3.1 Differential expression of plasma exosomal miRNAs in polycystic ovary syndrome patients

Deep sequencing analysis of the miRNA transcriptome (dataset GSE142819) was performed on plasma samples from 15 control subjects and 15 patients with polycystic ovary syndrome (PCOS). This differential expression analysis identified 15 miRNAs with significant expression changes (Fig. 1A). Among them, 12 miRNAs were up-regulated in PCOS patients, including miR-483, miR-206, miR-142, miR-330, miR-3120, miR-203a, miR-1228, miR-1246, miR-3656-p3, miR-4792, miR-193b, and miR-1 (Fig. 1B). In contrast, three miRNAs—miR-331, let-7f, and miR-424—were found to be down-regulated in PCOS samples.

Fig. 1

Differential analysis identified 15 differentially expressed plasma exosomal miRNAs in PCOS patients. A Volcano plot showing differential expression of plasma exosomal miRNAs. B Top ten miRNAs with the most significant fold changes

3.2 miRNAs as prognostic markers in UCEC tissue

PCOS has been associated with an elevated risk of developing UCEC. To explore the prognostic significance of the 15 differentially expressed miRNAs identified in PCOS, we analyzed their association with overall survival in UCEC patients. Notably, 12 of the 15 miRNAs were significantly linked to overall survival, except for miR-1, let-7f, and miR-483. Among these, miR-142 (up-regulated in PCOS), miR-424, and miR-331 (both down-regulated in PCOS) were identified as protective factors. In contrast, the remaining nine miRNAs, all up-regulated in PCOS, were associated with increased risk (Fig. 2). These findings suggest that these miRNAs may play a critical role in the heightened risk of developing UCEC in PCOS patients.

Fig. 2

Survival analysis of tissue miRNAs in UCEC patients grouped by miR expression. Survival curves for UCEC patients stratified by low and high expression levels of miR-142, miR-3120, miR-330, miR-206, miR-3656, miR-4792, miR-1246, miR-1228, miR-193b, miR-203a, miR-424, and miR-331

3.3 miRNA and target gene network analysis identifies key genes

To elucidate the relationships between miRNAs and their target genes, two miRNA-gene interaction networks were constructed—one for up-regulated miRNAs and the other for down-regulated miRNAs. Through these networks, several key genes were identified. In Network A, genes such as PHF8, LCOR, and SFT2D3 were strongly associated with three specific miRNAs (Fig. 3A). In Network B, genes like E2F1, ESR1, C5orf51, VEGFA, and others showed significant connections with various miRNAs (Fig. 3B). To validate the relevance of these genes, survival analysis was conducted, revealing that most of the key genes in both networks were prognostic for overall survival in UCEC patients (Fig. 3C). These findings suggest that these genes may play critical roles in the progression and prognosis of UCEC.

Fig. 3

miRNA-gene interaction network reveals key genes. A miRNA-gene interaction network for down-regulated miRNAs. B miRNA-gene interaction network for up-regulated miRNAs. C Survival curves for key genes SFT2D3, E2F1, ESR1, C5orf51, and VEGFA

3.4 Differential gene expression analysis and co-expression analysis of UCEC stratified by tissue miR-424 expression

Given that miR-424 is a protective factor but is down-regulated in PCOS, we explored the differences between UCEC patients with high versus low miR-424 expression. Differential gene expression analysis identified 550 genes (250 up-regulated and 300 down-regulated) in patients with high miR-424 expression compared to those with low expression (Fig. 4A). Notably, the most significantly up-regulated genes included SFRP4, ALDH1A2, APOD, CCL21, TWIST2, PAMR1, ECM1, CD248, WNT2, and LTBP4 (Fig. 4B). These up-regulated genes were mainly involved in extracellular matrix organization and response to TGF-β signaling (Fig. 4C), while the down-regulated genes were associated with ribosome function and mitochondrial gene expression (Fig. 4D). Additionally, pathways related to the up-regulated genes were enriched in focal adhesion processes (Fig. 4E).

Fig. 4

Differential gene expression and co-expression analysis based on tissue miR-424 expression in UCEC tissues. A Differential gene expression analysis identified 550 genes in samples with high miR-424 compared to low miR-424. B Boxplots illustrating the top ten significantly changed genes. C–E Gene Ontology (GO) biological processes and pathway enrichment analyses for up-regulated and down-regulated genes. F Heatmap depicting the relationship between four co-expression modules and miR-424. Each cell represents the correlation between the expression of specific modules and distinct clinical parameters, with numerical values in brackets indicating the statistical significance of the correlation

Gene co-expression analysis revealed four distinct gene co-expression modules (Fig. 4F). Functional enrichment analysis showed that up-regulated module ME1 was related to focal adhesion, while down-regulated module ME2 was linked to the cell cycle. Modules ME3 and ME5 were both associated with ncRNA metabolism. Intriguingly, eight of the top ten significantly up-regulated genes—SFRP4, APOD, CCL21, TWIST2, PAMR1, ECM1, CD248, and WNT2—were all correlated with longer overall survival (Figure S1) and were enriched in the regulation of T cell migration.

Further, correlation analysis identified ten genes most significantly associated with miR-424 expression, several of which, including SFRP4, ALDH1A2, and CD248, were linked to the positive regulation of apoptosis. This may help explain why UCEC patients with high miR-424 expression had improved survival rates.

3.5 Differential gene expression analysis and co-expression analysis of UCEC stratified by tissue miR-330 expression

As miR-330 is a risk factor and is up-regulated in PCOS, we explored the differences between UCEC patients with high and low miR-330 expression. Differential gene expression analysis identified 1289 genes (715 up-regulated and 574 down-regulated) in patients with high miR-330 expression compared to those with low expression (Fig. 5A). The most significantly altered genes were predominantly up-regulated, including AURKA, ABHD3, NCAPG, YARS, RRM2, AUNIP, and MTFR2 (Fig. 5B). These up-regulated genes were primarily involved in nuclear division processes (Fig. 5C), while down-regulated genes were associated with the cellular response to TGF-β signaling (Fig. 5D). In an independent ovarian cancer cell line dataset [14], we observed that all the most significantly altered genes, except AUNIP, displayed consistent directional changes following treatment with the miR-330 mimic.

Fig. 5

Differential gene expression and co-expression analysis based on tissue miR-330 expression in UCEC tissues. A Differential gene expression analysis identified 1289 genes in high miR-330 samples compared to low miR-330 samples. B Boxplots showing the top ten significantly changed genes. C–F Gene Ontology (GO) biological processes and pathway enrichment analyses for up-regulated and down-regulated genes

Notably, seven of the top ten significantly altered genes were up-regulated, and six of them (AURKA, NCAPG, YARS, RRM2, AUNIP, and MTFR2) were linked to shorter overall survival (Figure S2). These genes were enriched in pathways related to the positive regulation of the cell cycle (Fig. 5E). In contrast, the three down-regulated genes—WBP1L, WFS1, and AXIN2—were associated with longer overall survival and were involved in the positive regulation of protein ubiquitination. Pathway enrichment analysis revealed that up-regulated genes were primarily associated with the cell cycle, while down-regulated genes were linked to focal adhesion (Fig. 5F).

Gene co-expression analysis identified five distinct co-expression modules (Fig. 5F). Functional enrichment analysis revealed that the up-regulated modules ME1 and ME3 were involved in cell cycle regulation and immune system processes, while the down-regulated modules ME2, ME4, and ME5 were associated with focal adhesion, TGF-β signaling, and cilium assembly. Interestingly, the top ten genes most strongly correlated with miR-330 expression, identified through correlation analysis, were predominantly involved in the positive regulation of the cell cycle, including AURKA. These findings provide a potential explanation for the poorer survival outcomes in UCEC patients with high miR-330 expression.

3.6 Validation of the prognostic gene AURKA in UCEC by Mendelian randomization

The Mendelian randomization (MR) analysis revealed a significant positive association between AURKA eQTL SNPs and PCOS (Fig. 6A). A funnel plot showed a generally symmetric distribution of the genetic variants' estimates, suggesting minimal evidence of directional pleiotropy (Fig. 6B). The robustness of these results was further confirmed through sensitivity analyses (Fig. 6C). Leave-one-out analysis was performed to assess whether any single SNP disproportionately influenced the overall causal estimate. After sequentially excluding each SNP, the effect estimates remained consistent, demonstrating the stability of the results (Fig. 6C). This suggests that no single genetic variant was driving the association between AURKA and PCOS. The similar analysis was also performed for casual relationship between AURKA and UCEC (Fig. 6D, F). Taken together, these analyses support robust causal and positive relationships between AURKA and PCOS and UCEC, with minimal evidence of bias from pleiotropy or outlier SNPs.

Fig. 6

Two-sample Mendelian randomization analysis illustrating the causal effects of AURKA gene expression on PCOS and UCEC. A Scatter plot of effects of cis-eQTL SNPs on AURKA expression and PCOS occurence. B Funnel plot for PCOS. C Leave-one-out sensitivity analysis for PCOS. D Scatter plot of effects of cis-eQTL SNPs on AURKA expression and UCEC occurence. E Funnel plot for UCEC. F Leave-one-out sensitivity analysis for UCEC

3.7 Comparative analysis reveals differential cell subtypes in UCEC patients stratified by tissue miR-330 expression

To investigate the cellular composition differences between UCEC groups stratified by tissue miR-330 expression, we applied the state-of-the-art deconvolution algorithm Ecotyper to the TCGA UCEC samples. The analysis revealed significant heterogeneity in the cellular composition across UCEC samples, as illustrated by the heatmap (Fig. 7A). Notably, we identified distinct differences in cellular composition between groups with high and low miR-330 expression.

Fig. 7

Differences in tissue cellular composition in UCEC patients stratified by miR-330 expression. A Heatmap showing the abundance of cell subtypes in UCEC patients. B Identification of the top 12 differential cell subtypes in high miR-330 samples compared to low miR-330 samples. C Survival curve for samples stratified by abundance of fibroblasts.8. D Survival curve for samples stratified by the abundance of CD4.T.cells.1. In both survival curves, the green line represents high expression and the red line represents low expression

The top five significantly altered cell subtypes were epithelial.cells.4, epithelial.cells.2, fibroblasts.8, CD8.T.cells.3, and dendritic.cells.5 (Fig. 7B). According to cell state annotation, epithelial.cells.4 is a pro-inflammatory subset, fibroblasts.8 is a pro-migratory-like subset, and CD8.T.cells.3 represents an exhausted T cell state. All three of these subtypes were up-regulated in miR-330 high samples. In contrast, epithelial.cells.2 and dendritic.cells.5, which are normal-enriched subtypes, were down-regulated in miR-330 high samples. Additionally, CD4.T.cells.1, annotated as another exhausted T cell subtype, was up-regulated.

Moreover, survival analysis revealed that fibroblasts.8 and CD4.T.cells.1 were both significantly linked to poorer overall survival in UCEC patients (Fig. 7C, D). These findings suggest that the aberrant immune cell composition in miR-330 high patients, characterized by increased pro-inflammatory and exhausted cell types, may contribute to the poorer clinical outcomes observed in this group.

3.8 Identification and validation of UCEC-related drugs based on DEGs from miR-330 high samples

We utilized the differentially expressed genes (DEGs) from miR-330 high samples to screen for candidate drugs associated with UCEC. A total of twenty candidate drugs were identified, several of which were hormone-related (Table 1). Notably, testosterone was correlated with the up-regulated genes, while two phytoestrogens, enterolactone and coumestrol, were also associated with the up-regulated gene set. Additionally, hormone-related drugs such as calcitriol, retinoic acid, progesterone, and estradiol were identified. These findings highlight the potential involvement of aberrant hormone signaling pathways in the progression of UCEC.

Table 1 Identification of candidate drugs based on differential gene expression in high miR-330 samples

To validate the associations of the candidate drugs, experimental data from the Ishikawa cell line treated with six selected drugs was analyzed. Our findings revealed that testosterone down-regulated estrogen 16-alpha-hydroxylase activity while up-regulating cholesterol metabolism. Estradiol was observed to down-regulate extracellular exosome production while up-regulating developmental processes. These effects suggest potential adverse roles of testosterone and estradiol in UCEC progression. In addition, troglitazone was found to down-regulate artery development while up-regulating cholesterol metabolism. Valproic acid down-regulated nucleoplasm activity and up-regulated cell junction formation. Retinoic acid down-regulated cell migration while up-regulating programmed cell death, indicating its potential as an anti-tumor agent. Lastly, progesterone down-regulated the cell cycle while up-regulating primary metabolism (Fig. 8). These experimental data suggest that these drugs could modulate critical biological pathways in UCEC, potentially impacting disease progression and providing insights into therapeutic strategies.

Fig. 8

Impact of six candidate drugs on gene expression in the Ishikawa cell line of endometrial carcinoma. The bars represent the number of down-regulated and up-regulated genes following 6-h drug treatment. Functional enrichment analysis terms are annotated around the bars, with numbers in parentheses indicating the statistical significance of the enrichment