KDM6A deficiency promotes tumor progression and resistance to cabozantinib treatment in clear cell renal cell carcinoma

Abstract

Lysine (K)-specific demethylase (KDM) 6A, a histone modifier with inhibitory roles in many types of cancers, is frequently mutated in clear cell renal cell carcinoma (ccRCC). Here, we investigated the role of KDM6A in ccRCC progression, including its effect on resistance to tyrosine kinase inhibitors (TKIs). The clinical impact of KDM6A expression was examined by immunohistochemical analysis of nephrectomized tissues from patients with ccRCC. Upon generation of KDM6A-deficient RCC cells by CRISPR/Cas9-mediated gene editing, in vitro cancer cell property analysis, RNA sequencing, and in vivo murine xenograft model examination were performed. The relationship between KDM6A expression and efficacy of TKIs was analyzed using data from public databases and in vitro pharmacological assessments. Patients with KDM6A-low ccRCC had poor prognoses. Promoted invasion, migration, cancer stemness, epithelial-mesenchymal transition (EMT), and in vivo tumor progression were observed in KDM6A-deficient RCC cells. A reanalysis of previous clinical trial data revealed lower efficacy of sunitinib in patients with KDM6A-low ccRCC. Consistently, KDM6A-deficient cells showed resistance to cabozantinib and decreased expression of target molecules of TKIs. KDM6A deficiency contributes to ccRCC progression and TKI resistance, suggesting that targeting KDM6A-related pathways may offer new therapeutic strategies for patients with KDM6A-deficient ccRCC.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-22561-7.

Introduction

Metastatic renal cell carcinoma (RCC) is a lethal disease with only around 10% 5-year survival rate¹. Clear cell RCC (ccRCC) is the main subtype of RCC². Increased hypoxia and upregulation of angiogenesis-associated molecules³, including vascular endothelial growth factor (VEGF), due to loss of von Hippel-Lindau (VHL), are frequently observed in ccRCC⁴. Tyrosine kinase inhibitors (TKIs), such as sunitinib and cabozantinib, offer a survival benefit to metastatic ccRCC patients by suppressing the VEGF signaling pathway and angiogenesis of tumors^5,6. However, the clinical utility of TKIs is compromised when primary and acquired resistance are encountered⁷. Therefore, understanding the mechanisms underlying cancer progression and resistance to TKIs in ccRCC patients would help establish novel therapeutic strategies for metastatic ccRCC.

Mutations in various genes responsible for epigenetic modifications have been identified in ccRCC through whole-exome and targeted sequencing studies^8,9. Epigenome modifiers affect cancer hallmarks, such as invasion¹⁰, cell cycle regulation¹¹, cell metabolism¹², and several important signaling pathways, such as the VHL and hypoxia-inducible factor-related pathway and pathways involved in epithelial-mesenchymal transition (EMT)¹³. However, the contribution of epigenome modifiers and their genetic alterations to cancer progression and resistance to TKIs in ccRCC patients has been largely unclarified.

Dysregulation of histone demethylation, a key chromatin modification, results in abnormalities in a wide array of nuclear activities related to genome integrity maintenance, transcriptional regulation, and epigenetic inheritance¹⁴. The relevance histone H3 lysine 27 (H3K27) methylation level has been proposed as a prognostic marker for patients with RCC¹⁵. Lysine (K)-specific demethylase 6 A (KDM6A), also known as the ubiquitously transcribed tetratricopeptide repeat X chromosome¹⁶, functions as a demethylase for H3K27 tori-methylation (H3K27me3) and play important roles in the regulation of gene expression¹⁷. KDM6A also regulates histone H3 lysine 4 (H3K4) methylation and H3K27 acetylation as a component of a complex of proteins associated with set 1 (COMPASS)-like complexes¹⁸. Consistent with its indispensable role in chromatin modification, KDM6A is considered a tumor suppressor in various cancers¹⁹. However, it has been reported that KDM6A promotes the resistance of chronic myelogenous leukemia cells to treatment with a TKI, imatinib²⁰.

Although the KDM6A gene is sporadically mutated in patients with ccRCC²¹, the contribution of KDM6A and its genetic alterations to cancer progression and resistance to TKI treatment in ccRCC has not been fully elucidated. In this study, we demonstrated the pivotal role of KDM6A deficiency in ccRCC progression and TKI resistance. Low KDM6A expression correlates with poor prognosis in ccRCC patients. KDM6A deficiency in ccRCC cells is associated with enhanced invasion, migration, and cancer stemness in vitro and tumor progression in vivo. In addition to the positive correlation between KDM6A expression and the expression of TKI target molecules, a decreased response to cabozantinib was observed in KDM6A-deficient ccRCC cells.

Results

Nephrectomized patients with KDM6A-low ccRCC showed worse pathological outcomes and poorer prognosis

We evaluated the expression of KDM6A in specimens from 77 patients who underwent total or partial nephrectomy for ccRCC. As the neoplastic tissues in individual specimens were substantially stained with the anti-KDM6A antibody at almost the same level, an apparent boundary between nonneoplastic and neoplastic tissues was observed (Fig. 1a). As shown in Fig. 1b and c, and Table 1, all regions of the neoplastic tissues of the 50 patients were KDM6A-positive, whereas some regions (> 5%) of the tissues of 27 patients were KDM6A-negative. KDM6A expression was principally detected in the tumor nucleoplasm, which was consistent with the histone modifier profile. Compared with approximately 50 ccRCC patients with entirely KDM6A-positive tumors (classified as KDM6A-high), the metastatic status was significantly higher in partially KDM6A-negative (classified as KDM6A-low) patients (Table 1). The higher male rate in the KDM6A-low group, consistent with previous reports demonstrating loss-of-function mutations in KDM6A more frequently in males²², is likely explained by its X chromosome location. Kaplan–Meier analysis revealed worse cancer-specific and progression-free survival rates in KDM6A-low cases (p = 0.003 and p = 0.014, respectively) (Fig. 1d and e).

Fig. 1.

Relationship between KDM6A deficiency and clear cell renal cell carcinoma (ccRCC) progression. (a) Immunohistochemical KDM6A staining in nonneoplastic kidneys of ccRCC patients. The results are representative of 77 independent samples. Original magnification, ×100. Scale bar, 200 μm. The broken line indicates the boundary between nonneoplastic and cancerous tissues. (b) Representative immunohistochemical KDM6A staining for specimens obtained from patients with KDM6A-low ccRCC. Original magnifications: ×200 (upper panel) and ×400 (lower panel). Scare bar, 50 μm. (c) Representative immunohistochemical KDM6A staining for specimens obtained from patients with KDM6A-high ccRCC. Original magnifications: ×200 (upper panel) and ×400 (lower panel). Scare bar, 50 μm. (d,e) Kaplan–Meier plots of cancer-specific (d) and progression-free (e) survival rates for KDM6A-high (n = 50) and -low (n = 27) nephrectomized ccRCC patients. p values calculated by the log-rank test are also indicated. (f,g) Kaplan–Meier plots of cancer-specific (f) and progression-free (g) survival rates for patients with nephrectomized ccRCC with low and high KDM6A expression. p values calculated by the log-rank test are also indicated. (h,i) Kaplan–Meier plots of disease-specific survival (h) and progression-free interval (i) rates for KDM6A-low and -high cases in The Cancer Genome Atlas Kidney Clear Cell Carcinoma Database. p values calculated by the log-rank test are also indicated.

Table 1.

Comparison of clinicopathological factors between KDM6A-negative and -positive nephrectomized clear cell renal cell carcinoma patients. *p < 0.05.

	Case (% of total)		p value
	KDM6A-negative	KDM6A-positive
	n = 27	n = 50
Age: n (%)			0.333
≤ 65	15 (55.6)	22 (44.0)
> 65	12 (44.4)	28 (56.0)
Gender: n (%)			0.006*
Male	25 (92.6)	33 (66.0)
Female	2 (7.4)	17 (34.0)
Pathological T: n (%)			0.559
< pT3	16 (59.3)	33 (66.0)
≥ pT3	11 (40.7)	17 (34.0)
Tumor grade: n (%)			0.905
G1/2	18 (66.7)	34 (68.0)
G3/4	9 (33.3)	16 (32.0)
Vascular invasion: n (%)			0.284
Negative	18 (66.7)	39 (78.0)
Positive	9 (33.3)	11 (22.0)
Metastasis status: n (%)			0.027*
M0	18 (66.7)	44 (88.0)
M1	9 (33.3)	6 (12.0)

The relationship of KDM6A expression with the prognosis was further evaluated in the independent 76 ccRCC patient group by assessing KDM6A mRNA expression levels in the tumor tissues. Upon dividing patients around the median KDM6A expression level (0.75/HPRT), significantly worse cancer-specific and progression-free survival rates were observed in KDM6A-low cases (0.25 ± 0.04/HPRT, n = 38) compared to KDM6A-high (1.83 ± 0.30/HPRT, n = 38) (p = 0.040 and p = 0.042, respectively; Fig. 1f and g). There were no significant differences in pathological factors between KDM6A-low and -high cases (Supplementary Table S1). Thus, the disease-specific survival and progression-free interval rates were significantly worse in KDM6A-low cases than in KDM6A-high cases (p = 0.046 and 0.017, respectively; Fig. 1h and i) in the data from The Cancer Genome Atlas (TCGA) Kidney Clear Cell Carcinoma (KIRC). Consistent with the observation in the immunohistochemical examination, TCGA KIRC dataset analysis revealed that KDM6A-low cases showed more frequent lymph node or distant metastases than KDM6A-high cases (Table 2). Furthermore, KDM6A-low tumors were significantly associated with advanced pathological T stage and higher pathological grade. These results suggest that KDM6A is associated with tumor progression in ccRCC.

Table 2.

Comparison of clinicopathological factors between KDM6A-low and -high cases in the cancer genome atlas kidney clear cell carcinoma database.

	Case (% of total)		p value
	KDM6A-low	KDM6A-high
	n = 266	n = 267
Age: n (%)			0.553
≤ 65	170 (63.9)	163 (61.0)
> 65	96 (36.1)	104 (39.0)
Gender: n (%)			< 0.001*
Male	230 (86.5)	115 (43.1)
Female	36 (13.5)	152 (56.9)
Pathological T: n (%)			0.017*
< pT3	157 (59.0)	185 (69.3)
≥ pT3	109 (41.0)	82 (30.7)
Tumor grade: n (%)			< 0.001*¶
G1/2	99 (37.2)	144 (53.9)
G3/4	165 (62.0)	117 (43.8)
unknown	2 (0.8)	6 (2.2)
Metastasis status: n (%)			0.021*¶
N0M0	85 (32.0)	115 (43.1)
N1 or M1	26 (9.8)	27 (10.1)
unknown	155 (58.3)	125 (46.8)

*p < 0.05.

^¶The chi-square test was performed omitting unknown.

Human ccRCC cell lines lacking KDM6A exhibit enhanced invasion, migration, cancer stemness, and EMT properties in vitro

To determine the functional significance of KDM6A in ccRCC, we investigated its contribution to cell invasion, migration, proliferation, and the expression of several marker molecules by CRISPR/Cas9-mediated disruption of the KDM6A gene in 786-O and Caki-1 cell lines. We confirmed the successful disappearance of KDM6A expression in all KDM6A-knockout (KO) cells compared to that in the corresponding wild-type (WT) and scrambled sgRNA-expressing viruses-infected (SV) cells by western blotting (Fig. 2a). Besides the comparative H3 expression, as a positive loading control in those cells, KDM6A-KO cells exhibited increased H3K27me3 expression, indicating the reduction of KDM6A-mediated demethylation activity in these cells. In addition to promoted invasion (Fig. 2b), KDM6A-KO cells showed enhanced migration compared to WT and SV (Fig. 2c). Although the basal levels of sphere formation activity differed among the ccRCC cell lines, the size and number of sphere colonies increased in KDM6A-KO cells (Fig. 2d). The proliferation of 786-O and Caki-1 cells was slightly, but significantly suppressed by the deficiency of KDM6A (Fig. 2e). KDM6A-KO cells showed enhanced expression of the cancer stem cell marker, CD44, and²³ and EMT markers, N-cadherin and vimentin (Fig. 2f)²⁴. These results suggest that KDM6A deficiency positively regulates the invasion, migration, cancer stemness, and EMT properties of ccRCC cells.

Fig. 2.

Functional role of KDM6A deficiency in clear cell renal cell carcinoma cells in vitro. (a) Wild-type (WT), scrambled sgRNA-expressing virus-infected (SV), and KDM6A-knockout (KO) 786-O and Caki-1 cells were subjected to determine protein levels of KDM6A, H3K27me3, H3, and β-actin by western blot. The positions of molecular weight standards are indicated on the right side. The results shown are representative of three separate experiments. (b) KDM6A-KO cells and the corresponding control cells (WT and SV) were subjected to a transwell invasion assay. Results are expressed as the mean ± SEM (n = 4). *p < 0.05 and **p < 0.001 by one-way ANOVA with Dunnett’s test. Representative Diff-Quick-stained Transwell membranes are shown in the left panels. Scale bar, 100 μm. (c) KDM6A-KO cells were subjected to wound healing assay with corresponding control cells (WT and SV). Results are expressed as the mean ± SEM (n = 3). *p < 0.05 and **p < 0.001 by one-way ANOVA with Dunnett’s test. Representative images of migrated cells are shown in the left panels. Scale bar, 200 μm. (d) KDM6A-KO cells were subjected to a sphere-forming assay with corresponding control cells (WT and SV). Results are expressed as representative photographs (left panels) and mean ± SEM (n = 4) of sphere numbers/10 mm² (right panel).*p < 0.05 and **p < 0.001 by one-way ANOVA with Dunnett’s test. (e) KDM6A-KO cells were subjected to cell proliferation analysis with corresponding control cells (WT and SV). Results are expressed as the mean ± SEM (n = 3) of absorbance at 450 nm. *p < 0.05 and **p < 0.001, compared to SV by one-way ANOVA with Dunnett’s test. (f) KDM6A-KO cells with corresponding control cells (WT and SV) were subjected to determine protein levels of CD44, N-cadherin, vimentin, and β-actin by western blot. The positions of molecular weight standards are indicated on the right side. The results shown are representative of three separate experiments.

KDM6A regulates in vivo tumor progression

To examine the role of KDM6A in tumor progression in vivo, a xenograft model was generated by subcutaneous implantation of KDM6A-KO and SV 786-O cells into BALB/c-background nude (BALB/c nu/nu) mice. The estimated volumes of KDM6A-KO tumors were significantly larger than those of control tumors starting from 5 weeks after transplantation (Fig. 3a). The weight of KDM6A-KO tumors was significantly greater than that of the SV tumors at 8 weeks post-implantation (Fig. 3b). Consistent with the upregulation of EMT in vitro, enhanced expression of N-cadherin and vimentin was observed in KDM6A-KO tumors (Fig. 3c). It is suggested that KDM6A deficiency promotes ccRCC progression partly by activating EMT-related pathways.

Fig. 3.

Tumor progression by KDM6A deficiency in vivo. (a) Estimated tumor volume of scrambled sgRNA-expressing virus-infected (SV) and KDM6A-knockout (KO) 786-O cell-transplanted mice. Results are expressed as the mean ± SEM (n = 6). *p < 0.05 by unpaired t-test. (b) Weight of excised 786-O/SV and 786-O/KDM6A-KO tumors at the end-point. Results are expressed as the mean ± SEM (n = 6). *p < 0.05 by unpaired t-test. (c) The top two larger tumor samples were subjected to western blot analysis for N-cadherin, vimentin, and β-actin. The positions of molecular weight standards are indicated on the right side.

Gene expression alteration in KDM6A-KO ccRCC cells

Next, to assess the molecular pathways associated with KDM6A deficiency, we performed a comparative RNA sequencing analysis using KDM6A-KO and control 786-O and Caki-1 cells. A series of Reactome gene sets were extracted by processing the data using Gene Set Enrichment Analysis (GSEA) (Supplementary Figs. S1, S2, Supplementary Tables S2, S3, S4, and S5). Among the 53 and 47 gene sets extracted from 786-O and Caki-1 cells (false discovery rate (FDR)-adjusted p value (q value) < 0.25 and nominal (NOM) p value < 0.05), respectively, 35 gene sets were positively enriched in both KDM6A-KO cells (Supplementary Fig. S1). Among the 78 and 85 gene sets extracted from 786-O and Caki-1 cells (FDR q-value < 0.25 and NOM p value < 0.05), respectively, 65 gene sets were negatively enriched in both KDM6A-KO cells (Supplementary Fig. S2). Notably, the gene set involved in histone modification, the polycomb repressive complex 2 (PRC2) (Supplementary Fig. S1), and epigenetic regulation of ribosomal RNA (rRNA) (Supplementary Fig. S1) were significantly associated with KDM6A deficiency. In the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway database, the gene sets extracted in both KDM6A-KO cell lines were similar to Reactome database analysis (Supplementary Fig. S3). We also performed GSEA using TCGA-derived transcriptomic profiles and compared KDM6A-low and -high cases (Supplementary Fig. S4, Supplementary Tables S6 and S7). Among the 108 gene sets positively enriched in KDM6A-low cases (FDR q value < 0.25 and NOM p value < 0.05), 24 were also positively enriched in KDM6A-KO cells (Supplementary Figs. S4a and S4b). Notably, gene sets associated with histone-modifying complex PRC2, epigenetic regulation of rRNA, and base excision repair (BER) were positively enriched in both KDM6A-KO cells and KDM6A-low patient samples (Supplementary Figs. S4c, S4d, and S4e). In contrast, only two gene sets were negatively enriched in KDM6A-low cases (Supplementary Fig. S4f).

Association of KDM6A expression with the efficacy of TKIs and related molecule expressions

In the RNA sequencing analysis in 786-O and Caki-1 cells, a significant association between the gene sets involved in VEGF- (Fig. 4a) and mesenchymal-epithelial transition factor (MET)- (Fig. 4b) related pathways and KDM6A deficiency was indicated. Therefore, the association of KDM6A expression with a group of genes associated with the efficacy of TKIs was examined using TCGA KIRC dataset. The expression of KDM6A in ccRCC was positively correlated with that of FLT1, KDR, and FLT4, which encode VEGFR1, 2, and 3, respectively (upper panels of Fig. 4c), PDGFRA and PDGFRB, which encode platelet-derived growth factor receptor (PDGFR) α and β, respectively, and phosphatase and tensin homolog (PTEN), a tumor suppressor gene (lower panels of Fig. 4c). TKIs exhibit their efficacy, at least in part, by suppressing the signaling pathways through VEGFRs and PDGFRs²⁵, suggesting that they may demonstrate reduced effectiveness against KDM6A-deficient tumors, where VEGFR and PDGFR expression is downregulated. Furthermore, PTEN has been implicated in resistance to TKI treatment²⁶. Therefore, the expression of KDM6A, which correlates with the expression of these receptors and enzymes, is potentially related to TKI resistance in RCC patients.

Fig. 4.

The relationship between expressions of KDM6A and target genes of tyrosine kinase inhibitors. (a,b) The patterns of REACTOME_SIGNALING_BY_VEGF (b) and REACTOME_MET_ACTIVATES_PTK2_SIGNALING (b) gene sets enriched in KDM6A-KO 786-O and Caki-1 cells compared to those of control cells. The corresponding normalized enrichment scores (NES), nominal (NOM) p values, and false discovery rate (FDR)-adjusted p values (q values) are also shown. (c,d) The correlation of log2 transcripts per million (TPM) level of KDM6A with that of FLT1, KDR, FLT4, PDGFA, PDGFB, and PTEN (c) and that of MET and AXL (d) was analyzed using Gene Expression Profile Interactive Analysis-processed data in The Cancer Genome Atlas Kidney Clear Cell Carcinoma database. p values and Pearson’s correlation coefficients (R) were calculated. (e) Kaplan–Meier plots of progression-free survival rate for KDM6A-low and -high cases of sunitinib-treated clear cell renal cell carcinoma patients in IMmotion151 study. p values calculated by the log-rank test are also indicated. (f) KDM6A-KO cells and the corresponding control cells (WT and SV) were subjected to determine FLT1, KDR, FLT4, AXL, and MET mRNA levels by quantitative RT-PCR. Results are expressed as the mean ± SEM (n = 3). *p < 0.05 and **p < 0.001 by one-way ANOVA with Dunnett’s test.

Cabozantinib, a new-generation TKI that addresses a broader spectrum of tyrosine kinases, displays enhanced antitumor efficacy⁶. In addition to VEGFR and PDGFR signaling, cabozantinib inhibits MET and Anexelekto (AXL) receptor tyrosine kinases⁶. The expression of those genes was positively correlated with KDM6A expression (Fig. 4d).

Among those target molecules of TKIs, significant correlations in the expression of FLT1, KDR, FLT4, PDGFRB, PTEN, and AXL with KDM6A were also observed by analyzing the data in the previously reported IMmotion151 study (Supplementary Fig. S5)²⁷. Clinicopathological factors of KDM6A-low and -high cases of sunitinib-treated clear cell renal cell carcinoma patients in IMmotion 151 study were shown in Supplementary Table S8. Furthermore, the association between the therapeutic effects of TKIs and KDM6A expression was examined using IMmotion151 study data. Upon dividing 416 sunitinib-treated patients in that study around the median KDM6A expression level, KDM6A-low patients showed significantly shorter progression-free survival compared to KDM6A-high patients (Fig. 4e). In addition, we performed quantitative RT-PCR to evaluate the mRNA expression levels of FLT1, KDR, FLT4, AXL, and MET in WT, SV, and KDM6A-KO cell lines (Fig. 4f). The expression profiles aligned with findings from TCGA and IMmotion151 datasets, reinforcing the robustness of our transcriptomic analysis. Although some gene expression differences were observed between 786-O and Caki-1 cells, most genes exhibited reduced expression in KDM6A-KO cells compared to WT and SV, with several genes showing statistically significant differences.

Reduced cytotoxic effect of cabozantinib on ccRCC cells by KDM6A deficiency

Finally, we assessed the cytotoxic effects of cabozantinib in the KDM6A-KO ccRCC cells and their SV and WT. Cabozantinib treatment decreased the viability of 786-O and Caki-1 cells in a concentration-dependent manner. The dose-response curves shifted to the right in KDM6A-KO cells (Supplementary Fig. S6). When evaluated by IC50 values, a significant decrease in the cabozantinib-induced cytotoxic effect due to KDM6A deficiency was observed in KO 786-O cells, but not in Caki-1 cells (Supplementary Fig. S6). Next, we determined the effect of cabozantinib on the clonogenicity of KDM6A-KO and their SV and WT cells. A substantial number of colonies developed following a 2-week culture of 786-O and Caki-1 cells. Cabozantinib suppressed colony formation in a concentration-dependent manner (Fig. 5a), although this effect was significantly weaker in KDM6A-KO cells (Fig. 5b and c). The expression of γ-H2AX and cleaved PARP, markers of DNA damage and apoptosis, respectively, was further examined in KDM6A-KO cells. Treatment with cabozantinib at 15 µM that showed a half-maximum cytotoxic effect augmented the expression of γ-H2AX in 786-O and Caki-1 cells and that of cleaved PARP in 786-O cells. The cabozantinib-evoked γ-H2AX and cleaved PARP expression was mostly weaker in KDM6A-KO 786-O cells (Fig. 5d) than that of SV and WT cells, while γ-H2AX expression was also weaker in KDM6A-KO Caki-1 cells. These results suggest that KDM6A-deficient ccRCC cells are resistant to the cytotoxic effects of cabozantinib.

Fig. 5.

Cytotoxic effect of cabozantinib on KDM6A-deficient ccRCC cells. The effect of cabozantinib on the colony-forming activity of Wild-type (WT), scrambled sgRNA-expressing virus-infected (SV), and KDM6A-knockout (KO) 786-O and Caki-1 cells. Representative Diff-Quick-stained 786-O colonies on 6-well plates (a) and mean ± SEM of a percent of colony number obtained without cabozantinib treatment (b) are shown (n = 3). *p < 0.05 and **p < 0.001, compared to SV by one-way ANOVA with Dunnett’s test. (c) The IC50 values were calculated from the concentration-dependent curves drawn in (b). Results are expressed as mean ± SEM (n = 3). *p < 0.05 and **p < 0.001 by one-way ANOVA with Dunnett’s test. (d) Protein levels of cleaved PARP, γ-H2AX, and β-actin before (0 h) and after 48 h of cabozantinib treatment (15 µM) were analyzed by western blot. The positions of molecular weight standards are indicated on the right side. The results shown are representative of three separate experiments.

Discussion

The present study elucidated the molecular and clinical consequences of the functional deletion of KDM6A in ccRCC. KDM6A deficiency in ccRCC cells induced enhanced invasion, migration, cancer stemness, and EMT abilities, which is consistent with its correlation with poor prognosis in ccRCC patients with low KDM6A expression. The positive correlation between KDM6A and TKI target gene expression, and the weaker cytotoxic effect of cabozantinib on KDM6A-KO ccRCC cells, suggested the link between KDM6A deficiency and resistance to TKIs. Schematic representation of the impact of KDM6A deficiency in ccRCC is summarized in Fig. 6.

Fig. 6.

Schematic representation of the impact of KDM6A deficiency in clear cell renal cell carcinoma (ccRCC). The mutation of the KDM6A gene, under the influence of von Hippel-Lindau (VHL) loss and hypoxia/hypoxia-inducible factor (HIF) signaling with in the surrounding tumor microenvironment (purple dashed square), leads to the deficiency of KDM6A expression and function (purple starburst) in ccRCC (beige area outlined in red), which arises in the kidney (purple background). This deficiency influences epigenetic alterations, including upregulation of the polycomb repressive complex 2 (PRC2) and increases histone H3 lysine 27 trimethylation (H3K27me3) levels (red upward arrow in white box), ultimately disrupting transcriptional regulation. These changes affect signaling cascades, notably downregulating the vascular endothelial growth factor (VEGF) and mesenchymal-epithelial transition factor (MET) pathways (blue downward arrows). Consequently, several aggressive cellular phenotypes, including enhanced invasion, migration, cancer stemness, epithelial–mesenchymal transition (EMT), and base excision repair (BER), are promoted (red arrows in blue box). These phenotypes contribute to increased tumor proliferation and metastasis (peach-colored circle). Alterations in signaling and tumor behavior, influenced by the surrounding microenvironment (gray area with green dashed outline), also impact treatment response, particularly characterized by increasing resistance to tyrosine kinase inhibitors (TKIs), such as sunitinib and cabozantinib (yellow circle). Dashed outline—tumor microenvironment and hypoxia/HIF signaling—linking KDM6A deficiency to clinical outcomes and TKI resistance. The yellow arrows represent established biological relationships, while brown dashed arrows indicate hypothetical pathways requiring further investigation. Collectively, these molecular and cellular events contribute to the poor prognosis (red starburst) observed in KDM6A-deficient ccRCC.

Comprehensive molecular characterization of ccRCC has demonstrated that KDM6A is one of the most frequently mutated genes in ccRCC¹⁶. The significance of the functional mutation of KDM6A in ccRCC pathogenesis was elucidated in this study. Thus, patients with ccRCC exhibiting low KDM6A expression at the protein or mRNA level showed a poor prognosis. These findings were supported by the phenotypes of KDM6A-KO ccRCC cells, represented by enhanced invasion, migration, cancer stemness, and EMT in vitro. In contrast, tumor growth of ccRCC cells was upregulated by KDM6A deficiency in vivo, consistent with TCGA dataset analysis demonstrating the negative association of KDM6A expression and advanced pathological T stage, an indicator of tumor volume. Previous studies on ccRCC carcinogenesis proposed that specific genetic alterations contribute to tumor progression in vivo through functional interactions between cancer cells and the surrounding microenvironment²⁸. Furthermore, the increased tumor size observed in vivo may also be attributed to enhanced EMT and reduced tumor cell density, rather than increased proliferative activity, as suggested by the greater variability in tumor weight compared to size (Fig. 3). KDM6A deficiency potentially promotes tumor growth in vivo and in ccRCC patients, potentially through its impacts on both cancer cells and the tumor microenvironment. Similar properties based on KDM6A-deficient conditions have been observed in other types of cancers. Activated cytokine and chemokine pathways, promoted M2 macrophage polarization, and consequently, bladder cancer development was induced in urothelium-specific Kdm6a-KO mice upon crossing with mice heterozygous for p53²⁹. Malignant lung cancer model developed in Kras^G12D/+ mice was further promoted by crossing the conditional Kdm6a-KO allele³⁰. KDM6A-KO in pancreatic ductal adenocarcinoma cells resulted in their enhanced aggressive traits³¹. These findings suggest that KDM6A broadly works as a tumor–suppressor in various organs.

The enhanced sphere-forming activity and CD44 expression in KDM6A-KO cells suggested a role of KDM6A in the cancer stemness of ccRCC cells. Cancer stemness, characterized by self-renewal and differentiation capabilities, contributes to the development of primary cancer cells, expansion of tumor dimensions, and treatment resistance³². Reprogramming of epigenetic modifications, including histone methylation conditions, is essential for producing and maintaining cancer stemness properties³³. Therefore, KDM6A might affect the stemness of ccRCC cells by regulating H3K27 methylation levels. We observed the enhanced H3K27me3 level in KDM6A-KO cells. Consistently, the KDM6A deficiency has been reported to promote cancer stemness in bladder cancer, another urological malignancy²⁹. However, Lu et al. reported conflicting findings that KDM6A upregulates the stemness of breast cancer by demethylating regulatory regions of pluripotency factor genes, such as NANOG, SOX2, and KLF4³⁴. Notably, another member of KDM6 subfamily, KDM6B, which also demethylates H3K27, participated in the loss of stem cell properties, partly by upregulating CD44³⁵. The reason for the controversial findings including the detailed underlying mechanisms deserves further examination.

EMT is an important process for cancer progression via the loss of cell polarity, decreased adhesion, and changes in the cytoskeleton that enhance cell migration and motility³⁶. The involvement of EMT in promoted ccRCC progression in KDM6A deficient conditions was suggested by the upregulation of N-cadherin and vimentin, mesenchymal markers, in KDM6A-KO cells in vitro and in vivo. Although whether KDM6A regulates these marker genes through histone modification needs to be clarified, TGF-β-dependent expression of EMT-related molecules, including vimentin, was downregulated by KDM6A in lung cancer cells³⁷. Involvement of the Wnt/β-catenin signal transduction pathway in EMT proposed by Feng et al.³⁸ further needs to be considered as the mechanism of KDM6A deficiency-mediated ccRCC progression.

The results of RNA sequencing analysis using KDM6A-KO cells reasonably reflected the role of KDM6A in ccRCC cells. In particular, the association of KDM6A deficiency with pathways involved in histone modification, the PRC2 complex, and epigenetic regulation of rRNA was identified. Because PRC2 catalyzes the trimethylation of H3K27, thereby downregulating gene transcription, KDM6A acts antagonistically on PRC2^39,40. Notably, Shao et al. demonstrated that the knock-down of KDM6A in 786-O cells leads to ribosomal protein-coding gene downregulation through increased methylation of H3K27 and H3K9⁴¹. Moreover, we identified several other pathways, most of which have not been reported to be linked with KDM6A. The extracted gene sets were mostly enriched in two independent cell lines, suggesting the reliability of our findings and the possible identification of new biological functions of KDM6A in future studies.

We obtained a series of findings suggesting that KDM6A expression influences the efficacy of TKIs on ccRCC. Upon reanalyzing the IMotion151 study data, a poor prognosis was observed in sunitinib-treated ccRCC patients with low KDM6A expression. Consistently, the cytotoxic effects and inhibition of colony formation by cabozantinib, along with DNA damage and apoptosis induction, were alleviated in KDM6A-KO ccRCC cells. The possible involvement of cancer stem cell properties and EMT in TKI resistance has been previously reported^7,32. GSEA revealed activation of BER-related pathways in KDM6A-KO cells. Given that BER is a key mechanism by which cancer cells develop resistance to radiation and chemotherapy⁴², this finding suggests that upregulated BER may also contribute to reduced sensitivity to TKIs in KDM6A-KO cells. We consistently revealed a positive correlation between KDM6A and VEGFRs, PDGFRs, MET, and AXL at the gene expression level using TCGA KIRC dataset and IMmotion 151 data. The association of VEGF- and MET-related pathways with KDM6A deficiency was indicated by RNA sequencing analysis. Although the regulatory mechanisms of these genes by KDM6A in ccRCC cells need to be explored, treatment of endothelial colony-forming cells with an inhibitor of enhancer of zeste homolog 2, a component of PRC2, leads to the activation of multiple pro-angiogenic signaling pathways, including VEGFR⁴³. Our findings suggest that KDM6A deficiency disrupts the antagonistic interplay between KDM6A and PRC2 complex, thereby influencing the transcriptional regulation of VEGFRs, MET, and AXL. However, the expression levels of these genes do not consistently correlate with KDM6A expression, which may reflect gene-specific recruitment of PRC2 and the chromatin context of each target. PRC2 preferentially binds to nucleosome-depleted, Cytosine-phosphate-Guanine (CpG)-rich regions, and its activity is modulated by factors such as DNA shape and chromatin density, leading to gene-specific repression profiles^44,45. Therefore, differential epigenetic accessibility likely underlies the selective regulation of these receptor tyrosine kinases in KDM6A-deficient contexts. The contribution of KDM6A to histone methylation conditions in the genes related to the efficacy of TKIs, including VEGFRs, PDGFRs, MET, and AXL, needs to be further elucidated.

Our study has several limitations. First, our in vivo subcutaneous xenograft implantation model is well-suited for evaluating the proliferative capacity of ccRCC cells; however, it is suboptimal for accurately assessing the metastatic potential and may not fully recapitulate the native tumor microenvironment of ccRCC. Future studies employing syngeneic orthotopic transplantation models of KDM6A-deficient tumors, as well as dedicated metastatic models, will be essential to further elucidate tumor-microenvironment interactions and comprehensively evaluate metastatic behavior. Second, the VHL–hypoxia inducible factor (HIF) axis, a key driver in ccRCC, was not examined. Given prior reports linking hypoxia to KDM6A inactivation⁴⁶, future studies are warranted to explore this regulatory interplay. Third, while we examined sunitinib response using IMmotion 151 dataset, in vitro functional validation using KDM6A-deficient cells was not performed. These experiments are prioritized for future investigation. Additionally, our functional assays were limited to two ccRCC cell lines (786-O and Caki-1), which may not fully capture the molecular and genetic heterogeneity of ccRCC. Further validation using a broader panel of cell lines is warranted. Fourth, the effects of TKIs were evaluated solely in terms of cell proliferation and survival. However, the precise mechanisms by which KDM6A modulates TKI efficacy remain unclear. KDM6A may influence resistance through downstream signaling pathways, regulation of drug-metabolizing enzymes, or alterations in the tumor microenvironment. Although our RNA-seq–based GSEA identified several pathways potentially affected by KDM6A deficiency, these associations remain correlative. Further validation through functional assays and integration with additional omics layers, such as proteomics and epigenomics, is warranted to clarify the biological relevance of these pathways. Moreover, the impact of KDM6A deficiency on cellular behaviors such as migration, invasion, EMT, and stemness, as well as in vivo resistance, remains to be explored. In addition, the limited generalizability of our findings, stemming from the single-center design, relatively small sample size, and absence of an independent validation cohort, highlights the need for future studies involving larger multicenter cohorts. Clinical datasets used in this study (TCGA, IMmotion151) are predominantly derived from Western populations. To confirm the universality and specificity of KDM6A-related findings, future research should incorporate racially and geographically diverse cohorts, accompanied by detailed stratified analyses. Mechanistic studies using overexpression models, rescue experiments, and pathway inhibitors will be essential to confirm causal downstream pathways. Although we explored immune infiltration via correlation analysis, functional studies are needed to assess how KDM6A influences the tumor immune microenvironment and its contribution to TKI response. Finally, while our findings suggest the potential of KDM6A as a biomarker for treatment stratification, further prospective studies and clinical trials are needed to establish its utility in personalized therapy for ccRCC.

In conclusion, we have elucidated the essential role of KDM6A deficiency in promoting ccRCC progression by upregulating its invasion, migration, cancer stemness, and EMT properties. In addition to prognosis, the efficacy of TKIs in patients with ccRCC is closely associated with the expression level of KDM6A. Our findings provide novel insights into the pathogenesis of this heterogeneous and complex disease and may provide a novel therapeutic approach for ccRCC with KDM6A dysfunction.

Methods

Patients and tissue samples

Seventy-seven patients diagnosed with ccRCC who underwent nephrectomy at Hiroshima University Hospital between April 2002 and December 2012 were included in the immunohistochemical staining analysis. The independent 76 patients diagnosed with ccRCC, who underwent nephrectomy at Hiroshima University Hospital between April 2019 and March 2022, were included in the quantitative reverse transcription-polymerase chain reaction (RT-PCR) analysis. Tumor tissues used for RT-PCR were immediately frozen in liquid nitrogen after surgical resection and stored at − 80 °C until RNA extraction. Pathological diagnosis was carried out based on the World Health Organization criteria (2016)⁴⁷. The tumor staging was based on the tumor node metastasis (TNM) classification system established by the International Union Against Cancer⁴⁸. Upon evaluating the patients’ background, we obtained relevant clinicopathologic data from the medical records, such as age, sex, pathological TNM stage, tumor grade, and metastasis status.

Immunohistochemical staining

Immunohistochemical staining was performed as described previously⁴⁹. Briefly, formalin-fixed and paraffin-embedded tissue sections from 77 patients with ccRCC were deparaffinized in xylene and rehydrated using a gradient ethanol series. Tissue sections were incubated for 30 min in Target Retrieval Solution (pH9) (Agilent Technologies, Santa Clara, CA, USA) in a microwave oven (500 W, 30 min). Peroxidase activity was blocked with 3% H₂O₂-methanol (hydrogen peroxide and methanol, both from FUJIFILM Wako Pure Chemical Corporation, Osaka, Japan) for 10 min. The sections were incubated with anti-KDM6A (1:200; #33510; Cell Signaling Technology, Danvers, MA, USA) at 4 °C overnight. After washing with phosphate-buffered saline (PBS) (PHC Corpolation, Tokyo, Japan) containing 0.05% Tween-20 (FUJIFILM Wako Pure Chemical Corporation), the sections were incubated with an anti-rabbit horseradish peroxidase polymer (Agilent Technologies) for 1 h at room temperature, followed by incubation with diaminobenzidine Substrate-Chromogen Solution (Dako, Glostrup, Denmark) for 8 min for the color reaction, and then counterstained with 0.1% hematoxylin (MUTO PURE CHEMICALS, Tokyo, Japan). Negative controls were created by omitting the primary antibodies. The staining intensity of the resulting specimens was determined by two independent uropathologists who were blinded to patient parameters. Representative snapshots of immunohistochemical staining from 27 KDM6A-low and 50 KDM6A-high patients are provided in Supplementary Figs. S7 and S8. Kaplan–Meier survival analyses were conducted using the log-rank test to compare groups.

Quantitative RT-PCR

Frozen tumor tissues from 76 patients with ccRCC were used for quantitative RT-PCR analysis. Total RNA was isolated using ISOGEN (NIPPON GENE, Tokyo, Japan), and 500 ng of total RNA was reverse-transcribed to cDNA using PrimeScript RT Master Mix (Takara Bio, Shiga, Japan). Quantitative PCR analysis was performed for cDNA with SYBR Select Master Mix (Applied Biosystems, Waltham, MA, USA) using the StepOnePlus Real-Time PCR System (Applied Biosystems). Each patient sample was analyzed in technical duplicate, and all 76 samples were independently processed. The reactions were carried out using the following quantitative PCR primers: KDM6A (5’-GATGACCTGTCCAGTCCTGC-3’ and 5’-CTATTTGCGCGGAGCTGTTC-3’), HPRT (5’-CCTCATGGACTAATTATGGACAG-3’ and 5’-GCAGGTCAGCAAAGAATTTATAG-3’), FLT1 (5’-GCAACCTTCACCTACCGAGTTC-3’ and 5’-TCTCCTCCGAGCCTGAAAGT-3’), KDR (5’-ATGGGAACCGGAACCTCACTATC-3’ and 5’-GTCTTTTCCTGGGCACCTTCTATT-3’), FLT4 (5’-CAACTGGGTGTCCTTTCC-3’ and 5’-CTTGTCTATGCCTGCTCTC-3’), AXL (5’-GCAACCTTCACCTACCGAGTTC-3’ and 5’-GGCCAACATGGTGAAACCCT-3’), MET (5’-CCATCCAGTGTCTCCAGAAGTG-3’ and 5’-TTCCCAGTGATAACCAGTGTGTAG-3’). The Ct values were normalized to the expression of an endogenous housekeeping gene, HPRT, and 2(−ΔΔCt) values were calculated for relative quantification. Based on the mRNA expression data, patients were classified into high (≥ median) and low (< median) expressions of KDM6A. Kaplan–Meier survival analyses were conducted using the log-rank test to compare groups. Additionally, quantitative RT-PCR was conducted on WT, SV, and KDM6A-KO 786-O and Caki-1 cells. Total RNA was isolated using NucleoSpin RNA kit (Takara Bio) following the manufacturer’s instructions, and 500 ng of RNA was reverse-transcribed into cDNA using PrimeScript RT Master Mix (Takara Bio). Each condition was assessed in biological triplicate.

Bioinformatics analysis

The clinicopathological data of 533 ccRCC patients from TCGA KIRC database were downloaded from the UCSC XENA DATA HUBS (hub: https://tcga.xenahubs.net). Upon processing mRNA expression z-scores relative to diploid samples (RNA‐Seq V2 RSEM), patients were divided into KDM6A-high and -low groups. Kaplan–Meier analyses of disease-specific survival and progression-free interval for the two groups were performed. The correlation between the expression levels of KDM6A and other genes in TCGA KIRC database was analyzed using data processed by Gene Expression Profile Interactive Analysis, a web-based tool for delivering fast and customizable functionalities based on TCGA and Genotype-Tissue Expression data (http://gepia2.cancer-pku.cn/#analysis). Correlation analysis between the expression levels of KDM6A and other genes and Kaplan–Meier analyses of progression-free survival for KDM6A-high and -low groups were also carried out using data from a randomized phase III study of atezolizumab plus bevacizumab vs. sunitinib in untreated metastatic RCC (IMmotion151)²⁷. For both TCGA and IMmotion151 datasets, KDM6A-high was defined as samples with mRNA expression levels equal to or greater than the median, while KDM6A-low was defined as those with expression levels below the median.

Cell lines and cell cultures

Human ccRCC cell lines 786-O (RRID: CVCL_1051) and Caki-1 (RRID: CVCL_0234) were purchased from the Japanese Collection of Research Bioresources Cell Bank and maintained at 37 °C in a humidified atmosphere containing 5% CO₂ in RPMI-1640 medium (Sigma-Aldrich, St. Louis, MO, USA) supplemented with 10% fetal bovine serum (FBS) (Gibco, Tokyo, Japan) and 1% Penicillin-Streptomycin Solution (FUJIFILM Wako Pure Chemical Corporation). For each assay, cells cultured for 2–3 passages were used and seeded into 6–10 cm culture dishes (Corning, Corning, NY, USA), depending on the experiment requirements.

CRISPR/Cas9-mediated gene editing

To knock out KDM6A expression in 786-O and Caki-1 cells, we used CRISPR/Cas9-mediated gene editing. Commercially available KDM6A sgRNA CRISPR/Cas9 All-in-One Lentivector set (human) and CRISPR scrambled sgRNA All-in-One Lentivector (with spCas9) were purchased from Applied Biological Materials (Richmond, BC, Canada). Lenti-X 293 T cells (Takara Bio) were seeded in 10 cm culture dishes (Corning) and co-transfected with 1 µg of plasmid DNA and a third-generation packaging mix (Applied Biological Materials) using Lipofectamine 2000 transfection reagent (Thermo Fisher Scientific, Waltham, MA, USA). After 48 h, the lentiviral supernatant was collected, filtered, and used to infect 786-O and Caki-1 cells seeded in 6-well culture plates (Thermo Fisher Scientific). For infection, polybrene (8 µg/mL, Sigma-Aldrich) was added to the viral supernatant, and cells were subjected to spin infection at 1,000 × g for 30 min at 23 °C. Following overnight incubation, infected cells were selected using puromycin (2.0 µg/mL; FUJIFILM Wako Pure Chemical Corporation) for 48 h.

Western blot analysis

Western blot analysis was performed as described previously⁵⁰. Briefly, cells and tumor samples were lysed using RIPA buffer (FUJIFILM Wako Pure Chemical Corporation) containing protease and phosphatase inhibitors. After equalizing the protein concentration, the cell extract samples were electrophoresed on 5%–20% precast polyacrylamide gels (SuperSep Ace; FUJIFILM Wako Pure Chemical Corporation) and transferred to nitrocellulose blotting membranes (GE Healthcare Life Sciences, Chicago, IL, USA). The membranes were blocked for 30 min with 5% nonfat dry milk in TBST (10 mM Tris, pH 8.0, 150 mM NaCl, and 0.05% Tween-20) and incubated with the following primary antibodies at 4 °C overnight. Anti-human KDM6A (1:500; #33510; Cell Signaling Technology), anti-H3K27me3 (1:1000, #9733; Cell Signaling Technology), anti-H3 (1:1000, ab1791; abcam, Cambridge, UK), anti-CD44 (1:1000, #37259; Cell Signaling Technology), anti‐N‐cadherin (1:1000, #13116; Cell Signaling Technology), anti‐vimentin (1:1000, #5741; Cell Signaling Technology), anti‐Cleaved PARP (1:1000, #5625; Cell Signaling Technology), anti‐γ‐H2AX (1:1000, #9718; Cell Signaling Technology), and anti-β-actin (1:2000, #A2228; Sigma-Aldrich). After three washes with TBST, the membranes were incubated with horseradish peroxidase-conjugated secondary antibodies (anti-mouse, 1:1000; MBL Life Science, Tokyo, Japan; anti-rabbit, 1:1000; MBL Life Science) at room temperature for 1 h. After another washing step, immunoreactive bands were detected with an enhanced chemiluminescence kit (Thermo Fisher Scientific, Waltham). Cleaved PARP was selected as a marker of caspase‑mediated apoptosis, and γ‑H2AX was used to indicate DNA double‑strand breaks following TKI treatment, as both markers have been previously employed to evaluate TKI effect in vitro using RCC cell lines^51,52. Original western blot images are provided in Supplementary Figs. S9–S13. The membranes were cut prior to hybridisation with different antibodies (This was the reason for the absence of images of adequate length).

Transwell invasion assay

For the Transwell invasion assay, 24-well cell culture inserts (Corning) were used. After pre-coating the filter with 50 µL Matrigel (Corning) for 2 h, cells suspended in serum-free medium were seeded in the top chamber, and serum-containing medium (10% FBS) was added to the bottom chamber. A total of 2.0 × 10⁴ 786-O cells or 1.0 × 10⁵ Caki-1 cells were seeded into each insert. Twenty-four hours later, the cells that translocated to the bottom surface of the membrane were recovered and counted under a light microscope following Diff-Quick staining (Sysmex Corporation, Kobe, Japan). The assay was conducted in biological quadruplicate. Invaded cells were counted in randomly selected fields per insert under a light microscope at 200× magnification. Representative images were obtained, and quantification was performed using ImageJ software (NIH, Bethesda, MD, USA).

Wound healing assay

The cells were seeded in a culture insert (ibidi culture-insert 2 well; ibidi GmbH, Bayern, Germany) at a density of 5.0 × 10⁵ cells/ml. After allowing the cells to adhere overnight, the culture insert was recovered and washed with PBS to remove non-adherent cells. The cells were then cultured in fresh medium for 12 h. After photographing the plate at the start and end of the culture period, the cells that migrated into the wound space were manually counted in three fields per well under a light microscope. Cell migration areas were quantified using the ImageJ software (https://imagej.nih.gov/ij/).

Sphere-formation assay

Single-cell suspensions were seeded in a 6-well cell culture plate (Ultra-Low Attachment Surface, Corning) at a density of 1.0 × 10⁵ cells/well in serum-free DMEM/F12 (Thermo Fisher Scientific) supplemented with 20 ng/mL epidermal growth factor (EGF, Thermo Fisher Scientific), 20 ng/mL basic fibroblast growth factor (Thermo Fisher Scientific), 1% N-2 Supplement (Thermo Fisher Scientific), 2% B-27 Supplement (Thermo Fisher Scientific), and 1% penicillin-streptomycin solution. After the cultivation for 2 weeks, the number of tumorspheres over 50 μm diameters in 786-O and Caki-1 cells developed in each well was counted.

Cell proliferation and colony formation assay

For the cell proliferation assay, cells (3–5 × 10³ cells/well) were incubated in a 96-well cell culture plate (Thermo Fisher Scientific) for 24, 48, or 72 h. Cell proliferation was assessed by measuring the absorbance of each well at 450 nm using the Premix WST-1 Cell Proliferation Assay System (Takara Bio), following the manufacturer’s protocol. In some experiments, increasing concentrations of cabozantinib were administered at the beginning of the culture period. For the colony formation assay, cells were treated with increasing concentrations of cabozantinib for 48 h, harvested, plated at 500 cells/well in 6-well plates (Thermo Fisher Scientific), and cultured for 2 weeks until visible colonies formed. Colonies were stained using Diff-Quick and manually counted under a light microscope. For each measurement, wells with untreated cells and media without cells served as controls and blanks, respectively. Following cell proliferation and colony formation assays at 72 h and 2 weeks, respectively, concentration-response curves and IC50 values were calculated using Microsoft Excel. Based on preliminary dose-response experiments, cabozantinib with a concentration of 15 µM was employed for western blotting analysis of γ-H2AX and cleaved PARP, as this condition induced reliable responses in 48 h treatment.

RNA sequencing and GSEA

Total RNA was extracted from KDM6A-KO and WT cells using NucleoSpin RNA kit (Takara Bio). RNA quality was assessed with the Agilent 2100 Bioanalyzer (Agilent Technologies), and only samples with RNA integrity number (RIN) greater than 8.0 were used for sequencing. Libraries were prepared using the SureSelect Strand-Specific RNA Library Preparation Kit (Agilent Technologies), and transcriptome analysis was performed using the HiSeq 2500 next-generation sequencing platform (Illumina, San Diego, CA, USA). The generated sequence tags were mapped to human genomic sequences (hg38). To identify differentially expressed gene sets, data were analyzed using GSEA (version 4.3.2; Broad Institute, http://software.broadinstitute.org/gsea/downloads.jsp), referencing the Reactome (https://reactome.org/) and KEGG (https://www.genome.jp/kegg/pathway.html) pathway databases. Additionally, GSEA was performed on TCGA-KIRC RNA-seq data, stratifying patients into KDM6A-high and -low groups based on median z-score–normalized expression levels as processed using the RNA-Seq V2 RSEM pipeline.

Animal studies

In a xenograft model, KDM6A-KO and control (SV) 786-O cells (1.0 × 10⁴) in 50 µl HBSS (FUJIFILM Wako Pure Chemical Corporation) and 50 µl Matrigel (Corning) were injected subcutaneously into 8-week-old female BALB/c nu/nu mice (Charles River, Wilmington, MA, USA). A total of six mice per group (n = 6) were used. The estimated tumor volumes were measured weekly using calipers and calculated as (w² × l)/2, where w is the width and l is the length. Eight weeks after the cell transplantation, the tumors were excised from mice following euthanasia. Cervical dislocation, a physical method of euthanasia, was employed for small rodents in accordance with the guidelines of the American Veterinary Medical Association (AVMA). To ensure a humane and painless process, mice were anesthetized with a combination of three anesthetic agents (midazolam, medetomidine, and butorphanol) administered intraperitoneally to achieve a deep anesthetic state. The euthanasia procedure was performed by trained personnel to minimize potential pain or distress and ensure compliance with ethical standards. After weighing, the tumor samples were subjected to western blot analysis.

Statistical analysis

Statistical analyses were performed using R software (version 4.3.2; The R Foundation for Statistical Computing, Vienna, Austria). Multiple group comparisons in quantitative RT-PCR, transwell invasion, wound healing, sphere-formation, cell proliferation, and colony formation assays were conducted using one-way ANOVA followed by Dunnett’s multiple comparison test. Two-group comparisons for Kaplan–Meier survival analysis and animal studies were performed using the log-rank test and unpaired t-test, respectively. Statistical significance was defined as p < 0.05.

To address potential false positives in GSEA, FDR-adjusted p values (q values) were calculated, with statistical significance defined as q < 0.25 and NOM p < 0.05. In addition to p values, Pearson’s correlation coefficients (R) were calculated in linear regression analyses of bioinformatics data. All graphical representations were generated using GraphPad Prism (GraphPad Software, San Diego, CA, USA).

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Y.K. performed formal analysis, investigation, methodology, visualization, and writing—original draft preparation. K.K. contributed to conceptualization, methodology, project administration, and writing—review & editing. K.T. and T.B. were responsible for methodology. Y.S. performed formal analysis, methodology, and resource acquisition. H.N. performed resource acquisition. K.M. worked on methodology. R.Y. and R.T. carried out investigation. T.F., H.K., K.G., A.G., and K.H. handled data curation. O.K. contributed to visualization, writing, and review & editing. N.H. provided supervision.

Funding

This research was supported by JSPS KAKENHI (Grant Number 24K12509) and a Joint Research Grant from the Research Center for Radiation Disaster Medical Science to Kohei Kobatake.

Data availability

RNA-seq datasets analyzed during the current study are available in the DNA Data Bank of Japan (DDBJ) BioProject repository under accession number PRJDB20188 (https://ddbj.nig.ac.jp/search/entry/bioproject/PRJDB20188). Other data supporting the findings of this study are available from the corresponding author (KK) upon reasonable request.

Declarations

Competing interests

The authors declare no competing interests.

Ethics declarations

Approval for human experiments: All experimental procedures involving human participants were conducted in accordance with the ethical standards of the Declaration of Helsinki and were approved by the Ethics Committee of Hiroshima University Hospital (approval no. E-588-2). Approval for animal experiments: All animal experiments were carried out in strict accordance with the recommendations of the Guide for the Care and Use of Laboratory Animals of Hiroshima University Animal Research Committee (permission no. A22-42-2).

Consent to participate

All human participants provided written informed consent, with guarantees of confidentiality.