DCLK1 drives malignant progression and chemoresistance of bladder cancer by deubiquitinating HDAC6

Abstract

Background

Bladder cancer (BC) is the most common malignancy of the urinary system, with rising incidence and mortality. Advanced BC frequently recurs or metastasizes and is often refractory to curative surgical intervention. Although cisplatin-based chemotherapy remains the standard first-line treatment, its clinical efficacy is frequently compromised by the development of drug resistance. Elucidating the molecular mechanisms underlying chemoresistance and metastasis is therefore critical for improving therapeutic strategies.

Methods

Single-cell RNA sequencing (scRNA-seq) was conducted to assess intratumoral heterogeneity and identify expression programs associated with bladder cancer progression. DCLK1 emerged as a key cancer-related hub gene. Its role in metastasis, cisplatin resistance, and immune evasion was evaluated using sphere formation, CCK-8, Transwell, CFSE staining, and flow cytometry assays. The interaction between DCLK1, USP10, and HDAC6 was confirmed through RNA pull-down, co-immunoprecipitation, mass spectrometry, cell localization, and molecular docking. Finally, the therapeutic potential of DCLK1, cisplatin, and ACY-1215 (an HDAC6 inhibitor) was tested in vivo.

Results

We identified multiple cell types, including cancer cells, lymphocytes, myeloid cells, fibroblasts, and other stromal components. In cancer cells, six biologically relevant expression programs were revealed. Among key cancer-related genes, DCLK1 was notably enriched and promoted bladder cancer metastasis, cisplatin resistance, and stemness. Mechanistically, DCLK1 activated the Notch pathway to upregulate PD-L1, suppress CD8⁺ T cell activity, and promote immune evasion. It also facilitated USP10-HDAC6 interaction, removing K48-linked ubiquitin at Lys116 to prevent HDAC6 degradation. Importantly, HDAC6 depletion abrogated the oncogenic effects of DCLK1, whereas inhibition of DCLK1 suppressed tumor progression and enhanced the antitumor efficacy of combined cisplatin and ACY-1215 treatment.

Conclusions

DCLK1 is a critical driver of bladder cancer progression, chemoresistance, and immune escape. Single-cell analysis and functional assays revealed that DCLK1 enhances metastasis and stemness by activating the Notch/PD-L1 axis and stabilizing HDAC6 through USP10 interaction. Targeting DCLK1, alone or in combination with cisplatin and HDAC6 inhibition, represents a promising therapeutic strategy for advanced bladder cancer.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12943-025-02560-y.

Background

Bladder cancer (BC) is the most common malignancy of the urinary system and is associated with high incidence and mortality. Approximately 25% of patients present with muscle-invasive or metastatic disease at diagnosis [1, 2]. Transurethral resection of bladder tumors and radical cystectomy are indicated for patients with early-stage disease; however, progression to advanced stages is frequently accompanied by local recurrence or distant metastasis [3–5], and these tumors are no longer suitable for curative surgical resection. Cisplatin-based chemotherapy is the standard first-line therapy for patients with advanced BC [6]. However, patients invariably develop drug resistance after several treatment cycles, limiting overall clinical efficacy [7]. Therefore, identifying the molecular determinants underlying chemoresistance is critical for improving therapeutic outcomes in BC. Single-cell RNA sequencing (scRNA-seq) is a powerful tool that can provide single-cell resolution expression profiling of human cancers, enabling the identification and characterization of specific subclusters that bear unique biological effects [8, 9]. Using scRNA-seq, tumor cell states, clonal evolution, and interactions within the tumor microenvironment can be systematically characterized, facilitating the discovery of novel therapeutic targets [10–12].

In this study, we comprehensively delineated the intratumoral heterogeneity and immunosuppressive microenvironment of bladder cancer at single-cell resolution. Multiple cell populations were identified, including cancer cells, lymphocytes, myeloid cells, fibroblasts, and other stromal components. Notably, several immunosuppressive cell types were enriched in the tumor microenvironment, such as FOXP3 + regulatory T (Treg) cells, APOC1 + M2-like macrophages, WNT2 + cancer-associated fibroblasts, and LAMP3 + tolerogenic dendritic cells, suggesting extensive immune remodeling in BC. Further analysis revealed six conserved transcriptional programs in cancer cells, representing distinct biological states, including proliferation, metastasis, immune evasion, stress resistance, metabolic reprogramming, and inflammatory activation. Among these cancer-related hub genes, DCLK1 (doublecortin-like kinase 1) was markedly enriched in tumor cells and has been found to potentially be an effective target for the inhibition of metastasis in bladder cancer [13]. DCLK1 is a microtubule-associated protein kinase that is highly expressed in multiple malignancies and has attracted increasing attention as a therapeutic target in cancer research [14]. Previous studies have shown that DCLK1 plays a role in modulating tumor cell pluripotency [15], the stemness of cancer stem cells (CSCs) [16], drug resistance [17], epithelial-to-mesenchymal transition (EMT) [18], DNA damage response [19], and the tumor microenvironment [20, 21]. Moreover, DCLK1 has been shown to promote metastatic behavior and chemotherapy resistance through the activation of EMT-related and TGF-β signaling pathways [22].

In the present study, we demonstrate that DCLK1 promotes bladder cancer progression by functioning as a molecular scaffold that facilitates the interaction between the deubiquitinating enzyme USP10 and HDAC6, thereby stabilizing HDAC6 through inhibition of ubiquitin-mediated proteasomal degradation. In parallel, DCLK1 enhances tumor immune evasion by activating the Notch signaling pathway, leading to upregulation of PD-L1 expression and suppression of CD8⁺ T cell proliferation. Furthermore, DCLK1 promotes USP10-mediated removal of K48-linked ubiquitin chains from Lys116 of HDAC6, preventing its degradation via the ubiquitin–proteasome pathway. Importantly, knockdown of HDAC6 reversed the oncogenic effects induced by DCLK1. Collectively, our study identifies DCLK1 as a crucial driver of cisplatin resistance and provides mechanistic insights into potential therapeutic targets for overcoming chemoresistance in bladder cancer patients.

Materials and methods

Patient samples

This study involved four patients who underwent surgery for bladder cancer. A total of ten fresh surgical specimens were analyzed, including two normal tissues, four paracancerous tissues, and four tumor tissues with lymph node metastasis. Tumor specimens were verified by pathologists via cytological inspection during surgical procedures and through post-operative paraffin sections. Tumor staging followed the American Joint Committee on Cancer (AJCC) 8th edition at diagnosis.

Tissue dissociation and single-cell suspension for scRNA-Seq

Fresh tissues were promptly kept on ice in sCelLive™ Tissue Preservation Solution (Singleron, China). Samples were washed three times with HBSS before digesting with 2 mL of sCelLive™ Tissue Dissociation Solution at 37 °C for 15 min. Following tissue mincing, red blood cells were lysed using a red blood cell lysis solution (MACS), and the cell suspension was centrifuged at 1,500 rpm for 5 min. The pellet was washed twice with RPMI 1640 containing 0.04% BSA and then resuspended in sorting buffer (PBS with 0.04% FBS). The cell concentration and vitality were measured using a Luna cell counter.

Library Preparation and single-cell RNA sequencing

Chromium barcodes were used to uniquely label transcripts from individual cells. Cells were divided into Gel Beads in Emulsion using Chromium, where cell lysis and barcoded reverse transcription of RNA occurred. Single-cell cDNA libraries were constructed using the 10× Genomics Chromium system. Sequencing was performed on an Illumina HiSeq X instrument using 150 bp paired-end reads.

Raw data processing and quality control

Raw sequencing reads were processed using Cell Ranger (v3.0.2). After removal of low-quality reads, the remaining reads from 10×Genomics libraries were aligned to the human reference genome GRCh38 (Ensembl v92), and gene expression matrices were generated using the Cell Ranger count pipeline. Additional filtering was performed to exclude low-quality cells, unassigned cells, and potential doublets. Cells with > 200 detected genes, > 1,000 UMIs, log10GenesPerUMI > 0.7, a mitochondrial gene UMI proportion < 10%, and a hemoglobin gene UMI proportion < 5% were retained as high-quality cells. Potential doublets were further identified and removed using DoubletFinder, and the remaining cells were used for downstream analyses.

Differentially expressed gene (DEG) analysis and cell type annotation

DEGs were identified using the FindMarkers function implemented in Seurat (v3.1.2) based on the Wilcoxon rank-sum test. Genes expressed in more than 10% of cells within a cluster and with an average log2 fold change > 0.25 were considered as DEGs. Cell types for each cluster were annotated using the SynEcoSys database based on the expression of canonical marker genes derived from the DEGs. The expression patterns of these marker genes were visualized using Seurat’s DoHeatmap, VlnPlot, and DotPlot functions.

Pseudotime trajectory analysis (Monocle2)

Pseudotime trajectory analysis was performed using Monocle2 (version 2.9.0) [23]. Raw UMI count matrices were converted into a CellDataSet (CDS) object according to the Monocle2 workflow. Ordering genes were identified using differentialGeneTest with a q-value < 0.01. Dimensionality reduction was performed using the reduceDimension function with the DDRTree method, followed by cell ordering using orderCells. Gene expression dynamics along pseudotime were visualized using plot_genes_in_pseudotime.

Cell-cell communication analysis (CellChat)

Cell-cell communication networks were analyzed using the CellChat R package (version 2.1.2) [24]. Normalized matrices were imported using createCellChat. Overexpressed genes and interactions were detected using identifyOverExpressedGenes, identifyOverExpressedInteractions, and projectData. Communication probabilities were computed using computeCommunProb, filtered with filterCommunication (min.cells = 10), and computeCommunProbPathway functions were then used to determine any potential ligand-receptor interactions. Finally, the cell-cell communication network was aggregated using the aggregateNet function.

Copy number variation (CNV) analysis

Copy number variation profiles were inferred using the inferCNV R package (version 1.0.4) [25]. CNV patterns were estimated from single-cell RNA-seq expression data using a gene expression cutoff of 0.1. Genes were ordered according to their chromosomal positions, and a moving average of gene expression was calculated using a window size of 101 genes. Expression values were subsequently centered to zero by subtracting the mean expression across reference cells. Epithelial cells with high CNV burden were defined as putative malignant cells, whereas all other cell types were used as reference normal cells. Denoising was performed to generate the final CNV profiles.

Pathway enrichment analysis

KEGG pathway enrichment analysis was conducted to explore the biological functions and signaling pathways associated with each cell type using the clusterProfiler R package. Pathways with an adjusted P value < 0.05 were considered significantly enriched. Gene set variation analysis (GSVA) was performed using hallmark gene sets, with average gene expression values calculated for each cell cluster as input data.

Cell culture

Human bladder cancer cell lines (T24, 5637, RT4, SW780, HT1376), the immortalized normal ureter epithelial cell line (SV-HUC-1), and HEK-293T cells were sourced from ATCC (Manassas, VA, USA). Cells were cultivated in DMEM (Gibco) supplemented with 10% FBS and 1% penicillin/streptomycin in an incubator with a 5% CO2 at 37 °C.

Constructs

To construct the pLK0.1-sh-DCLK1#1/2, sh-USP10#1/2, and sh-HDAC6#1/2 plasmids, shRNA sequences inserted into the pLKO.1-RFP vector. The coding sequences of human DCLK1, USP10, and HDAC6 were cloned into the pSin-EF2-puro vector. The PRK-HA-Ub plasmid was acquired as previously reported [26]. For stable DCLK1 knockdown in BC cells, lentiviral vectors carrying shRNA targeting DCLK1 or a control scrambled shRNA were synthesized. T24 and 5637 cells were incubated in six-well plates from NEST Biotechnology and transduced with viral infection at predetermined viral titers. After puromycin selection (1 µg/mL) for 1 week, the knockdown efficiency was confirmed using RT-qPCR and western blotting. For RNA isolation, total RNA was extracted using TRIzol reagent (Invitrogen). cDNA was synthesized using reverse transcriptase (RT) and random primers. cDNA amplification was performed using Platinum SYBR Green qPCR Super Mix-UDG (Invitrogen) on a CFX96 Touch system (Bio-Rad), with tubulin as the reference control. Relative gene expression was determined using the ^2−ΔΔCT method. Primers are listed in Table S1.

Western blotting

The proteins were extracted using RIPA lysis buffer (Beyotime, Shanghai, China), and the concentration was measured using a BCA kit (Beyotime). SDS-PAGE was performed, and proteins were transferred to PVDF membranes (Merck Millipore, MA, USA). After blocking with 5% nonfat milk, the membranes underwent overnight incubation at 4 °C in the presence of initial antibodies. DCLK1 (1:1000; Cell Signaling, 62257), HDAC6 (1:1000; Cell Signaling, 7558), USP10 (1:1000; Proteintech, 8501), E-cadherin (1:1000; Cell Signaling, 3195), Vimentin (1:1000; Cell Signaling, 5741), β-catenin (1:1000; Cell Signaling, 9562), Flag (1:1000; Sigma, F1804), HA (1:1000; Sigma, H6908), Myc-tag (1:1000; Proteintech, 16286-1-AP), α-tubulin (1:1000; Proteintech, 66031-1-Ig), ALDH1 (1:1000; Cell Signaling, 54135), CD133 (1:1000; Cell Signaling, 64326), CD44 (1:1000; Cell Signaling, 37259),SOX2 (1:1000; Cell Signaling, 23064),Nanog (1:2000; Cell Signaling, 4903), Notch1 (1:1000; Cell Signaling, 3608), c-Myc (1:1000; Cell Signaling, 13987), MAML1 (1:1000; Cell Signaling, 12166), HES1 (1:1000; Cell Signaling, 11988), HEY1 (1:1000; abcam, 154077), PD-L1 (1:1000; Cell Signaling, 13684), β-Actin (1:10000; Cell Signaling, 4967) and GAPDH (1:5000; Abcam, ab128915). Signals were detected using the ECL detection system (Thermo Fisher Scientific) after incubation with secondary antibodies.

Isolation of peripheral blood CD8⁺T cells

Venous blood (10 mL) was collected from fasting healthy donors, diluted with an equal volume of PBS, and layered over Ficoll (4550-OP, Sigma-Aldrich) for low-speed centrifugation. The white PBMC layer at the plasma interface was collected. CD8⁺T cells were isolated using a CD8⁺T cell sorting kit (11348D, Invitrogen). Briefly, 1 × 10⁷ PBMCs were centrifuged, resuspended in PBS, incubated with 10 µL biotin-labeled antibody at 4 °C for 5 min, followed by 30 µL PBS and 20 µL CD8⁺ microbeads for 10 min. Unlabeled cells passing through the magnetic column were collected as CD8⁺ T cells. Following Cui et al. [27], BC cells transfected with different plasmids for 24 h were co-cultured with CD8⁺T cells at a 1:3 ratio in Transwell chambers for 48 h for subsequent experiments.

Carboxyfluorescein diacetate succinimidyl ester (CFSE) staining

Following Wang et al. [28], CD8⁺ T cells were labeled by adding 1µL CFSE probe (C1031, Beyotime) per milliliter of cell suspension and incubating at 37 °C for 10 min, achieving a final CFSE concentration of 5 µM. Staining was stopped by adding five volumes of complete medium, mixing, then cells were centrifuged and washed twice with PBS to remove excess CFSE. A small sample was checked under a fluorescence microscope to confirm labeling. Labeled CD8⁺ T cells were co-cultured with transfected BC cells for 48 h. CFSE intensity was measured by flow cytometry (BD FACSCalibur™, BD Biosciences) and analyzed using FlowJo software (v10.8).

ELISA

Human Interferon-γ (IFN-γ) levels in CD8⁺ T cell culture supernatants were measured using the Beyotime ELISA kit (PI511). Samples were added to ELISA wells and incubated for 2 h, followed by incubation with specific antibodies for 1 h. After washing three times, streptavidin-HRP was added for 30 min. TMB substrate was then applied for 10 min before adding 50 µL stop solution. The absorbance at 450 nm was measured to calculate IFN-γ concentration.

Immunofluorescence (IF) staining

Immunofluorescence staining was performed as described [29].Cells grown on glass coverslips were fixed with methanol (for 5 min). After blocking with 5% BSA, the cells were incubated overnight at 4 °C with the indicated primary antibodies. After washing, the cells were incubated with fluorescent dye-conjugated secondary antibodies (Alexa Fluor 488, 594, FITC, or Cy5) for 2 h at room temperature in the dark. Nuclei were counterstained with DAPI or Hoechst 33,342. Fluorescence images were captured using a confocal laser scanning microscope(Olympus FV1000, Tokyo, Japan).

Tumorsphere formation assay

T24 and 5637 cells with different treatments were seeded in ultra-low attachment dishes (Corning) at the density of 5,000 cells per well and cultured in serum- free DMEM/F12 medium (Invitrogen) supplemented with factors as previously described [30]. About 2 weeks later, images were taken at 4 × magnification and counted.

Cell viability assay

BC cells were seeded into 96-well plates at a density of 3 × 10³ cells per well and allowed to adhere overnight. For drug sensitivity assays, cells were treated with cisplatin (0.15–10 µg/mL) for 48 h. Cell viability was assessed using the Cell Counting Kit-8 (CCK-8) assay (TargetMol) by adding 10 µL CCK-8 solution to each well, incubating for 2 h at 37 °C, and reading absorbance at 450 nm. For T-cell co-culture experiment, cell viability was evaluated using the MTT assay. Briefly, 20 µL MTT solution (5 mg/mL in PBS) was added to each well 4 h before the end of incubation. The medium was then removed, and 150 µL DMSO was added to dissolve the formazan crystals. Absorbance was measured at 570 nm.

Transwell assay

For the Transwell assay, 5 × 10⁴ cells (migration) or 1 × 10⁵ cells (invasion) were seeded in 100 µL of serum-free medium in Transwell chambers (Corning, NY, USA). The membrane was either precoated with Matrigel (invasion) or left uncoated (migration). The lower chambers contained additional medium. After incubation (12 h for migration and 24 h for invasion), cells were fixed, stained, and the number of migrated or invaded BC cells was counted.

Flow cytometric apoptosis assay

For apoptosis analysis, cells were treated as indicated and subsequently harvested and washed with PBS. Cells were then resuspended in 500 µL of 1× binding buffer and incubated with 5 µL Annexin V/FITC and 5 µL propidium iodide (PI) for 15 min at room temperature in the dark. Apoptosis was analyzed using a CytoFLEX flow cytometer and quantified with CytExpert 2.2 software, with FITC+/PI- and FITC+/PI + cells classified as apoptotic.

Co-IP and MS

Cells were lysed using Pierce IP Lysis Buffer (Thermo Fisher Scientific) supplemented with protease inhibitor mixture. Immunoprecipitation was performed overnight at 4 °C with either anti-FLAG antibody (3 µg; Sigma, F1804) or IgG (3 µg; Invitrogen, 10500 C). Protein A/G magnetic beads (Thermo Scientific) were incorporated into the immune complexes, which were then washed and analyzed by western blot or mass spectrometry as described in a prior study [31].

Denaturing IP assay

For the denaturing ubiquitination assay as outlined before [32], BC cells were transfected and treated with 10 µM MG132 for 6 h. After 24 h, cells were lysed in buffer containing an EDTA-free protease inhibitor cocktail (Roche). The lysates were heated at 95 °C for 5 min in the presence of 1% SDS to denature the proteins, followed by dilution with lysis buffer to lower the SDS concentration to < 0.1%. Immunoprecipitation was performed using the indicated antibodies, and ubiquitination levels were assessed by Co-IP with anti-Flag antibody and western blotting with anti-HA antibody.

In vitro HDAC6 ubiquitination assay

His-tagged HDAC6 was expressed in E. coli BL21 cells and purified using Ni-NTA agarose beads. In addition, Flag-tagged RNF168, Flag-DCLK1, and Flag-USP10 were purified from HEK293T cells. For the in vitro ubiquitination reaction, purified His-HDAC6 (bound to Ni-NTA beads) was incubated with Flag-RNF168 together with 200 ng of E1 (UBE1; R&D Systems), 400 ng of E2 (UBE2D3; R&D Systems), and 2 µg of recombinant ubiquitin (R&D Systems) in reaction buffer (50 mM Tris-HCl [pH 7.4], 2 mM MgCl₂, and 4 mM ATP [Sigma]) at 37 °C for 1 h. Where indicated, Flag-DCLK1 or Flag-USP10 was added to assess its regulatory effect on HDAC6 ubiquitination. After incubation, the beads were washed, the supernatant was removed, and the reaction was terminated by adding 2× protein loading buffer. The ubiquitinated HDAC6 were detected by SDS-PAGE followed by immunoblotting using anti-ubiquitin and anti-His antibodies.

In vivo mouse models

Five-week-old female BALB/c nude mice were obtained from the Hunan Sleek Jingda Laboratory Animal Co., LTD. Guizhou Provincial People’s Hospital’s Experimental Animal Care and Use Committee approved the animal experiments.

For the subcutaneous tumor formation, 2 × 10⁶ stably transfected DCLK1-knockdown or control T24 cells were mixed with 20% Matrigel and injected subcutaneously into the dorsal flanks of nude mice. Once tumors reached approximately 100 mm³, mice were randomly divided into two subgroups (n = 4 per group) and treated intraperitoneally with either cisplatin (4 mg/kg), ACY-1215 (50 mg/kg).

or saline every 3 days. Tumor size was measured every 3 days using the formula V = 1/2 × length × width². After 28 days, tumors were excised, and tumor growth inhibition (TGI) was calculated using the formula: TGI (%) = (Vc − Vt) / (Vc − V0) × 100.

For the in situ model of bladder cancer, T24 cells (2 × 10⁶) stably transduced with sh-Ctrl or sh-DCLK1 were injected into the bladder tissues of mice. Meanwhile, cisplatin (4 mg/kg) or normal saline was intraperitoneally injected every 3 days. After 35 days, thebladder tumors were collected and subjected to further analysis.

In the lung metastasis assay, 2 × 10⁶ T24 cells that had been genetically modified to express either sh-Ctrl or sh-DCLK1 were injected into mice via the tail vein. Cisplatin (4 mg/kg) or normal saline was intraperitoneally injected every 3 days. After 8 weeks, the mice were sacrificed, and lung specimens were harvested for the evaluation of metastatic nodules. Paraffin-embedded tissue samples were sectioned and stained with hematoxylin and eosin.

CT image analysis and acquiring data

The mice were intraperitoneally injected with iodixanol injection (370 mg I/ml) at a dose of 200 µl per mouse. CT scanning was performed 15 min after injection. The mice were placed into an anesthesia chamber, where they were pre-anesthetized with 3% isoflurane at an air flow rate of 1 L/min. After pre-anesthesia, the mice were placed in the scanning bed of the MOLECUBES x-cube system. The tumor-bearing mice were fixed in a prone position on the scanning bed and continuously anesthetized with 1.5% isoflurane at an air flow rate of 1 L/min. Physiological monitoring was initiated during imaging to monitor the life signs of the tumor-bearing mice, and the anesthetic dosage (isoflurane) was adjusted as necessary.

The MOLECUBES x-cube system’s acquisition workstation is MOLECUBES Acquisition/ Reconstruction. A new workflow was created before data acquisition, which included CT acquisition. The parameters for MOLECUBES x-cube acquisition were as follows: tube voltage: 50 kV; tube current: 30 µA; exposure time: 3 min. The data collected by the MOLECUBES x-cube system were reconstructed into voxels with a resolution of 200 μm using an iterative image reconstruction algorithm. The reconstructed images were corrected for attenuation and analyzed.

IHC staining

Immunohistochemical staining was performed as previously described [33].Paraffin-embedded BC tissue sections were deparaffinized, rehydrated, and treated to block endogenous peroxidase activity. After preventing non-specific protein binding, sections were incubated overnight at 4 °C with primary antibodies. The primary antibodies were then detected using HRP-conjugated secondary antibodies. Haematoxylin was used to stain the cell nuclei. IHC staining was visualized and analyzed using the AxioVision Rel.4.6 image analysis system (Carl Zeiss).

Statistical analysis

Statistical analyses were performed using GraphPad Prism 9.0 and R 4.2.0. Data are presented as mean ± SD from at least three independent experiments. Differences between two groups were analyzed using unpaired two-tailed Student’s t-tests, while multiple-group comparisons were evaluated using one-way or two-way ANOVA with Tukey’s post-hoc test. Time-course experiments were analyzed using two-way repeated-measures ANOVA. The P-value of < 0.05 denoted statistical significance.

Results

Tumor ecosystem of bladder cancer characterized by single-cell transcriptomic sequencing

To decipher the cellular architecture within the tumor microenvironment in metastases of bladder cancer, we performed single-cell RNA sequencing of patient-matched tissue samples, including normal tissue (N), para-cancer (BPa), and paired carcinoma tissues with lymph node metastasis (BCa) (Fig. 1A). Following strict quality control filters and doublet elimination, 147,969 cells were retained for downstream analysis. On average, we identified 2978 genes and 7905 unique molecular identifiers (UMIs) in each cell. The Seurat tool was then used to perform unsupervised clustering analysis, which defined significant groupings of cells possessing comparable expression profiles. We identified nine main cell populations: plasma cells (N = 1291), endothelial cells (N = 1859), B cells (N = 2272), mast cells (N = 3059), neutrophil cells (N = 6771), mono_macrophage_DC (N = 8356), T_NK cells (N = 18381), interstitial cells (N = 30142), and cancer cells (N = 65544) (Fig. 1B). According to the expression of canonical markers and variable genes, a specific cell subpopulation was confirmed for each cluster: T_NK cells (gene markers: PTPRC, CD3D, NKG7 and IFNG), B cells (gene markers: CD79A, MS4A1 and CD19), plasma cells (gene markers: JCHAIN, XBP1 and CD38),mono_macrophage_DC (gene markers: CD68, AIF1,CD300E and CD1C ), neutrophil cells (gene markers: CXCR2, SORL1 and CSF3R), mast cells (gene markers: TPSB2,CPA3 and KIT), interstitial cells (gene markers: COL1A1,DCN and MYH11), endothelial cells (gene markers: VWF, CDH5 and FLT4) and cancer cells (gene markers: CDH1 and KRT18). Dot plots show the scaled expression levels and proportions of cells expressing cluster-specific markers in each cell subsets (Fig. 1C, D).The markers and rates of each cell subtype are shown in histograms (Fig. 1E).

Fig. 1.

Single-cell transcriptomic landscape of the bladder cancer tumor ecosystem. A Schematic overview of the experimental workflow, including tissue dissociation, single-cell suspension preparation, and scRNA-seq using the 10x Genomics platform. B UMAP visualization of single cells colored by major cell types. C Feature plots showing the expression of canonical marker genes across different cell populations. D Dot plot illustrating the expression levels and proportions of key marker genes in each major cell type. E Relative proportions of different cell types in normal tissues (N), adjacent tissues (BPa), and paired lymph node metastatic bladder cancer tissues (BCa)

Identification of common expression programs of cancer cells in bladder cancer

To characterize tumor heterogeneity, we analyzed transcriptomic patterns and identified distinct cancer cell subpopulations. The InferCNV method was used to distinguish neoplastic cells from normal epithelial cells. A total of 65,544 malignant epithelial cells in bladder cancer tumors were identified.Differential gene expression analysis revealed nine major cancer cell subclusters (Supplementary Fig. 1A), with subtype-specific markers visualized in different tissue types (Supplementary Fig. 1B) and their expression patterns quantified in dot plots (Supplementary Fig. 1C). Histograms display the markers and proportions in different tissue types (Supplementary Fig. 1D). To investigate shared expression patterns among cancer cells, we developed a meta-clustering approach. Tumor subclusters were analyzed using hierarchical clustering to compare transcriptional programs. This revealed six functionally distinct expression modules, including proliferation sustaining, metastasis activation, immune evasion, stress resistance, metabolic reprogramming, and inflammation promotion, the distribution of each expression program is displayed in a reduced-dimension map (Supplementary Fig. 1E), proliferation (e.g., PSMA1, TAZ, CD36, GRB2, USP10, ZNF750 and MMP12). Metastasis (e.g., DCLK1, MMP9, MMP2 and PDGFRB).Immune evasion (e.g., CRH, HLA-DRB5, CD74, STAT1, CD47 and CD274).Stress resistance (e.g., FGFR4, IGF1R, PVT1, ERBB3 and CDK12). Metabolic reprogramming (e.g., SAA1, SAA2, SLC3A2, ANXA1 and FABP5). Inflammation promotion (e.g., CCL2, CCL21, CXCL12, CXCR3, FCGBP and FCGR2B). We assessed the program scores based on the fraction of matched program cells to evaluate the activity of each sample (Supplementary Fig. 1F). Building on the established utility of CNV analysis in scRNA-seq for studying disease progression [34], We evaluated the CNV level of all cancer cells and identified significant copy number alterations in cancer cell populations. The inferred CNV profiles revealed pronounced inter-lesional and intra-lesional heterogeneity within bladder cancer tissues (Supplementary Fig. 1G, H). To further characterize tumor evolution, heatmaps were created to visualize the marker genes of cancer cells (Supplementary Fig. 2A). Trajectory analysis was subsequently performed using the Monocle 2 algorithm to reconstruct the developmental progression of cancer cells (Supplementary Fig. 2B). Dynamic gene expression patterns along tumor progression trajectories revealed four functionally distinct modules (Supplementary Fig. 2C, D). Module 1 was characterized by genes associated with cell proliferation and migration, reflecting core oncogenic transcriptional programs. Module 2 was enriched for pathways related to focal adhesion, lysosomal function, and actin cytoskeleton organization, suggesting enhanced migratory and invasive capacity of tumor cells. Module 3 was associated with regulation of cell motility, epithelial development, and cell migration, indicative of epithelial remodeling during tumor progression. Module 4 showed enrichment of the PI3K–Akt signaling pathway, ubiquitin-mediated proteolysis, and protein processing in the endoplasmic reticulum, representing aberrant activation of oncogenic signaling networks (Supplementary Fig. 2E). KEGG pathway enrichment analysis of the top 10 differentially expressed genes revealed distinct functional patterns across groups (Supplementary Fig. 2F). The N group was enriched in lipid (linoleic/alpha-linolenic acid, ether lipids) and vitamin (retinol/ascorbate) metabolism, along with hormone biosynthesis. The BPa group was enriched in core metabolic pathways (TCA cycle, fatty acid synthesis, insulin signaling) and translational regulation. The BCa group exhibited pathways related to drug metabolism, lipid regulation, ROS, and angiogenesis. These functions are linked to tumor malignancy traits including proliferation, metastasis, drug resistance, and immune evasion. These findings highlight the role of tumor cells in cancer colonization and the development of an immunosuppressive microenvironment during bladder cancer progression.

The landscape of T_NK cells in bladder cancer

To further delineate the immunological landscape of bladder cancer, T and NK cells were reclustered to identify distinct subpopulations. Five CD4⁺ T cell and five CD8⁺ T cell subsets were identified across all patients. CD4⁺ T cell populations included naïve/central memory T cells (CD4_T_c1_CCR7), effector memory–like T cells (CD4_T_c2_RUNX2 and CD4_T_c3_MCAM), and regulatory T cells (CD4_T_c4_FOXP3). CD8⁺ T cell subsets comprised cytotoxic effector T cells (CD8_T_c1_GZMB), GZMK-expressing effector memory–like T cells (CD8_T_c2_GZMK), tissue-associated T cells (CD8_T_c3_GPR15), FGFBP2⁺ highly cytotoxic T cells (CD8_T_c4_FGFBP2), and CCL3-expressing inflammatory T cells (CD8_T_c5_CCL3). In addition, NK cells, proliferating T cells, and γδ T cells were identified based on canonical marker expression (Supplementary Fig. 3A). with subtype-specific markers visualized in different tissue types (Supplementary Fig. 3B). Dot plots show the relative expression levels and proportions of cluster-specific markers in each cell population (Supplementary Fig. 3C), while histograms display the proportion and markers of each cell subtype (Supplementary Fig. 3D). Trajectory analysis of T/NK cells was performed using the Monocle 2 algorithm, revealing four transcriptional modules distributed along the developmental trajectory (Supplementary Fig. 3E, F). Module 1 showed Ribosome biogenesis, RNA transport, DNA replication. Module 2 was associated with phagosome formation, apoptosis, and endocytosis, suggesting involvement in cytotoxic and clearance-related processes. Module 3 was enriched in antigen processing and presentation, nucleotide excision repair, and immune surveillance–related pathways. Module 4 encompassed TNF signaling, JAK–STAT signaling, cell adhesion, and cytokine–cytokine receptor interactions, reflecting activated immune response programs in T/NK cells (Supplementary Fig. 3G). The heatmap showed the top 10 most highly expressed genes of T_NK cells (Supplementary Fig. 3H). KEGG enrichment analysis revealed that the N cluster was primarily enriched in processes including amino acid metabolism (glycine, serine, threonine), lipid and carbohydrate metabolism, signal transduction, and disease-related mechanisms. The BPa cluster showed enrichment in cytokine-cytokine receptor interactions, folate biosynthesis, glycosphingolipid biosynthesis, histidine metabolism, glycerolipid metabolism, and aminoacyl-tRNA biosynthesis. The BCa cluster was enriched in pathways related to IL-17, NF-kappa B signaling, interferon-gamma response, antigen processing and presentation, drug metabolism, and RIG-I-like receptor signaling (Supplementary Fig. 3I).

Characterization of tumor-associated myeloid cells in bladder cancer

Myeloid cells in bladder cancer tissues comprised tumor-associated macrophages (TAMs), monocytes, neutrophils, dendritic cells (DCs), mast cells, and B cells. Differential gene expression analysis identified multiple myeloid cell subclusters (Supplementary Fig. 4A), with subtype-specific markers visualized in different tissue types (Supplementary Fig. 4B).Dot plots and histograms illustrate the relative expression levels and proportions of each macrophage subtype (Supplementary Fig. 4C, D). Macrophage-specific markers, such as APOC1, EGFL7, TGM2, NRG1, FOLR3, DTNA, and LYPD2, were differentially expressed among TAM subsets and are potentially associated with macrophage functions related to lipid metabolism, angiogenesis, tissue remodeling, and inflammatory regulation.Trajectory analysis of TAMs was performed using the Monocle 2 algorithm, revealing four transcriptional modules distributed along the pseudotime trajectory (Supplementary Fig. 4E, F). Module 1 was enriched in spliceosome, ribosome, and proteasome pathways, reflecting enhanced transcriptional and translational activity in macrophages. Module 2 was associated with thermogenesis, antigen processing and presentation, and chemokine signaling pathways, suggesting metabolically active and immune-responsive macrophage states. Module 3 was enriched in arachidonic acid, fatty acid, and inositol phosphate metabolism, indicating regulation of lipid-derived inflammatory mediators and intracellular signaling. Module 4 encompassed TGF-β, TNF, and MAPK signaling pathways, which are implicated in macrophage polarization, inflammatory responses, and immunomodulatory functions (Supplementary Fig. 4G). A heatmap displaying the top 10 most highly expressed genes of myeloid cells (Supplementary Fig. 4H). KEGG analysis of each tissue type revealed distinct enrichments. The N cluster was enriched in bile acid biosynthesis, steroid hormone biosynthesis, tryptophan metabolism, nitrogen metabolism, backbone biosynthesis, butanoate metabolism, and tyrosine metabolism. The BPa cluster was associated with retinol metabolism, pentose and glucuronate interconversions, taurine and hypotaurine metabolism, the renin-angiotensin system, ascorbate and aldarate metabolism, folate biosynthesis, and glycosaminoglycan degradation. The BCa cluster was enriched in mTOR signaling pathway, basal cell carcinoma, interferon-gamma response, histidine, arginine and proline metabolism, glycolysis, and the unfolded protein response. These functions are closely linked to the pro-inflammatory response and macrophage metabolism (Supplementary Fig. 4I).

Diversity of cancer-associated fibroblasts (CAFs), smooth muscle cells (SMCs)in the tumor microenvironment

Based on highly variable genes, cancer-associated fibroblasts (CAFs) were subdivided into five distinct subpopulations, including CAFs_c1_SLC66A1L, CAFs_c2_WNT2, CAFs_c3_FGF14, CAFs_c4_NRG1, and CAFs_c5_PTGDS. In parallel, smooth muscle cells (SMCs) were classified into four mural cell subtypes: Mural_cells_c1_KRT19, Mural_cells_c2_NALT1, Mural_cells_c3_RBFOX3, and Mural_cells_c4_S100B. The distribution of CAF and SMC subtypes is shown in feature plots (Supplementary Fig. 5A), with subtype-specific marker expression visualized across different tissue types (Supplementary Fig. 5B), Dot plots and histograms summarize the scaled expression levels and relative proportions of cluster-specific markers (Supplementary Fig. 5C, D). A heatmap displaying the top 10 highly expressed genes across CAF subpopulations (Supplementary Fig. 5E). KEGG enrichment analysis revealed that The N group was enriched in pathways related to antigen processing and presentation, extracellular matrix–receptor interaction, retinol metabolism, ribosome biogenesis, and tyrosine metabolism, suggesting active stromal remodeling and immune-related functions. The BPa cluster was olfactory transduction, lysosome, taste transduction, cell adhesion molecules, steroid hormone biosynthesis, and primary bile acid biosynthesis. The BCa cluster showed enrichment in pyruvate metabolism, the citrate cycle, antigen receptor-mediated signaling, immune response activation, VEGF, FOXO, and calcium signaling pathways. These functions highlight the role of fibroblasts in ECM synthesis, immunomodulatory remodeling, energy metabolism, and growth factor signaling (Supplementary Fig. 5F).

DCLK1 is highly expressed in bladder cancer and promotes cancer cell stemness

Our previous study demonstrated that DCLK1 is preferentially enriched in malignant epithelial cells, suggesting a potential role in bladder cancer development. However, the functional significance of.

DCLK1 in tumor metastasis remains unexplored, and its precise mechanisms in BC pathogenesis remain to be elucidated. To investigate the function of DCLK1 in BC progression, we analyzed the expression level of DCLK1 in BC tissues. Compared with normal tissues, DCLK1 was highly expressed in BC tissues (Fig. 2A). We assessed the clinical relevance of DCLK1 in BC through IHC staining of DCLK1 in 150 tumor samples, which were divided into negative, weak, moderate, and strong groups according to staining intensity (Supplementary Fig. 6A). Moreover, correlation analysis showed that high DCLK1 expression was markedly associated with distant metastasis (Supplementary Fig. 6B). Furthermore, DCLK1 expression was markedly elevated in BC cells relative to SV-HUC-1 cells (Fig. 2B), High DCLK1 expression was positively correlated with tumor metastasis and histological grade (Table S2). qRT-PCR analysis revealed a decrease in DCLK1 mRNA levels in BC cells following DCLK1 knockdown, whereas ectopic overexpression of DCLK1 led to increased DCLK1 expression (Fig. 2C, Supplementary Fig. 6C). Sphere formation assays, which assess the self-renewal capacity of stem-like cells, revealed that DCLK1 interference markedly reduced the size of BC cell spheres compared to the control group (Fig. 2D). This was accompanied by a significant downregulation of stemness-related genes, including ALDH1, CD133, CD44, Nanog, and SOX2 (Fig. 2E), as well as a decreased proportion of ALDH1⁺ and CD44⁺ CSC populations (Fig. 2F). Further KEGG enrichment analysis of differentially expressed genes between DCLK1-high and DCLK1-low cells in the single-cell transcriptome revealed significant enrichment of drug resistance-related pathways (Supplementary Fig. 6D, E).Relevant research reports cisplatin-based chemotherapy has become an important adjuvant treatment for BC patients. Notably, a subset of patients develop tumor recurrence or metastasis following cisplatin treatment. To explore the role of DCLK1 in cisplatin resistance, we treated the DCLK1 overexpressed and knocked-down BC cell lines with cisplatin, CCK-8 assays showed a significant reduction in the IC50 value in the DCLK1 knockdown group compared to controls (Supplementary Fig. 6F). Flow cytometric analysis showed that DCLK1 knockdown significantly increased the proportion of apoptotic cells following cisplatin treatment (2.5 µg/mL for 24 h) (Fig. 2G). TUNEL assays showed a significant rise in TUNEL-positive cells in the DCLK1 knockdown group after cisplatin treatment (10 µg/mL for 24 h) (Fig. 2H). Conversely, DCLK1 upregulation decreased cisplatin sensitivity, reduced apoptosis, and fewer TUNEL-positive cells (Supplementary Fig. 6G, H). These findings demonstrate that DCLK1 contributes to enhanced stemness and cisplatin resistance in bladder cancer cells.

Fig. 2.

DCLK1 knockdown suppresses stemness and enhances chemosensitivity in bladder cancer cells. A mRNA expression levels of DCLK1 in bladder cancer tissues compared with normal tissues. B qRT-PCR analysis of DCLK1 expression in BC cell lines and SV-HUC-1 cells. C qRT-PCR analysis confirming the knockdown efficiency of DCLK1. D Representative images of tumorsphere formation in DCLK1-depleted BC cells (scale bar = 100 μm). E Western blot analysis of stemness-associated markers (ALDH1, CD133, CD44, Nanog, and SOX2) in DCLK1-silenced BC cells. F The proportions of ALDH1⁺ and CD44⁺ cells were significantly higher in the control group than in the DCLK1-knockdown group under cancer stem cell-enriched conditions. G Cisplatin-induced apoptosis in BC cells analyzed by Annexin V/PI staining followed by flow cytometry. H Cisplatin-induced apoptosis in BC cells examined by the TUNEL assay (scale bar = 20 μm). Data are presented as mean ± SD. **P < 0.01, ***P < 0.001

DCLK1 facilitates bladder cancer cell migration, invasion and epithelial-mesenchymal transition (EMT)

To evaluate the role of DCLK1 on BC cell metastasis, we conducted transwell assays in DCLK1-knockdown BC cells. As anticipated, DCLK1 knockdown decreased cell migration and invasion (Supplementary Fig. 6I), while DCLK1 overexpression enhanced these behaviors (Supplementary Fig. 6J). Given the critical role of EMT in tumor metastasis, we further investigated the effect of DCLK1 on EMT in bladder cancer cells.DCLK1 knockdown resulted in increased expression of the epithelial marker E-cadherin and decreased expression of the mesenchymal marker Vimentin (Supplementary Fig. 6K, L), while DCLK1 overexpression produced the opposite effects (Supplementary Fig. 6M, N). These findings were further confirmed by immunofluorescence staining. Collectively, these results indicate that DCLK1 promotes bladder cancer cell migration, invasion, and EMT.

DCLK1 upregulates PD-L1 through notch signaling pathway to mediate CD8 + T cell activity

Single-cell analysis indicated that the immunosuppressive microenvironment in bladder cancer may influence immune cell activity within the tumor niche. A detailed investigation was carried out to explore the interplay between DCLK1 expression in tumor cells and the immune response. KEGG pathway analysis revealed significant enrichment of the Notch signaling pathway and the T cell receptor signaling pathway (Supplementary Fig. 6E), suggesting a potential role for DCLK1 in modulating the tumor immune microenvironment. Given the critical role of CD8⁺ T cells in antitumor immunity [35, 36], we hypothesize that DCLK1 may enhance PD-L1 expression via the Notch pathway, thereby suppressing CD8⁺ T cell activity, promoting immune evasion, and driving tumor progression.

To assess the functional consequences of DCLK1 expression in BC cells on CD8⁺ T cells, a 48-hour co-culture system was established (Fig. 3A). CFSE staining demonstrated that CD8⁺ T cells cultured with DCLK1-overexpressing BC cells showed a significant decrease in CFSE-positive populations, indicating impaired proliferation. Conversely, silencing DCLK1 led to a marked increase in CFSE-positive CD8⁺ T cells, reflecting enhanced proliferative capacity (Fig. 3B). Apoptosis analysis further revealed that CD8⁺ T cell apoptosis was significantly elevated following co-culture with DCLK1-overexpressing BC cells, while DCLK1 knockdown resulted in reduced apoptosis (Fig. 3C), suggesting that high DCLK1 expression compromises CD8⁺ T cell viability. Moreover, DCLK1 overexpression decreased IFN-γ and IL-2 secretion and was associated with increased BC cell viability, whereas DCLK1 knockdown exerted opposite effects. (Fig. 3D, F). Together, these results indicate that DCLK1 overexpression suppresses CD8⁺ T cell function, facilitating immune escape and supporting tumor progression in bladder cancer.

Fig. 3.

DCLK1 upregulates PD-L1 through the Notch signaling pathway to mediate CD8 + T cell activity. A CD8⁺ T cells were isolated from peripheral blood and co-cultured with bladder cancer cells. B CFSE staining demonstrated that CD8⁺ T cell proliferation was significantly reduced following DCLK1 overexpression, while knockdown of DCLK1 markedly enhanced their proliferative capacity. C Flow cytometry analysis revealed an increased apoptosis rate in CD8⁺ T cells co-cultured with DCLK1-overexpressing cancer cells, whereas DCLK1 knockdown led to decreased apoptosis. D, E ELISA assays were used to measure IFN-γ and IL-2 levels secreted by CD8⁺ T cells, showing reduced cytokine secretion with DCLK1 overexpression and elevated levels after DCLK1 knockdown. F MTT assays assessed the proliferation of bladder cancer cells after 48 h of co-culture with CD8⁺ T cells. G TCGA analysis revealed a positive correlation between DCLK1 and PD-L1 expression. H Immunohistochemistry demonstrated higher levels of DCLK1 and PD-L1 in bladder cancer tissues compared to adjacent normal tissues. Scale bar: 100 μm. I qRT-PCR analysis (n = 35) showed that PD-L1 mRNA expression was significantly lower in normal tissues than in bladder cancer tissues. J Western blotting indicated that DCLK1 knockdown reduced PD-L1 protein levels, while DCLK1 overexpression increased them. K PD-L1 knockdown in BC cells was confirmed by qRT-PCR following sh-PD-L1 transfection. L CFSE staining showed enhanced proliferation of CD8⁺ T cells after co-transfection with DCLK1 and sh-PD-L1. M Flow cytometry measured CD8⁺ T cell apoptosis after 48 h of co-culture. N ELISA assessed secretion levels of IFN-γ and IL-2 from CD8⁺ T cells. MTT assays evaluated BC cell proliferation after 48 h of co-culture with CD8⁺ T cells. O oe-DCLK1 or sh-DCLK1 was transfected in BC cells, and western blot measured the levels of Notch1, c-Myc, MAML1, HES1, and HEY1. P BC cells treated with the Notch inhibitor Crenigacestat for 2 h followed by DCLK1 transfection showed reduced levels of PD-L1 and Notch pathway proteins by Western blot. Results are shown as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001

Next, TCGA database analysis showed a positive correlation between DCLK1 and PD-L1 expression (Fig. 3G). Immunohistochemistry of bladder cancer tissues further confirmed a significant positive association between DCLK1 and PD-L1 protein levels (Fig. 3H). PD-L1 mRNA expression was markedly higher in tumor tissues compared to adjacent normal tissues, mirroring the expression pattern of DCLK1 (Fig. 3I). Western blot analysis demonstrated that DCLK1 overexpression increased PD-L1 levels, while DCLK1 knockdown had the opposite effect (Fig. 3J). Additionally, sh-PD-L1 transfection significantly reduced PD-L1 expression in BC cells (Fig. 3K).

To investigate whether DCLK1 modulates CD8⁺ T cell function via PD-L1, BC cells were transfected with Vector, DCLK1, or DCLK1 plus sh-PD-L1 and co-cultured with CD8⁺ T cells. Co-transfection with sh-PD-L1 reversed the immunosuppressive effects of DCLK1 overexpression increasing CFSE-positive T cells (Fig. 3L), reducing apoptosis (Fig. 3M), and enhancing IFN-γ and IL-2 secretion, while decreasing BC cell viability (Fig. 3N). These results suggest that DCLK1 impairs CD8⁺ T cell activity through PD-L1 upregulation. Further analysis showed that DCLK1 overexpression activated the Notch signaling pathway, increasing the expression of Notch1, c-Myc, MAML1, HES1, and HEY1, whereas DCLK1 knockdown suppressed these markers (Fig. 3O). Treatment with the Notch inhibitor Crenigacestat reversed the effects of DCLK1 overexpression, reducing both Notch-related proteins and PD-L1 expression (Fig. 3P). Collectively, these findings demonstrate that DCLK1 promotes immune evasion by upregulating PD-L1 via the Notch signaling pathway.

Knockdown of DCLK1 facilitates the degradation of HDAC6 by increasing its K48-linked polyubiquitination

To investigate the mechanisms of DCLK1 in BC progression, we performed co-immunoprecipitation followed by mass spectrometry (IP-MS) to identify potential DCLK1 targets (Fig. 4A), with HDAC6 being the highest-ranked protein in the cluster (Fig. 4B, Table S3). Co-IP assays confirmed that both exogenous and endogenous DCLK1 interact with HDAC6 in BC cells (Fig. 4C). This interaction was further validated by the colocalization of DCLK1 and HDAC6 in the cytoplasm shown by IF staining, We generated a predicted three-dimensional interaction model of the DCLK1–HDAC6 complex based on molecular docking analysis using PyMOL (Fig. 4D). Knockdown of DCLK1 significantly reduced HDAC6 protein levels, while DCLK1 overexpression increased HDAC6 expression (Fig. 4E, Supplementary Fig. 7A). However, mRNA levels of HDAC6 remained unchanged with either DCLK1 knockdown or overexpression (Fig. 4F, Supplementary Fig. 7B). Additionally, after cells were treated with cycloheximide (CHX), DCLK1 knockdown facilitated the degradation of endogenous HDAC6 (Fig. 4G, H), indicating that DCLK1 prolongs the half-life of the HDAC6 protein.These results show that DCLK1 inhibits HDAC6 protein degradation in BC cells. To investigate whether DCLK1 is involved in the degradation of HDAC6 via the ubiquitin-proteasome or autophagy-lysosomal pathway, BC cells were treated with a proteasome inhibitor (MG132) or a lysosome inhibitor (chloroquine, CQ). MG132, but not CQ, reversed the reduction in HDAC6 protein levels caused by DCLK1 knockdown (Fig. 4I), indicating that DCLK1 regulates HDAC6 degradation through the ubiquitin-proteasome pathway. Ubiquitination modification is usually required before the target protein is recognized by the proteasome. Therefore, we first demonstrated through in vivo ubiquitination experiments in HEK-293T cells that overexpression of DCLK1 could significantly reduce the polyubiquitination level on the HDAC6 protein (Fig. 4J). Furthermore, it has also been demonstrated in BC cells that knockdown of DCLK1 leads to an enhanced polyubiquitination level of HDAC6 protein, while overexpression of DCLK1 results in a weakened polyubiquitination level of HDAC6 protein (Fig. 4K, Supplementary Fig. 7C). To further determine the type of polyubiquitination of HDAC6 protein regulated by DCLK1, we screened different types of Ub in HEK293T cells through in vivo ubiquitination experiments, including: WT, K6, K11, K27, K29, K33, K48 and K63. The results showed that DCLK1 mainly weakened the K48-type polyubiquitination chain on the HDAC6 protein, and this type of polyubiquitination mainly regulated the recognition of the substrate protein by the proteasome, which was consistent with the previous results (Fig. 4L). These findings suggest that DCLK1 stabilizes HDAC6 by inhibiting its ubiquitination.

Fig. 4.

DCLK1 knockdown promotes HDAC6 degradation by enhancing K48-linked polyubiquitination. A T24 cells transfected with Flag-DCLK1 or empty vector were subjected to immunoprecipitation using an anti-Flag antibody, followed by SDS–PAGE and silver staining. Proteins in the indicated bands were analyzed by mass spectrometry (MS). B Mass spectrometry analysis identified HDAC6 as a potential binding partner of DCLK1 (top), with representative IP–MS spectra of HDAC6 shown (bottom). C Immunoprecipitation (IP) using anti-Flag or anti-HDAC6 antibodies, followed by immunoblotting (IB) for HDAC6, Flag, and DCLK1, was performed in HEK-293T, T24, and 5637 cells transfected with or without Flag-DCLK1. D Confocal microscopy images showing the colocalization of DCLK1 and HDAC6 in T24 and 5637 cells (top). Predicted three-dimensional structure of the DCLK1-HDAC6 complex based on molecular docking analysis (bottom). E, F Western blotting and qRT-PCR analyses of HDAC6 protein and mRNA levels in T24 and 5637 cells following DCLK1 knockdown. G, H Immunoblot analysis of HDAC6 and DCLK1 in T24 and 5637 cells transduced with sh-DCLK1 or sh-control after cycloheximide (CHX) treatment for the indicated times. Quantification of normalized HDAC6 protein levels is shown on the right. I Immunoblot analysis of HDAC6 and DCLK1 in T24 and 5637 cells transduced with sh-control or sh-DCLK1 following treatment with the proteasome inhibitor MG132 or the lysosome inhibitor chloroquine (CQ). J, K Denaturing immunoprecipitation using an anti-Myc antibody followed by immunoblotting for HA, Myc, and DCLK1 in HEK-293T and bladder cancer cells transfected with the indicated plasmids after MG132 treatment (10 µM, 6 h). L Denaturing immunoprecipitation using an anti-Myc antibody followed by immunoblotting for HA-Ub, Flag-DCLK1, and Myc-HDAC6 in HEK-293T cells transfected with the indicated plasmids after MG132 treatment (10 µM, 6 h). Data are presented as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001

DCLK1 recruits USP10 to deubiquitylate and stabilize HDAC6

To investigate whether DCLK1 stabilizes HDAC6 by recruiting a deubiquitinating enzyme, we examined candidate interactors identified in the IP-MS analysis. USP10, a deubiquitinating enzyme previously reported to regulate HDAC6 stability, emerged as a potential DCLK1-interacting protein (Table S3). Co-IP assays demonstrated that DCLK1 has an interaction with both exogenous and endogenous USP10 (Fig. 5A). Next, we explored whether USP10 impacts HDAC6 stability, knockdown of USP10 significantly decreased HDAC6 protein levels without affecting its mRNA levels, while USP10 overexpression increased HDAC6 protein levels, with no effect on its mRNA level (Fig. 5B, Supplementary Fig. 7D, E). Moreover, in USP10-knockdown cells, the half-life of HDAC6 was shortened and its stability was reduced (Fig. 5C).We next examined whether DCLK1 recruits USP10 to stabilize HDAC6. DCLK1 deficiency significantly reduced the interaction between USP10 and HDAC6 (Fig. 5D), suggesting that DCLK1 is essential for this interaction. Furthermore, in DCLK1-knockdown BC cells, the decrease in HDAC6 levels mediated by USP10 knockdown was completely reversed (Fig. 5E), indicating that DCLK1 acts as a scaffold to recruit USP10, thereby enhancing HDAC6 stability, HDAC6 degradation mediated by DCLK1 occurs via the proteasome pathway. Further experiments confirmed that MG132 treatment restored HDAC6 expression in cells with USP10 knockdown.(Fig. 5F).Consistent with these results, USP10 overexpression reduced HDAC6 ubiquitination, while a catalytically inactive USP10 mutant (Cys424Ala; C424A) had no effect (Fig. 5G), demonstrating that the deubiquitylase activity of USP10 is essential. Additionally, USP10 knockdown significantly increased HDAC6 ubiquitination (Fig. 5H).

Fig. 5.

DCLK1 recruits USP10 to deubiquitinate and stabilize HDAC6. A Immunoprecipitation (IP) was performed using anti-FLAG antibody or IgG control to assess the interaction between DCLK1 and USP10 in HEK-293T, T24, and 5637 cells transfected with Flag-DCLK1. B Protein and mRNA levels of HDAC6 were examined in T24 and 5637 cells with or without USP10 knockdown. C Protein stability of HDAC6 was assessed in USP10-knockdown T24 and 5637 cells after treatment with cycloheximide (CHX, 100 µg/mL) for the indicated time points. D T24 and 5637 cells transduced with sh-control or sh-DCLK1 were treated with MG132 (10 µM, 6 h), followed by IP using anti-USP10 or anti-HDAC6 antibodies and IB for HDAC6, DCLK1, and USP10. E Immunoblotting analysis of HDAC6 and USP10 in DCLK1-knockdown T24 and 5637 cells transduced with sh-control or sh-USP10. F IB analysis of HDAC6 and USP10 in T24 and 5637 cells transduced with sh-control or sh-USP10 after MG132 treatment (10 µM, 6 h). G, H IB analysis of HDAC6 and USP10 in T24 and 5637 cells transduced with Flag-USP10 or sh-USP10 following MG132 treatment (10 µM, 6 h). I Denaturing IP was performed using an anti-Myc antibody, followed by IB for HA-Ub, USP10, and Myc-HDAC6 in T24 and 5637 cells transfected with the indicated plasmids after MG132 treatment (10 µM, 6 h). J Protein levels of Flag-HDAC6 were analyzed in T24 and 5637 cells transfected with the indicated siRNAs and plasmids following CHX treatment (100 µg/mL). K FISH analysis of DCLK1, USP10, and HDAC6 expression in cisplatin-sensitive and cisplatin-resistant bladder cancer tissues.Scale bar, 50 μm. L Differential expression of DCLK1, USP10, and HDAC6 between bladder cancer and normal tissues in the TCGA cohort. M Association of DCLK1, USP10, and HDAC6 expression with clinical stage in the TCGA-BLCA cohort. Data are presented as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001

Next, we use PhosphoSitePlus (https://www.phosphosite.org), which is a public database of protein modification after translation, determines the HDAC6 known seven known ubiquitination sites on HDAC6: K32, K51, K116, K849, K854, K860, and K1206. We constructed seven Lys/Arg (K/R) substitution mutants of HDAC6 to screen for ubiquitination sites. After USP10 knockdown, polyubiquitination level of HDAC6 mutant was increased, except for the K116R mutant, indicating that USP10 primarily removes the polyubiquitin chain from K116 of HDAC6 (Fig. 5I). To investigate whether the HDAC6 K116R mutant would affect the stability of HDAC6, further analysis in BC cells that transfected with siRNA targeting HDAC6 showed that the K116R mutation significantly enhanced HDAC6 protein stability upon CHX treatment (Fig. 5J). These findings suggest that DCLK1 recruits USP10 to deubiquitinate HDAC6 at K116, preventing its degradation via the ubiquitin-proteasome pathway. To further demonstrate that the polyubiquitin chain on HDAC6 can be removed by USP10 in a DCLK1 dependent manner, we performed an in vitro ubiquitination assay. A His tagged HDAC6-K116 construct was generated by mutating all known ubiquitination sites except K116, and this construct together with His tagged HDAC6-WT was purified from E. coli (BL21). Flag tagged RNF168, which has been reported to catalyze ubiquitination at K116 [37], as well as Flag-USP10 and Flag-DCLK1 were purified from HEK293T cells. USP10 alone failed to remove the RNF168 mediated polyubiquitin chain, whereas in the presence of DCLK1, USP10 efficiently eliminated the ubiquitination at K116. These results indicate that DCLK1 recruits USP10 to removes RNF168-mediated K116-linked polyubiquitin chain on HDAC6 (Supplementary Fig. 7F). Furthermore, we conducted a comprehensive analysis of DCLK1, USP10, and HDAC6 expression and their clinical relevance in BC using TCGA data. FISH analysis revealed increased expression of DCLK1, USP10, and HDAC6 in cisplatin-resistant BC tissues compared to cisplatin-sensitive ones (Fig. 5K). The expression levels of these three genes were significantly higher in BC tissues compared to normal tissues and correlated positively with advanced TNM stage (Fig. 5L, M). Furthermore, Kaplan-Meier survival analysis showed that elevated expression of DCLK1, USP10, and HDAC6 was linked to poorer overall survival in BC patients (Supplementary Fig. 7G). To further evaluate the spatial coherence of the proposed DCLK1-USP10-HDAC6 regulatory axis at the single-cell level, UMAP analysis was performed to visualize gene-level expression of DCLK1, USP10, HDAC6, and PD-L1 in cancer cell clusters. DCLK1, USP10, and HDAC6 exhibited highly overlapping expression domains, whereas PD-L1 was expressed in a smaller subset of cancer cells that were localized within similar UMAP regions. These spatial expression patterns support the coherence of the DCLK1-USP10-HDAC6 regulatory axis and are consistent with DCLK1-associated PD-L1 upregulation (Supplementary Fig. 7H).

HDAC6 is required for the oncogenic effects of DCLK1 on bladder cancer progression

To determine whether HDAC6 mediates the oncogenic effects of DCLK1 in bladder cancer, a series of rescue experiments were performed. HDAC6 knockdown significantly abrogated the inhibitory effect of DCLK1 overexpression on cisplatin-induced apoptosis, as determined by flow cytometric analysis (Supplementary Fig. 8A). Consistently, TUNEL assays further confirmed that HDAC6 knockdown restored cisplatin-induced apoptosis in DCLK1-overexpressing cells (Supplementary Fig. 8B).Additionally, HDAC6 knockdown reversed the increased migration and invasion of BC cells induced by DCLK1 overexpression (Supplementary Fig. 8C). Western blotting and IF staining confirmed that HDAC6 knockdown also restored the protein levels of E-cadherin and Vimentin altered by DCLK1 (Supplementary Fig. 8D, E). Together, these results suggest that HDAC6 mediates the oncogenic effects of DCLK1 in BC.

Knockdown of DCLK1 increases the chemosensitivity and suppresses the metastasis of bladder cancer cells in vivo

To further evaluate the role of DCLK1 in chemoresistance and metastasis in BC, we employed subcutaneous xenograft, lung metastasis, and orthotopic bladder cancer models. In the subcutaneous xenograft model, DCLK1 knockdown significantly enhanced sensitivity to cisplatin, reducing tumor size and weight in treated mice (Fig. 6A-C). In the orthotopic model, the DCLK1-knockdown combined with cisplatin group exhibited and smaller tumor sizes, mean CT value (HU) of ROI reduced, immunohistochemical demonstrated tumor was significantly diminished (Fig. 6D-F). In the lung metastasis model, DCLK1 knockdown also enhanced sensitivity to cisplatin, and reduced lung nodule formation, immunohistochemical analysis demonstrated that metastasis was significantly diminished (Fig. 6G, H). These results indicate that knockdown of DCLK1 can increase chemosensitivity and reduce the progression of BC.

Fig. 6.

Knockdown of DCLK1 enhances chemosensitivity and suppresses bladder cancer metastasis in vivo. A Representative images of subcutaneous xenograft tumors in nude mice subjected to the indicated treatments. B, C Tumor growth curves (B) and tumor weights (C) of xenografts from different treatment groups. D Representative images of tumors from the orthotopic bladder cancer mouse model. E Representative CT images of orthotopic bladder tumors (arrows indicate tumor regions) and quantitative analysis of average CT values within the regions of interest (ROI). F Representative hematoxylin and eosin (H&E)-stained sections of orthotopic bladder cancer tissues. Scale bar, 100 μm. G Schematic illustration of the lung surface metastasis model. H Representative H&E-stained lung tissue sections and quantification of microscopic metastatic foci in the lungs. Scale bar, 100 μm. All data are presented as mean ± SD. **P < 0.01, ***P < 0.001

DCLK1 knockdown synergizes with HDAC6 inhibitor ACY-1215 and cisplatin to enhance antitumor efficacy in vitro and in vivo

ACY-1215 is a selective HDAC6 inhibitor that has demonstrated promising therapeutic activity in hematological malignancies and is currently under clinical evaluation; however, its efficacy in solid tumors remains largely unexplored [38, 39]. To investigate the antitumor effects of ACY-1215 in bladder cancer (BC) and its potential synergy with cisplatin, CCK-8, Transwell migration, and flow cytometry assays were performed. BC cells were treated with cisplatin (2.5 µg/mL for 24 h) and/or ACY-1215 (4 µM for 48 h). The CCK-8 assay revealed that, compared with the sh-Ctrl group, DCLK1 knockdown combined with ACY-1215 and cisplatin markedly suppressed cell proliferation (Fig. 7A). In addition, Transwell migration and flow cytometry analyses demonstrated that this triple combination significantly reduced cell migratory capacity and increased apoptotic rates (Fig. 7B, D). These results indicate that DCLK1 knockdown enhances the antitumor effects of ACY-1215 and cisplatin in vitro.

Fig. 7.

Antitumor effects of DCLK1 knockdown synergize with the HDAC6 inhibitor ACY-1215 and cisplatin. A Cell proliferation was assessed using the CCK-8 assay under the indicated treatment conditions. B Cell migratory ability was evaluated by Transwell migration assays. C, D Apoptotic cell populations were analyzed by flow cytometry across different treatment groups. E Representative images of subcutaneous xenograft tumors, along with quantification of tumor volume and tumor weight in each group. F FISH analysis of DCLK1, USP10, and HDAC6 expression in tumors from different treatment groups. Scale bar, 50 μm. G Representative hematoxylin and eosin (H&E) staining of tumor tissues, immunohistochemical (IHC) analysis of Ki-67 expression, and immunofluorescence staining for TUNEL across different groups. Scale bar, 50 μm. H Representative H&E staining images of major organs (heart, liver, spleen, lung, and kidney) from tumor-bearing mice, showing no overt histopathological toxicity. Scale bar, 50 μm. I Schematic illustration of the proposed working model. DCLK1 is upregulated in bladder cancer and recruits USP10 to deubiquitinate and stabilize HDAC6, thereby promoting chemoresistance and metastatic progression. All data are presented as mean ± SD. **P < 0.01, ***P < 0.001

To further validate these findings in vivo, a subcutaneous bladder cancer xenograft model was established. When tumor volumes reached approximately 100 mm³, mice were treated with intraperitoneal injections of ACY-1215 and cisplatin or vehicle control every two days, and tumor growth was monitored regularly. On day 28, mice were sacrificed for further analysis. Notably, the combination of DCLK1 knockdown with ACY-1215 and cisplatin resulted in the smallest tumor volumes and the slowest tumor growth among all groups (Fig. 7E). Fluorescence in situ hybridization analysis revealed markedly reduced expression of DCLK1, USP10, and HDAC6 in tumors from the triple-treatment group (Fig. 7F). Hematoxylin and eosin staining showed a substantial decrease in tumor cell density, accompanied by reduced Ki-67 expression and increased apoptosis, as indicated by TUNEL staining (Fig. 7G). Importantly, histopathological examination of major organs, including the heart, liver, spleen, lung, and kidney, revealed no overt treatment-related toxicity (Fig. 7H). Collectively, these findings demonstrate that DCLK1 is highly expressed in bladder cancer and promotes tumor progression by recruiting USP10 to deubiquitinate and stabilize HDAC6. Knockdown of DCLK1 markedly enhances the antitumor efficacy of combined treatment with the HDAC6 inhibitor ACY-1215 and cisplatin, resulting in suppressed tumor growth and metastatic potential both in vitro and in vivo (Fig. 7I).

Discussion

In this study, we performed high-resolution profiling of paired clinical samples, including normal tissues, paracancerous tissues, and carcinoma tissues with lymph node metastasis. Our analysis revealed diverse cell populations, such as T/NK cells, B cells, plasma cells, mono_macrophage_DC, neutrophils, mast cells, interstitial cells, endothelial cells, and cancer cells. We further characterized the molecular signatures, regulatory networks, dynamic changes, and functional roles of these cell clusters in tumor progression.Within cancer cells, we identified six distinct expression programs associated with different cellular states and biological functions: proliferation maintenance, metastatic activation, immune evasion, stress resistance, metabolic reprogramming, and inflammation promotion. Among the key tumor-related hub genes, DCLK1 is regarded as a key regulator of bladder cancer metastasis. The research results reveal DCLK1 is highly expressed in BC patients, promoting to both cisplatin resistance and metastasis of BC cells. Upregulated DCLK1 suppresses CD8⁺T cell activation by inducing PD-L1 expression via the Notch pathway, thereby promoting immune evasion and malignant progression in bladder cancer. These findings highlight DCLK1 as a key regulator of tumor progression and a potential target for immunotherapy.

Mechanistically, DCLK1 recruits USP10 to deubiquitinate and stabilize HDAC6, promoting tumor metastasis. Our research results establish DCLK1 as a crucial gene in cisplatin resistance and provide insight into the underlying mechanisms in BC patients, offering potential therapeutic targets for treatment.DCLK1 is widely expressed in various cancers, with high expression levels linked to tumor progression [40]. It plays a crucial role in carcinogenesis and metastasis through signaling pathways like Notch [41], WNT/β-catenin, and NF-κB [42, 43], DCLK1 contributes to the maintenance of tumor stem cells [44]. In addition, increased levels of DCLK1 levels in cancers such as renal [45], bladder [46], pancreatic [47], colorectal [48], and liver [49] are associated with processes like epithelial-to-mesenchymal transition, cell proliferation, and migration. However, the precise biological functions and underlying mechanisms of DCLK1 in BC remain unclear.

Our study reveals DCLK1 recruits USP10 in BC cells, promotes the deubiquitination and stabilization of HDAC6, and emphasizes the importance of ubiquitination in tumor metastasis. HDAC6 is a structurally and functionally unique deacetylase that targets both histone and non-histone substrates [50, 51], such as Hsp90 [52], cortactin [53], peroxiredoxin [54], α-tubulin [55], and HSF-1 [56], influencing various biological processes. HDAC6 has the ability to promotes cell motility, migration, and invasion, and is overexpressed in various tumor types [57], including ovarian [58], oral squamous cell carcinoma [59], hepatocellular carcinoma [60], and acute myeloid leukemia [61], Beyond these oncogenic functions, HDAC6 has emerged as an important modulator of the cancer immune microenvironment through regulation of inflammatory mediators, including NLRP3, IL-1β, and IL-6 [62]. However, the functional significance and mechanistic contribution of HDAC6 in bladder cancer remain largely undefined. Our study reveals that DCLK1 overexpression deubiquitinates HDAC6, thereby preventing its degradation, while HDAC6 knockdown reduces the oncogenic effects of DCLK1 in BC cells.However, the role of ubiquitination in regulating HDAC6 stability remains unclear. In this work, we offer new insights by showing that HDAC6 undergoes degradation through K48-linked ubiquitination and highlighting the essential role of the DCLK1-USP10 axis in blocking its ubiquitination and degradation.

USP10, a ubiquitin-specific protease, removes ubiquitin chains from substrates, ensuring protein stability [63]. It has a pivotal role in diverse cellular processes and the development of tumors, with the ability to act as either an oncogenic or tumor suppressive factor [64, 65]. USP10 regulates p53 by deubiquitinating and stabilizing it, enhancing its tumor-suppressing function [66]. Similarly, USP10 promotes hepatocellular carcinoma cell proliferation through YAP/TAZ deubiquitination [67]. In lung cancer, USP10 knockdown inhibits PTEN ubiquitination, promoting cell proliferation and invasion [68]. Moreover, USP10 has been reported to contribute to immune regulation through deubiquitination of key immune signaling molecules [69]. Notably, USP10 attenuates NF-κB and IL-1R/TLR pathway activation by deubiquitinating TRAF6 [70], suggesting a role in the modulation of inflammatory and immune signaling. Our study shows that DCLK1 recruits USP10, which prevents HDAC6 degradation by removing its K48-linked polyubiquitin chain at Lys116. HDAC6 knockdown reversed DCLK1’s oncogenic effects. These findings uncover a new binding partner for HDAC6 and offer valuable insight into its regulation, highlighting its potential as a therapeutic target. Therefore, DCLK1 silencing suppresses tumor progression, synergizing with HDAC6 inhibitor ACY-1215 and cisplatin to enhance antitumor efficacy.

Despite the strengths of this study, several limitations should be acknowledged. First, the scRNA-seq analysis was based on a relatively small patient cohort, which may not fully capture the complexity of tumor heterogeneity and immune features in bladder cancer. Although stringent quality control and complementary analyses were applied, validation in larger and independent single-cell or spatial transcriptomic cohorts will be important in future studies. Second, while UMAP-based single-cell analyses revealed highly overlapping expression of DCLK1, USP10, and HDAC6 within malignant epithelial clusters, PD-L1 was detected in a more restricted subset of tumor cells localized within similar regions. These findings provide supportive spatial transcriptomic evidence but do not establish direct causality, which was instead addressed through complementary in vitro functional and pathway perturbation experiments. Third, mechanistic validation was primarily conducted in vitro and in xenograft models that do not fully recapitulate an intact immune microenvironment; therefore, future studies using immune-competent or genetically engineered mouse models will be required to more comprehensively assess the role of the DCLK1-USP10-HDAC6 axis in immune evasion. In addition, although the current study focused mainly on genetic inhibition of DCLK1, pharmacological targeting of DCLK1 (e.g., with small-molecule inhibitors such as DCLK1-IN-1) represents an important direction for future investigation and will further strengthen the translational relevance of targeting DCLK1 in bladder cancer.

In conclusion, our study identifies DCLK1 as a key driver of cisplatin resistance, immune evasion, and metastasis in bladder cancer. By recruiting USP10 to stabilize HDAC6, DCLK1 promotes tumor progression through both molecular and immunological mechanisms. These findings offer valuable insights into the DCLK1-USP10-HDAC6 axis and suggest DCLK1 as a potential therapeutic target for bladder cancer treatment.

Authors’ contributions

Ashuai Du, Xinpei Deng and Yuzheng Zhou conceived the study and designed the experiments. Ashuai Du, Xinpei Deng and Yuzheng Zhou carried out most of the experiments and analyzed the data. Dongbo Yuan, Kai Li, Yuanyuan Luo, Songsong Tan, Xuchao Dai and Bo Yu helped with the experiments and provided technical assistance. Kehua Jiang and Xingliang Tan provided equipment. Kehua Jiang, Xingliang Tan and Jianguo Zhu provided funding support and reagents. Ashuai Du, Xinpei Deng, Yuzheng Zhou, Xingliang Tan and Jianguo Zhu wrote and revised the manuscript. All authors reviewed and approved the final manuscript.

Funding

This work was supported by Guizhou Provincial Basic Research Program (ZK[2025]-505).The National Natural Science Foundation of China(82160551).Guizhou Provincial Key Technology R&D Program(ZSYS[2025]-031). Guizhou Provincial Science and Technology Support Plan Program( [2025]-132).Medical Research Union Found for High-quality health development of Guizhou Province (2024GZYXKYJJXM0054). Guizhou Provincial People’s Hospital Talent Fund Project ([2023]-17). The National Natural Science Foundation of China (82303301). The National Natural Science Foundation of China (82160551). Medical Research Union Found for High-quality health development of Guizhou Province (2024GZYXKYJJXM0054).

Data availability

The raw sequence data reported in this paper have been deposited in the Genome Sequence Archive (Genomics, Proteomics & Bioinformatics 2025) in National Genomics Data Center (Nucleic Acids Res 2025), China National Center for Bioinformation / Beijing Institute of Genomics, Chinese Academy of Sciences (GSA-Human: HRA015249) that are publicly accessible at [https://ngdc.cncb.ac.cn/gsa-human](https:/ngdc.cncb.ac.cn/gsa-human).

Declarations

Ethics approval and consent to participate

All human tissues and clinical information were collected using protocols approved by the Ethics Committee of the Guizhou Provincial People’s Hospital. Written informed consent was obtained from each patient. All animal procedures used in the study were approved by the Animal Ethics Committee of Guizhou Provincial People’s Hospital and conducted according to the Guidelines for the Care and Use of Laboratory Animals at Guizhou Provincial People’s Hospital.

Consent for publication

All authors have reviewed the final version of the manuscript and approve it for publication.

Competing interests

The authors declare no competing interests.