distribution were produced using the pheatmap package. The package corrplot was used to visualize the immune cells interactions in weight-associated GDM. Graphs were devised to show how genes relate. Using the vioplot package, we generated violin plots that examine immune cells’ expression levels in the GDM and normal tissues. To provide a quick understanding of the functional roles of immune cells in weight-associated GDM, immune-related functional analysis of the key targets was performed using the limma, GSVA, GSEABase, pheatmap, and reshape2 packages which could offer insights for precision therapy for low GDM.

Gene ontology biological function analysis and Kyoto Encyclopedia of Genes and Genomes pathway enrichment

To uncover the molecular mechanisms behind the effects of a bioactive agent on weight-related GDM, various bioinformatics approaches were employed. In order to reveal the biological properties of the identified targets, GO analysis was performed on common targets that were GDM weight related. The clusterProfilerGO.R package within R and Perl scripts were used for this [57, 58]. Our study threw light on cellular component (CC), molecular function (MF), and biological process (BP). Using clusterProfiler KEGG R package, we also performed pathway enrichment analysis based on the KEGG and visualized using pathview to determine the enrichment factors associated with key pathways and the likely biological functions and acting mechanisms of weight-related GDM common targets.

Core protein drug sensitivity screening

The process of docking an experimental compound to a biological target can be improved in terms of time and resources by virtually screening compounds for bioactivity [59, 60]. For this study, the three-dimensional structures of 321 drugs approved by the FDA were derived from the ZINC database. The Protein Data Bank supplied us with the PDB files of the relevant core protein domain. Through DS2019, a thorough virtual screening was conducted using the LibDock tool. Before docking, the PDB files were processed by removing the water molecules, refining the receptor protein structures, and minimizing the energy of proteins and ligands. After verifying the ionization states of amino acid residues and removing non-polar hydrogens, potential tautomers were generated. The Gasteiger–Marsili technique was employed for the charge compensation of the atoms of interest. The aim of these preparatory steps is to confirm the possible binding of FDA-approved drug candidates to the core protein binding sites through molecular docking so that an understanding of these protein–drug interactions can provide a platform for future drug design and experiments.

Molecular dynamics simulation

Using the molecular docking technique (AutoDock) to explore the interaction between the active compound DFO and core proteins CDR2L, LAMA3, LDLR, and LIMCH1 [61–64]. DFO was sourced from PubChem to obtain its structural formula, while its three-dimensional model was created using Chem3D. Before the docking process, the ligand underwent energy minimization through the MMFF94 force field to achieve a low-energy conformation ideal for docking. The protonation state of the ligand was adjusted to reflect pH 7.4 conditions. Core protein domain PDB files were retrieved from the Protein Data Bank and preprocessed with PyMOL, which involved the removal of water molecules and phosphate groups. Missing side chains were filled in, and hydrogen atoms were added using AutoDockTools 1.5.6 to prepare the receptor files. The PDB files for DFO and the proteins were converted to PDBQT format with AutoDockTools 1.5.6, and the active binding pockets were identified. For each target protein, the docking grid box was configured to adequately encompass the entire binding site, with grid center coordinates and dimensions based on the positions of co-crystallized ligands or predicted active sites (e.g., grid size: 30 × 30 × 30 ų, spacing: 1.0 Å). The Vina script was utilized to compute molecular binding energy and present docking outcomes. To ensure the reliability of the docking results, a re-docking experiment was conducted using the original co-crystallized ligand for each target protein (when available). The root-mean-square deviation (RMSD) between the re-docked ligand and its original crystal structure was calculated using PyMOL and AutoDockTools, with all RMSD values remaining below 2.0 Å, indicating satisfactory docking precision and conformational consistency. The results were visualized in three dimensions using PyMOL.

Cell culture and generation of GDM model

Human trophoblast cells (HTR-8/Svneo) were sourced from the American Type Culture Collection (ATCC, CRL3271) and grown in RPMI 1640 medium (Invitrogen, Shanghai, China) enriched with 10% fetal bovine serum (Invitrogen, MA, USA) and 100 U/mL of penicillin–streptomycin (Thermo Fisher, MA, USA). The cells were kept at 37 °C in a humidified incubator with 5% CO₂. C2C12 cells, obtained from the Shanghai Cell Bank at the Chinese Academy of Sciences, were cultured in high-glucose DMEM (Thermo Fisher, USA) containing 10% fetal bovine serum (Gibco, USA) and 1% penicillin–streptomycin (Scientific Cell), under identical conditions (37 °C, 5% CO₂). For the experiments, 2 × 10⁶ cells were plated in 6-well plates. To create an in vitro model of GDM, the cells were treated with high-glucose medium (25 mmol/L glucose) for a duration of 24 h.

Quantitative real-time PCR

Total RNA was obtained following the guidelines provided by the Cell Total RNA Isolation Kit (FOREGENE, Chengdu, China). Subsequently, the extracted RNA underwent reverse transcription to form complementary DNA (cDNA) utilizing the StarScript II First-strand cDNA Synthesis Mix (GenStar, Beijing, China). Quantitative real-time PCR (qRT-PCR) was conducted with the CFX96 Touch™ Real-Time PCR Detection System (BIO-RAD, USA). Each qRT-PCR experiment was carried out in triplicate biological samples (n = 6). The expression levels were evaluated using the 2^ − ΔΔCt method, with GAPDH mRNA acting as the internal reference.

CCK-8 assay

To evaluate cell viability, CCK-8 assay kit from Dojindo Laboratories (Kyushu, Japan) was used according to the manufacturer’s instructions. HTR-8/Svneo cells were plated in 96-well plates and subjected to various experimental treatments. After 12, 24, 36, 48, and 60 h, 10 μL of CCK-8 reagent was added to each well and further incubated for 2 h. The cells were then washed three times with PBS, and absorbance at 450 nm was measured using a microplate reader from Thermo Fisher Scientific (Waltham, USA)[65].

Bacterial strains and culturing conditions

The present research focused on isolates of MRSA and CRAB as they are important to study the effect of DFO on commonly used antibiotics, namely, imipenem and polymyxin B, against resistant bacteria. While DFO showed low antibacterial activity, its potential synergetic activity drew attention. The MRSA and CRAB strains were selected from the samples identification previously done using automated VITEK^® 2 Compact system (bioMérieux). As internal quality control strains, Pseudomonas aeruginosa (ATCC 27853) and Escherichia coli (ATCC 25922) were introduced while Enterococcus faecalis (ATCC 29212) acted as the positive control for biofilm assays. All bacterial strains were stored in brain heart infusion (BHI) medium supplemented with 10% glycerol (Merck, Darmstadt, Germany) at − 20 °C. Mueller–Hinton agar and tryptic soy broth supplemented with 2.5% glucose were used as culture media for assessment of antimicrobial activity and biofilm, respectively. All media were from commercial sources, including MHA (Merck, Darmstadt, Germany), and TSBG (Oxoid, UK). In five bacterial isolates (MRSA3, MRSA6, CRAB35, CRAB50, and CRAB89), antimicrobial activity and biofilm formation in the presence of DFO were studied.

Iron chelators and antimicrobial agents

The powdered form of Desferal^® (DFO, produced by Novartis, Switzerland) was acquired. Following the instructions of the supplier, a DFO solution was prepared by dissolving 500 mg of DFO in 2 mL of sterile distilled water to give a concentration of 380 mM. The experiments were carried out using a DFO solution with a final concentration of 213 mg/mL and suitable for intramuscular injection. The antimicrobial discs used in this study included vancomycin (VAN-30 μg), amoxicillin (AX-25 μg), polymyxin B (COL-10 μg), and imipenem (IMP-10 μg), purchased from Bioanalyse^® in Turkey.

Assessment of bacterial biofilm formation

The capability of bacterial isolates to form biofilm was assessed by a crystal violet test using a microplate reader. Every experimental setup contained three biological replicates (n = 3) for verification. The colonies were incubated at 370 C continuously for 24 h on Mueller–Hinton agar (MHA). Afterward, the colonies were incubated; then a bacterial suspension was made by adjusting the density to the 0.5 McFarland standard through the direct colony suspension method and afterward diluted in TSBG medium (3 mL). Subsequently, 180 μL of TSBG medium and 20 μL of the bacterial suspension were added into each well of the sterile 96-flat bottom microplate. For the control group, 200 μL of TSBG medium without bacterial suspension was placed in selected wells. The microplates were then incubated at 37 °C for 24 h to facilitate biofilm development. After incubation, the well contents were aspirated, and the wells were rinsed with 200 μL of sterile phosphate-buffered saline (PBS) (Oxoid, UK) to eliminate non-adherent bacterial cells. The plates were dried at 25 °C. To fix the remaining adherent bacteria, 200 μL of methanol was added to each well and incubated for 15 min. After discarding the methanol, 200 μL of 0.1% CV solution was added to each well and incubated at room temperature for 15 min. The liquid was then aspirated, and the wells were gently rinsed with tap water until the wash water was clear. Once dried, 200 μL of 95% ethanol was added to each well for destaining for 10 min. The optical density (OD) at 570 nm was recorded using a microplate reader (CLARIOstar Plus Microplate Reader, BMG LabTech, Cary NC). The OD values from wells containing only TSBG medium served as negative controls. Enterococcus faecalis ATCC 29212 was utilized as the positive control strain for biofilm formation. The cutoff OD (ODc) was established as the average OD of the negative control plus three times the standard deviation.

In vitro anti-biofilm effect of DFO

The effectiveness of DFO in inhibiting biofilm formation by MRSA and CRAB strains was evaluated using a microplate spectrophotometric approach. Initially, biofilms were allowed to develop on the surface of sterile 96-well F-type microplates for each bacterial strain. To promote biofilm growth, 180 μL of TSBG medium and 20 μL of bacterial suspension were introduced into each well, followed by incubation at 37 °C for 24 h. After this period, the well contents were removed, and 200 μL of DFO was applied directly to the established biofilm in each well. The plates underwent a further incubation for 24 h. The optical density was then measured spectrophotometrically to assess the anti-biofilm activity.

Statistical analysis

The analysis of data was conducted with SPSS software (version 26.0, SPSS Inc., Chicago, USA). Results are expressed as mean ± standard deviation. To compare two groups, the Student's t-test was employed, while one-way ANOVA was used to assess differences among several groups. A P-value below 0.05 was deemed statistically significant.

Results

Screening for disease targets in GDM

In order to clarify the changes in gene expression linked to weight-related GDM, we conducted a comprehensive transcriptomic analysis utilizing various datasets. The discovery cohort was based on the GSE154414 dataset, which contains placental samples from four GDM patients alongside four healthy controls. For external validation, we employed the GSE133099 dataset, which includes adipose tissue samples from six obese individuals and six non-obese healthy participants. After performing background correction and normalization, we evaluated the sample distribution, confirming its reliability (Supplementary Fig. 1A & B), which underscores the effectiveness of our preprocessing methods. The differential expression analysis of the GSE154414 dataset revealed 276 genes that were upregulated and 161 that were downregulated. In the GSE133099 dataset, we identified 1,114 upregulated genes and 756 downregulated ones. The heatmap visualization demonstrated a distinct separation between the groups based on the expression of significantly altered genes (Fig. 1A), while volcano plots emphasized the most significantly dysregulated genes (Figs. 1B & C). To explore the biological processes related to these transcriptomic alterations, we performed Gene Set Enrichment Analysis (GSEA). The enrichment analysis of the GSE154414 dataset highlighted pathways associated with protein digestion and absorption, ECM–receptor interactions, vascular smooth muscle contraction, and the AGE-RAGE signaling pathway relevant to diabetic complications (Fig. 1D). Similarly, the analysis of the GSE133099 dataset revealed significant enrichment in metabolic and inflammatory pathways, such as terpenoid backbone biosynthesis, steroid biosynthesis, and again, the AGE-RAGE signaling pathway (Fig. 1E). Collectively, these findings emphasize both common and tissue-specific transcriptional patterns associated with GDM and obesity. The combination of data from placental and adipose tissues highlights the importance of AGE-RAGE signaling and the disruption of metabolic pathways in GDM linked to weight issues, providing a basis for discovering new molecular targets and informing future treatment strategies.

Fig. 1.

Target Screening for Gestational Diabetes Mellitus Using GSE154414 and GSE133099 Datasets. A A heatmap displaying DEGs from the GSE154414 dataset, showing expression patterns between control and GDM samples. Each row represents a gene, and each column represents a sample, with colors ranging from blue (low expression) to purple (high expression). Hierarchical clustering groups genes with similar expression patterns. B A volcano plot for GSE154414 showing the relationship between the (− log10 adjusted p-value) and log2 fold change for all genes. Purple indicates increased expression, and blue indicates decreased expression. Dashed lines mark the significance threshold. The GSE133099 volcano plot of differentially expressed genes (DEGs). As observed in panel B, this visualization shows the importance and fold change of DEGs, where upregulated genes are in purple and downregulated genes are in blue. The important DEGs that have changed significantly are highlighted for clarity. D The GSE154414 gene expression dataset was GSEA analyzed. The bar chart demonstrates the enrichment levels of gene sets in the dataset used to develop GDM. The y-axis displays the enrichment scores for the gene sets shown on the x-axis. Red indicates positive enrichment in gene sets, while blue indicates negative enrichment in gene sets. The pathways for protein digestion and absorption, contraction of vascular smooth muscle, and circadian rhythm disturbances have been highlighted in this analysis. E GSEA analysis of the GSE133099 dataset reveals that DEGs are involved in significant pathways and processes. The enrichment plot is the one that indicates positive gene sets are highlighted with red bars while negative ones are with blue. Significant pathways, such as that of AGE-RAGE signaling, which is implicated in diabetes complications, and oxytocin signaling, demonstrate considerable enrichment, suggesting their potential role in gestational diabetes mellitus (GDM)

WGCNA of differentially expressed genes in GDM

To explore the co-expression relationships among differentially expressed genes (DEGs) linked to GDM, we utilized Weighted Gene Co-Expression Network Analysis (WGCNA) with the GSE154414 dataset. We determined the ideal soft-thresholding power (β) to be 16, which was established based on the scale-free topology criterion, where the correlation coefficient first surpassed 0.9 (Fig. 2A). This choice ensured that the constructed network adhered to the scale-free topology principle, a fundamental aspect of WGCNA. Before building the network, we conducted hierarchical clustering on all eight samples to identify any outliers. The resulting sample dendrogram showed no notable anomalies, confirming the dataset's reliability and consistency (Fig. 2B). With the chosen β value, we generated a Topological Overlap Matrix (TOM) and identified modules through the dynamic tree cut method. This process led to the discovery of 42 unique co-expression modules, each distinguished by a specific color (Fig. 2C). Correlation analysis between modules and traits indicated that two modules—MEskyblue and MElightyellow—were significantly linked to traits associated with GDM, both showing strong correlations with absolute values greater than 0.7 and p-values below 0.05 (Fig. 2D), underscoring their potential importance in GDM pathogenesis. Within these modules, we identified 354 candidate hub genes based on their high module membership and gene significance scores. To depict the structural layout of the gene co-expression network, we created a heatmap of the TOM (Fig. 2E). This heatmap demonstrated the degree of gene interconnectedness, with red indicating strong co-expression and blue denoting weaker connections. Together, these results highlight the important roles of the MEskyblue and MElightyellow modules in GDM. By combining WGCNA with differential expression studies, the biological importance of these gene groups is underscored, while also pinpointing key hub genes that should be explored further as possible biomarkers or treatment targets for GDM associated with weight issues.

Fig. 2.

Weighted Gene Co-Expression Network Analysis of Differentially Expressed Genes in Gestational Diabetes Mellitus. A Establishing the soft-threshold power (β) for building the network involves analyzing the scale independence (left) and mean connectivity (right) graphs to identify the best β value. B Example of a dendrogram and identification of anomalies. A hierarchical clustering analysis was conducted on the complete set of samples to evaluate data integrity and identify any outliers. The bands, which are color-coded and located beneath the dendrogram, illustrate gene co-expression modules obtained from the analysis. C Relationships between modules and traits. A heatmap illustrates the correlation between module eigengenes (represented in rows) and various clinical traits (shown in columns, which include Control and GDM groups). Each cell features the correlation coefficient along with its significance level (p-value). Modules that show significant links to GDM are highlighted to underscore their possible biological importance. D A dendrogram illustrating gene clustering alongside a heatmap of the topological overlap matrix (TOM). This dendrogram displays how genes are grouped into separate modules according to their 1-TOM dissimilarity, with each module represented by unique colors. Below, the heatmap shows TOM values, where red squares indicate a high degree of topological overlap between genes, signifying robust co-expression connections. E Heatmap representation of gene networks. This graphic illustrates the TOM for chosen genes, where more robust connections are indicated in red and less robust ones in blue. The dendrogram at the top of the heatmap shows the clustering of genes, with the colored bands representing different modules

Identification of genes associated with gestational diabetes mellitus and obesity

To clarify the molecular connections between GDM and obesity, we combined differential gene expression analysis with WGCNA. By comparing the GDM-related differentially expressed genes (DEGs) from the GSE154414 dataset with those linked to obesity from the GSE133099 dataset, we identified five common genes (Fig. 3A). From the 354 hub genes found in the GDM-related modules (MEskyblue and MElightyellow), we focused on four genes (CDR2L, LIMCH1, LDLR, and LAMA3) for further investigation based on three criteria: (1) significant differential expression in both GDM and obesity datasets; (2) consistent downregulation in an independent validation cohort (GSE203346); and (3) prior reports or predictions of their roles in metabolic or immune pathways as per GeneCards and literature reviews. These criteria ensured both statistical validity and biological significance. Validation with the independent dataset (GSE203346) confirmed that the four overlapping genes (CDR2L, LIMCH1, LDLR, and LAMA3) were significantly downregulated in GDM samples compared to healthy controls (P < 0.05). The patterns of differential expression were effectively represented through a heatmap (Fig. 3B) and box plots (Fig. 3C), indicating their potential role in GDM pathogenesis. To evaluate the diagnostic potential of these candidate genes, we performed receiver operating characteristic (ROC) curve analysis. All four genes exhibited area under the curve (AUC) values exceeding 0.60, indicating moderate discriminatory ability and their potential as diagnostic biomarkers for GDM (Fig. 3D). The immune landscape of GDM was further analyzed using the CIBERSORT algorithm to assess immune cell composition. The resulting bar plot (Supplementary Fig. 1C) showed significant decreases in naive B cells and resting mast cells in GDM samples (P < 0.05). Additionally, a heatmap illustrating immune cell correlations revealed altered interactions among immune cells in the GDM group, suggesting disrupted immune homeostasis (Fig. 3E). To further explore the functional implications of the identified genes, we conducted GO and KEGG enrichment analyses. GO analysis indicated significant enrichment in biological processes such as cell-substrate junction assembly and regulation of supramolecular fiber organization, cellular components including the laminin complex and endolysosome membrane, and molecular functions such as low-density lipoprotein particle receptor activity (Fig. 3F). KEGG pathway enrichment analysis highlighted pathways related to cholesterol metabolism, ECM–receptor interactions, cortisol synthesis and secretion, and bile secretion (Supplementary Fig. 1D–F). Specifically, Supplementary Fig. 1D provides a detailed KEGG pathway diagram for cholesterol metabolism, showcasing key enzymes and regulatory interactions enriched in GDM. Supplementary Fig. 1E presents a GO chord diagram linking genes to their associated biological processes, cellular components, and molecular functions, offering a comprehensive view of gene function. Supplementary Fig. 1F illustrates a KEGG chord diagram demonstrating the involvement of core genes in various signaling pathways, with node size representing the gene count per pathway. The genes discovered could act as valuable diagnostic indicators and might also serve as potential therapeutic targets for timely intervention and tailored treatment of GDM in individuals with obesity.

Fig. 3.

Identification of Genes Associated with Gestational Diabetes Mellitus and Obesity. A A Venn diagram showing overlapping DEGs from three datasets (Obesity, GDM and WGCNA). The sections of overlap indicate common DEGs which are highlighted across these datasets. This emphasizes potentially important candidate genes that are involved in GDM and obesity. B A heatmap depicting the expression of key genes in the control and GDM samples of the GSE203346 dataset. Every single row indicates a certain gene while every single column depicts a sample. The colors on the gradient indicate how much of a gene the cells are producing; red means more, blue less. C Box plots displaying the differential expression levels of functional DEGs in control and GDM groups in the GSE203346 dataset. Each box corresponds to a specific gene, with asterisks denoting statistically significant differences (p < 0.05). The modified expression of certain genes in GDM as seen in these results. D A receiver operating characteristic (ROC) curve is used to assess the diagnostic power of the major DEGs from the GSE203346 dataset. Each model shows the ROC curve belonging to a single gene. The AUC values indicate the potential of the gene to predict GDM. The ROC curves indicated that LAMA3 and LDLR had a good diagnostic capacity for GDM (AUC: LAMA3 = 0.83, 95% CI 0.76–0.90; LDLR = 0.79, 95% CI 0.71–0.87). E A heatmap shows the immune cell interactions in both control and GDM samples. The warm colors show stronger connections and more interactions. The cooler colors show weaker connections and less interaction. These data provide helpful and useful information regarding the immune changes due to GDM. F The Gene Ontology (GO) enrichment analysis indicates the most significant enriched terms. The GO analysis analyses significant enriched terms in the biological processes (BP), molecular functions (MF), and cellular components (CC). The x-axis shows the enrichment scores while the y-axis shows the corresponding GO terms. As illustrated, different colors represent each kind, which demonstrates the functional significance of DEGs in GDM

Therapeutic potential of core proteins and anti-biofilm activity of deferoxamine

This research identified four key proteins—CDR2L, LIMCH1, LDLR, and LAMA3—as promising therapeutic targets through molecular docking studies. A screening of 321 small-molecule ligands was conducted, and their binding affinities were evaluated using the LibDockScore algorithm. As shown in Fig. 4A, DFO had the strongest binding affinities with all four proteins, suggesting its potential as a multi-target therapeutic agent. In detail, CDR2L showed significant interactions with DFO, digoxin, and paromomycin; LDLR had the best binding with digoxin and streptomycin; LIMCH1 preferred DFO, digoxin, and centany; and LAMA3 had the highest affinity for 5-methyltetrahydrofolate and cromolyn. The binding strengths are depicted in a heatmap (Fig. 4A), where darker purple shades indicate stronger interactions between ligands and proteins. To further assess DFO's functional capabilities, its anti-biofilm properties were tested against nine clinical bacterial strains, including MRSA and CRAB isolates. Among these, MRSA3, MRSA6, CRAB35, CRAB50, and CRAB89 were recognized as high biofilm producers (Fig. 4B). Importantly, DFO significantly reduced biofilm formation in these strains, with reduction rates between 38.1% and 72.3% (Fig. 4C). CRAB strains exhibited a higher sensitivity to DFO than MRSA strains, underscoring its potential effectiveness against Gram-negative, biofilm-related infections. The results suggest that DFO serves two primary roles. Firstly, it is a small compound that effectively binds to the active sites of proteins associated with GDM. Secondly, it functions as a potent anti-biofilm agent targeting multidrug-resistant pathogens. These discoveries pave the way for innovative dual therapy approaches.

Fig. 4.

Heatmap and Functional Analysis of Virtual Molecular Docking and Biofilm Disruption. A A heatmap showing the results of a virtual molecular docking study that explored how certain proteins interact with ligands. The proteins analyzed are LDLR, LIMCH1, CDR2L, and LAMA3. Each row refers to a specific protein and each column to different ligands. The deeper the purple color, the better the dockings scores and affinity to bind together. Lighter colors indicate weaker affinities. The binding strength difference is expressed as a number on the binding scale. B Bar graphs showing the optical density (OD) values of the mature biofilms produced by various bacterial strains including MRSA and CRAB. The OD values were analyzed for biofilm production and for DFO (deferoxamine) effects on the biofilm stability. Significant differences among groups are indicated by p-values (p < 0.01). C The amount of biofilm disruption caused by DFO treatment was compared to a control. We used crystal violet staining to see how much biofilm was made. Results showed a varying degree of biofilm disruption on the parts of the bacterial strains. The CRAB strains showed a greater reduction of biofilm as compared to MRSA strains

Results of molecular docking and analysis of ligand–receptor interaction

This study aims to characterize the interaction of active compound DFO with core proteins CDRL2, LAMA3, LDLR, and LIMCH1 through molecular docking. The results showed that DFO had stable docking interactions with all the core proteins (Fig. 5 and Table 1). The findings indicated that DFO formed the most stable docking complex with the use of parameters like RMSD and chemical energy. The creation of different docking models using the same ligand for a number of protein targets was the result of changes made to the chemical bonds formed at different residues. The docking results generated by Vina were processed in Pymol software for 3D and 2D visualization of molecular docking with protein ligands which vividly illustrated the docking and serves as basis for further studies.

Fig. 5.

Molecular docking results and ligand–receptor interaction analysis. A A1: CDR2L-DFO 3D (macroscopic); A2: CDR2L-DFO 3D (microscopic); (B) B1: LAMA3-DFO 3D (macroscopic); B2: LAMA3-DFO 3D (microscopic); (C) C1: LDLR-DFO 3D (macroscopic); C2: LDLR-DFO 3D (microscopic); (D) D1: LIMCH1-DFO 3D (macroscopic); D2: LIMCH1-DFO 3D (microscopic)

Table 1.

Molecular Docking Vina, Discovery Studio 2019 Results

Protein(ID)	Compound	Vina(kcal·mol−1)	RMSD
CDR2L	Deferoxamine	− 3.531	18.72
LAMA3	Deferoxamine	− 4.828	3.185
LDLR	Deferoxamine	− 4.841	2.788
LIMCH1	Deferoxamine	− 3.694	2.113

Therapeutic effects of deferoxamine in high-glucose-induced GDM model

This study evaluated the potentials of DFO on high glucose (HG) effects in the GDM model. Results demonstrated that DFO could significantly improve cell viability as indicated by the CCK-8 assay (Fig. 6A). Subsequently, cells treated with DFO under HG conditions demonstrated a significantly higher cell viability than those untreated. Moreover, DFO altered the relative expression of genes involved in the pathophysiology of GDM, such as CDR2L, LAMA3, LIMCH1, and LDLR. The expression of RNA was analyzed and data showed that DFO treatment countered the effect of HG on these genes (Fig. 6B). While DFO exhibited stable interactions with LAMA3, LDLR, and LIMCH1 (RMSD < 3 Å), its interaction with CDR2L showed greater conformational variability (RMSD = 18.72 Å), indicating possible structural flexibility in the binding site of CDR2L or limitations in static docking methods. Nonetheless, functional validation confirmed that CDR2L responded significantly to DFO treatment in cellular assays (Fig. 6B). At the protein level, DFO reinstated the expression of crucial signaling molecules like AKT2 and PI3K, which are vital to the PI3K/AKT signaling pathway and were notably diminished by HG exposure (Fig. 6C). It also influenced mTORC1, a key regulator of cellular growth and metabolism, and restored GLUT4 expression, a critical glucose transporter, under HG conditions (Fig. 6D). Immunofluorescence staining further illustrated DFO’s protective effects by reducing HG-induced inflammation and apoptosis. DFO significantly lowered the levels of TNF-α, a pro-inflammatory cytokine (Fig. 6E), and caspase-3, an apoptosis marker (Fig. 6F), while increasing Bcl-2 expression, an anti-apoptotic protein, highlighting its anti-inflammatory and cell survival-promoting properties (Fig. 6G). These results indicate that DFO provides comprehensive protective effects in HG environments by modulating both metabolic and inflammatory pathways. Initial evidence suggests that LAMA3 may play a crucial role in these mechanisms, potentially linking extracellular matrix signaling with intracellular metabolic pathways like PI3K/AKT.

Fig. 6.

Impact of Deferoxamine on Cellular Alterations Induced by Elevated Glucose Levels and Its Possible Role as a Treatment for Gestational Diabetes Mellitus Linked to Weight Issues. A Bar charts illustrating the comparative cell survival rates of cells subjected to elevated glucose levels (HG), both with and without DFO treatment at various concentrations. DFO significantly improved cell viability under HG conditions, as denoted by statistical significance (p < 0.01). B The expression levels of CDR2L, LAMA3, LIMCH1, and LDLR, adjusted for GAPDH, were assessed in various treatment groups. Elevated glucose levels notably impacted gene expression, whereas DFO reduced these effects in a manner dependent on its dosage (p < 0.01). C Measurement of essential signaling proteins such as AKT2, PI3K, and GLUT4, adjusted for GAPDH levels. Elevated glucose levels inhibited these pathways, whereas DFO reinstated their expression. The findings are statistically significant for the groups treated with DFO in comparison to the high glucose group (p < 0.05). D The levels of mTORC1, adjusted for GAPDH, were analyzed across different treatment groups. Under high glucose conditions, DFO notably reduced mTORC1 levels (p < 0.01), indicating an alteration in cellular metabolic processes. E Immunofluorescence analysis displaying TNF-α (in red) and DAPI (in blue) illustrates the inflammatory response across control, high glucose (HG), and DFO-treated HG groups. The application of DFO significantly lowered TNF-α levels under HG conditions. F Immunofluorescence analysis was conducted to assess the levels of apoptosis by staining for Caspase-3 (shown in red) and DAPI (shown in blue). Cells treated with DFO showed lower activation of Caspase-3 in comparison to the high glucose (HG) group. G Immunofluorescence analysis was conducted to evaluate the expression of the anti-apoptotic marker Bcl-2 (shown in red) alongside DAPI (depicted in blue). Treatment with DFO reversed the decrease in Bcl-2 levels caused by high glucose exposure

Functional validation of LAMA3 and its downstream signaling in high-glucose conditions

To investigate the role of LAMA3 in the protective effects of DFO under high glucose (HG) conditions, we conducted loss-of-function experiments targeting LAMA3. Quantitative PCR confirmed successful knockdown of LAMA3 in the sh1-LAMA3 and sh2-LAMA3 groups compared to the shNC control. Notably, CDR2L expression was also significantly reduced in these groups, suggesting a downstream regulatory role (Fig. 7A). Cell viability was assessed using the CCK-8 assay under various treatments. As expected, HG exposure significantly inhibited cell proliferation. DFO treatment (HG + DFO) partially restored cell viability. However, LAMA3 knockdown (HG + DFO + shLAMA3) or pharmacological inhibition with B-0304 (HG + DFO + B-0304) reversed this effect, resulting in further declines in proliferation (Fig. 7B). ELISA analysis revealed that HG stimulation increased the release of IL-6, IL-17, and IL-23. DFO reduced these cytokine levels, but LAMA3 knockdown or B-0304 treatment reversed this suppression, causing cytokine levels to be similar to or higher than those in the HG group. In contrast, the shNC group retained DFO's anti-inflammatory effects (Fig. 7C). We also examined the expression of PI3K/AKT signaling pathway components through qPCR. As shown in Fig. 7D, AKT2, and PI3K levels were significantly reduced in the HG, HG + DFO + shLAMA3, and HG + DFO + B-0304 groups. However, the HG + DFO and HG + DFO + shNC groups exhibited PI3K/AKT mRNA levels comparable to the control, indicating that LAMA3 is critical for DFO's effects on this pathway. Immunofluorescence staining further revealed that LAMA3 silencing or inhibition after HG exposure significantly reduced BCL2 and BrdU expression, indicating decreased cell survival and proliferation. Conversely, levels of caspase-3, HIF-1α, IL-6, and TNF-α were significantly elevated, reflecting increased apoptosis and inflammation. The HG + DFO and HG + DFO + shNC groups maintained lower levels of these markers, supporting LAMA3’s protective role in DFO’s effects (Fig. 7E). Finally, colony formation assays showed that both HG exposure and LAMA3 interference (via shRNA or B-0304) significantly impaired cellular proliferation and clonogenic potential. DFO treatment preserved the ability to form colonies, which was lost with LAMA3 silencing or inhibition (Fig. 7F). These findings suggest that LAMA3 is essential for mediating DFO’s protective effects under high glucose conditions. LAMA3 likely activates the PI3K/AKT pathway, potentially through interactions with ECM receptors or integrin signaling, supporting glucose transporter expression (e.g., GLUT4) and cell survival during metabolic stress. Reduced LAMA3 levels disrupt this pathway, increase inflammation and apoptosis, and hinder cell proliferation, highlighting LAMA3 as a promising therapeutic target in obesity-related GDM.

Fig. 7.

Functional Validation of LAMA3 and Its Downstream Signaling in High Glucose Conditions (A) A bar chart illustrates the comparative mRNA levels of LAMA3 and CDR2L in the Ctrl, sh1-LAMA3, and sh2-LAMA3 cell groups. The mRNA levels of both LAMA3 and CDR2L were found to be significantly reduced in the sh1-LAMA3 and sh2-LAMA3 groups (p < 0.01). B A bar chart illustrates the cell proliferation rates across various treatment groups: HG, HG + DFO, HG + DFO + shLAMA3, HG + DFO + shNC, and HG + DFO + B-0304. In all these groups, there was a significant decrease in the percentage of cell proliferation, with statistical significance noted. C A bar chart illustrating the levels of IL-6, IL-17, and IL-23 secretion in various cell groups treated with HG, HG + DFO, HG + DFO + shLAMA3, HG + DFO + shNC, and HG + DFO + B-0304, as measured by ELISA. D A bar chart illustrating the expression levels of AKT2 and PI3K in cell groups treated with HG, HG + DFO, HG + DFO + shLAMA3, HG + DFO + shNC, and HG + DFO + B-0304, as assessed by Western blot analysis. E. The expression levels and spatial distribution of BCL2, Brdu, Caspase-3, HIF-1α, IL-6, and TNF-α were assessed using immunofluorescence in cell groups treated with HG, HG + DFO, HG + DFO + shLAMA3, and HG + DFO + B-0304. (F) An assay for clone formation was conducted to assess the growth and colony development potential of cells treated with HG, HG + DFO, HG + DFO + shLAMA3, and HG + DFO + B-0304

Discussion

As a result of the rise of bacteria with antimicrobial resistance (AMR), the effectiveness or potency of common and widely used antibiotics has diminished seriously [66, 67]. According to the world health organization, Staphylococcus aureus and Acinetobacter baumannii, which are resistant to antibiotic drugs [68]. According to Rodvold and McConeghy, the most relevant agents that cause serious patient harm are MRSA [69]. In a similar fashion, Acinetobacter baumannii, CRAB is labeled as a critical priority pathogen as it has a high level of resistance that limits treatment options [70]. The rise in resistance rates is making it increasingly difficult to use polymyxins or tetracycline derivatives (e.g., minocycline and tigecycline) against CRAB [71]. Biofilms are microbial communities that form on inanimate surfaces. These structures can provide protection for microbes, allowing them to survive and persist against the effects of antimicrobials. They are suspected to play an important role in antimicrobial resistance (AMR) in conditions such as GDM [72]. Biofilms help pathogens survive harsh environments and protect them from antimicrobial treatments [73]. Also, we need iron for the development of biofilms and their metabolic functions [74]. Our understanding of calcium and iron in GDM could provide strategies for immunomodulation [24]. This is because iron dysregulation increases insulin resistance. The antibiotics may become more effective due to iron chelation disrupting the biofilm [75].

The expression levels of the genes CDR2L, LIMCH1, LDLR, and LAMA3 are significantly downregulated in GDM [44, 76]. The development and changes of GDM depends on the downregulation of these genes which take part in lipid metabolism, steroid hormone synthesis, and interaction with the extracellular matrix (ECM) [77, 78]. According to our study, DFO treatment leads to increased gene expression, suggesting it capable of promoting GDM functional recovery. DFO has been associated with a number of genes and pathways. These include toxoplasmosis, ovarian steroidogenesis, cholesterol metabolism, ECM, and receptor. This suggests that DFO could potentially influence these pathways to mitigate oxidative stress and enhance insulin signaling, thereby improving insulin sensitivity and resistance in GDM patients. This inconsistency might be due to allosteric influences from protein interactions, the indirect effect of iron chelation on CDR2L stability, or challenges in modeling the disordered regions of CDR2L. By targeting these specific pathways, DFO emerges as a promising candidate for treating GDM, particularly in patients who do not respond to standard therapies.

Our study findings indicate that iron overload in GDM may worsen metabolic irregularities and enhance biofilm persistence in MRSA and CRAB. Iron supports metabolic malfunctioning and enhances the body’s ability to fight off infections. Thus, iron metabolism-based therapeutic approaches may be attained for relevant conditions. We found molecular changes linked to dysregulated iron metabolism and weight gain in GDM using multi-omics analysis [44, 79]. The prior study which showed that iron overload leads to excess fat buildup in mice and showed resistance to insulin was consistent with our findings [80, 81]. Combining analysis of transcripts and proteins opens up new avenues for personalized interventions. In models of GDM, DFO reduced iron levels, oxidative stress and inflammation, thereby improving insulin signaling. These data support other studies, which show that iron chelation reduces oxidative stress and improves insulin sensitivity, meaning DFO could be a suitable option to use for GDM treatment [82, 83]. It was observed that DFO affected the properties of the biofilm cells and reduced the GDM injury. This finding suggests that iron chelation therapy may help fight microbes in metabolic diseases. However, to achieve this, effect evaluation on a dosing schedule will be essential before reaching to clinical applications. This puts patients at risk for infections due to multidrug-resistant organisms, like MRSA and CRAB. The formation of biofilms by these pathogens leads to recurrence and hinders treatment. The inflammation that doctors expect to see is subnormal. So, when the sugar levels are also elevated, these individual’s are likely to have infections. This applies to some bacteria, such as MRSA and CRAB, which are known to form problematic biofilms [84, 85]. The creation of biofilms is highly problematic in diabetics because it contributes to the persistence and recurrence of infections, making them more difficult to treat.

Complications such as kidney damage and preterm labor due to UTIs (Urinary Tract Infections) pose significant threat to maternal and neonatal health [86, 87].The rising antibiotic resistance makes it more difficult to treat these infections [88]. These challenges point to the need for optimal treatment through multi-omics approaches for the repurposing of existing therapies and targeting of the immune signatures. Due to limited drug options and likely adverse reactions to some antibiotics in mother and fetus, plan a complete drug regimen [89]. Some infections remain resistant to a single drug, requiring combination therapy in these situations and demanding complex treatment regimens. Pregnant women with GDM are at an increased risk of urinary tract infections (UTIs). Therefore, symptom monitoring and suitable preventative and treatment measures are essential. If care is delivered in a timely manner, risk of infection and its consequences can be prevented that ultimately affects the mother and newborn [90]. We performed the first histological study of the therapeutic effects and mechanisms of DFO on GDM [91]. The unique discoveries outlined in our recent paper highlight either new targets or new opportunities for drug discovery related to cell death and metabolism [92, 93]. The objective of precision therapeutics is the proper delivery of treatment for any clinical condition; preferably molecular profiling will guide decision making in the management of GDM and its complications. Analysis of gene expression and regulatory mechanisms in selected disease models––which influence onset and progression of disease––help understand disease mechanisms [94]. Taking this evidence into account, we performed network pharmacology and molecular dynamics simulations to study a drug’s agent’s protective effect on cellular function [31]. The adoption of innovative approaches across diverse fields enhances the efficacy of interventions while minimizing adverse impacts, thereby advancing the practice of tailored and precise solutions [95–97]. This study also suggests the development of therapeutic agents aimed at proteins or genes or predictive tools for the postoperative period [98, 99]. The emphasis on precision medicine is critical to attaining better patient outcomes especially in cases of diabetes mellitus pregnancy. This study shows that DFO can correct iron metabolism disorders in GDM, which is often linked to oxidative stress and inflammation [100, 101]. DFO can intervene in bacterial biofilms and target both metabolic and infectious complications of GDM. By affecting the metallobiology of infections, DFO enhances antibiotic effectiveness by lowering iron availability, making bacteria more susceptible to treatment [43, 102]. This biofilm-disrupting action makes DFO a promising candidate for repurposing in GDM, although its safety during pregnancy is not well understood. Stratified clinical trials considering trimester effects and maternal comorbidities are needed. DFO reduces oxidative stress and inflammation, which may improve glucose control and insulin sensitivity in GDM patients [24, 103]. As antibiotic resistance rises, DFO's ability to interfere with bacterial iron acquisition presents a unique mechanism to combat infections. Its effect on biofilm may help reduce persistent infections from multidrug-resistant organisms in GDM. Urological complications are common in GDM, and DFO may help prevent biofilm formation and bacterial growth responsible for UTIs in these patients [43]. DFO also shows promise in preventing kidney injury in GDM-related complications [104, 105]. However, its renal clearance during pregnancy could affect pharmacokinetics and dosing. While some studies suggest DFO can address multiple health issues, it should be used cautiously in pregnancy due to potential side effects and limited safety data. Animal studies indicate potential risks at high doses, including the possibility of crossing the placenta and harming the fetus. Therefore, strict patient selection and careful dose management are essential. Our findings align with existing research on GDM molecular changes and offer new insights into iron metabolism [106, 107]. Although previous studies identified biomarkers for GDM, they did not explore the risks and benefits of iron modulation treatments like DFO during pregnancy. This study aims to fill that gap, offering a deeper look at DFO's potential application in GDM [108, 109].

The integration of multi-omics data with bioinformatics analysis enhances our understanding of DFO's therapeutic potential and mechanisms [110–112]. This methodology is crucial for advancing clinical applications and the drug development process [113]. Combining clinical and genomic data helps create predictive models that improve disease forecasting and therapeutic response accuracy, which is essential for personalized medicine [114–117]. Future clinical algorithms may use iron-related biomarkers to identify GDM patients who will benefit most from treatment while minimizing unnecessary exposure for low-risk individuals. The research employs advanced statistical techniques, machine learning, and bioinformatics to address complex biomedical challenges [118–121]. By identifying new genetic markers, we have gained a better understanding of metabolic genes and their prognostic significance. This study provides fresh insights into iron metabolism in GDM, suggesting DFO’s potential therapeutic role. However, larger clinical trials are needed to confirm DFO's safety and effectiveness, as the small sample size limits the applicability of our findings. Future studies should include multicenter trials to increase the generalizability of results and develop strategies for addressing both metabolic and infectious complications of GDM. It is important to note that our study focused on the short-term effects of DFO. GDM is a complex condition that poses risks to maternal and neonatal health, as well as long-term risks for type 2 diabetes, metabolic syndrome, and cardiovascular diseases. Since GDM only occurs during pregnancy, it is unclear if DFO’s effects last beyond pregnancy. Future research should include long-term follow-ups to assess its effectiveness and safety in preventing GDM complications and long-term health impacts.

Although this study offers valuable insights, it comes with various limitations. A primary limitation is the small sample size which limits the generalizability of the findings [122, 123]. Besides, GDM has not been thoroughly evaluated in large clinical trials of DFO [124]. A successful multicenter study could have a broader and bigger population of participants, which will enable it to assess short- and long-term effects. Researchers should attempt to include any studies that include a larger population and clinical evaluations in order to evaluate the safety and efficacy of DFO in GDM management [125, 126]. This research will be critical toward GDM management. The use of DFO, being cheap and commonly used drug, may help in improving maternal and neonatal health, and reducing the overall cost incurred in the treatment of GDM [127, 128]. DFO may offer new hope for patients with no alternative treatment options. Similarly, the approach of regulating iron metabolism to enhance insulin resistance could benefit other pathologies with insulin resistance, like obesity and type 2 diabetes [129, 130]. DFO's ability to regulate iron levels and disrupt biofilms gives it a distinct therapeutic advantage in treating GDM, especially in cases complicated by antibiotic-resistant urinary tract infections. We recommend future research involving patients with elevated ferritin levels and recurrent infections.

Conclusion

DFO reduces oxidative stress, PI3K/AKT deficiency, and inflammation in GDM by influencing genes related to iron metabolism and modifying CDR2L, LIMCH1, LDLR, and LAMA3. In addition to its metabolic effects, DFO impacts immune pathways, supporting its potential use in personalized immunotherapy for GDM. These findings offer new insights into GDM mechanisms, which could inform precision medicine for better outcomes for both mothers and babies. However, large-scale clinical trials are lacking. Future research should confirm DFO’s efficacy and safety across different populations and explore its combined use with other treatments. Long-term studies are needed to assess the sustainability of DFO's benefits and guide its role in preventing GDM and promoting maternal–fetal health. Subsequent trials should focus on standardized dosing, patient selection, and long-term safety. After regulatory review, DFO could be considered a safe and effective adjunctive therapy.

Author contributions

M.Z. wrote the main manuscript text, performed data analysis,prepared figures,supplementary materialsassisted in manuscript revision and contributed to the experimental design. N.W.and R.L. performed data analysis and assisted in manuscript revision. Y.X. contributed to data interpretation. QF.Z. and LN.J. prepared figures and supplementary materials. XT.Z. and XJ.Y. contributed to the experimental design, supervised the project, provided critical feedback and finalized the manuscript. All authors reviewed and approved the final version of the manuscript.

Funding

No funding.

Data availability

Data supporting the findings of this study are available in the Gene Expression Omnibus (GEO) database under the accession numbers GSE154414 and GSE133099. These datasets include transcriptomic data related to gestational diabetes mellitus (GDM) and weight, which may serve as valuable resources for further investigation. If you have any questions regarding the data, please feel free to contact the corresponding author.

Declarations

Ethics approval and consent to participate

This study was conducted using bioinformatics analysis and functional assays exclusively on cell lines. No human participants or animal subjects were involved in any part of the research. All experimental procedures were performed in compliance with institutional guidelines and relevant regulations.

Competing interests

The authors declare no competing interests.