Characterization of KLK5-high epithelial cells and their interactions with the tumor microenvironment in high-grade serous ovarian cancer.

To identify specific markers of ovarian cancer, we analyzed genes upregulated in ovarian cancer relative to normal tissues across 29 cancer types using TCGA datasets accessed via GEPIA2. Among the 70 genes specifically upregulated in ovarian cancer, KLK5 emerged as the most significantly overexpressed transcript (Log 2 FC = 5.759, P  < 0.001; Fig.  1 A-B; Supplementary Table 2). Further analysis using GEPIA2 data demonstrated that high KLK5 expression in tumor tissues was significantly associated with worse disease-free survival (DFS; Fig.  1 C) and advanced-stage disease (Fig.  1 D). These findings suggest that KLK5 may serve as a specific biomarker for the early detection of ovarian cancer. Fig. 1 Elevated KLK5 expression correlates with poor survival in ovarian cancer. A . Among the up-regulated genes identified across 29 cancer types, 70 genes were specifically overexpressed in ovarian cancer, as determined from the TCGA datasets analyzed using GEPIA2 ( http://gepia2.cancer-pku.cn/#index ). B . KLK5 emerged as the most significantly up-regulated gene (Log 2 FC = 5.759; P-value < 0.001) among the 70 ovarian cancer-specific genes. C . High levels of KLK5 in tumor tissues were significantly associated with reduced disease-free survival. D . KLK5 expression in tumor tissues was notably increased in advanced stages of ovarian cancer (P-value = 0.0455) Elevated KLK5 expression correlates with poor survival in ovarian cancer. A . Among the up-regulated genes identified across 29 cancer types, 70 genes were specifically overexpressed in ovarian cancer, as determined from the TCGA datasets analyzed using GEPIA2 ( http://gepia2.cancer-pku.cn/#index ). B . KLK5 emerged as the most significantly up-regulated gene (Log 2 FC = 5.759; P-value < 0.001) among the 70 ovarian cancer-specific genes. C . High levels of KLK5 in tumor tissues were significantly associated with reduced disease-free survival. D . KLK5 expression in tumor tissues was notably increased in advanced stages of ovarian cancer (P-value = 0.0455) To confirm the direct effects of KLK5 on ovarian cancer cells, we conducted cellular assays to evaluate its impact on the proliferation, migration, and invasion of ovarian cancer cell HO-8910 and SKOV3. It was shown that KLK5 (400 ng/mL) significantly increased the migration rate of HO-8910 cells (16 h) and SKOV3 cells (24 h) in the wound healing assay (Figure. 2 A and Supplementary Fig. 1 A). In addition, for HO-8910 cells, KLK5 significantly enhanced cell invasion in a dose-dependent manner only when it was present in the upper chamber (16 h, p  < 0.05 for 400 ng/mL compared to the 0 ng/mL control; Fig.  2 B). In contrast, the presence of KLK5 in the lower chamber failed to promote invasion, showing no statistically significant effect compared to the control ( p  > 0.05; Fig.  2 C). This result demonstrates that the pro-invasive function of KLK5 is primarily attributed to its direct activation of cellular invasion pathways upon local contact. For SKOV3 cells, the KLK5 also significantly enhanced cell invasion when it was present in the upper chamber (24 h, p  < 0.05 for 400 ng/mL compared to the 0 ng/mL control; Supplementary Fig. 1B). Importantly, the observed enhancement in migration and invasion was not attributed to increased cell numbers, as KLK5 treatment showed no significant effects on the proliferation or apoptosis of either cell line within the 48-hour assay period (Fig.  2 C-E and Supplementary Fig. 1 C, D). Fig. 2 The effect of KLK5 on migration, invasion, apoptosis, and proliferation of HO8910 cells. A . The significantly enhanced of migration in the HO8910 cell line with the addition of KLK5 (400ng/mL; P  < 0.05). B .The significantly enhanced of invasion in the HO8910 cells with the addition of KLK5 (400 ng/mL in upper chamber; P  < 0.05). C . No significantly differences of invasion in the HO8910 cell line with the addition of KLK5 in lower chamber. D . The apoptosis of HO8910 cells treated with or without KLK5 (200 and 400 ng/mL). E , F . The proliferation of HO8910 cells treated with or without KLK5. The line plot in Fig. 2E and F show the mean ± SEM The effect of KLK5 on migration, invasion, apoptosis, and proliferation of HO8910 cells. A . The significantly enhanced of migration in the HO8910 cell line with the addition of KLK5 (400ng/mL; P  < 0.05). B .The significantly enhanced of invasion in the HO8910 cells with the addition of KLK5 (400 ng/mL in upper chamber; P  < 0.05). C . No significantly differences of invasion in the HO8910 cell line with the addition of KLK5 in lower chamber. D . The apoptosis of HO8910 cells treated with or without KLK5 (200 and 400 ng/mL). E , F . The proliferation of HO8910 cells treated with or without KLK5. The line plot in Fig. 2E and F show the mean ± SEM To rigorously exclude any potential long-term proliferative effect that might confound our interpretation, we performed an extended cell counting assay aligned with the reported doubling time of HO-8910 cells (~ 34.5 h, Cellosaurus: CVCL_0310). Consistent with the short-term results, KLK5 treatment did not significantly increase cell numbers at 35, 70, or 105 h (covering 1–3 doublings) compared to the control (Fig.  2 F). Furthermore, to establish the generalizability of this finding, we assessed the impact of KLK5 on proliferation in the second HGSOC cell line, SKOV3. Similarly, no significant increase in cell number was observed in SKOV3 cells over a 96-hour period following KLK5 treatment (Supplementary Fig. 1D). These findings collectively demonstrate that the pro-migratory and pro-invasive effects of KLK5 are direct and specific, operating independently of any influence on cell proliferation or survival. To investigate the functional role of KLK5 in ovarian cancer, we analyzed single-cell RNA sequencing (scRNA-seq) datasets encompassing five ovarian cancer subtypes (carcinosarcoma [OCS], clear-cell carcinoma [OCCC], endometrioid carcinoma [OEC], high-grade serous ovarian cancer [HGSOC], and low-grade serous ovarian cancer [LGSOC]) alongside normal controls. The datasets were sourced from a published study (sample details and cell counts are provided in Supplementary Table 1). Using cell type-specific markers, we classified 151,729 cells into seven major populations: B cells (B), endothelial cells (Endo), epithelial cells (Epi), fibroblasts (Fib), myeloid cells (Mye), plasma cells (Plasma), and a combined group of T cells and NK cells (T_NK) (Fig.  3 A-B). KLK5 expression was uniquely enriched in epithelial cells compared to other cell types (Fig.  3 C). Notably, KLK5 levels were markedly elevated in HGSOC tissues relative to normal controls (Fig.  3 D; Supplementary Fig. 2), while no significant differences were observed in the other subtypes. Furthermore, spatial transcriptomics analysis confirmed KLK5 expression in HGSOC tumor tissue (Fig.  3 E). These results suggest that KLK5 plays a critical role within the epithelial compartment of high-grade serous ovarian cancer and may contribute to its pathological progression. Fig. 3 KLK5 is significantly elevated in epithelial cells of high-grade serous ovarian cancer based on single-cell RNA sequencing. A . Uniform Manifold Approximation and Projection (UMAP) visualization of 151,729 cells from 60 samples in publicly available single-cell RNA sequencing datasets from GEO ( https://www.ncbi.nlm.nih.gov/geo/ ). Cells were clustered into seven major cell types, including B cells (B), endothelial cells (Endo), epithelial cells (Epi), fibroblasts (Fib), myeloid cells (Mye), plasma cells (Plasma), and T cells/NK cells (T_NK), based on cell type-specific markers. B . Dot plot illustrating the expression of known marker genes for each cell type. C . UMAP plot displaying the expression levels of KLK5 across the cell clusters. D . KLK5 expression was significantly elevated in epithelial cells of high-grade serous ovarian cancer (HGSOC) compared to normal controls (Log2FC = 0.502, adjusted P  < 0.001). E . Spatial transcriptomics shows the expression of KLK5 in HGSOC. OCS, ovarian carcinosarcoma; OCCC, ovarian clear-cell carcinoma; OEC, ovarian endometrioid carcinoma; HGSOC, high-grade serous ovarian cancer; LGSOC, low-grade serous ovarian cancer KLK5 is significantly elevated in epithelial cells of high-grade serous ovarian cancer based on single-cell RNA sequencing. A . Uniform Manifold Approximation and Projection (UMAP) visualization of 151,729 cells from 60 samples in publicly available single-cell RNA sequencing datasets from GEO ( https://www.ncbi.nlm.nih.gov/geo/ ). Cells were clustered into seven major cell types, including B cells (B), endothelial cells (Endo), epithelial cells (Epi), fibroblasts (Fib), myeloid cells (Mye), plasma cells (Plasma), and T cells/NK cells (T_NK), based on cell type-specific markers. B . Dot plot illustrating the expression of known marker genes for each cell type. C . UMAP plot displaying the expression levels of KLK5 across the cell clusters. D . KLK5 expression was significantly elevated in epithelial cells of high-grade serous ovarian cancer (HGSOC) compared to normal controls (Log2FC = 0.502, adjusted P  < 0.001). E . Spatial transcriptomics shows the expression of KLK5 in HGSOC. OCS, ovarian carcinosarcoma; OCCC, ovarian clear-cell carcinoma; OEC, ovarian endometrioid carcinoma; HGSOC, high-grade serous ovarian cancer; LGSOC, low-grade serous ovarian cancer To characterize the tumor microenvironment associated with high KLK5 expression in ovarian cancer, we performed cell-cell interaction network analysis across ovarian cancer cell types (Fig.  4 A). This analysis revealed 313 ligand-receptor pairs uniquely enriched in high-grade serous ovarian carcinoma (HGSOC) (Fig.  4 A). Notably, HGSOC epithelial cells exhibited upregulated expression of several interaction-related genes compared to normal controls, including SDC1 , AGRN , DAG1 , FN1 , LGALS9 , and DSC2 (Figs.  3 D and 4 B), with associated signaling pathways detailed in Fig.  4 C. Fig. 4 Cell-cell interaction network between epithelial cells and the other cells. A . Venn diagram illustrating the cell-cell interaction network among different cell types in each subtype of ovarian cancer. B . Specific cell-cell interactions in HGSOC that correspond to up-regulated genes in epithelial cells from HGSOC. C . Corresponding pathway of the specific cell-cell interactions shown in Fig. 4B. D . Expression levels of KLK5 (Log 2 FC = 1.720; adjusted P  < 0.001) in KLK5 -high HGSOC, KLK5 -low HGSOC, and normal controls. E. Expression levels of COL1A1 (Log 2 FC = 1.153; adjusted P  = 0.0085), COL1A2 (Log 2 FC = 0.728; adjusted P  < 0.001), and COL6A2 (Log 2 FC = 0.546; P  < 0.001) were significantly elevated in fibroblast cells from the KLK5 -high HGSOC compared to those in KLK5 -low HGSOC or normal controls. *** indicates the P value < 0.001 Cell-cell interaction network between epithelial cells and the other cells. A . Venn diagram illustrating the cell-cell interaction network among different cell types in each subtype of ovarian cancer. B . Specific cell-cell interactions in HGSOC that correspond to up-regulated genes in epithelial cells from HGSOC. C . Corresponding pathway of the specific cell-cell interactions shown in Fig. 4B. D . Expression levels of KLK5 (Log 2 FC = 1.720; adjusted P  < 0.001) in KLK5 -high HGSOC, KLK5 -low HGSOC, and normal controls. E. Expression levels of COL1A1 (Log 2 FC = 1.153; adjusted P  = 0.0085), COL1A2 (Log 2 FC = 0.728; adjusted P  < 0.001), and COL6A2 (Log 2 FC = 0.546; P  < 0.001) were significantly elevated in fibroblast cells from the KLK5 -high HGSOC compared to those in KLK5 -low HGSOC or normal controls. *** indicates the P value < 0.001 Significantly enhanced interactions mediated by collagen-SDC1 ligand-receptor pairs, including COL1A1 - SDC1 , COL1A2 - SDC1 , COL6A1 - SDC1 , COL6A2 - SDC1 , and COL6A3 - SDC1 , were observed between fibroblasts and epithelial cells in HGSOC (Fig.  4 B). To further delineate fibroblast-specific contributions within the KLK5-high epithelial microenvironment, we analyzed differentially expressed genes (DEGs) in fibroblasts by comparing epithelial KLK5 -high HGSOC samples to both epithelial KLK5 -low HGSOC and normal controls. The expression levels of KLK5 in epithelial cells across these three groups are illustrated in Fig.  4 D. Fibroblasts derived from epithelial KLK5 -high HGSOC exhibited significant upregulation of collagen-related genes, including COL1A1 (Log 2 FC = 1.153; adjusted P  = 0.0085), COL1A2 (Log 2 FC = 0.728; adjusted P  < 0.001) and COL6A2 (Log 2 FC = 0.546; adjusted P  < 0.001) (Fig.  4 E). The elevated collagen expression ( COL1A1 , COL1A2 , and COL6A2 ) in fibroblasts may interact with SDC1 in epithelial cells, potentially contributing to tumor progression within the KLK5-high microenvironment. To delineate the upstream mechanism through which KLK5-high epithelial cells stimulate collagen production in stromal fibroblasts, we performed gain-of-function experiments. Stable KLK5 overexpression was established in two distinct ovarian cancer cell lines, SKOV3 and HO8910, with transduction efficiency confirmed by qPCR (Fig.  5 A, D). Transcriptomic profiling of KLK5-overexpressing SKOV3 cells showed marked activation of innate immune signaling pathways (Fig.  5 B). Gene set enrichment analysis further revealed that the most significantly upregulated genes in these cells were overwhelmingly enriched for biological processes related to type I interferon signaling, including type I interferon production and cellular response to type I interferon (Fig.  5 C). These results indicate that KLK5 overexpression is sufficient to trigger the canonical type I interferon signaling cascade within epithelial cancer cells. We propose that this activated interferon pathway may remodel the immune microenvironment and thereby promote collagen synthesis in fibroblasts. Fig. 5 Transcriptomic changes induced by KLK5 overexpression in ovarian cancer cell lines and the positive relationship between AGRN and KLK5 in single cell RNA datasets of HGSOC. A . Quantitative PCR analysis revealed a significant difference in KLK5 gene expression between the KLK5 -overexpressing SKOV3 cell line (KLK5_OE) and the empty vector control group (Control) ( P  < 0.05). B . Transcriptome sequencing analysis identified genes with significant differential expression between KLK5 -overexpressing SKOV3 cells and the empty vector control, where differences with |logFC| > 0.585 and P  < 0.05 were considered statistically significant. C . Genes significantly upregulated in KLK5 -overexpressing SKOV3 cells were enriched in the type I interferon synthesis and activation pathways, as indicated by GO enrichment analysis. D . Quantitative PCR analysis revealed a significant difference in KLK5 gene expression between the KLK5 -overexpressing HO-8910 cell line (KLK5_OE) and the empty vector control group (Control) ( P  < 0.05). E . Transcriptome sequencing analysis identified genes with significant differential expression between KLK5 -overexpressing HO-8910 cells and the empty vector control, where differences with |logFC| > 0.585 and P  < 0.05 were considered statistically significant. F . Venn diagram showing differentially expressed genes (DEGs) between KLK5 -high HGSOC and KLK5 -low HGSOC/normal controls in single cell RNA datasets. G . GO pathway enrichment analysis of common DEGs in epithelial cells from KLK5 -high HGSOC compared to KLK5 -low HGSOC and normal controls, respectively. H . DEGs of epithelial cells between the KLK5 -expression group (Counts ≥ 2) and the null KLK5 -expression group (Counts = 0 in HGSOC. I . KEGG (left) and GO (right) pathway enrichment analyses of DEGs of epithelial cells between the KLK5 -expression group (Counts ≥ 2) and the null KLK5 -expression group (Counts = 0 in HGSOC. J . Positive correlation between KLK5 and AGRN , as demonstrated using the GEPIA2 database ( http://gepia2.cancer-pku.cn/ ) Transcriptomic changes induced by KLK5 overexpression in ovarian cancer cell lines and the positive relationship between AGRN and KLK5 in single cell RNA datasets of HGSOC. A . Quantitative PCR analysis revealed a significant difference in KLK5 gene expression between the KLK5 -overexpressing SKOV3 cell line (KLK5_OE) and the empty vector control group (Control) ( P  < 0.05). B . Transcriptome sequencing analysis identified genes with significant differential expression between KLK5 -overexpressing SKOV3 cells and the empty vector control, where differences with |logFC| > 0.585 and P  < 0.05 were considered statistically significant. C . Genes significantly upregulated in KLK5 -overexpressing SKOV3 cells were enriched in the type I interferon synthesis and activation pathways, as indicated by GO enrichment analysis. D . Quantitative PCR analysis revealed a significant difference in KLK5 gene expression between the KLK5 -overexpressing HO-8910 cell line (KLK5_OE) and the empty vector control group (Control) ( P  < 0.05). E . Transcriptome sequencing analysis identified genes with significant differential expression between KLK5 -overexpressing HO-8910 cells and the empty vector control, where differences with |logFC| > 0.585 and P  < 0.05 were considered statistically significant. F . Venn diagram showing differentially expressed genes (DEGs) between KLK5 -high HGSOC and KLK5 -low HGSOC/normal controls in single cell RNA datasets. G . GO pathway enrichment analysis of common DEGs in epithelial cells from KLK5 -high HGSOC compared to KLK5 -low HGSOC and normal controls, respectively. H . DEGs of epithelial cells between the KLK5 -expression group (Counts ≥ 2) and the null KLK5 -expression group (Counts = 0 in HGSOC. I . KEGG (left) and GO (right) pathway enrichment analyses of DEGs of epithelial cells between the KLK5 -expression group (Counts ≥ 2) and the null KLK5 -expression group (Counts = 0 in HGSOC. J . Positive correlation between KLK5 and AGRN , as demonstrated using the GEPIA2 database ( http://gepia2.cancer-pku.cn/ ) Notably, this effect exhibited cell-type specificity. Although HO8910 cells achieved comparable levels of KLK5 overexpression (Fig.  5 D), they did not show significant enrichment of interferon-related gene sets (Fig.  5 E), in contrast to the robust interferon pathway activation observed in SKOV3 cells. However, the upregulation of the LAYN gene may explain how KLK5 directly enhances the invasive capacity of ovarian cancer (Fig.  5 E). To characterize features of KLK5 -high epithelial cells, we compared differentially expressed genes (DEGs) in epithelial cells between epithelial KLK5 -high HGSOC and KLK5 -low HGSOC/normal controls (KLK5 expression levels across these groups are shown in Fig.  4 D). A total of 194 upregulated DEGs were identified, including AGRN (Fig.  5 F). Functional enrichment analysis revealed that these DEGs were significantly associated with pathways related to extracellular matrix (ECM) disassembly (Fig.  5 G). We further stratified all HGSOC epithelial cells based on KLK5 expression: KLK5 -positive (Counts ≥ 2) and KLK5 -null (Counts = 0). AGRN expression was significantly higher in KLK5 - positive cells compared to KLK5 -null cells (Fig.  5 H). Beyond AGRN , other upregulated DEGs in KLK5 - positive cells were also increased in KEGG and GO pathways related to ECM-receptor interaction and cell-matrix adhesion (Fig.  5 I), suggesting potential roles in cancer progression. Furthermore, analysis of TCGA datasets via the GEPIA2 platform confirmed a positive correlation between AGRN and KLK5 expression (Fig.  5 J). Collectively, these findings suggest that AGRN and KLK5 may act cooperatively within epithelial cells, potentially contributing to HGSOC progression through ECM-related pathways. To investigate the impact of KLK5 on immune cell infiltration in ovarian cancer, we retrieved bulk RNA sequencing data from tumor and normal control samples using the TCGA and GTEx databases via the R package TCGAbiolinks (v2.34.0). Samples were stratified based on KLK5 expression: the top 100 expressors were designatedTCGA-KLK5 High , while the bottom 100 expressors were designated TCGA-KLK5 Low . Analysis revealed significantly enhanced infiltration of myeloid-derived suppressor cells (MDSCs) in the TCGA-KLK5 High group compared to TCGA-KLK5 Low , as quantified by ssGSEA scores (Fig.  6 A). This increased MDSC infiltration may facilitate tumor immune evasion by suppressing anti-tumor immune responses. Fig. 6 Enhanced tumor-associated macrophages in KLK5-high HGSOC. A. Immune cell infiltration profiles of TCGA-KLK5 High , TCGA-KLK5 Low and normal controls, analyzed using the ssGSEA algorithm. B . Subtypes and their proportions of monocytes and macrophages. C . Unsupervised trajectory analysis illustrating state transitions between monocytes and macrophages. The branched trajectory is color-coded by cell states and cell subtypes. D . GO terms enriched from up-regulated genes in macrophage subtypes. E . Differentially expressed genes (DEGs) in THP-1-derived macrophages after co-culture with KLK5-overexpressing (versus empty vector control) HO-8910 cells. (|log₂FC| > 0.585 and P   0.585 and P  < 0.05) Enhanced tumor-associated macrophages in KLK5-high HGSOC. A. Immune cell infiltration profiles of TCGA-KLK5 High , TCGA-KLK5 Low and normal controls, analyzed using the ssGSEA algorithm. B . Subtypes and their proportions of monocytes and macrophages. C . Unsupervised trajectory analysis illustrating state transitions between monocytes and macrophages. The branched trajectory is color-coded by cell states and cell subtypes. D . GO terms enriched from up-regulated genes in macrophage subtypes. E . Differentially expressed genes (DEGs) in THP-1-derived macrophages after co-culture with KLK5-overexpressing (versus empty vector control) HO-8910 cells. (|log₂FC| > 0.585 and P   0.585 and P  < 0.05) To further characterize myeloid cell heterogeneity, we analyzed scRNA-seq datasets and identified ten distinct myeloid subtypes, including dendritic cells (DC-CD1C), monocytes (Mono-S100A9), and macrophages (Macro-C1QC, Macro-C3, Macro-CD163, Macro-CXCL10, Macro-EGR1, Macro-IGF2, Macro-MARCO, Macro-STMN1; Fig.  6 B). Notably, KLK5 -high HGSOC samples exhibited elevated proportions of specific tumor-associated macrophages (TAMs) subsets, including Macro-STMN1, Macro-IGF2, Macro-C3, and Macro-C1QC (Fig.  6 B), which are associated with poor prognosis in ovarian cancer. To elucidate the transcriptional dynamics during myeloid differentiation, we performed trajectory analysis, which revealed a pseudotemporal progression from monocytes (State-1) towards diverse macrophage subtypes (Fig.  6 C). States 2, 3, and 4 were predominantly composed of Macro-C3, a subset linked to inflammatory responses. In contrast, State-5, which was significantly enriched in KLK5-high HGSOC samples, contained high proportions of TAMs (Macro-C1QC and Macro-IGF2). Genes upregulated in these State-5 TAMs were enriched in Gene Ontology (GO) pathways related to cell junction disassembly, cell adhesion, and other processes implicated in tumor progression (Fig.  6 D). To directly validate the ability of KLK5 to reprogram immune cells within the tumor microenvironment, we performed co-cultured experiments using KLK5-overexpressing HO-8910 or SKOV3 cells with THP-1-derived macrophages. RNA-seq analysis of macrophages co-cultured with KLK5-overexpressing HO-8910 cells revealed extensive transcriptional reprogramming, including significant upregulation of genes linked to TAM characteristics, such as NODAL and SERPIND1 (Fig.  6 E). Elevated expression of WNT10B was also observed, which may further support cancer progression (Fig.  6 E). In contrast, macrophages co-cultured with KLK5-overexpressing SKOV3 cells showed pronounced upregulation of the innate immune regulator FCN3 (encoding Ficolin-3; Fig.  6 F), a change that could promote the establishment of an immunosuppressive tumor microenvironment. Collectively, these findings indicate that KLK5 contributes to the formation of an immunosuppressive microenvironment, likely through the recruitment and/or differentiation of specific TAM subsets. These TAM populations may act as key mediators of immune evasion and disease progression in ovarian cancer. Given the elevated regulatory T cell (Treg) ratio in the TCGA-KLK5 High group identified by CIBERSORT analysis (Supplementary Fig. 3), we further characterized the T and NK (T/NK) cell populations in KLK5 -high HGSOC, KLK5 -low HGSOC, and normal controls using scRNA-seq data presented in Fig.  3 A. The T and NK cells were classified into 11 distinct subtypes (Fig.  7 A and Supplementary Table 3). Fig. 7 Immune cell-cell interactions in KLK5 -high HGSOC. A . Uniform Manifold Approximation and Projection (UMAP) plot of T and NK cells, color-coded by cell subtypes. B . Relative proportions of T and NK cell subsets in each group. C . GO terms enriched from up-regulated genes in T_CD8_Cytotoxic cells. D . GO terms enriched from up-regulated genes in NK_XCL1 cells. E . Unique Cell-cell interactions between myeloid cells (Mye) and T-NK cells in KLK5 -high HGSOC Immune cell-cell interactions in KLK5 -high HGSOC. A . Uniform Manifold Approximation and Projection (UMAP) plot of T and NK cells, color-coded by cell subtypes. B . Relative proportions of T and NK cell subsets in each group. C . GO terms enriched from up-regulated genes in T_CD8_Cytotoxic cells. D . GO terms enriched from up-regulated genes in NK_XCL1 cells. E . Unique Cell-cell interactions between myeloid cells (Mye) and T-NK cells in KLK5 -high HGSOC Strikingly, the cytotoxic CD8 + T cell (T_CD8_Cytotoxic) subtype was specifically enriched in KLK5 -high HGSOC. Genes upregulated in this subtype were significantly associated with pathways including ATP biosynthetic process and adaptive immune response (Fig.  7 B-C). Furthermore, the T_CD8_Cytotoxic subtype displayed a pronounced exhaustion phenotype, notably expressing a suite of checkpoint inhibitor genes, including LAG3 , TIGIT , PDCD1 , CTLA4 and HAVCR2 (Supplementary Fig. 4). Similarly, the NK_XCL1 subtype was predominantly enriched in KLK5 -high HGSOC, and its up-regulated genes were enriched in pathways related to adaptive immune response and antigen processing and presentation via MHC class II (Fig.  7 B and D). To explore the communications between immune cell populations, we compared cell-cell interactions between T/NK and myeloid (Mye) cell subtypes in epithelial KLK5 -high versus KLK5 -low HGSOC samples (Supplementary Fig. 5). The analysis revealed unique signaling pathways specific to KLK5 -high HGSOC within T/NK-Mye interactions, including IGF, BAG, BTLA, and NOTCH (Fig.  7 E). Of particular interest was the IGF signaling pathway, which was prominent between T_CD8_Cytotoxic cells (specifically enriched in KLK5-high HGSOC) and macrophages within the KLK5-high microenvironment (Fig.  7 E). This specific IGF-mediated interaction may critically modulate macrophage function, potentially contributing to carcinogenesis and tumor progression. Collectively, these findings reveal complex immune dynamics in KLK5 -high HGSOC, highlighting the potential roles of cytotoxic CD8 + T cells, and their interactions (notably via IGF signaling), in shaping the pro-tumorigenic microenvironment. Further studies are warranted to elucidate the functional consequences of these interactions and their therapeutic relevance in HGSOC.

Transcriptome data from 29 types of cancers and their paired normal tissues were analyzed using the GEPIA2 website ( http://gepia2.cancer-pku.cn/#index ). Differentially expressed genes (DEGs) were identified based on the criteria of adjusted P-value  1 between cancer and para-carcinoma tissues. Additionally, disease-free survival analysis was conducted on GEPIA2, utilizing gene expression data. The ovarian cancer cell lines HO-8910 and SKOV3 were cultured with RPMI 1640 medium (R5886, Sigma-Aldrich, UK) supplemented with 10% fetal calf serum (F0193, Sigma-Aldrich, UK) and incubated at 37 °C and 5% CO 2 . HO-8910 or SKOV3 ovarian cancer cells were seeded in 24-well plates at 2 × 10 5 cells per well. After 12 h, a mechanical wound was introduced into the cell monolayer. Following wound induction, cells were immediately treated with KLK5 (0, 200, or 400 ng/mL; HY-P70939A, MedChemExpress, USA) and the initial (0 h) images were acquired. Phase-contrast images were then captured every 4 h for a total of 24 h. For HO8910 cells, quantitative analysis was performed using the 16-hour time point, as it optimally captured differential migration rates prior to wound closure, which occurred at approximately 20 h. For SKOV3 cells, the wound remained unclosed at the 24-hour time point; therefore, the 24-hour data were used for quantitative analysis. The wound area was quantified using ImageJ software (National Institutes of Health, USA; https://imagej.nih.gov/ij/ ). The wound closure rate (%) was calculated using the following formula: [(Wound area at 0 h - Wound area at 16–24 h)/Wound area at 0 h] × 100%. All experiments were performed in at least triplicate. Cell invasion was assessed using transwell chambers (3422, Corning, USA). Matrigel (C0372, Beyotime, China) was diluted 1:8 in serum-free medium and applied to the upper surface of the transwell membrane. Following the manufacturer’s instructions, the coated chambers were incubated at 37 °C for 2 h to allow the Matrigel to solidify. After solidification, the transwell chambers were placed into a 24-well plate. Then, 200 µL of serum-free medium containing 5 × 10⁴ cells (HO-8910 or SKOV3) was added to the upper chamber on top of the Matrigel layer. To distinguish between the direct activating effect and the chemoattractant function of KLK5, parallel experiments were set up for HO-8910 cells, where KLK5 (0, 200, or 400 ng/mL) was added either to the upper chamber (to assess direct activation) or to the lower chamber (to assess chemotaxis). Cells were incubated for 16 h at 37 °C in 5% CO, which was optimized according to preliminary wound-healing assay observations. Subsequently, non-invading cells and Matrigel on the upper surface of the membrane were removed. Invading cells on the lower surface were fixed with 4% paraformaldehyde (P0099, Beyotime), stained with crystal violet (C0121, Beyotime), and visualized. For SKOV3 cells, we employed only the direct stimulation setup to further validate the effect of KLK5 on invasion observed in HO-8910 cells: KLK5 (0, 200, or 400 ng/mL) was added exclusively to the upper chamber. Cells were incubated for 24 h at 37 °C in 5% CO₂, followed by fixation, staining, and examination as described above. For the cellular proliferation assays, ovarian cancer cells were plated in 96-well plates (2 × 10³ cells/well) and treated with KLK5 (200 ng/mL or 400 ng/mL) or left untreated. After the indicated treatment intervals, CCK-8 reagent (C0042, Beyotime) was added to each well and the cells were incubated for 2 h at 37 °C according to the manufacturer’s instructions. Absorbance was then measured at 450 nm using a multifunctional microplate reader (Cytation 5, BioTek, USA). To exclude the possibility that the effects of KLK5 on migration and invasion were due to changes in cell number, we performed cell proliferation assays within 48 h. Furthermore, to investigate the long-term impact of KLK5 on cell proliferation, we conducted assays over 105 h (at 0, 35, 70, and 105 h) for HO‑8910 cells and over 96 h (at 0, 24, 48, 72, and 96 h) for SKOV3 cells, respectively, based on their reported doubling times. The doubling times of the cell lines used (HO‑8910 and SKOV3) were verified with reference to the Cellosaurus database ( https://www.cellosaurus.org/ ). All experiments were performed in at least triplicate. Cells were treated with KLK5 (0, 200, and 400 ng/mL) for 48 h, a duration consistent with our 48-hour proliferation assay, to ensure that any potential effects on apoptosis could be distinguished from alterations in cell number. Following the manufacturer’s instructions (Annexin V-FITC Apoptosis Detection Kit, C1062L, Beyotime), the apoptotic rate (percentage of early and late apoptotic cells) in each group was examined using a CytoFLEX flow cytometer (Beckman Coulter, USA). HO-8910 or SKOV3 cells were seeded in 6-well plates at a density of 300,000 cells per well. After 24 h, plasmid transfection was performed using Lipofectamine™ 3000 (Thermo Fisher Scientific, Waltham, MA, USA; Catalog # L3000015) following the standardized protocol. The transfection mixture consisted of 2.5 µg of either the KLK5 overexpression plasmid (pCMV-KLK5(human)−3×FLAG-Neo, Miaoling, China, Cat# P45004 ) or the empty vector control (pCMV-T7-MCS-3×FLAG-Neo, Miaoling, China, Cat# P1303), supplemented with 5 µL of P3000™ Enhancer Reagent and 7.5 µL of Lipofectamine™ 3000 Reagent. After 8 h of transfection, the medium was replaced with complete growth medium. Cells were harvested 72 h post-transfection for total RNA extraction, followed by quantitative PCR (qPCR) and transcriptome sequencing analysis. Total RNA was extracted using RNAiso Plus reagent (9108; Takara, Japan) following the manufacturer’s protocol. For quantitative real-time PCR (qPCR), total RNA was reverse-transcribed into cDNA with the PrimeScript RT reagent Kit (RR037A; Takara, Japan). Quantitative real-time PCR was performed using THUNDERBIRD SYBR qPCR Mix (QPS-201; TOYOBO, Japan). The human qPCR primer pairs GAPDH (Cat No. HP205798 ) and KLK5 (Cat No. HP228990 ) from Origene were used. The relative mRNA expression levels were calculated using the comparative 2^(-ΔΔCt) method and normalized to the endogenous control GAPDH. Data are presented as fold change relative to the control group. Co-culture experiments were performed using 6-well plate transwell inserts (LABSELECT, Cat# 14112). THP-1 monocytes (3 × 10⁵ cells) were seeded in the upper chamber and differentiated into macrophages by treatment with 100 ng/mL phorbol 12-myristate 13-acetate (PMA; Sigma-Aldrich, Cat# P8139) for 48 h. Meanwhile, HO-8910 or SKOV3 cells were seeded in the lower chamber of the 6-well plate and transfected with either KLK5 overexpression plasmid or empty vector control (as described in Sect. 2.8) for 24 h. Following the 48-hour differentiation period, the THP-1-derived macrophages in the upper chamber were co-cultured with the transfected HO-8910 or SKOV3 cells in the lower chamber. After 48 h of co-culture, macrophages from the upper chamber were carefully harvested for total RNA extraction and subsequent transcriptome sequencing analysis. RNA quality and integrity were verified before library preparation. For each sample, 2 µg of total RNA was used for cDNA library construction. Sequencing was performed on a DNBSEQ platform (MGI, China), generating 150-bp paired-end reads. Raw sequencing data in FASTQ format underwent quality control assessment using FastQC (v0.12.1). Adapter sequences and low-quality bases were trimmed with trim_galore (v0.6.10). The resulting high-quality reads were aligned to the human reference genome (GRCh38) using STAR (v2.7.10b). Gene-level quantification was performed with featureCounts (v2.0.1). Differential expression analysis was conducted using DESeq2 (v1.46.0), with genes showing |log₂FoldChange| > 0.585 and adjusted p-value < 0.05 considered differentially expressed. Functional enrichment analysis of Gene Ontology (GO) terms and pathways was performed using the clusterProfiler package (v4.14.6) to identify biological processes and pathways significantly enriched among the differentially expressed genes. To further investigate ovarian cancer at the single-cell level, we analyzed single-cell RNA sequencing datasets comprising five subtypes of ovarian cancer (carcinosarcoma, OCS; clear-cell carcinoma, OCCC; endometrioid carcinoma, OEC; high-grade serous ovarian cancer, HGSOC; and low-grade serous ovarian cancer, LGSOC) along with normal controls. These single-cell RNA sequencing datasets were obtained from the Gene Expression Omnibus (GEO), The accession number and sample information is summarized in Supplementary Table 1 (Brand et al. 2024 ; Denisenko et al. 2024 ; Kan et al. 2022 ; Kwon et al. 2023 ; Regner et al. 2021 ; Stur et al. 2022 ; Wang et al. 2022 ; Xie et al. 2023 ; Xu et al. 2022 ). Briefly, using the R package Seurat (v4.4.0), the gene expression matrix containing 151,729 cells was normalized using the NormalizeData function. The top 2,000 highly variable genes were identified and utilized for dimensionality reduction through principal component analysis (PCA). Subsequently, t-SNE and UMAP algorithms were applied to project the cell populations into two-dimensional space for visualization. To identify signature genes specific to different cell clusters, the FindAllMarkers function was employed with a statistical threshold of Log 2 FC ≥ 0.58 and adjusted P-value < 0.001. The cell types were annotated based on known markers. For example, CD79A and MS4A1 for B cell (B); CD34 , PECAM1 and VWF for endothelial cell (Endo); KRT18 , KRT8 and EPCAM for epithelial cell (Epi); LUM , DCN and ACTA2 for Fibroblasts cell (Fib); LYZ , C1QB and C1QA for myeloid cell (Mye); IGHG1 , JCHAIN and MZB1 for plasma cell (Plasma); CD3E , CD3D and NKG7 for T cell and NK cell (T_NK). The R package ggplot2 (v3.4.4) were used to visualize the gene expression of these markers. For the analyses of the single cell RNA, in the 42 single-cell transcriptome datasets of HGSOC, we defined the 10 samples with the highest KLK5 expression in epithelial cells as KLK5 -high HGSOC, and the 10 samples with the lowest KLK5 expression in epithelial cells as KLK5 -low HGSOC, as shown in Supplementary Table 1. Cell-cell communication analysis was performed using the R package CellChat (v 2.1.0) through the following workflow: standardized single-cell RNA sequencing data were preprocessed to retain genes expressed in at least 10% of cells, after which significant cell-cell communication networks were identified using CellChat’s built-in database containing 2,021 validated ligand-receptor interaction pairs, with analysis parameters set to filter out ligand-receptor pairs showing gene expression in less than 5% of the total cell population. P-values were calculated using the default permutation test (1,000 permutations), and signal strength was computed based on the geometric mean of ligand and receptor gene expression levels. Pseudotime analysis was conducted using the R package Monocle2 (version 2.18.0) to investigate developmental trajectories and transitional relationships among cells. To identify state- and sub-cell-type-specific genes, the FindAllMarkers function from the Seurat R package (version 4.4.0) was applied, with a threshold of average Log2FC ≥ 0.25 and adjusted P-value < 0.01, as determined by the Wilcoxon Rank-Sum test. Subsequently, the identified specific genes were functionally enriched using the R package clusterProfiler (version 3.18.1). Bulk RNA sequencing data for ovarian cancer and normal controls were obtained from the TCGA and GTEx databases using the R package TCGAbiolinks (version 2.34.0). Immune cell composition was analyzed using the R package CIBERSORT (version 0.1.0) with the LM22 signature gene file, which was downloaded from the CIBERSORT website ( https://cibersort.stanford.edu/runcibersort.php ). The GSVA R package (version 2.1.3) was utilized to calculate the enrichment scores of tumor-infiltrating immune cells through the single-sample gene set enrichment analysis (ssGSEA) method. This was performed using the gsva function with the parameter method = ‘ssgsea’. The signature gene sets were obtained from TISIDB ( http://cis.hku.hk/TISIDB/ ). To conduct bulk-RNA sequencing analysis of ovarian cancer using data from the TCGA website, we divided the expression matrix into two groups. The top 100 samples with the highest KLK5 expression were categorized as TCGA- KLK5 High , whereas the 100 samples with the lowest KLK5 expression were labeled as TCGA- KLK5 Low . Meanwhile, the datasets obtained from the GTEx website were used as normal controls. Spatial transcriptomics sequencing data for ovarian cancer were obtained from the GEO databases ( https://www.ncbi.nlm.nih.gov/geo/ ) including GSM6506113 , GSM6506114 , GSM6506115 and GSM6506116 (Denisenko et al. 2024 ). The function SpatialFeaturePlot was used to visual the expression of KLK5 . All data were analyzed using the R software (v4.2.1). The unpaired Student’s t-test was used compare the differences between two groups. All data are shown as the mean ± standard deviation (SD). P value < 0.05 was onsidered as statistically significant differences.

Ovarian cancer remains one of the most lethal gynecological malignancies, characterized by high mortality rate largely attributed to asymptomatic early stages, delayed diagnosis, and frequent presentation at advanced stages (Jamali et al. 2025 ; Siegel et al. 2024 ). These challenges underscore the urgent need for reliable biomarkers to enable early detection, prognostication, and prediction of therapeutic response. KLK5 has emerged as a promising candidate biomarker in this context, with studies highlighting its significant role in ovarian cancer pathogenesis (Dorn et al. 2011 ; Kim et al. 2001 ). For instance, Dorn et al., (Dorn et al. 2011 ) demonstrated markedly elevated KLK5 levels in serum and ascitic fluid of ovarian cancer patients compared to those with benign ovarian tumors, and higher KLK5 expression correlates with poorer clinical outcomes. These findings strongly support KLK5 as a potential biomarker for early diagnosis. However, the molecular mechanisms underlying its role in ovarian cancer remain incompletely understood. In this study, we validated KLK5 as a specific biomarker for early ovarian cancer detection based on bulk RNA sequencing data, aligning with earlier reports (Dorn et al. 2011 ; Kim et al. 2001 ; Loessner et al. 2012 ). Beyond its diagnostic relevance, our in vitro functional assays suggest that KLK5 may actively promote ovarian cancer cell metastasis. Transwell invasion assay revealed that KLK5 enhanced cell invasion only when added to the upper chamber; no pro‑invasive effect was observed when it was restricted to the lower compartment. This distinct chamber‑specific outcome indicates that KLK5 promotes invasion primarily through direct, local activation of cancer cells rather than by acting as a long-range chemoattractant. We hypothesize that such localized activity likely involves proteolytic cleavage and activation of cell-surface receptors or nearby pericellular substrates. To further investigate functional impact of KLK5 on ovarian cancer initiation and progression, we analyzed published single-cell transcriptomic datasets (Brand et al. 2024 ; Denisenko et al. 2024 ; Kan et al. 2022 ; Kwon et al. 2023 ; Regner et al. 2021 ; Stur et al. 2022 ; Wang et al. 2022 ; Xie et al. 2023 ; Xu et al. 2022 ). Our results revealed that KLK5 expression is predominantly elevated in epithelial cells of HGSOC, the most aggressive and prevalent ovarian cancer subtype. To characterize the cell-cell interactions associated with KLK5 -high phenotypes, we mapped ligand-receptor networks across cell types. Notably, we identified specific interactions between fibroblasts and epithelial cells in HGSOC involving the collagen- SDC1 axis, including COL1A1 - SDC1 , COL1A2 - SDC1 , COL4A2 - SDC1 , COL6A1 - SDC1 , COL6A2 - SDC1 and COL6A3 - SDC1 pairs. SDC1 , a heparan sulfate proteoglycan upregulated in HGSOC, is known to drive tumorigenesis by regulating cell proliferation (Beauvais et al. 2004 ), migration (Endo et al. 2003 ), apoptosis (Yang et al. 2022 ), angiogenesis (Purushothaman et al. 2010 ), and metastasis (Ishikawa and Kramer 2010 ; Matsumoto et al. 1997 ) in multiple cancers. Importantly, we observed significant increases in COL1A1 , COL1A2 , and COL6A2 expression in fibroblasts from KLK5 -high HGSOC compared to KLK5 -low HGSOC and normal controls. These findings align with recent work byHuang et al. (2023) (Huang et al. 2023 ), who demonstrated that COL1A1 / COL1A2 / COL6A2 -expressing cancer-associated fibroblasts (CAFs) interact with SDC1 + epithelial cells to suppress immune infiltration, thereby contributing to poor prognosis (Huang et al. 2023 ). Our data suggest that similar collagen- SDC1 crosstalk between CAFs and epithelial cells may drive tumor progression in KLK5 -high HGSOC patients. We next interrogated the upstream mechanism through which KLK5-high epithelial cells enhance collagen production in fibroblasts. Gain-of-function models revealed that KLK5 overexpression initiates a strong type I interferon (IFN-I) response in epithelial cancer cells. We propose that this IFN-I signature remodels the local immune milieu to favor fibroblast activation, potentially through the upregulation of pro-fibrotic mediators. Harcourt and Offermann (Harcourt and Offermann 2000 ),, Ito (Ito et al. 1996 ),, Zimmermann (Zimmermann et al. 2016 ),. For example, IL-6 emerges as a critical candidate, known to directly drive fibroblast activation and collagen synthesis (Dufour et al. 2018 ; Johnson et al. 2020 ; O'Reilly et al. 2014 ), thereby linking the epithelial KLK5/IFN-I axis to stromal matrix deposition. Furthermore, we identified a positive correlation between AGRN and KLK5 expression in the epithelial cells of KLK5 -high HGSOC patients. AGRN , a basement membrane protein, has been implicated in tumor migration and progression through mechanisms involving extracellular matrix remodeling and signaling pathway activation (Adamiok-Ostrowska et al. 2023 ; Chakraborty et al. 2015 ; Wang et al. 2021 ). This association highlights a potential synergistic role for KLK5 and AGRN in promoting aggressive tumor behavior, while warranting further mechanistic exploration. Previous studies implicate KLK5 , a member of the KLK family, in ovarian cancer progression through extracellular matrix remodeling and immune modulation (Srinivasan et al. 2022 ). Our observation that KLK5 -high HGSOC exhibits enriched infiltration of immunosuppressive myeloid cells, including MDSCs and specific TAM subsets (e.g., Macro-C1QC and Macro-IGF2), corroborates earlier reports linking KLKs to myeloid cell recruitment and polarization (Song et al. 2025 ; Srinivasan et al. 2022 ). For instance, KLK4-7 overexpression in ovarian cancer promotes TGF-β signaling and TAM-mediated immune suppression (Srinivasan et al. 2022 ; Wang et al. 2018 ), while elevated KLK5 level is associated with poor prognosis and ascites formation (Dorn et al. 2011 ), further supporting its role in shaping a pro-tumorigenic microenvironment. Pseudotime trajectory analysis revealing monocyte-to-TAM differentiation in KLK5 -high tumors resonates with studies highlighting the plasticity of myeloid cells in cancer (Guimaraes et al. 2024 ). Notably, the TAM subsets Macro-C1QC and Macro-IGF2, which we found enriched in KLK5 -high tumors, have been linked to ECM remodeling, angiogenesis, and immune tolerance in prior pan-cancer analyses (Coulton et al. 2024 ; Lv et al. 2021 ; Yu et al. 2024 ). Our pathway analysis further connects these TAMs to cell junction disassembly and adhesion processes, consistent with their role in facilitating metastasis. To directly establish the causal role of KLK5 in macrophage reprogramming, we performed co-culture experiments. Transcriptomic analysis of macrophages co-cultured with KLK5-overexpressing HO-8910 cells revealed extensive transcriptional reprogramming toward a TAM-like phenotype. This was marked by the significant upregulation of genes defining TAM characteristics and functions, including NODAL (Wang et al. 2014 ) and SERPIND1 (Fu et al. 2025 ). Elevated expression of WNT10B was also observed, suggesting a potential additional role in promoting cancer progression (Bui et al. 1997 ; Wu et al. 2017 ). Similarly, KLK5-overexpressing SKOV3 cells induced profound transcriptomic changes in co-cultured macrophages, most notably a significant upregulation of FCN3 , a pivotal innate immune regulator implicated in fostering an immunosuppressive tumor microenvironment (Wang et al. 2024 ). The observed enrichment of cytotoxic CD8 + T cells (T_CD8_Cytotoxic) and NK_XCL1 cells in KLK5 -high tumors initially appears paradoxical, given their typical association with an anti-tumor phenotype (Bottcher et al. 2018 ; Farhood et al. 2019 ). However, further in-depth analysis of the T_CD8_Cytotoxic cluster revealed a definitive exhaustion signature, characterized by elevated expression of multiple inhibitory immune checkpoint molecules, including LAG3 , TIGIT , PDCD1 , CTLA4 , and HAVCR2 ( TIM-3 ) - a hallmark of dysfunctional T cells in cancer (Thommen and Schumacher 2018 ). This finding elegantly resolves the apparent contradiction: although cytotoxic lymphocytes are recruited or activated within the KLK5-high TME, they concurrently acquire an exhausted, functionally impaired state that likely subverts their anti-tumor efficacy. The co-existence of these exhausted lymphocytes with immunosuppressive TAMs suggests a dynamic interplay, where KLK5 -driven mechanisms - such as IGF signaling between TAMs and T/NK cells - may actively suppress cytotoxic effector functions (Guo et al. 2025 ; Zuyin et al. 2024 ). Collectively, these results underscore the multifaceted role of KLK5 in orchestrating a complex, immune-evasive microenvironment.

Our study showed the role of KLK5 as a diagnostic biomarker while uncovering novel interactions within the tumor microenvironment that likely contribute to its pro-tumorigenic effects. The collagen- SDC1 axis and AGRN - KLK5 synergy represent promising therapeutic targets for mitigating disease progression in KLK5 -high ovarian cancer. Furthermore, KLK5 modulates myeloid cell differentiation and intercellular crosstalk to foster a permissive tumor microenvironment. Targeting KLK5 or its downstream effectors could disrupt these networks and potentially enhance immunotherapy efficacy. Future studies should validate these mechanisms in preclinical models and explore combinatorial strategies to counteract KLK5 -mediated immune evasion.

Although ovarian cancer mortality has declined at an accelerating rate - from approximately 1% per year in the 1990 s to 2.4% per year between 2004 and 2021, attributed to factors like increased oral contraceptive use and decreased menopausal hormone therapy (Siegel et al. 2024 )- it remains the eighth most common cancer and cause of cancer death among women globally as of 2022 (Bray et al. 2022 ). The American Cancer Society projects approximately 19,680 new ovarian cancer (OC) diagnoses and 12,740 °C-related deaths in the United States for 2024 (Jamali et al. 2025 ). The lack of effective screening methods and non-specific symptoms, often mimicking benign conditions, contribute to the majority of patients being diagnosed at advanced stages (FIGO III-IV), when the disease has spread beyond the pelvis (Colombo et al. 2019 ; Stewart et al. 2019 ). Consequently, ovarian cancer remains one of the most lethal gynecological malignancies, highlighting the urgent need to identify biomarkers for early diagnosis and elucidate the molecular mechanisms driving its pathogenesis. Ovarian cancer is broadly categorized into three main types: epithelial, germ cell, and sex cord-stromal tumors. Among these, epithelial ovarian cancer (EOC) is the most prevalent, encompassing several histological subtypes: high-grade serous ovarian cancer (HGSOC), low-grade serous ovarian cancer (LGSOC), endometrioid carcinoma (OEC), clear-cell carcinoma (OCCC), and carcinosarcoma (OCS; also known as malignant mixed Müllerian tumor) (del Carmen et al. 2012 ; Dareng et al. 2024 ). LGSOC accounts for approximately 10% of all serous ovarian cancers, typically originates within the ovary, and is associated with a relatively favorable prognosis. OEC frequently arises from endometriosis, which is usually diagnosed at an early stage and has a better prognosis (Stewart et al. 2019 ). OCCC is also linked to endometriosis and is frequently detected at an early stage (stage I), but advanced OCCC carries a poor prognosis and exhibits resistance to standard chemotherapy (Fujiwara et al. 2016 ). OCS is a rare and highly aggressive subtype, characterized by a dismal prognosis, with most patients experiencing recurrences within one year of initial treatment (Boussios et al. 2019 ). HGSOC, the most common and aggressive subtype of EOC, often originates in the fallopian tube and subsequently spreads to the ovary or peritoneum. Serous tubal intraepithelial carcinoma (STIC), considered a precursor lesion, is associated with up to 60% of HGSOC cases, a feature rarely seen in other EOC subtypes. The prognosis for HGSOC is poor, with over 85% of patients diagnosed at advanced stages and a 10-year mortality rate as high as 70% (Colombo et al. 2019 ; Shih et al. 2021 ). The human tissue kallikrein (KLK) family comprises at least 15 genes located in a cluster on chromosome 19q13.4 (Yousef and Diamandis 2001 ). These genes encode secreted serine proteases that are widely expressed in various tissues (e.g., skin, salivary glands, prostate, central nervous system, and breast) and biological fluids (e.g., seminal plasma, breast milk, ovarian cancer ascites, cerebrospinal fluid, and urine) (Shaw and Diamandis 2007 ). Members of the KLK family share significant structural and functional homology. KLKs exert their biological functions by cleaving specific substrates, thereby either activating or terminating signaling events mediated by hormones, growth factors, receptors, and cytokines. KLKs play critical roles in regulating physiological processes such as blood pressure control, skin desquamation, semen liquefaction, and neural development (Kryza et al. 2016 ). However, dysregulation of KLK expression and excessive protease activity have been implicated in various diseases, including hypertension, inflammation, skin disorders, and Alzheimer’s disease (Yousef and Diamandis 2001 ). In cancer, many KLK family members exhibit altered expression patterns. For instance, KLK3 , also known as prostate-specific antigen (PSA), is a well-established biomarker for prostate cancer screening (Penney et al. 2011 ). In breast cancer, the most common cancer among women, overexpression of KLK6 and KLK11 , alongside loss of KLK10 expression, have been reported (Srinivasan et al. 2022 ). Elevated levels of KLK5 and KLK14 are associated with higher tumor grade and poorer prognosis in breast cancer (Papachristopoulou et al. 2013 ). In ovarian cancer, overexpression of KLK4 , KLK5 , and KLK10 has been linked to diagnosis and poor prognosis, while KLK11 and KLK13 may serve as independent predictors of favorable overall survival (Paliouras et al. 2007 ). Notably, KLK5, especially its tumor-specific variants are significantly overexpressed in epithelial ovarian cancer cells compared to very low or undetectable levels in normal ovarian tissue or benign conditions (Dong et al. 2003 ; Dorn et al. 2011 ),, reinforcing its potential as a cancer-specific biomarker. Emerging evidence suggests that kallikreins may promote tumor progression by remodeling extracellular matrix (ECM) components and activating signaling pathways that regulate angiogenesis, invasion, and metastasis (Srinivasan et al. 2022 ). Conversely, certain kallikreins, such as KLK10 , may exhibit tumor-suppressive effects, as evidenced by their frequent downregulation in acute lymphoblastic leukemia and breast cancer (Luo et al. 2003 ). KLKs are thought to influence tumorigenesis through multiple hallmarks of cancer, including sustaining proliferative signaling, evading growth suppressors, inhibiting apoptosis, enabling replicative immortality, inducing angiogenesis, promoting invasion and metastasis, evading immune destruction, and disrupting cellular energy metabolism (Filippou et al. 2016 ). Additionally, kallikreins may modulate cancer development and progression by remodeling the tumor microenvironment (TME) as well (Srinivasan et al. 2022 ). Experimental studies indicate that KLK5 overexpression is most pronounced during the transition from grade I to grade II ovarian tumors (Kim et al. 2001 ). Furthermore, KLK5 protein is generally undetectable in the serum and ascites of healthy women and is weakly expressed in patients with benign ovarian tumors. In contrast, KLK5 levels are significantly elevated in the serum of ovarian cancer patients, and higher expression levels are inversely associated with overall survival (Dorn et al. 2011 ). These findings suggest serum KLK5 as a potential biomarker for the early diagnosis of ovarian cancer. Despite these advances, the precise mechanisms by which KLK5 influences the initiation, progression, and metastasis of ovarian cancer remain poorly understood. This article aims to explore the tumor-promoting mechanisms of KLK5 and provide new insights for developing early screening strategies and identifying potential therapeutic targets for ovarian cancer.

Supplementary Material 1. Supplementary Figure 1. The effect of KLK5 on migration, invasion, apoptosis, and proliferation of SKOV3 cells. A. The significantly enhanced of migration in the SKOV3 cell line with the addition of KLK5. B.The significantly enhanced of invasion in the SKOV3 cells with the addition of KLK5. C. The apoptosis of SKOV3 cells treated with or without KLK5. D. The proliferation of SKOV3 cells treated with or without KLK5. Supplementary Material 1. Supplementary Figure 1. The effect of KLK5 on migration, invasion, apoptosis, and proliferation of SKOV3 cells. A. The significantly enhanced of migration in the SKOV3 cell line with the addition of KLK5. B.The significantly enhanced of invasion in the SKOV3 cells with the addition of KLK5. C. The apoptosis of SKOV3 cells treated with or without KLK5. D. The proliferation of SKOV3 cells treated with or without KLK5. Supplementary Material 2. Supplementary Figure 2. UMAP plot showed the expressions of KLK5 in different types of ovarian cancer. Supplementary Material 2. Supplementary Figure 2. UMAP plot showed the expressions of KLK5 in different types of ovarian cancer. Supplementary Material 3. Supplementary Figure 3. Immune cell infiltration profiles of TCGA-KLK5High, TCGA-KLK5Low and normal controls, analyzed using the CIBERSORT algorithm. Supplementary Material 3. Supplementary Figure 3. Immune cell infiltration profiles of TCGA-KLK5High, TCGA-KLK5Low and normal controls, analyzed using the CIBERSORT algorithm. Supplementary Material 4. Supplementary Figure 4. Visualization of genes in the T_CD8_Cytotoxic subset. Supplementary Material 4. Supplementary Figure 4. Visualization of genes in the T_CD8_Cytotoxic subset. Supplementary Material 5. Supplementary Figure 5. Cell-cell interactions in KLK5-high HGSOC and KLK5-low HGSOC. A. Differential interaction strength between KLK5-high HGSOC and KLK5-low HGSOC. B. The cell-cell interactions in KLK5-high HGSOC (H) and KLK5-low HGSOC (L). Supplementary Material 5. Supplementary Figure 5. Cell-cell interactions in KLK5-high HGSOC and KLK5-low HGSOC. A. Differential interaction strength between KLK5-high HGSOC and KLK5-low HGSOC. B. The cell-cell interactions in KLK5-high HGSOC (H) and KLK5-low HGSOC (L). Supplementary Material 6. Supplementary Table 1. Single-cell RNA sequencing datasets and corresponding cell numbers used in this study. Supplementary Material 6. Supplementary Table 1. Single-cell RNA sequencing datasets and corresponding cell numbers used in this study. Supplementary Material 7. Supplementary Table 2. Seventy ovarian cancer-specific genes identified among the up-regulated genes in 29 cancers, based on RNA sequencing data from TCGA datasets in GEPIA2 (http://gepia2.cancer-pku.cn/#index). Supplementary Material 7. Supplementary Table 2. Seventy ovarian cancer-specific genes identified among the up-regulated genes in 29 cancers, based on RNA sequencing data from TCGA datasets in GEPIA2 (http://gepia2.cancer-pku.cn/#index). Supplementary Material 8. Supplementary Table 3. The up-regulated genes in subtypes of T and NK cells. Supplementary Material 8. Supplementary Table 3. The up-regulated genes in subtypes of T and NK cells.