25-day-old female ICR mice were purchased from Liaoning Changsheng Biotechnology Co., Ltd and housed under controlled conditions (12-hour light/dark cycle, temperature 22°C ± 2°C) with free access to food and water. DHEA was dissolved in 10% sesame oil (in 95% ethanol) and administered daily to each mouse in the model group at 6 mg/100 g body weight in 0.1 mL for 20 days. Control mice were injected with vehicle in the same way. Ovaries from mice in the model group were collected at the end of the DHEA administration period. For the control group, ovaries were obtained from mice at the same estrous stage as those in the treatment group to ensure physiological comparability. All samples were immediately frozen in OCT compound for ST.

Testosterone levels were quantified in both mouse serum and ovarian tissue lysates to evaluate the androgenic status of the DHEA-induced PCOS model. Serum was collected via orbital bleeding, and ovarian tissues were homogenized in PBS followed by centrifugation to obtain supernatants. Testosterone concentrations were measured using commercial ELISA kits (Cusabio. China), following the manufacturer’s instructions.

As describe previously, the 10 × Genomics Spatial Transcriptomics platform was used to map gene expression across entire tissue sections, revealing cellular spatial organization and function (Luo et al., 2024). Ovaries from three biological replicates per group (n = 3) were collected, embedded in OCT, sectioned at 10 μm thickness using a Leica CM1950 cryostat, and mounted onto Visium spatial gene expression slides according to the manufacturer’s protocol.

Following tissue permeabilization, mRNA transcripts were captured by spatially barcoded oligonucleotides on the slide surface. Reverse transcription was then performed to synthesize cDNA incorporating spatial barcodes, enabling localization of gene expression to specific tissue regions. Sequencing libraries were generated and subjected to high-throughput sequencing on the Illumina platform.

Raw data processing, including alignment, barcode assignment and gene quantification was performed using Space Ranger (10x Genomics). Quality control filtering was applied to exclude spots with >10% mitochondrial gene content or fewer than 100 detected genes. The number of retained and excluded spots for each sample is summarized in Supplementary Table S1. Unsupervised clustering was conducted in Seurat (v4.0) to identify distinct transcriptional clusters. UMAP was used for dimensionality reduction and visualization of spatially distinct cell populations across control and PCOS ovaries. Key quality control metrics, including total spots under tissue, median reads per spot, and sequencing saturation, are summarized in Supplementary Table S2.

Differential gene expression analysis between control and PCOS-like ovary was conducted using Seurat’s FindMarkers function. Genes with an adjusted p-value <0.05 and log2 fold change (FC) > 1 were considered significantly differentially expressed. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses were performed using the clusterProfiler package (v4.0) in R. Gene Set Enrichment Analysis (GSEA) was conducted using the fgsea package to identify significantly enriched biological processes and signaling pathways.

TCs proliferation was assessed using the EdU Cell Proliferation Kit (Beyotime, China). TCs were incubated with 10 μM EdU for 2 h, followed by fixation and detection of EdU incorporation according to the manufacturer’s instructions. Images were captured using a fluorescence microscope (Nikon, Japan), and the percentage of EdU-positive cells was quantified using ImageJ software.

Cell cycle progression in thecal cells was assessed by flow cytometry using a cell cycle and apoptosis detection kit (Beyotime, China), following the manufacturer’s instructions. Briefly, adherent cells were harvested by trypsinization, centrifuged, and washed with pre-chilled PBS. Cells were fixed in 70% ethanol at −20°C for 2 h, rinsed with PBS, and stained with 50 μg/mL propidium iodide and 100 μg/mL RNase A. The distribution of cells across G1, S, and G2/M phases was analyzed on a BD Biosciences flow cytometer using FlowJo software. Three independent biological replicates (n = 3) were included for each experimental group. Each replicate was processed and analyzed separately, without averaging across technical replicates.

Total RNAs from ovaries or TCs were applied to synthesize cDNA and then amplified to determine the expression levels of Inhba, Betaine-homocysteine S-methyltransferase (Bhmt), Hydroxysteroid (17-beta) dehydrogenase 7 (Hsd17b7), Inhibin subunit beta B (Inhbb), 7-dehydrocholesterol reductase (Dhcr7), Smad2 and E2f4. Relative gene expression was calculated using the 2^−ΔΔCt method, normalized to Gapdh as the internal control. The primer sequences were listed in Table 1.

After extraction and quantification, proteins from TCs were subjected to SDS-PAGE and subsequently transferred onto PVDF membranes. The membranes were then incubated overnight with primary antibodies against Inhba (1:1,000, Proteintech, United States), Smad2 (1:2,000, Proteintech, United States), E2f4 (1:2,000, Proteintech, United States), or Gapdh (1:5,000, Proteintech, United States), followed by treatment with an HRP-conjugated secondary antibody. Protein signals were visualized using an enhanced chemiluminescence substrate.

All data are presented as mean ± SD. Statistical analyses were performed using GraphPad Prism (v9.0). Differences between groups were analyzed using unpaired Student’s t-tests or one-way ANOVA followed by Tukey’s post-hoc test, as appropriate. pvalue <0.05 were considered statistically significant.

Spatial transcriptomics was employed to map cell type alterations in ovarian tissues from the PCOS mice (Figure 1A). Unsupervised clustering identified nine distinct transcriptional clusters across (Figure 1B). Both serum and ovarian testosterone levels were significantly elevated in the PCOS group, confirming a hyperandrogenic state. Moreover, estrous cycle monitoring showed that PCOS mice remained in the proestrus phase, indicating disrupted ovulatory function (Figures 1C–E). In the PCOS samples, the clusters 2, 4, and 6 were markedly expanded, while the clusters 1, 3, and 8 were reduced in sizes, indicating alterations in cell composition within the ovarian microenvironment. Heatmap analysis of the top 10 marker genes for each cluster revealed distinct expression patterns, highlighting potential functional heterogeneity (Figure 1F). UMAP visualization further confirmed this possibility, with PCOS samples enriched in the clusters 2, 4, and 6, and reduced in the clusters 1, 3, and 8 (Figure 1G). The dot plot of cluster markers reinforced the distinct gene expression signatures across clusters (Figure 1H). These results highlight substantial molecular and cellular alterations in ovarian tissue in response to DHEA-induced PCOS, providing insights into the reorganization of cell populations in the disorder.

To investigate the molecular mechanisms driving cluster alterations, we conducted differential gene expression and functional enrichment analysis between the control and PCOS samples. GO enrichment analysis revealed that genes involved in steroid biosynthesis, steroid metabolism, and cholesterol biosynthesis were significantly upregulated in the PCOS group (Figure 2A). Conversely, downregulated genes were primarily enriched in pathways related to immune responses, including leukocyte migration and chemotaxis. KEGG pathway analysis further identified significant upregulation of pathways involved in steroid biosynthesis and various metabolic processes in PCOS samples (Figure 2B). GSEA analysis supported these findings, revealing enrichment of cholesterol homeostasis and epithelial-to-mesenchymal transition pathways in the PCOS group (Figure 2C). Differential expression analysis between PCOS and control ovaries highlighted several key upregulated genes, including Inhba, Bhmt, Hsd17b7, Inhbb and Dhcr7 (Figure 2D). Co-expression network analysis of these upregulated and downregulated genes revealed strong associations, particularly among genes involved in steroidogenesis (Figures 2E,F). Cross-referencing with the DisGeNET PCOS database identified five overlapping genes Inhba, Bhmt, Hsd17b7, Inhbb and Dhcr7, which were notably elevated in the PCOS group and confirmed by both ST and qRT-PCR validation (Figures 2G–I). Among these, Inhba was the most significantly upregulated gene, suggesting a pivotal role in the pathogenesis of DHEA-induced PCOS.

In PCOS ovaries, cluster 6 was identified as a TCs population through AUC scoring with a TC-specific gene set (Morris et al., 2022) (Figures 3A,B). This cluster, defined as Lrp2^high TC, displayed significantly elevated expression of Lrp2 compared to other clusters. Notably, Lrp2^high TC was significantly expanded in PCOS samples relative to controls, as demonstrated by cluster composition analysis (Figures 3C,D). Gdf9 and Bmp15 are considered as markers of primary follicles, with their expression being minimal in primordial follicles but undergoing significant upregulation during the primary follicle stage. At this stage, these markers play pivotal roles in granulosa cell proliferation, differentiation, and oocyte-granulosa cell communication, emphasizing their importance in follicle activation and development (Sanfins et al., 2018). The co-localization of Gdf9, Bmp15 and Lrp2 in PCOS samples, shown through quantification and spatial visualization (Figures 3E,F), indicated that the proliferation of Lrp2^high TC is concentrated predominantly within primary follicles.

The Lrp2^high TC population demonstrated enriched expression of key androgen biosynthesis genes, including Cyp17a1 and Hsd3b1, with Cyp17a1 being predominantly expressed (Figure 4A). This suggests that Lrp2^high TC are significant contributors to androgen production in PCOS ovaries. KEGG pathway analysis further revealed a notable enrichment of cell cycle-related pathways in Lrp2^high TC, indicating a high proliferative potential of this cluster (Figure 4B). Intersection analysis identified 14 core cell cycle-related genes, and STRING network analysis highlighted the Inhba/Smad2/E2f4 signaling axis as the primary driver of Lrp2^high TC proliferation (Figures 4C,D). AUC analysis confirmed that Lrp2^high TC exhibit the highest co-expression levels of Inhba, Smad2, and E2f4, emphasizing the axis’s role in promoting cell cycle progression within this cluster (Figure 4E). Spatial transcriptomics further corroborated the interactions among Inhba, Smad2, and E2f4 in Lrp2^high TC through co-localization analysis. Quantitative analysis of co-localized spots revealed a significantly expanded region of co-localization in PCOS ovaries compared to controls (Figures 4F,G), highlighting the spatial and functional specificity of this proliferative, androgenic cell population.

To investigate the spatial dynamics of the Inhba/Smad2/E2f4 signaling axis in thecal cells under androgenic stimulation, immunofluorescence staining was conducted to assess subcellular localization. As shown in Figure 5A, Inhba was predominantly localized in the cytoplasm, whereas Smad2 and E2f4 were mainly detected in the nucleus, reflecting their canonical roles in intracellular signal relay and transcriptional regulation. Following DHEA treatment, expression levels of all three proteins were notably increased, with enhanced cytoplasmic Inhba signal and corresponding nuclear accumulation of Smad2 and E2f4. This spatial pattern supports the activation of a cytoplasm-to-nucleus signaling cascade in response to androgen stimulation.

To determine the functional relevance of this axis, EdU incorporation assays were performed after siRNA-mediated knockdown of Inhba, Smad2, or E2f4. As shown in Figures 5B,C, DHEA significantly increased the proportion of EdU-positive nuclei, indicating augmented DNA synthesis and proliferation. This proliferative effect was substantially attenuated upon silencing any of the three genes, with Inhba knockdown producing the most pronounced suppression. These findings underscore the critical role of the Inhba/Smad2/E2f4 axis in mediating androgen-driven thecal cell proliferation.

Flow cytometry was next employed to assess cell cycle changes (Figures 6A,B). DHEA treatment significantly decreased the proportion of cells in G1 phase while increasing S and G2/M phases, consistent with cell cycle progression and active proliferation. These effects were reversed upon knockdown of Inhba, Smad2, or E2f4, confirming the requirement of this axis for androgen-driven cell cycle advancement.

To further support these findings, qRT-PCR and Western blot analyses were performed to evaluate gene expression levels (Figures 6C–I). DHEA significantly elevated the mRNA and protein expression of Inhba, Smad2, and E2f4, while gene-specific siRNA knockdown effectively suppressed both transcript and protein levels. Together, these results establish the Inhba/Smad2/E2f4 axis as a critical mediator of androgen-induced TCs proliferation and cell cycle progression in the PCOS model.