ANGPTL4 induces granulosa cells proliferation dysfunction in PCOS by regulating the immune microenvironment and WNT signaling pathway.

A total of 343 patients were included in this study, conprising 168 patients with PCOS and 175 control women. The clinical baseline characteristics and outcomes of IVF/ICSI treatment were recorded and summarized. Results showed that PCOS patients had significantly higher infertility duration, AFC and higher levels of AMH, LH, T, ANGPTL4, LDL and TG, and lower levels of FSH and HDL (Table  1 ). Table 1 Baseline characteristics of control group and PCOS group Parameter Control Group ( n  = 175) PCOS Group ( n  = 168) P Age (year) 30.31 ± 4.06 29.79 ± 3.71 0.215 BMI (kg/m²) 20.96 ± 1.73 23.26 ± 2.00 0.090 AFC (n) 16.60 ± 6.02 28.33 ± 9.62 < 0.001 * AMH (ng/mL) 4.19 ± 2.00 8.34 ± 3.90 < 0.001 * ANGPTL4 in follicular fluid (ng/mL) 24.51 ± 4.31 36.86 ± 6.08 < 0.001 * FSH (IU/L) 6.32 ± 1.52 5.60 ± 1.29 < 0.001 * LH (mIU/mL) 5.10 ± 2.30 8.90 ± 4.89 < 0.001 * E2 (pg/mL) 40.58 ± 27.34 39.91 ± 18.12 0.789 PRL (ng/mL) 15.93 ± 5.90 14.93 ± 6.55 0.140 T (ng/dL) 24.61 ± 12.51 37.89 ± 19.18 < 0.001 * TSH (µIU/mL) 2.16 ± 0.97 2.37 ± 0.90 0.037 FBG (mmol/L) 5.27 ± 0.37 5.22 ± 0.58 0.284 TG (mmol/L) 1.19 ± 0.68 1.38 ± 0.73 0.011 * LDL (mmol/L) 1.88 ± 0.61 2.55 ± 1.61 < 0.001 * HDL (mmol/L) 1.36 ± 0.32 1.28 ± 0.38 0.040 * Information on continuous variables that conformed to normal distribution was analyzed using the two-independent samples t-test and is expressed as mean ± standard deviation(SD). All tests were two-tailed, and a P -value of less than 0.05 was considered statistically significant.*, P  < 0.05 Baseline characteristics of control group and PCOS group Information on continuous variables that conformed to normal distribution was analyzed using the two-independent samples t-test and is expressed as mean ± standard deviation(SD). All tests were two-tailed, and a P -value of less than 0.05 was considered statistically significant.*, P  < 0.05 In this study, we quantified the expression of ANGPTL4 in follicular fluid from all patients. Our findings indicate that ANGPTL4 expression in follicular fluid from the PCOS group was significantly higher compared to the control group (24.51 ± 4.31 vs. 36.86 ± 6.08, P  < 0.001, Table  1 ). To investigate the relationship between ANGPTL4 and pregnancy outcomes in patients with PCOS, We categorized patients with PCOS into high and low ANGPTL4 expression groups based on median follicular fluid ANGPTL4 levels (37.6 ng/mL). The baseline characteristics of the two groups of patients were compared (Table  2 ). The levels of BMI (22.53 ± 1.75 vs. 24.90 ± 1.99, P  < 0.001) and LDL (2.39 ± 1.11 vs. 3.29 ± 1.00, P  < 0.001) were significantly higher in the high ANGPTL4 group. The clinical characteristics of the two groups of patients were compared (Table  3 ). The high ANGPTL4 group required higher initial gonadotropin (Gn) dosage (131.65 ± 28.44 vs. 147.75 ± 29.65, P  < 0.001) and hCG administration (5620.25 ± 2260.61 vs. 6606.74 ± 2565.59, P  < 0.05), with lower LH (5.49 ± 6.92 vs. 3.52 ± 2.90, P  < 0.05), E₂(4325.94 ± 1910.36 vs. 3515.64 ± 2644.60, P  < 0.05), P (0.91 ± 0.56 vs. 0.55 ± 0.44, P  < 0.001) levels on the hCG day (Table  3 ). Key reproductive indicators such as the number oocyte retrieved (16.67 ± 7.40 vs. 12.43 ± 7.53, P  < 0.001), 2PN fertilized oocytes (9.85 ± 4.39 vs. 6.47 ± 3.53, P  < 0.001), blastocyst formation(7.19 ± 4.64 vs. 4.44 ± 3.49, P  < 0.001), and high-quality embryos on Day 3 (5.52 ± 3.69 vs. 3.64 ± 2.90, P  < 0.001) were significantly lower in the high ANGPTL4 group, along with reduced cleavage rates (100% ± 0% vs. 96% ± 18%, P  < 0.05) and ICSI fertilization rates (83% ± 13% vs. 70% ± 24%, P  < 0.05)(Table  3 ). Additionally, the high ANGPTL4 group exhibited significantly higher biochemical pregnancy loss rates (3.1% [2/65] vs. 12.2 [9/74], P  < 0.05) and abortion rates (32.7% [1/37] vs. 16.7 [6/36], P  < 0.05) after fresh embryo transfer (Table  3 ). No significant differences were observed in clinical pregnancy rates and live birth rates between the two groups, suggesting that high ANGPTL4 expression may be associated with increased demand for ovulation induction drugs, compromised oocyte/embryo quality, and higher early pregnancy loss risk, though its impact on clinical pregnancy rate requires further investigation. Table 2 Baseline characteristics of low ANGPTL4 group and high ANGPTL4 group in PCOS patients Parameter Low ANGPTL4 ( n  = 79) High ANGPTL4 ( n  = 89) P Age(year) 29.86 ± 4.05 29.73 ± 3.40 0.821 BMI(kg/m 2 ) 22.53 ± 1.75 24.90 ± 1.99 < 0.001 * AMH(ng/mL) 8.22 ± 3.72 9.08 ± 4.05 0.455 AFC(n) 28.78 ± 11.12 25.16 ± 5.96 0.067 FSH(IU/L) 5.40 ± 1.20 5.77 ± 1.33 0.370 LH(mIU/mL) 8.38 ± 5.66 9.01 ± 4.06 0.067 E2(pg/mL) 41.0 ± 13.56 39.69 ± 21.41 0.179 PRL(ng/mL) 15.65 ± 6.97 14.29 ± 6.12 0.408 T(ng/dL) 37.38 ± 22.11 40.56 ± 16.15 0.904 TSH(µIU/mL) 2.59 ± 0.95 2.36 ± 0.81 0.285 LDL(mmol/L) 2.39 ± 1.11 3.29 ± 1.00 < 0.001 * TG(mmol/L) 1.40 ± 0.75 1.36 ± 0.70 0.340 HDL(mmol/L) 1.30 ± 0.37 1.25 ± 0.38 0.702 Information on continuous variables that conformed to normal distribution was analyzed using the two-independent samples t-test and is expressed as mean ± SD. All tests were two-tailed, and a P -value of less than 0.05 was considered statistically significant.*, P  < 0.05 Baseline characteristics of low ANGPTL4 group and high ANGPTL4 group in PCOS patients Information on continuous variables that conformed to normal distribution was analyzed using the two-independent samples t-test and is expressed as mean ± SD. All tests were two-tailed, and a P -value of less than 0.05 was considered statistically significant.*, P  < 0.05 Table 3 ANGPTL4 expression and assisted reproductive outcomes in PCOS patients Parameter Low ANGPTL4 ( n  = 79) High ANGPTL4 ( n  = 89) P Initial gonadotropin dose (IU) 131.65 ± 28.44 147.75 ± 29.65 < 0.001 * Days of ovarian stimulation (d) 11.01 ± 2.11 10.51 ± 2.68 0.178 Total gonadotropin dose (IU) 2071.84 ± 902.39 1852.67 ± 799.95 0.097 hCG trigger dose(IU) 5620.25 ± 2260.61 6606.74 ± 2565.59 0.009 * LH on hCG trigger day (mIU/mL) 5.49 ± 6.92 3.52 ± 2.90 0.016 * E2 on hCG trigger day (pg/mL) 4325.94 ± 1910.36 3515.64 ± 2644.60 0.026 * P on hCG trigger day (ng/mL) 0.91 ± 0.56 0.55 ± 0.44  14 mm, n) 13.90 ± 5.08 12.34 ± 5.40 0.056 Number of oocytes retrieved (n) 16.67 ± 7.40 12.43 ± 7.53 < 0.001 * Cleavage rate(%) 100.00 ± 0.00 96.00 ± 18.00 0.042 * Number of 2PN fertilized oocytes (n) 9.85 ± 4.39 6.47 ± 3.53 < 0.001 * IVF fertilization rate(%) 75.00 ± 20.00 72.00 ± 24.00 0.415 ICSI fertilization rate(%) 83.00 ± 13.00 70.00 ± 24.00 0.027 * Number of blastocysts cultured (n) 9.94 ± 5.33 6.29 ± 3.92 < 0.001 * Number of blastocysts formed (n) 7.19 ± 4.64 4.44 ± 3.49 < 0.001 * Blastocyst formation rate(%) 67.00 ± 29.00 60.00 ± 35.00 0.227 Number of good-score embryos on day 3(n) 5.52 ± 3.69 3.64 ± 2.90 < 0.001 * Number of fresh embryo transfer (n) 65 74 — Biochemical miscarriage rate (%) 3.10 (2/65) 12.20 (9/74) 0.048 * Clinical pregnancy rate (%) 56.90 (37/65) 48.60 (36/74) 0.330 Ectopic pregnancy rate (%) 2.70 (1/37) 2.80 (1/36) 0.984 Abortion rate (%) 2.70 (1/37) 16.70 (6/36) 0.043 * Live birth rate (%) 53.80 (35/65) 39.20 (29/74) 0.084 Information on continuous variables that conformed to normal distribution was analyzed using the two-independent samples t-test and is expressed as mean ± SD. Categorical variable information was analyzed using the chi-square test to analyze differences between groups and expressed as a rate (%). All tests were two-tailed, and a P -value of less than 0.05 was considered statistically significant.*, P  < 0.05 ANGPTL4 expression and assisted reproductive outcomes in PCOS patients Information on continuous variables that conformed to normal distribution was analyzed using the two-independent samples t-test and is expressed as mean ± SD. Categorical variable information was analyzed using the chi-square test to analyze differences between groups and expressed as a rate (%). All tests were two-tailed, and a P -value of less than 0.05 was considered statistically significant.*, P  < 0.05 To further validate the importance of gene ANGPTL4 in PCOS, a total of 4,818 DEGs were identified from the GSE155489 dataset, including 3,478 downregulated genes and 1,340 upregulated genes. Notably, ANGPTL4 was found to be significantly upregulated in the GSE155489 dataset (Fig.  1 A). GO enrichment analysis showed that DEGs were primarily associated with GO terms such as detection of chemical stimuli (involving sensory perception), olfactory sensory perception, hormone metabolic processes, regulation of leukocyte migration, monoatomic anion transport, and chloride transport (Fig.  1 B). In the cellular component (CC) category, genes were enriched in neuronal cell body, collagen-containing extracellular matrix, synaptic membrane, postsynaptic membrane, monoatomic ion channel complex, endoplasmic reticulum lumen, presynaptic membrane, basement membrane, GABA-A receptor complex, and GABA receptor complex. In the molecular function (MF) category, enrichment was observed in olfactory receptor activity, channel activity, passive transmembrane transporter activity, monoatomic ion channel activity, odorant binding, structural constituent of extracellular matrix, transmitter-gated monoatomic ion channel activity, transmitter-gated channel activity, ligand-gated monoatomic anion channel activity, and GABA-gated chloride channel activity. KEGG pathway analysis revealed that DEGs were significantly enriched in olfactory transduction, neuroactive ligand-receptor interaction, PI3K-Akt signaling pathway, cytokine-cytokine receptor interaction, glutamatergic synapse, morphine addiction, drug metabolism – cytochrome P450, GABAergic synapse, viral protein interaction with cytokines and cytokine receptors, protein digestion and absorption, xenobiotic metabolism by cytochrome P450, retinol metabolism, nicotine addiction, complement and coagulation cascades, ECM-receptor interaction, taste transduction, bile secretion, malaria, and steroid hormone biosynthesis (Fig.  1 C-D). Fig. 1 Screening and enrichment of DEGs in PCOS GCs. A Volcano plot illustrating DEGs between PCOS and control GCs from GSE155489 dataset. Red points represent upregulated genes blue points denote downregulated genes, and gray points indicate non-significant genes. B Gene Ontology (GO) enrichment analysis of DEGs, highlighting significant BP, CCs and MFs. Top enriched terms are labeled, with bubble size representing the number of enriched genes and color gradient indicating the adjusted P-value. C Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment bubble plot showing significantly altered pathways. X-axis denotes enrichment factor, Y-axis lists pathway names, bubble size reflects gene count, and color indicates significance. D KEGG enrichment sankey diagram visualizing the relationship between DEGs and enriched pathways. Arrows represent gene-pathway associations, with width proportional to the number of genes involved Screening and enrichment of DEGs in PCOS GCs. A Volcano plot illustrating DEGs between PCOS and control GCs from GSE155489 dataset. Red points represent upregulated genes blue points denote downregulated genes, and gray points indicate non-significant genes. B Gene Ontology (GO) enrichment analysis of DEGs, highlighting significant BP, CCs and MFs. Top enriched terms are labeled, with bubble size representing the number of enriched genes and color gradient indicating the adjusted P-value. C Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment bubble plot showing significantly altered pathways. X-axis denotes enrichment factor, Y-axis lists pathway names, bubble size reflects gene count, and color indicates significance. D KEGG enrichment sankey diagram visualizing the relationship between DEGs and enriched pathways. Arrows represent gene-pathway associations, with width proportional to the number of genes involved To investigate interactions between proteins, pathways, and co-expression relationships, a PPI network of 4,818 DEGs was constructed using the STRING database. The PPI network demonstrated complex interaction patterns among most DEGs, with a minimum interaction score of 0.7, 403 nodes, 554 edges, an average node degree of 2.75, an average local clustering coefficient of 0.317, 123 expected edges, and a PPI enrichment p-value of < 1.0e-16 (Fig. S1). Enrichment analysis of PPI network genes showed that significant enrichment in hormone metabolic processes, leukocyte migration, ERK1/ERK2 cascade and its regulation, negative regulation of immune system processes in Biological processes (BP), enrichment in synaptic membrane, collagen-containing extracellular matrix, neuronal cell body, endoplasmic reticulum lumen in CC, and focus on ligand-gated monoatomic ion channel activity, neurotransmitter receptor activity, transporter complex MF(Fig.  2 A). KEGG pathway analysis revealed significant enrichment in neuroactive ligand-receptor interaction, PI3K-Akt signaling pathway, cytochrome P450 drug metabolism, complement and coagulation cascades, and cAMP signaling pathway, involving physiological and disease-related pathways such as nicotine addiction, steroid hormone biosynthesis, and glutamatergic synapse (Fig.  2 B-C). These findings suggest that PPI network genes may participate in pathological and physiological mechanisms by regulating neural signaling, metabolic processes, and immune responses. Fig. 2 Construction of protein-protein interaction (PPI) network for DEGs. A GO enrichment analysis of PPI network genes. B KEGG enrichment bubble plot of PPI network genes. C KEGG enrichment sankey diagram of PPI network genes Construction of protein-protein interaction (PPI) network for DEGs. A GO enrichment analysis of PPI network genes. B KEGG enrichment bubble plot of PPI network genes. C KEGG enrichment sankey diagram of PPI network genes Through LASSO regression analysis, we systematically explored the dynamic relationship between model complexity and the logarithmic penalty parameter (logλ). As the logλ value gradually increased, model complexity showed a significant downward trend, and the coefficients of various features gradually shrank to zero. When the logλ value was approximately − 2.7, the mean squared error (MSE) curve tended to stabilize, indicating that the model achieved a stable and effective fitting state. Based on this, we successfully screened five core feature genes: CXCL2, SLC2A1, KLF4, ANGPTL4, and HSPB3 (Fig.  3 A-B). Fig. 3 Identification of hub genes. A Coefficient distribution map for the logarithmic (λ) sequence in the LASSO model. B LASSO coefficient spectrum from LASSO cox analysis. C Feature gene selection plot by SVM-RF. D Venn diagram of genes screened by the two algorithms Identification of hub genes. A Coefficient distribution map for the logarithmic (λ) sequence in the LASSO model. B LASSO coefficient spectrum from LASSO cox analysis. C Feature gene selection plot by SVM-RF. D Venn diagram of genes screened by the two algorithms In the exploration of the SVM-RF recursive feature elimination algorithm, we found that when the number of features was optimized to 7, the model performance reached the optimal level. At this point, the cross-validated root mean square error (RMSE) dropped to the lowest value (approximately 0.27), and the screened gene set included COL8A2, JUNB, PEX5L, SLC2A1, CHAC1, NTS, and ANGPTL4 (Fig.  3 C). This result strongly demonstrates the superiority of this feature combination in predictive performance, providing a reliable direction for gene screening in subsequent studies. Further analysis using a Venn diagram for the gene sets screened by LASSO and SVM-RF algorithms revealed that two genes, ANGPTL4 and SLC2A1, were identified in both results (Fig.  3 D). This finding not only reflects the consistency of feature selection across different algorithms but also highlights the unique specificity of each algorithm in gene screening. Although the SLC2A1 gene detected in the same batch showed significant differences at the sequencing level, subsequent PCR validation revealed no statistically significant difference in the mRNA expression of SLC2A1 in human ovarian granulosa cells(Fig. S4 ). And the expression trend of ANGPTL4 was completely consistent with the sequencing results(Fig.  7 A). Therefore, SLC2A1 was not included in the core research. Immune cell infiltration scores for each sample were calculated using the ssGSEA method. Pearson correlation analysis was performed between hub genes and these immune scores.ANGPTL4 expression was significantly negatively correlated with infiltration levels of CD8⁺ T cells, macrophages, a Dendritic Cells(DCs), B cells, T helper type 1 (Th1) cells, pDCs, and Th2 cells ( P  < 0.001), and significantly positively correlated with the activity of immune pathways such as Type II interferon(IFN) response, Checkpoint, MHC class I, Parainflammation, Type I IFN response, Neutrophils, regulatory T cell(Treg), and APC co-inhibition ( P  < 0.001) (Fig.  4 A). Fig. 4 Correlation between hub genes and immune scores. A The upper bar chart displays results sorted by ANGPTL4 expression, where the height of each bar represents the corresponding gene expression level; the lower heat map presents the Z-score values of immune cells and related pathways in the matched samples. B The upper bar chart shows results sorted by SLC2A1 expression, with bar height indicating the gene expression level; the lower heat map illustrates the Z-score values of immune cells and related pathways in the corresponding samples.Pearson correlation analysis was used for statistical analysis, with * indicating P  < 0.05, ** indicating P  < 0.01, and *** indicating P  < 0.001 Correlation between hub genes and immune scores. A The upper bar chart displays results sorted by ANGPTL4 expression, where the height of each bar represents the corresponding gene expression level; the lower heat map presents the Z-score values of immune cells and related pathways in the matched samples. B The upper bar chart shows results sorted by SLC2A1 expression, with bar height indicating the gene expression level; the lower heat map illustrates the Z-score values of immune cells and related pathways in the corresponding samples.Pearson correlation analysis was used for statistical analysis, with * indicating P  < 0.05, ** indicating P  < 0.01, and *** indicating P  < 0.001 SLC2A1 expression was significantly negatively correlated with infiltration levels of CD8⁺ T cells, macrophages, B cells, pDCs, and Th2 cells ( P  < 0.001), and significantly positively correlated with Type II IFN response, MHC class I, Type I IFN response, Neutrophils, APC co-inhibition, Treg, and Parainflammation ( P  < 0.05) (Fig.  4 B). Specific information on the correlation and P-values of genes associated with immunity is shown in the attached Fig. S2 and S3. Additionally, a significant positive correlation was observed between the hub genes ANGPTL4 and SLC2A1. Collectively, the expression levels of ANGPTL4 and SLC2A1 are significantly associated with infiltration of multiple immune cells and activity of inflammatory pathways, suggesting their involvement in disease progression by regulating the immune microenvironment (Fig.  4 A-B). In both the training dataset ( GSE155489 ) and validation dataset ( GSE10946 ), the expression levels of hub genes ANGPTL4 and SLC2A1 were higher in PCOS groups than in control groups ( P  < 0.05) (Fig.  5 A-B), indicating its robust overexpression in PCOS across datasets and potential as a core biomarker. Its function may be linked to pathological mechanisms such as regulating immune cell infiltration and metabolic pathways. The significance of SLC2A1 was only observed in the training dataset, requiring further validation. Fig. 5 Validation and GSEA of hub genes. A Expression of ANGPTL4 in PCOS and control groups ( GSE155489 ).This dataset contains granulosa cells from 4 healthy individuals and 4 PCOS patients. B Expression of SLC2A1 in PCOS and control groups ( GSE10946 ; this dataset contains granulosa cells from 11 healthy individuals and 12 PCOS patients). C Gene Set Enrichment Analysis (GSEA) results for ANGPTL4 (key pathways shown). D GSEA results for SLC2A1 (key pathways shown).For ( A ) and ( B ), error bars represent the standard deviation (SD) of the expression levels. Statistical analysis was performed using the Wilcoxon rank-sum test, with ns indicating P  > 0.05, * indicating P  < 0.05, ** indicating P  < 0.01, and *** indicating P  < 0.001 Validation and GSEA of hub genes. A Expression of ANGPTL4 in PCOS and control groups ( GSE155489 ).This dataset contains granulosa cells from 4 healthy individuals and 4 PCOS patients. B Expression of SLC2A1 in PCOS and control groups ( GSE10946 ; this dataset contains granulosa cells from 11 healthy individuals and 12 PCOS patients). C Gene Set Enrichment Analysis (GSEA) results for ANGPTL4 (key pathways shown). D GSEA results for SLC2A1 (key pathways shown).For ( A ) and ( B ), error bars represent the standard deviation (SD) of the expression levels. Statistical analysis was performed using the Wilcoxon rank-sum test, with ns indicating P  > 0.05, * indicating P  < 0.05, ** indicating P  < 0.01, and *** indicating P  < 0.001 GSEA revealed that ANGPTL4 was significantly positively enriched in pathways such as piwi interacting RNA pirna biogenesis, medicus env factor iron to anterograde axonal transport, and medicus pathogen escherichia ESPG to microtubule RHOA signaling pathway (NES > 1), suggesting its involvement in BP via mechanisms like metabolic substrate supply, signal transduction, or energy homeostasis (Fig.  5 C). ANGPTL4 was significantly negatively enriched in WNT signaling pathway and KEGG GNRH signaling pathway (NES<-1), indicating inhibitory effects on these pathways or a negative correlation with pathway activity, potentially contributing to metabolic imbalance or signal pathway suppression. SLC2A1 (Fig.  5 D) was significantly positively enriched in pyrimidine metabolism, transforming growth factor-beta(TGF-β) signaling pathway, and galactose metabolism (NES > 1), implicating roles in metabolic reprogramming and signal transduction. SLC2A1 was significantly negatively enriched in the same metabolic and signaling pathways (NES<-1), suggesting complex regulatory roles. Notably, both hub genes were significantly negatively enriched in the WNT signaling pathway, highlighting WNT pathway dysregulation as a potential key mechanism in PCOS. Through NetworkAnalyst prediction, a total of 42 miRNAs and 33 TFs were identified to form complex interaction networks with the two hub genes (Fig.  6 ). Fig. 6 Hub gene-miRNA-TF regulatory network. The blue nodes on the left represent miRNAs, while the green nodes on the right denote TFs Hub gene-miRNA-TF regulatory network. The blue nodes on the left represent miRNAs, while the green nodes on the right denote TFs The results of PCR detection showed that the mRNA level of ANGPTL4 in ovarian GCs of PCOS patients was significantly highly expressed, suggesting that the abnormal expression of ANGPTL4 at the transcriptional level was involved in the pathological process of PCOS (Fig.  7 A).The ELISA detection results further indicated that the protein level of ANGPTL4 in the follicular fluid of PCOS patients was also significantly highly expressed, confirming that the abnormal expression of this gene at the protein level was equally associated with the pathological mechanism of PCOS (Fig.  7 B). Fig. 7 External experiments and pathway validation of ANGPTL4 gene. A The transcriptional level of ANGPTL4 gene in ovarian GCs of control patirnts( n  = 58) and PCOS patients( n  = 50) by PCR. B The protein expression of ANGPTL4 in the follicular fluid of control patirnts( n  = 175) and PCOS patients( n  = 168) by ELISA. C The effect of ANGPTL4 overexpression on the proliferation of KGN cells through CCK-8 assay. Every group has 5 samples. D The viability of KGN cells with ANGPTL4 overexpression at different time points. Every group has 5 samples. E The m RNA expression of WNT2 in ovarian GCs of control patirnts( n  = 58) and PCOS patients( n  = 50) by PCR. F The mRNA expression of β-CATENIN in ovarian GCs of control patirnts( n  = 58) and PCOS patients( n  = 50) by PCR. G The effect of ANGPTL4 overexpression and the addition of the WNT/β-catenin pathway agonist SKL2001 on the proliferation of KGN cells. Every group has 5 samples. H The effect of ANGPTL4 overexpression and the addition of the WNT/β-catenin pathway agonist SKL2001 on the viability of KGN cells.Every group has 5 samples. I Western blot analysis of ANGPTL4, WNT2, CTNNB1 (β-catenin) and β-ACTIN (loading control) protein levels in KGN cells treated with control (Ctrl), ANGPTL4-overexpressing adenovirus (Ad-ANGPTL4), Ad-ANGPTL4 combined with WNT/β-catenin pathway agonist SKL2001 (Ad-ANGPTL4 + SKL2001), or Ctrl combined with SKL2001(Ctrl + SKL2001). The right side of each protein band indicates the corresponding molecular weight (kDa). ( J ) Quantitative analysis of the relative protein expression (normalized to β-ACTIN) of ANGPTL4, WNT2 and CTNNB1 in KGN cells from the four treatment groups (Ctrl, Ad-ANGPTL4, Ad-ANGPTL4 + SKL2001, Ctrl + SKL2001). Error bars represent the ± SD. Statistical significance was determined by t-test.Data are presented as the mean ± SD of at least three independent experiments, with ns indicating P  > 0.05, * indicating P  < 0.05, ** indicating P  < 0.01, and *** indicating P  < 0.001 External experiments and pathway validation of ANGPTL4 gene. A The transcriptional level of ANGPTL4 gene in ovarian GCs of control patirnts( n  = 58) and PCOS patients( n  = 50) by PCR. B The protein expression of ANGPTL4 in the follicular fluid of control patirnts( n  = 175) and PCOS patients( n  = 168) by ELISA. C The effect of ANGPTL4 overexpression on the proliferation of KGN cells through CCK-8 assay. Every group has 5 samples. D The viability of KGN cells with ANGPTL4 overexpression at different time points. Every group has 5 samples. E The m RNA expression of WNT2 in ovarian GCs of control patirnts( n  = 58) and PCOS patients( n  = 50) by PCR. F The mRNA expression of β-CATENIN in ovarian GCs of control patirnts( n  = 58) and PCOS patients( n  = 50) by PCR. G The effect of ANGPTL4 overexpression and the addition of the WNT/β-catenin pathway agonist SKL2001 on the proliferation of KGN cells. Every group has 5 samples. H The effect of ANGPTL4 overexpression and the addition of the WNT/β-catenin pathway agonist SKL2001 on the viability of KGN cells.Every group has 5 samples. I Western blot analysis of ANGPTL4, WNT2, CTNNB1 (β-catenin) and β-ACTIN (loading control) protein levels in KGN cells treated with control (Ctrl), ANGPTL4-overexpressing adenovirus (Ad-ANGPTL4), Ad-ANGPTL4 combined with WNT/β-catenin pathway agonist SKL2001 (Ad-ANGPTL4 + SKL2001), or Ctrl combined with SKL2001(Ctrl + SKL2001). The right side of each protein band indicates the corresponding molecular weight (kDa). ( J ) Quantitative analysis of the relative protein expression (normalized to β-ACTIN) of ANGPTL4, WNT2 and CTNNB1 in KGN cells from the four treatment groups (Ctrl, Ad-ANGPTL4, Ad-ANGPTL4 + SKL2001, Ctrl + SKL2001). Error bars represent the ± SD. Statistical significance was determined by t-test.Data are presented as the mean ± SD of at least three independent experiments, with ns indicating P  > 0.05, * indicating P  < 0.05, ** indicating P  < 0.01, and *** indicating P  < 0.001 The CCK-8 assay results also showed that the OD values of the Ad-ANGPTL4 group were significantly lower than those of the control group at the detection points of 24, 48, and 72 h(Fig.  7 C), indicating that overexpression of ANGPTL4 can inhibit the proliferation ability of KGN cells. We further tested the cell viability and found that the relative cell viability of KGN cells with overexpressed ANGPTL4 showed a continuous downward trend with the extension of culture time (Fig.  7 D). It exerts an inhibitory effect on cells over time, gradually reducing cell viability. Previous enrichment analysis suggested that ANGPTL4 negatively regulates the WNT signaling pathway, and PCR results further verified this finding. We found that in the ovarian GCs of PCOS patients, the significant up-regulation of ANGPTL4 and the significant down-regulation of the mRNA expression of core genes ( WNT2 , β-CATENIN ) in the WNT signaling pathway ( P  < 0.0001) confirmed that the WNT signaling pathway is inhibited under the condition of ANGPTL4 overexpression (Fig.  7 E-F). In order to clarify whether ANGPTL4 overexpression affects cell proliferation by inhibiting the WNT pathway, four groups were set up according to whether the WNT pathway activator was added or not, which were Control, Control + SKL2001, Ad-ANGPTL4, and Ad-ANGPTL4 + SKL2001.The results showed that cells showed a proliferative trend over time. Compared with the Control group, the OD values of the Ad-ANGPTL4 group were significantly lower at 24, 48, and 72 h, indicating that overexpression of ANGPTL4 can inhibit the proliferation of KGN cells, while the OD values of the Ad-ANGPTL4 + SKL2001 group were significantly higher than those of the Ad-ANGPTL4 group (Fig.  7 G). This indicates that the inhibitory effect of ANGPTL4 overexpression on KGN cell proliferation can be partially reversed by the WNT/β-catenin pathway agonist SKL2001. The cell viability assay results also showed that at the same time point, the relative cell viability of the Ad-ANGPTL4 + SKL2001 group was higher than that of the Ad-ANGPTL4 group (Fig.  7 H). These results indicate that ANGPTL4 inhibits the proliferation and viability of KGN cells by suppressing the WNT signaling pathway, thereby participating in the pathological processes related to PCOS. To further validate the regulatory effect of ANGPTL4 on the WNT signaling pathway, we detected the expression levels of relevant proteins via Western blotting. After infecting KGN cells with ANGPTL4-overexpressing adenovirus (Ad-ANGPTL4), the expression levels of WNT2 and CTNNB1 (β-CATENIN), core proteins of the WNT signaling pathway, decreased significantly. When the WNT pathway agonist SKL2001 was added following Ad-ANGPTL4 infection, the protein expressions of WNT2 and CTNNB1(β-CATENIN) were notably restored compared with the Ad-ANGPTL4 group (Fig.  7 I-J). Quantitative analysis results showed that ANGPTL4 protein expression in the Ad-ANGPTL4 group was significantly upregulated compared with the control group, while the protein levels of WNT2 and CTNNB1 were significantly reduced. In the Ad-ANGPTL4 + SKL2001 group, the protein expressions of WNT2 and CTNNB1 were significantly increased compared with the Ad-ANGPTL4 group. These results confirm at the protein level that ANGPTL4 can inhibit the WNT signaling pathway, and this inhibitory effect can be partially reversed by SKL2001, further clarifying the causal relationship by which ANGPTL4 affects cell function via regulating the WNT signaling pathway.

Clinical samples were collected from follicular fluid and ovarian GCs of patients who underwent IVF or intracytoplasmic sperm injection (ICSI) for fresh embryo transfer cycles at Guangdong Provincial People’s Hospital from 2023 to 2024.There were no differences in ovarian stimulation protocols between groups. Inclusion criteria for patients in the PCOS group were as follows: female patients aged 18 to 40 years who met the Rotterdam diagnostic criteria [ 35 ], including oligo- or anovulation combined with either hyperandrogenism or polycystic ovaries, and other causes of hyperandrogenism and ovulation dysfunction were excluded. Inclusion criteria for patients in the control group were as follows: female patients aged 18 to 40 years, normal menstrual cycles, no endocrine abnormalities, and normal ovarian and uterine morphology confirmed by either ultrasound or histological examination. Exclusion criteria for all participants encompassed recurrent pregnancy loss, chromosomal abnormalities, diabetes mellitus, adenomyosis, endometriosis, and history of malignancy. PCOS patients were divided into two groups based on the median ANGPTL4 level. The number of oocytes obtained, fertilized oocytes, Day 3 high-quality embryos, biochemical pregnancy abortion rate, clinical pregnancy rate, miscarriage rate, and live birth rate were counted after ovulation induction treatment to determine the relationship between the ANGPTL4 and pregnancy outcomes in PCOS patients. The study was approved by the Ethics Review Committee of Guangdong Provincial People’s Hospital (ethical number KY-Z-2022-345-03), and informed consent was obtained from all participants. The baseline data of the patients included age, body mass index (BMI), antral follicle count (AFC), anti-Müllerian hormone (AMH), follicle-stimulating hormone (FSH), luteinizing hormone (LH), estradio(E₂), progesterone(P), testosterone(T), low-density lipoprotein (LDL), TG and high-density lipoprotein (HDL) levels in serum, as well as ANGPTL4 levels in follicular fluid. Laboratory and embryo development indexes included drug dosage, the number of oocytes obtained, the number of high-quality embryos formed on Day 3, the number of cystic embryos, and the number of blastocysts formed. Pregnancy outcome indicators included biochemical pregnancy miscarriage rate(number of biochemical pregnancy miscarriages/number of transfer cycles × 100%), clinical pregnancy rate(number of clinical pregnancies/number of transfer cycles × 100%), ectopic pregnancy rate (number of ectopic pregnancy cycles / number of clinical pregnancy cycles × 100%), miscarriage rate (number of miscarriage cycles after clinical pregnancy confirmation / number of clinical pregnancy cycles × 100%), and live birth rate (number of live birth cycles / number of clinical pregnancy cycles × 100%). When at least one follicle in both ovaries reached 2 cm or at least three follicles reached 1.8 cm, human chorionic gonadotropin (hCG) 4000–10,000 IU was administered intramuscularly to promote oocyte maturation, based on the patient’s BMI, AFC, AMH, E2 level, etc. Oocytes were retrieved 36–38 h after hCG injection. In our center, ultrasound-guided transvaginal puncture of mature follicles with a fine needle was typically used. The follicular fluid was collected into a 50 mL centrifuge tube and centrifuged at 2000 g for 10 min at 4 °C. The middle and upper layers of the follicular fluid were separated into cryopreservation tubes and stored at -80 °C for long-term preservation. The remaining follicular fluid was subjected to granulosa cell separation. The level of ANGPTL 4 in the follicular fluid was detected by enzyme-linked immunosorbent assay. The follicular fluid was dispensed, excess follicular fluid was aspirated, and hyaluronidase at a concentration of 1 mg/mL was added to the precipitate. After shaking and mixing, the mixture was heated in a 37 °C water bath for 20 min and then transferred to a new centrifuge tube. 4 mL of lymphocyte isolate was added, and centrifugation was performed at 4 °C. The intermediate layer of pellet cells was collected as much as possible for sieving. Subsequently, the sample was centrifuged at 6000 g for 3 min at 4℃, the supernatant was discarded, and 2mL of erythrocyte lysate was added to the tube. The mixture was blown and mixed well, incubated on ice, and protected from light for 10–15 min. Lysis was terminated by adding 1mL of pre-cooled PBS. Afterwards, the sample was centrifuged at 6000 g for 3 min at 4℃, the supernatant was aspirated, 500µL of PBS was added, mixed well, and transferred to a new microcentrifuge tube. The sample was centrifuged at 6000 g for 3 min at 4℃, and the supernatant was thoroughly discarded. Following removal of the supernatant, the GCs obtained from isolated ovaries were treated with Trizol reagent for lysis and subsequent extraction of total RNA. The concentration and purity of the extracted total RNA were assessed using UV spectrophotometry. Subsequently, the total RNA was reverse transcribed into complementary DNA and analyzed using real-time quantitative PCR. The expression levels of ANGPTL4 , SLC2A1, WNT2 and β-CATENIN mRNA were normalized to the levels of β-ACTIN mRNA, which was utilized as an internal control. The expression levels of A NGPTL4 , SLC2A1, WNT2 and β-CATENIN were determined using the 2 −∆∆Ct method. The related primer sequences are shown in Supplemental Table S1 . Two gene expression datasets for PCOS, GSE155489 and GSE10946 , were retrieved from the GEO database ( https://www.ncbi.nlm.nih.gov/geo/ ). GSE155489 served as the training set, consisting of GCs from 4 healthy tissues and 4 PCOS patients, while GSE10946 was used as the validation set, including GCs from 11 healthy tissues and 12 PCOS patients. The edgeR package was employed to identify differentially expressed genes(DEGs) between PCOS patients and healthy controls in the GSE155489 dataset, with the criteria of a P-value  1. The STRING database ( https://cn.string-db.org/ ) was utilized to construct a protein-protein interaction (PPI) network of 4,818 DEGs, setting a minimum interaction score of 0.7, aiming to explore interactions between proteins, pathways, and co-expression relationships. Genes in the PPI network were then imported into the clusterProfiler R package for GO and KEGG analyses, with a significance threshold of P  < 0.05. Through intersection analysis of the results from LASSO regression and SVM-RF algorithms, two core hub genes, SLC2A1 and ANGPTL4, were identified. The hub genes filtered by Cytoscape were input into the NetworkAnalyst platform ( https://www.networkanalyst.ca/ ), and potential miRNAs and TFs interacting with the hub genes were predicted using online databases (miRTarBase v8.0 and ENCODE), followed by the construction of a corresponding miRNA-TF-hub gene regulatory network. The single-sample Gene Set Enrichment Analysis (ssGSEA) method was used to calculate immune cell infiltration scores for each sample, and Pearson correlation analysis was performed to assess the correlation between hub genes and immune scores, with statistical significance set at P  < 0.05. The expression differences of hub genes were validated in both the training ( GSE155489 ) and validation datasets ( GSE10946 ) by R. Box plots comparing gene expression across groups were generated via the ggpubr package, and significance was determined by t-test. Finally, based on the hub gene list, GSEA was carried out to evaluate functional enrichment characteristics in external datasets, and the results were visualized using GSEA 4.3.3 software. The human ovarian granulosa cell line involved of the experiment was the KGN cell line. The complete culture medium was configured as 90% DMEM/F12 medium, 10% fetal bovine serum and 1% penicillin-streptomycin. The cells were cultured in complete medium at 37℃ with 5% CO2 in a constant temperature incubator. The complete medium was changed every two days, and the cell lines were passaged every three days. Cell morphology and proliferation status were closely observed under a light microscope, and appropriate cell status was selected for the next experimental step. We used adenovirus to construct ANGPTL4-overexpressing cell lines. Adenovirus expression ANGPTL4 and empty adenovirus were purchased from Shanghai Genechem Co., Ltd. To determine the dysregulated signaling pathways, KGN cells were treated with WNT-β-catenin agonist (SKL2001,Selleckchem, Houston, TX, USA) for 48 h after adenoviral infection. To clarify whether ANGPTL4 overexpression affects cell proliferation by inhibiting the WNT pathway, four groups were set up: Control, Control + SKL2001, Ad-ANGPTL4, and Ad-ANGPTL4 + SKL2001. Granulosa cells from cell lines after adenovirus transfection were washed with PBS and lysed in RIPA buffer containing 1mM phenylmethylsulfonyl fluoride (PMSF, Beyotime Biotechnology, Shanghai, China). Approximately 30 µg of protein in each lane was electrophoresed on a 4%–20% SDS-polyacrylamide gel (Boyi Biotech, China), and the bands were transferred onto polyvinylidene fluoride membranes (Millipore, Boston, MA, USA) before being blocked with 5% non-fat milk. Protein levels in each sample were normalized to β-ACTIN. At least three independent experiments were performed, and semi-quantitative analysis of each band was used with enhanced chemiluminescence kit (New Cell & Molecular Biotech) by fully automated gel imaging analysis system (BIO-RAD, CA, USA) and Image J software. The CCK8 experiment followed the experimental steps of Wuhan Servicebio Technology Co., Ltd. KGN cells were uniformly seeded in 96-well plates for culture, with a total volume of 100 uL cell suspension per well, averaging 2000 cells per well. Five replicate wells were set up in each group, and a blank control group was included to remove background effects. After cell attachment, the cells were divided for adenovirus infection, and CCK8 experiments were performed at 0 h, 24 h, 48 h, and 72 h time points of treatment. Under light protection, 10uL of CCK8 reagent was added to each well of each group in the 96-well plate, and mixed with slight shaking. After incubation for 2 h in a 37℃ cell incubator, the liquid in the wells turned orange. The absorbance value of each well was measured using an enzyme marker at 450 nm wavelength, which was used to calculate cell viability. Data analysis was performed using SPSS 27.0 (IBM, Germany) and Graphpad prism 10 (IBM, USA) software. Data on continuous variables that were not normally distributed were analyzed by the Mann-Whitney nonparametric test, and the results were expressed as median (M) and interquartile spacing (P25, P75). Information on continuous variables that conformed to normal distribution was analyzed using the two-independent samples t-test and is expressed as mean ± standard deviation. Categorical variable information was analyzed using the chi-square test to analyze differences between groups and expressed as a rate (%). All tests were two-tailed, and a P -value of less than 0.05 was considered statistically significant.

Polycystic Ovary Syndrome (PCOS) is a complex reproductive, endocrine, and metabolic disorder affecting women of reproductive age, significantly impacting their physical and mental health, as well as their quality of life [ 1 , 2 ]. It is characterized majorly by a raised level of androgens such as testosterone and a large number of ovarian cysts (more than 10) that cause anovulation, infertility, and irregular menstrual cycle [ 3 ]. The pathogenesis involves intricate interactions across genetic, environmental, neuroendocrine, and metabolic dimensions, posing significant challenges for prevention and treatment [ 4 – 9 ]. Granulosa cells (GCs) have emerged as central players in PCOS pathogenesis across multiple studies, with their dysfunction involving multidimensional mechanisms such as hormone synthesis, signaling pathways, cellular senescence, and environmental interactions [ 10 – 13 ]. It has been demonstrated that ovarian GCs proliferation dysfunction contributes to the development of PCOS [ 14 ]. GCs are interdependent with oocytes and can provide them with nutrients, thereby ensuring normal growth and development of oocytes and playing a key role in ovulation [ 15 ]. Abnormalities in ovarian granulosa cell proliferation or metabolic disorders can lead to impaired oocyte development [ 16 – 19 ]. Several studies have shown that patients with PCOS typically experience endocrine and glycolipid metabolism disorders, which may enhance communication between oocytes and GCs, ultimately leading to impaired follicular growth and a reduced number of high-quality embryos [ 20 , 21 ]. Clinical research has indicated that growth hormone (GH) inhibits GC apoptosis via PI3K/AKT pathway activation, reduces reactive oxygen species (ROS) production, and improves mitochondrial membrane potential, suggesting that targeting GC signaling pathways may represent a novel therapeutic strategy for PCOS [ 22 ]. In recent years, immune dysregulation has gained increasing attention in PCOS pathogenesis. Studies have identified a state of chronic low-grade inflammation in PCOS patients, characterized by significantly elevated levels of proinflammatory cytokines such as interleukin-6 (IL-6), tumor necrosis factor-α (TNF-α), interleukin-15 (IL-15), and interleukin-22 (IL-22) [ 13 , 23 – 28 ]. This inflammatory microenvironment disrupts normal ovarian physiology, impairs follicular development, maturation, and ovulation, and exacerbates endocrine and metabolic disorders. For example, IL-15 inhibits GC proliferation, promotes apoptosis, and interferes with steroid hormone synthesis via specific signaling pathways, leading to increased androgen production, reduced estrogen synthesis, and hormonal imbalance [ 29 , 30 ]. Moreover, the quantity and activity of immune cells [ 31 , 32 ] and natural killer cells [ 14 ] in the ovarian microenvironment of PCOS patients have been significantly altered. As a multifunctional protein, angiopoietin-like protein 4(ANGPTL4) have been reported to be abnormal in ovarian tissues and circulation of PCOS patients [ 28 – 30 ]. In ovarian GCs, high ANGPTL4 expression correlates closely with insulin resistance(IR) and glycolipid metabolic abnormalities in PCOS patients [ 14 , 33 , 34 ]. ANGPTL4 may influence triglyceride(TG) metabolism by inhibiting lipoprotein lipase activity, contributing to dyslipidemia. Concurrently, it may regulate ovarian angiogenesis, cell proliferation, and apoptosis, affecting follicular development and ovulation [ 14 , 34 ]. However, the precise role of ANGPTL4 in the pathogenesis of PCOS, as well as its impact on pregnancy outcomes in PCOS patients, remains controversial. Although ANGPTL4 levels have been showed to be significantly elevated in the serum of PCOS patients compared to the healthy population, a comprehensive understanding of how ANGPTL4 regulates the immune response and reproductive outcomes in PCOS is still lacking. Additionally, the regulatory network of ANGPTL4 expression, such as the miRNA-transcription factor (TF)-gene axis, and its impact on the WNT/β-catenin signaling pathway remain unclear. Despite ANGPTL4 being considered a potential biomarker for PCOS, its correlation with clinical indicators such as oocyte quality, embryonic development, and miscarriage during in vitro fertilization(IVF) cycles still requires systematic verification. Therefore, an in-depth exploration of the mechanism by which ANGPTL4-mediated immune regulation in GCs contributes to the pathogenesis of PCOS and reproductive dysfunction, as well as the identification of novel diagnostic and therapeutic targets, is essential for improving the clinical management of PCOS patients.

By integrating clinical sample and bioinformatics analysis validation, ANGPTL4 was found to be significantly overexpressed in PCOS GCs, and its expression level was closely associated with reduced oocyte retrieval, decreased embryo quality, and increased biochemical miscarriage rates and abortion rates in PCOS patients. This study confirms that ANGPTL4 can induce GCs dysfunction in PCOS. Mechanistically, ANGPTL4 inhibits the WNT signaling pathway, hindering granulosa cell proliferation and follicular development. Meanwhile, it regulates the immune microenvironment, showing a negative correlation with infiltration of multiple immune cells, while activating inflammatory pathways to exacerbate chronic inflammation in the follicular microenvironment. Additionally, ANGPTL4 participates in metabolic homeostasis regulation, affecting lipid metabolism and insulin sensitivity, further worsening metabolic disorders in PCOS patients. This study provides new evidence for elucidating the interactive disorder mechanism of metabolism-immunity-reproduction in PCOS, suggesting that targeting the ANGPTL4/WNT pathway may become a new direction for clinical intervention in PCOS. Future studies are needed to expand sample size and deeply explore its upstream and downstream regulatory networks to promote precision diagnosis and treatment.

The pathogenesis of PCOS is closely associated with granulosa cell dysfunction, with core mechanisms involving interactive disorders of metabolic, immune, and reproductive regulations [ 30 – 33 ] .This study found that ANGPTL4 is significantly overexpressed in GCs of PCOS patients, with its expression level being closely correlated with reduced oocyte retrieval, decreased embryo quality, and increased biochemical miscarriage rates and abortion rates. These findings are similar to previous studies which showed that ANGPTL4 overexpression may exacerbate granulosa cell dysfunction leading to the development of PCOS [ 14 ]. Clinical cohort studies have also demonstrated a positive correlation between serum ANGPTL4 levels and insulin resistance(IR) indices in PCOS patients, suggesting that this factor may serve as a bridge molecule for metabolic-reproductive crosstalk abnormalities [ 36 ]. The integration of public datasets and machine learning algorithms further validated these findings. Bioinformatics analysis identified ANGPTL4 and SLC2A1 as core hub genes, with ANGPTL4 exhibiting specific overexpression in GCs across both the training dataset ( GSE155489 ) and validation dataset ( GSE10946 ). The single-gene GSEA enrichment analysis in this study showed that ANGPTL4 was significantly negatively correlated with the WNT signaling pathway (NES < -1). These findings provide new evidence for understanding the interaction mechanism of the “metabolism-immunity-reproduction” axis in PCOS. The WNT pathway is a key regulatory pathway for follicular development, and its abnormal inhibition can lead to granulosa cell proliferation stagnation and follicular maturation disorders. Under normal physiological conditions, WNT signaling stabilizes β-catenin and activates its downstream target genes, such as Cyclin D1, to drive granulosa cell proliferation and follicular development [ 37 ].In this study, the mRNA expression of core genes of the WNT pathway, WNT2 and β-CATENIN , was significantly down-regulated in GCs of PCOS patients, suggesting that ANGPTL4 may accelerate the ubiquitination degradation of β-catenin by enhancing GSK-3β phosphorylation, thereby inhibiting the transcriptional activity of WNT target genes [ 38 ]. This mechanism is highly consistent with the pathological characteristics of a large number of small follicle accumulation and follicular development arrest in the ovaries of PCOS patients, which is also verified in animal models. The inhibition of the WNT pathway leads to follicular development block in mice, while activation of the WNT pathway improves PCOS-like phenotypes [ 37 ]. In addition, ANGPTL4-overexpression experiments showed that it could significantly reduce granulosa cell viability (such as decreased proliferation ability of KGN cells) by inhibiting the WNT pathway, and the WNT agonist (SKL2001) could partially reverse this inhibitory effect, further confirming that ANGPTL4 drives granulosa cell dysfunction by regulating the WNT pathway. At the level of interactive regulation between immunity and metabolism, this study found that ANGPTL4 expression was significantly negatively correlated with the infiltration levels of CD8⁺ T cells, macrophages, aDCs, B cells, Th1 cells, pDCs, and Th2 cells ( P < 0.05), and significantly positively correlated with the activity of immune pathways such as Type_II_IFN_Reponse, Checkpoint, MHC_class_I, Parainflammation, Type_I_IFN_Reponse, Neutrophils, Treg, and APC_co_inhibition ( P < 0.001). Other studies have found that high expression of ANGPTL4 can activate the NF-κB pathway, promote the secretion of pro-inflammatory factors such as IL-6 and TNF-α, and exacerbate the chronic inflammatory state of the follicular microenvironment [ 22 ]. It is worth noting that the concentration of IL-15 in the follicular fluid of PCOS patients is significantly increased, and this factor can further amplify hyperandrogenemia by inducing granulosa cell apoptosis and upregulating the expression of CYP17A1 (a key enzyme for androgen synthesis) [ 13 ]. ANGPTL4 may enhance the sensitivity of immune cells to inflammatory signals, forming a positive feedback loop of “inflammation-androgen imbalance”. In addition, ANGPTL4 inhibits lipoprotein lipase activity in lipid metabolism, leading to the accumulation of free fatty acids in GCs and exacerbating IR [ 36 , 39 , 40 ]. In the state of IR, the sensitivity of GCs to insulin decreases, inhibiting the activity of the PI3K/Akt pathway, which in turn affects the expression of FSH receptors and follicular development [ 41 – 43 ], which is closely related to the common metabolic abnormalities and reproductive dysfunction in PCOS patients. Granulosa cell senescence and apoptosis are critical in PCOS pathogenesis [ 44 , 45 ]. PCOS GCs exhibit telomere shortening and senescence-associated secretory phenotype (SASP) [ 46 ], and ANGPTL4 may accelerate granulosa cell senescence by inducing oxidative stress and mitochondrial DNA damage, and activating the γH2AX-mediated DNA damage response [ 47 ]. Meanwhile, ANGPTL4 can activate the caspase-3-dependent apoptotic pathway by upregulating the Bax/Bcl-2 ratio of pro-apoptotic proteins [ 48 ]. In a DHEA-induced PCOS mouse model, inhibition of ANGPTL4 significantly reduced the apoptosis rate of GCs and improved ovarian morphological abnormalities, further verifying its key role in the pathological process. This study confirms that ANGPTL4 overexpression correlates with poor reproductive outcomes, acting as a driver of PCOS granulosa cell dysfunction via WNT pathway inhibition and metabolic-immune dysregulation. Targeting the ANGPTL4/WNT axis may disrupt the “metabolism-immunity-reproduction” pathological cycle, offering new therapeutic directions for PCOS [ 49 ].

Supplementary Material 1. Supplementary Material 1. Supplementary Material 2. Supplementary Material 2. Supplementary Material 3. Supplementary Material 3. Supplementary Material 4. Supplementary Material 4.