A NEK2/ZWINT–NDC80 regulatory axis drives bladder cancer progression and chemoresistance

Abstract

Bladder cancer progression and chemoresistance remain major clinical challenges, yet the molecular determinants underlying these aggressive phenotypes are incompletely understood. Integrating TCGA transcriptomics, GEO single-cell transcriptomics, patient tissue profiling, flow cytometry–based cell sorting, functional perturbation assays, and xenograft models, we investigated the cooperative roles of NEK2, ZWINT, and the downstream effector NDC80 in bladder cancer biology. We identified a previously unrecognized NEK2/ZWINT–NDC80 regulatory axis that governs malignant cell behavior. NEK2 and ZWINT were markedly co-upregulated in high-grade tumors and strongly associated with EMT-related transcriptional programs. NEK2⁺ZWINT⁺ double-positive cells represented a stable and aggressive tumor subpopulation enriched in invasive lesions. Dual knockdown of NEK2 and ZWINT profoundly impaired proliferation, clonogenicity, motility, and in vivo tumorigenicity, whereas single-gene perturbation produced only partial effects. Mechanistically, NDC80 acted as a critical downstream effector, integrating NEK2–ZWINT signals and acting as a key downstream effector that integrates signals from NEK2 and ZWINT to drive malignant phenotypes. NDC80 depletion suppressed tumor growth and significantly increased gemcitabine sensitivity. Clinically, NDC80 expression correlated with advanced stage and chemoresistant bladder cancer tissues. Our findings reveal the NEK2–ZWINT–NDC80 axis as a central regulatory module driving bladder cancer progression and gemcitabine resistance. Targeting this axis may offer a therapeutic strategy for high-risk bladder cancer.

Supplementary Information

The online version contains supplementary material available at 10.1007/s12672-026-05435-w.

Introduction

Bladder cancer remains a major global health burden characterized by high recurrence rates, substantial morbidity, and marked clinical heterogeneity [1]. Although most patients initially present with non–muscle-invasive bladder cancer (NMIBC), up to one-third experience progression to muscle-invasive disease (MIBC), a transition that dramatically worsens survival outcomes [2]. Current prognostic systems based on histopathological and clinical indicators provide limited capacity to predict which tumors will progress [3]. Moreover, chemoresistance—particularly resistance to gemcitabine-based regimens—continues to undermine therapeutic efficacy in both localized and advanced settings [4]. These enduring clinical challenges highlight a fundamental gap in understanding the molecular determinants that govern bladder cancer progression and therapeutic refractoriness.

While substantial efforts have focused on canonical oncogenic pathways such as FGFR3 signaling, PI3K alterations, and immune–tumor interactions, comparatively little attention has been given to chromosome segregation regulators and kinetochore-associated factors [5], despite their well-established roles in promoting tumor cell survival, genomic instability, and therapy resistance in other malignancies [6]. Among these regulators, NEK2 (NIMA-related kinase 2) is a centrosomal serine/threonine kinase that regulates centrosome separation and bipolar spindle assembly during mitosis through phosphorylation of centrosomal linker proteins. Its activity is coordinated with mitotic regulators such as Aurora-A and Polo-like kinase 1 (PLK1), thereby contributing to proper mitotic progression [7]. ZWINT (ZW10-interacting kinetochore protein) is a kinetochore-associated factor involved in the recruitment of spindle assembly checkpoint (SAC) components and stabilization of kinetochore–microtubule attachment, thereby ensuring accurate chromosome alignment during mitosis [8]. Similarly, NDC80 is a core component of the NDC80 complex within the KMN kinetochore network, which directly mediates microtubule binding at the kinetochore and is essential for accurate chromosome segregation [9].

Critically, no studies have evaluated whether NEK2, ZWINT, and NDC80 operate as an integrated regulatory axis with coordinated biological effects on bladder cancer behavior or treatment response [10]. It also remains unclear whether NEK2–ZWINT co-upregulation identifies a distinct malignant subpopulation with enhanced aggressiveness [11], or whether downstream effectors such as NDC80 mediate tumor progression and chemoresistance [12].

In this study, we identify a previously unrecognized NEK2–ZWINT–NDC80 regulatory axis that governs bladder cancer aggressiveness. Through integrated transcriptomic analyses, patient tissue profiling, flow cytometry–based cell sorting, functional perturbation assays, and in vivo xenograft studies, we demonstrate that: (i) NEK2 and ZWINT are co-upregulated in high-grade bladder cancer and tightly associated with EMT-related transcriptional programs; (ii) NEK2⁺ZWINT⁺ double-positive cells constitute an aggressive tumor subpopulation enriched in high-grade lesions; (iii) dual loss of NEK2 and ZWINT profoundly impairs tumor cell proliferation, migration, and tumorigenicity; and (iv) NDC80 functions as a critical downstream effector that reinforces this axis and contributes to gemcitabine resistance. Collectively, these findings establish the NEK2–ZWINT–NDC80 axis as a mechanistic driver of bladder cancer progression and a potential therapeutic vulnerability (Fig. 1).

Fig. 1.

Flowchart. Identification of high-risk genes associated with bladder cancer progression based on public single-cell datasets; Isolation of double-positive cell populations with high target-gene expression from clinical bladder cancer tissues; Exploration of the effects of targeted gene knockout on the biological activity of these cell populations

Materials and methods

Single-cell RNA sequencing analysis

We obtained three single-cell RNA sequencing (scRNA-seq) datasets from the GEO database: GSE135337 (n = 8), GSE169379 (n = 26), and GSE222315 (n = 13) [13]. After integration, we performed quality control for subsequent analysis. For each dataset, used the CreateSeuratObject function in the Seurat package (version v4.3.3) to convert the gene expression matrix into a Seurat object (version R v4.4.0). To correct for batch effects, we applied the Harmony algorithm to process the PCA embedding [14]. The updated embedding information was used to reconstruct the neighborhood graph and was subjected to dimensionality reduction processing through the UMAP method. Subsequently, Leiden clustering was performed to partition the cell populations. Visualizations of clusters and gene expression patterns were conducted at different resolutions, with particular attention paid to the marker genes of epithelial, endothelial, fibroblast, and immune cell populations [15]. A marker gene dictionary was constructed to define cell types. Cell type annotations were manually assigned to the clusters, and the necessary cells were retained for subsequent analysis.

Differential gene expression and functional enrichment analysis

The differential expression analysis was conducted using the Wilcoxon rank sum test. The differentially expressed genes (DEGs) were defined by the following thresholds: p < 0.05, false discovery rate (FDR) < 0.05, and |log₂adjusted fold change| > 1.05, and |log₂ adjusted fold change| > 1. Use R software (version 4.3.0) and the ggplot2 package to draw a volcano plot, and label the NEK2 and ZWINT genes. For each cluster, the differentially expressed genes were identified and pathway enrichment analysis was performed using the Metascape platform. The analysis categories included GO molecular function, GO biological process, classical pathways, Reactome gene sets, KEGG pathways, BioCarta gene sets, and signature gene sets. A significance threshold of p < 0.01 was set to determine the terms with significant meaning.

Protein interaction network analysis

Using the STRING database (version 12.0), the predicted protein interaction network of ZWINT was retrieved [16]. The species was set as Homo sapiens, and the minimum interaction confidence threshold was set to 0.9 (high confidence). The top 10 functionally associated proteins were displayed.

TCGA data analysis

Download the RNA sequencing data of bladder urothelial carcinoma (BLCA) from the TCGA website (in TPM format), including 114 cases in early stage I-II and 81 cases in late stage III-IV [17]. Use the GEPIA2 and GEPIA3 databases for pan-cancer analysis, and retrieve the ZWINT and NEK2 expression data of 33 cancer types [18]. Differential expression analysis used the log₂ transformed TPM values, with the filtering condition being |log₂FC| > 1 and P < 0.05. Use Pearson correlation analysis to evaluate the correlation between NEK2 and ZWINT mRNA expression in the TCGA-BLCA cohort. Through the Wilcoxon rank sum test, compare the expression differences between tumors and adjacent tissues, with the significance threshold defined as *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001.

Study design and sample collection

Tissue samples were collected from 16 patients treated at Harbin Medical University Cancer Hospital between 2022 and 2025 (KY2024-100). The use of bladder cancer patient samples was approved by the Medical Ethics Committee of Harbin Medical University. Informed consent was obtained from all patients, and all experiments were conducted in compliance with relevant ethical regulations for human research participants. Ethical approval was obtained, and no clinical trial registration was required as this was a basic science study (Clinical trial number: not applicable). All patients were clinically or pathologically diagnosed with bladder cancer prior to admission, primarily through cystoscopy or pathological examination. Most samples were obtained from tumor-bearing bladders following radical cystectomy. Because some tumors presented with multiple lesions within the bladder, multiple sampling points were collected from individual patients according to research needs. After excision, tissues were immersed in sterile normal saline and transported to the laboratory at 0–4 °C for subsequent processing.

Flow cytometry and sorting of bladder cancer cell subpopulations

Bladder cancer tissues were collected and immediately treated with preconfigured tissue digestion solution. The preconfigured tissue digestion solution contains 5% fetal bovine serum (Gibco, A31608-02), 2 mg/ml collagenase I (ThermoFisherm, 17018-029), 2 mg/ml hyaluronidase (Solarbio, H8030) and 25 mg/ml DNase I (Solarbio, D8071). The mixture was incubated on an orbital shaker at 200 rpm for 45 min to facilitate tissue dissociation. Following digestion, the lower cell pellet was harvested by centrifugation at 500×g for 5 min and washed three times with PBS to remove residual digestion solution. The washed pellet was then filtered through a 200 μm cell strainer to obtain cell suspension. To eliminate contaminating red blood cells (RBCs), an appropriate volume of RBC lysis buffer (Solarbio, R1010) was added to the single-cell suspension for 5 min at 4℃, which was subsequently centrifuged at 500×g for 5 min. The resulting cell suspension was carefully collected and reserved for subsequent flow cytometric sorting.The prepared single-cell suspension was incubated with three fluorochrome-conjugated antibodies and a viability dye: FineTest^® Red780 (Viability Dye), Alexa Fluor^® 488 Anti-EpCAM antibody [EPR20532-225], and PE Anti-Human/Mouse CD44 Antibody [IM7]. Staining was performed at room temperature for 60 min in the dark. After staining, cell subpopulation identification and flow cytometric sorting were conducted using a BD FACSAria™ Ⅲ flow cytometer. The EpCAM⁺CD44⁺ cell subpopulation was defined as primary bladder cancer cells based on phenotypic markers [19]. The sorted primary bladder cancer cells were seeded into culture flasks containing RPMI 1640 medium with 10% FBS and 1% penicillin-streptomycin in a humidified incubator with 5% CO2 at 37℃.All flow cytometric data were acquired and analyzed using FlowJo 10.8.1 software.

Cell lines

Human bladder cancer cell lines RT112, T24, and HEK293T were obtained from the American Type Culture Collection (ATCC, USA). T24 and 5637 cells were maintained in RPMI-1640 medium (Gibco, USA) supplemented with 10% fetal bovine serum (FBS; Gibco, USA) and 1% penicillin–streptomycin (Gibco, USA). HEK293T cells were propagated in DMEM (Gibco, USA) containing 10% FBS and 1% penicillin–streptomycin. RT112 cells were cultured in MEM (Gibco, USA) with the same supplements. All cell lines were kept under standard incubator conditions at 37 °C in a humidified atmosphere with 5% CO₂. Each line was authenticated by short tandem repeat profiling within six months prior to use and confirmed to be free of Mycoplasma contamination.

Production of lentivirus

HEK293T cells were seeded in 10-cm dishes to achieve 85–90% confluency at the time of transfection. Co-transfection was performed using the calcium phosphate precipitation method, with the transfer plasmid along with the packaging plasmid psPAX2 (RRID: Addgene_12260) and the envelope plasmid pDM2.G (RRID: Addgene_12259). Specifically, 20 µg of the transfer plasmid, 15 µg of psPAX2, and 6 µg of pDM2.G were combined in a 15 mL tube. Then, 500 µL of nuclease-free water was added, followed by 50 µL of calcium chloride solution. While vortexing the mixture continuously, 500 µL of 2× HBS buffer (pH 7.05) was added dropwise. The mixture was incubated at room temperature for 15 min. The resulting transfection complex was then added dropwise to the 10-cm dish. After 6–8 h, the medium was replaced with 10 mL of fresh medium. Lentivirus-containing supernatant was harvested at 48- and 56-hours post-transfection, then filtered through a 0.45 μm filter, and purified by centrifugation at 19,400 rpm under cold conditions. The virus was resuspended in 100 µl of PBS, aliquoted, and stored at -80℃.

Plasmid construction and stable transduction

SgRNAs targeting ZWINT and NDC80 (Supplementary Table S1) were synthesized by Genesoul Technology (Harbin, China) and cloned into the lentiCRISPR v2 vector. sgRNAs targeting NEK2 (Supplementary Table) were synthesized by the same vendor and inserted into the LentiGuide-Neo vector. Lentiviral transduction was performed in the presence of Polybrene (Beyotime Biotechnology, China) according to the manufacturer’s recommendations.

Tumor cells were individually transduced with lentiviruses expressing sgRNAs targeting ZWINT or NDC80, and two corresponding stable cell lines were established through puromycin selection (Beyotime Biotechnology, China). Cells were also independently transduced with lentiviruses expressing sgRNAs targeting NEK2, followed by neomycin selection. For the generation of dual-knockdown lines, tumor cells were first stably transduced with ZWINT-targeting sgRNAs and selected with puromycin, after which NEK2-targeting sgRNAs were introduced and neomycin selection was applied, yielding cell lines with simultaneous depletion of ZWINT and NEK2.

Establishment of resistant cell line

To establish gemcitabine (GEM)-resistant DP cell lines, cells were exposed to RPMI-1640 medium supplemented with 10% fetal bovine serum (FBS) containing an initial GEM concentration of 0.1 µM and incubated for 24 h. Subsequently, the remaining viable cells were expanded in RPMI-1640 medium with 10% FBS for 5 days. The second and third rounds of selection were performed following the same protocol, with gradual increments of GEM concentrations (0.5 µM, 1.0 µM, 2.0 µM, 4.0 µM, 8.0 µM, and 20 µM). Ultimately, we found that the GEM-resistant DP cells could be stably cultured in RPMI-1640 medium supplemented with 10% FBS and 10 µM GEM.

RT‑qPCR

First, total RNA was extracted using the Cwbio Total RNA Extraction Kit (Jiangsu, China) according to the manufacturer’s instructions. cDNA synthesis was performed following the protocol provided by Toyobo (Tokyo, Japan). Finally, PCR amplification was conducted using ChamQ SYBR qPCR Master Mix (Vazyme, Nanjing, China) in accordance with the manufacturer’s instructions. All quantifications were normalized to GAPDH as an internal reference gene. All qRT-PCR assays were performed in at least three independent experiments. The primer sequences were as follows (5′→3′):

NEK2 forward, CGGAAGTTCCTGTCTCTGGCA;

NEK2 reverse, TTCAGGTCCTTGCACTTGGACT;

ZWINT forward, AGGACACTGCTAAGGGTCTCG;

ZWINT reverse, GCCTCTACGTGCTCCCTGTA;

GAPDH forward, GGAGCGAGATCCCTCCAAAAT;

GAPDH reverse, GGCTGTTGTCATACTTCTCATGG.

Cell counting kit-8 (CCK8) assay

After treatment, various cells were seeded into 96-well plates at a density of 2000 cells per well. Over a period of 0–3 days, 10 µl of Cell Counting Kit-8 (CCK-8) (Beyotime, Shanghai, China) was added to each well at the same time every day. Following a 2-hour incubation, the optical density (OD) values were measured at a wavelength of 450 nm using a microplate reader (BioTek, USA). Each experiment was repeated three times.

Cell viability assay and IC50 calculation

Four cell groups (DP, DP Ctrl, DP NDC80-KD, and DN) were seeded at identical density into 96-well plates. After 24 h of adherence, cells were treated with serial concentrations of gemcitabine (0, 0.001, 0.01, 0.1, 1, 5, 10 µM), with 5 replicate wells per group. Cell viability was assessed by CCK-8 assay, and relative viability was calculated with the 0 µM control group set as 100%. The half-maximal inhibitory concentration (IC50) for each group was determined by fitting dose-response curves using GraphPad Prism software.

Colony formation assay

After treatment, various cells were seeded into 6-well plates at a density of 300 cells per well. Following approximately 2 weeks of culture, the cells were fixed with 4% paraformaldehyde for 60 min, and subsequently stained with crystal violet solution (Solarbio, Beijing, China) for 30 min. After capturing images of cell colonies using a digital camera, the images were analyzed with ImageJ software. Each experiment was repeated three times.

Migration assays

Cell migration ability was evaluated using the wound healing assay. After treatment, various cells were seeded into 6-well plates at a density of 1 × 10⁵ cells per well. When the cell confluency reached 90%, wounds were created by scratching the cell monolayer with a 200 µL pipette tip. Following rinsing with PBS, the cells were incubated in serum-free medium. Images were captured at designated time points (0 and 24 h) using a microscope to record the wound closure process.

Xenograft tumor models

Nude mice aged 6–8 weeks (weighing 25–30 g) were purchased from Changchun Changsheng Biotechnology Co., Ltd. (Liaoning, China) for the establishment of xenograft tumor models. The treated stable cell lines (DN cells, DP cells, NDC80-KD and Contral) were subcutaneously implanted into the right flank of the mice. Tumor volume was measured every 7 days. when the tumors reached a volume of not exceed 1200 mm³ (for DP and shCtrl groups) (i.e., on day 21 after tumor inoculation), the mice were euthanized, followed by tumor imaging and weighing. Anesthesia and euthanasia methods Animals were anesthetized with sodium pentobarbital (50 mg/kg, intraperitoneal injection), and euthanasia was performed by overdose of pentobarbital, in accordance with institutional guidelines. All animal experiments were performed in accordance with the guidelines approved by the Animal Care and Use Committee of Harbin Medical University and complied with all relevant ethical regulations for animal use. At least 6 nude mice were used per experiment. Tumor volume was calculated using the formula: Volume = (length × width²) × 1/2.

Immunohistochemistry (IHC)

First, paraffin sections were dewaxed with xylene and gradient ethanol, followed by antigen retrieval using EDTA antigen retrieval buffer (Beyotime, Shanghai, China). Subsequently, endogenous peroxidase was inactivated with 3% hydrogen peroxide. After blocking with immunofluorescence blocking buffer (Beyotime, Shanghai, China), the sections were incubated with primary antibodies overnight at 4 °C. Following 1-hour incubation with secondary antibodies (goat anti-rabbit or anti-mouse IgG), the paraffin sections were stained with DAB Detection Kit (Zhong Shan -Golden Bridge, Beijing, China) and counterstained with hematoxylin (Solarbio, Beijing, China). The following antibodies were used for IHC: ZWINT (1:200, Abcam, AB252950), NEK2 (1:200, Abcam, AB227958), NDC80 (1:200, Proteintech, 18932-1-AP), Ki67 (1:200, Proteintech, 27309-1-AP), anti-rabbit IgG (1:200, Proteintech, SA00001-2), anti-mouse IgG (1:200, Proteintech, SA00001-1). ImageJ software was used for image analysis and quantification. Image analysis was performed using ImageJ software (version 1.53k) with the IHC Profiler plugin. The integrated optical density (IOD) and the total area of the tissue were measured for each image. The expression levels of target proteins were expressed as the Mean Optical Density (MOD = IOD/area), which reflects the average concentration of the stained antigen per unit area.

Immunofluorescence (IF)

Multiplex immunofluorescence staining was performed using the PANO 6-Color Immunofluorescence Kit (Panovue, Beijing, China). After dewaxing in xylene and rehydration with gradient ethanol, the sections were immersed in Tris-EDTA antigen retrieval buffer (pH 8.0) and subjected to microwave heating for antigen retrieval. The sections were allowed to cool naturally to room temperature. To block endogenous antigens, the sections were treated with 1% bovine serum albumin (BSA) for 30 min at room temperature. Primary antibodies were incubated for 1 h at room temperature, followed by 30 min of incubation with secondary antibodies at room temperature. Tyramide signal amplification (TSA) was carried out using fluorescent reagents (PPD 520, PPD 570, PPD 620, PPD 650, and PPD 780; Panovue, Beijing, China; 1:100) for 10 min at room temperature. The following antibodies were used for IF: ZWINT (1:200, Abcam, AB252950), NEK2 (1:200, Abcam, AB227958), NDC80 (1:200, Proteintech, 18932-1-AP), EPCAM (1:200, Abcam, ab223582). After staining all targets, nuclear staining was performed with DAPI (Panovue, Beijing, China; 1:100) for 15 min at room temperature. Following each step, the sections were washed three times with TBST solution for 2 min per wash. Finally, the sections were mounted with neutral balsam. Image acquisition was performed using a confocal laser scanning microscope (ZEISS, Germany). All immunofluorescence assays were independently repeated at least three times.

Statistical analysis

All quantitative data in this study are presented as the mean ± standard deviation (SD) from at least three independent biological replicates. Statistical analyses were performed using GraphPad Prism (version 9.0) and R software (version 4.2.0). Comparisons between two experimental groups were analyzed using the two-tailed Student’s t-test. For comparisons involving more than two groups (such as cell viability assays across multiple treatment arms), one-way analysis of variance (ANOVA) followed by Tukey’s post-hoc test was applied. Survival curves were constructed using the Kaplan-Meier method and compared via the log-rank test.

Results

Transcriptomic profiling identifies NEK2–ZWINT co-upregulation and its association with EMT activation in bladder cancer

To identify key molecular programs associated with bladder cancer progression, we first analyzed **single-cell RNA-sequencing data from the GEO database**, comparing bladder cancer tissues with adjacent normal urothelium. Differential expression analysis revealed that NEK2 and ZWINT were among the most consistently upregulated mitosis-related genes [12] (Fig. 2A). Correlation analysis in the TCGA bladder cancer cohort demonstrated a strong positive association between NEK2 and ZWINT expression (R = 0.803) (Fig. 2B), suggesting a coordinated regulatory relationship rather than independent activation [20].

Fig. 2.

NEK2⁺ZWINT⁺ double-positive cells define an aggressive tumor subpopulation that can be prospectively isolated from patient samples. A Volcano plot. The x-axis represents log₂(foldchange) and y-axis represents -log₁₀(P value). Red dots indicate significantly upregulated genes and blue dots indicate significantly downregulated genes (|log₂FC| > 1 and P < 0.05), with NEK2 and ZWINT specifically labeled. B Pearson correlation analysis of NEK2 and ZWINT mRNA expression in the TCGA bladder cancer cohort. C Pathway enrichment dot plot for differentially expressed genes in bladder tumors. D Expression difference of ZWINT between tumor and adjacent normal tissues across 33 cancer types in TCGA. The x-axis represents log₂(tumor/adjacent) fold change. Cancer types marked in red indicate positive correlation with tumor, green indicates negative correlation, and black indicates no significant difference. E Boxplot showing TPM expression difference of NEK2 between tumor and adjacent normal tissues in TCGA pan-cancers. F TPM expression difference of ZWINT and NEK2 between tumor and adjacent normal tissues in the TCGA-BLCA cohort. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001 (Wilcoxon rank-sum test)

To delineate the biological pathways associated with this co-upregulated module, we performed gene set enrichment analyses on single-cell RNA sequencing data comparing bladder tumors with normal bladder tissue.Analysis of tumors that co-express NEK2 and ZWINT revealed that the characteristics of epithelial-mesenchymal transition (EMT), cell cycle progression, and mitotic spindle assembly were significantly enriched [21] (Fig. 2C). EMT-related genes, including VIM, FN1, ITGA5, SNAI2, and ZEB1, were consistently elevated in the NEK2-high/ZWINT-high group, whereas epithelial markers tended to be downregulated, suggesting that NEK2–ZWINT upregulation is embedded within a broader EMT-like transcriptional state. Stratified analysis of The Cancer Genome Atlas (TCGA) data according to pathological features demonstrated that NEK2 and ZWINT were significantly upregulated in tumor tissues relative to adjacent normal counterparts, implying a strong link to tumorigenesis and progression [22] (Fig. 2D-F).

NEK2⁺ZWINT⁺ double-positive cells define an aggressive tumor subpopulation that can be prospectively isolated from patient samples

Given the strong transcriptional correlation of NEK2 and ZWINT in high-grade tumors, we next examined whether their co-expression defines a distinct malignant cell population in primary bladder cancer. Fresh tumor specimens were dissociated into single-cell suspensions, and tumor cells were enriched by flow cytometry. Because NEK2, ZWINT, and NDC80 are intracellular antigens, we applied a two-step strategy: bulk tumor-cell enrichment followed by molecular stratification based on intracellular expression. qPCR and Western blot analyses of sorted tumor cells revealed a discrete ZWINT+ NEK2 + subset consistently detected across independent patient samples (n = 3). These findings indicate that coordinated upregulation of NEK2 and ZWINT marks a reproducible double-positive (DP) malignant population enriched in advanced tumors, suggesting functional relevance of this mitotic axis in bladder cancer progression [23] (Fig. 3A).

Fig. 3.

Selection and characterization of the double-positive cell population. A Schematically illustrates the fluorescence-activated cell sorting workflow, encompassing fluorescent antibody labeling, flow cytometric detection via FSC/SSC analysis, electrostatic charging of target cells, and subsequent deflection-based sorting and collection using oppositely charged plates. B Flow cytometric sorting of the BC cell population. C, D, Western blot and qPCR analysis of NEK2 and ZWINT expression in various bladder cancer cells (cell1-3) and adjacent tumor tissues (PT1-3). E, F, Western blot and qPCR analysis of NEK2 and ZWINT expression in various bladder cancer cell lines and adjacent tumor tissues. G, H Expression levels of ZWINT and NEK2 in low-grade and high-grade human bladder cancer. Statistical significance was determined by two-tailed Student’s t-test. Asterisks indicate significance levels: *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001

Although the DP cells fraction accounted for only a minority of total tumor cells, its presence was highly reproducible and clearly separable from single-positive or double-negative populations, indicating a biologically stable phenotype. Importantly, the proportion of NEK2⁺ZWINT⁺ DP cells tended to be higher in high-grade tumors than in low-grade lesions, mirroring the expression patterns observed in bulk transcriptomic data [24] (Fig. 3B).

The double-positive (DP) cells (Cell1) we screened displayed significantly higher transcriptional and protein expression levels of NEK2 and ZWINT compared with primary bladder cancer cells (Cell2 and Cell3) and adjacent non-tumor bladder tissues (PT1, PT2, and PT3) (Fig. 3C and D). This further verified the accuracy of the cells sorted by flow cytometry. Meanwhile, these DP cells also showed markedly elevated NEK2 and ZWINT expression relative to low and high invasive potentials bladder cancer cell lines (RT112 and T24) with distinct invasive potentials, as well as tumor tissue samples derived from patients with early (BC1), middle (BC2), and advanced (BC3) stages bladder cancer. Meanwhile, based on previous studies on bladder cancer, T24 cells exhibit stronger invasive potential than RT112 cells, along with relatively higher expression levels of NEK2 and ZWINT, which is consistent with the established notion that NEK2 and ZWINT may promote the progression of bladder cancer [25] (Fig. 3E and F).

Immunofluorescence analysis further supported these findings. In low-grade tumors, NEK2 and ZWINT showed weak or focal staining. By contrast, high-grade tumors exhibited intense, overlapping NEK2 and ZWINT signals, often enriched at invasive edges and regions of disorganized epithelium (Fig. 3G). Statistical analysis revealed that the positive rates of NEK2 and ZWINT were 19.2% and 34.5%, respectively, in low-grade bladder cancer (BC); whereas the co-positive rate of NEK2 and ZWINT reached 95% in high-grade BC (Fig. 3H). These observations suggest that NEK2⁺ZWINT⁺ cells may represent a functionally aggressive subpopulation poised for invasion and progression [26].

The ability to prospectively isolate NEK2⁺ZWINT⁺ DP cells from patient tissues by flow cytometry provided a powerful platform to functionally characterize this subpopulation and to model their behavior in vitro and in vivo.

Dual loss of NEK2 and ZWINT suppresses oncogenic programs and markedly attenuates tumor growth in vivo

To functionally interrogate the contribution of the NEK2–ZWINT axis to bladder cancer malignancy, we generated stable double-negative (DN) bladder cancer cell line via lentiviral sgRNA-mediated CRISPR/Cas9 knockdown of both NEK2 and ZWINT. Multiple independent DN clones were established. qPCR and Western blot analyses confirmed robust and simultaneous suppression of NEK2 and ZWINT when compared with wild-type (WT), vector control, and single-gene knockdown lines (KD1, KD2, KD3), demonstrating efficient dual targeting (Fig. 4A and B).

Fig. 4.

Generation of the double-negative cell line and functional characterization. A, B, qPCR and Western blot analysis of NEK2 and ZWINT expression in WT cells, Control, KD1, KD2 and KD3 treated bladder cancer cells. C, D, qPCR and Western blot were used to analyze the expression levels of ZWINT and NEK2 in the double-negative cell line, adjacent tumor tissues, NMIBC and MIBC. E, Images of representative nude mouse xenograft model studies. Tumor xenografts excised from male BALB/c (nu/nu) nude mice after 21 days of treatments with DP and DN cells. F, The statistical graph showed the tumor weights and volume of mice bearing tumors treatments with DP and DN cells. G, H, IHC revealed the expression levels of ZWINT, NEK2, and Ki67 in xenografts derived from DP and DN cells. Statistical significance was determined by two-tailed Student’s t-test. Asterisks indicate significance levels: *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001

At the transcriptional and protein levels, the expression levels of NEK2 and ZWINT in DN cells were lower than those in the traditional cell lines RT112 and T24. When compared with clinical samples of NMIBC and MIBC, the molecular characteristics of DN cells were more similar to those of early or less invasive tumors, further supporting the causal role of this axis in driving disease progression [27] (Figs. 4C-D).When benchmarked against clinical samples, the molecular profile of DN cells more closely resembled that of early-stage or less aggressive tumors, whereas WT or NEK2/ZWINT-high conditions aligned with advanced disease, further supporting a causal role for this axis in driving progression.

To assess the impact of NEK2–ZWINT dual knockdown on tumorigenicity in vivo, we established subcutaneous xenograft models in male BALB/c nude mice using NEK2⁺ZWINT⁺ DP cells and their DN counterparts (Fig. 4E). During the 21-day observation period, tumors derived from DP cells exhibited rapid growth, forming large-sized masses with a steep growth kinetics profile. In contrast, the growth of xenografts derived from DN cells was significantly impaired. At the experimental endpoint, the subcutaneous tumors established from DP cells reached a weight of 0.58 g and a volume of 585 mm³, whereas those derived from DN cells showed marked reductions, with only 0.20 g in weight and 105 mm³ in volume [28] (Fig. 4F).

Histopathological and immunohistochemical analyses of xenografted tumors revealed pronounced differences between groups. DP tumors exhibited high cellular density, abundant mitotic figures, and strong staining for NEK2, ZWINT and the proliferation marker Ki67. Statistical analysis showed that the H-score of NEK2-ZWINT reached as high as 235, and that of Ki67 reached 268. In contrast, DN tumors displayed reduced cellular density, decreased mitoses, and a marked decline in the Ki67 index, accompanied by sustained suppression of NEK2 and ZWINT expression; the H-score of NEK2-ZWINT was only 35, and that of Ki67 was merely 32 (Fig. 4G and H). These data collectively demonstrate that NEK2 and ZWINT act cooperatively to sustain the proliferative and tumor-forming capacity of bladder cancer cells and that dual Knockdown effectively attenuates tumor growth in vivo.

NEK2–ZWINT co-expression confers proliferative and migratory advantages, whereas dual knockdown reverses these malignant phenotypes

To dissect how NEK2–ZWINT co-expression shapes the biological behavior of bladder cancer cells, we performed a series of in vitro functional assays comparing double-positive (DP) and double-negative (DN) cell lines in parallel with parental bladder cancer lines (RT112, 5637) and primary bladder cancer cells.

In the colony formation assay, the number and volume of colonies formed by DP cells were significantly higher than those formed by RT112, 5637 and primary bladder cancer cells, indicating a prominent clonogenic advantage. The number of colonies formed by DP cells reached as high as 125, compared with only 22 by RT112 cells, 25 by 5637 cells and 48 by primary bladder cancer cells (Fig. 5A). In contrast, both the number and volume of colonies formed by DN cells were markedly reduced, suggesting impaired long-term proliferative potential. Only 15 colonies were formed by DN cells, versus 33 by RT112 cells, 35 by 5637 cells and 58 by primary bladder cancer cells (Fig. 5C). Consistent with these findings, CCK-8 assays showed that DP cells had the steepest growth curves (Fig. 5B), whereas DN cells exhibited strikingly reduced proliferation over time [29] (Fig. 5D). These data indicate that NEK2–ZWINT co-expression is critical for sustaining high proliferative output.

Fig. 5.

Functional comparison between double-positive and double-negative cell lines. A, C, Representative images of colony formation assays and their quantification in DP cells, RT112, 5637 and bladder cancer cells (A). Representative images of colony formation assays and their quantification in DN cells, RT112, 5637 and bladder cancer cells (C). B, D, CCK-8 analysis showed the effect of DP or DN cells, RT112, 5637 and bladder cancer cells of the cell proliferation. E, F, Wound-healing assay highlighted the difference in migratory capacity between the double-positive and double-negative cell lines. G–I, CCK-8 analysis showed the effect of ZWINT-KD, NEK2-KD, DN cells on the cell proliferation of the tested bladder cancer cells. Statistical significance was determined by two-tailed Student’s t-test or one-way ANOVA followed by Tukey’s post-hoc test. Asterisks indicate significance levels: *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001

Subsequently, we evaluated the effect of the NEK2-ZWINT axis on cell motility. In the wound healing assay, DP cells efficiently closed the scratch gap, exhibiting a migratory phenotype reminiscent of epithelial-mesenchymal transition (EMT). The scratch closure rate of DP cells reached as high as 83%, compared with only 23% in RT112 cells, 26% in 5637 cells and 52% in primary bladder cancer cells (Fig. 5E). In contrast, DN cells showed extremely weak migratory capacity, with large residual wound areas remaining at the same time point. The scratch closure rate of DN cells was only 18%, versus 42% in RT112 cells, 45% in 5637 cells and 62% in primary bladder cancer cells (Fig. 5F). These observations dovetail with the EMT enrichment signature identified in NEK2–ZWINT-high tumors and suggest that this axis not only fuels proliferation but also facilitates migration, a key prerequisite for invasion and metastasis.

To parse the individual contributions of NEK2 and ZWINT, we examined single-gene knockdown lines. NEK2-KD and ZWINT-KD cells both showed reduced proliferation in CCK-8 assays compared with control cells (Fig. 5G and H), but neither single knockdown fully recapitulated the degree of functional impairment observed in DN cells. This partial overlap supports a cooperative or synergistic model in which NEK2 and ZWINT jointly regulate mitotic integrity, EMT-associated behavior, and overall malignant fitness [30] (Fig. 5I).

Collectively, these data demonstrate that NEK2–ZWINT co-expression provides bladder cancer cells with a dual advantage—enhanced proliferative capacity and increased motility—whereas dual knockdown reverses these malignant traits and shifts cells toward a less aggressive state.

NDC80 is embedded in the NEK2–ZWINT regulatory circuit and mediates tumor progression and gemcitabine resistance

To further elucidate the mechanistic circuitry downstream of the NEK2–ZWINT module, we focused on NDC80, a core component of the kinetochore complex that coordinates chromosome segregation and mitotic fidelity. We predicted via the online protein-protein interaction (PPI) database STRING that NEK2 and ZWINT interacts with NDC80 (Fig. 6A). Meanwhile, their strong correlations were further verified using bladder cancer data retrieved from The Cancer Genome Atlas (TCGA) database (Fig. 6B). Western blot analysis revealed that the expression of NDC80 was significantly decreased in NEK2⁺ and ZWINT⁺ single-positive cells compared with double-positive (DP) cells, albeit to a lesser extent, indicating that NDC80 acts downstream of these two regulators. Concurrently, we performed NDC80 knockdown and found that NDC80-KD failed to reduce the expression of NEK2 and ZWINT in DP cells, primary tumor (PT) tissues, and different subtypes of bladder cancer tissues (NMIBC and MIBC). These results collectively confirmed that NDC80 was localized downstream of NEK2 and ZWINT, thereby amplifying proliferation-related signaling pathways [31] (Fig. 6C).

Fig. 6.

Relationships among ZWINT, NEK2, and NDC80, and functional investigation of NDC80. A The protein interaction network of ZWINT and NEK2 predicted by the STRING database. B Pearson correlation of NEK2 and ZWINT with NDC80 in TCGA bladder cancer. C Western blot analysis showed the expression levels of NDC80 in different knockdown treatment groups (left), the levels of NDC80 in cell lines with individual knockdown of ZWINT or NEK2 (middle), and the changes in ZWINT and NEK2 expression following NDC80 knockdown (right). D Expression levels of NCD80 in early-stage and late-stage bladder cancer. E CCK-8 analysis showed the effect of NDC80-KD cells on the cell proliferation of the tested bladder cancer cells. F Gemcitabine sensitivity assay showed a higher 50% maximal inhibitory concentration (IC50) value of gemcitabine in shCtrl cells compared to shNDC80 cells. G Images of representative nude mouse xenograft model studies. Tumor xenografts excised from male BALB/c (nu/nu) nude mice after 21 days of treatments with shCtrl and shNDC80 cells. H The statistical graph showed the tumor weights and volume of mice bearing tumors treatments with shCtrl and shNDC80 cells. I Representative IHC staining images of NDC80 and Ki67 were presented in xenograft nude mice tissues. Scale bar, 100 μm. J–M, Tissue immunofluorescence analysis of NDC80 expression in early (Ⅰ+Ⅱ) (J), late (Ⅲ+Ⅳ) (K)grades, drug-sensitive (L) and drug-resistant (M) of bladder cancer. EPCAM stains bladder cancer tissues (green), NDC80 (red) are expressed in the tissues, and DAPI stains nuclei (blue). Scale bar, 100 μm. J, K, L, M, IF confirmed the expression levels of NDC80 in pathological samples from patients with different characteristics. Statistical significance was determined by two-tailed Student’s t-test. Asterisks indicate significance levels: *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001

Clinically, analysis of the Phase II atezolizumab trial cohort demonstrated that NDC80 expression was significantly higher in late-stage (III–IV) than in early-stage (I–II) bladder cancer tissues, aligning with its role in aggressive disease (Fig. 6D). Functional assays showed that NDC80 knockdown markedly impeded bladder cancer cell proliferation in CCK-8 assays, mirroring the effects of NEK2–ZWINT Knockdown and supporting the notion that NDC80 is a critical downstream effector within this axis [32] (Fig. 6E).

Because mitotic regulators are often linked to treatment response, we next examined the impact of NDC80 on chemotherapy sensitivity [33]. Gemcitabine dose-response assays demonstrated that the half-maximal inhibitory concentration (IC50) of gemcitabine in the DP Ctrl group reached approximately 4.5 µM, comparable to the parental DP cells (IC50 ≈ 4.6 µM). In contrast, DP NDC80-KD cells exhibited a markedly lower IC50 of only ~ 0.9 µM compared with DP Ctrl cells, while DN cells showed the lowest IC₅₀ (≈ 0.1 µM), indicating significantly enhanced sensitivity to gemcitabine following NDC80 knockdown (Fig. 6F). These findings implicate NDC80 as a key mediator of chemoresistance in this model.

In the in vivo experiments, after a 21-day observation period, the xenografts derived from NDC80 KD cells were significantly smaller and lighter than those derived from control cells. The specific values were as follows: the subcutaneous tumors established from DP Ctrl cells reached a weight of 0.78 g and a volume of 785 mm³, whereas the tumors derived from NDC80 KD cells were markedly reduced, with a weight of only 0.20 g and a volume of 103 mm³ (Fig. 6G and H), with reduced Ki67 staining and diminished NDC80 expression, confirming the in vitro observations (Fig. 6I). Importantly, immunofluorescence analyses of human bladder cancer specimens demonstrated that NDC80 expression increased with histological grade and was enriched in gemcitabine-resistant tumors compared with drug-sensitive ones (Fig. 6J-M). In these tissues, NDC80 positivity overlapped with EPCAM-marked tumor epithelium, underscoring its tumor cell–intrinsic nature and its association with aggressive, treatment-refractory disease.

Together, these data establish NDC80 as an integral component of the NEK2–ZWINT oncogenic circuit, linking dysregulated mitotic control to tumor progression and chemoresistance. The NEK2–ZWINT–NDC80 axis thus emerges as a coherent signaling module that coordinates proliferation, EMT-associated behavior, and therapeutic response in bladder cancer.

Discussion

This study identifies a previously unrecognized NEK2–ZWINT–NDC80 regulatory axis that governs bladder cancer aggressiveness and chemoresistance. Through an integrative approach incorporating patient tissues, flow-sorted subpopulations, molecular perturbation, and in vivo models, we uncover a regulatory framework in which NEK2 and ZWINT cooperatively sustain malignant phenotypes, and NDC80 serves as a downstream amplifier of this axis [34].

A major conceptual advance is the identification of NEK2⁺ZWINT⁺ double-positive cells as a stable and aggressive tumor subpopulation enriched in high-grade tumors. Their EMT-associated transcriptomic profile and spatial localization at invasive fronts point to a functionally distinct malignant state, one that has likely been under-recognized in bladder cancer biology [35].

We further demonstrate that dual loss of NEK2 and ZWINT produces far more profound phenotypic suppression than individual knockdowns, highlighting their cooperative role in regulating tumor proliferation and motility. This synergy reflects a deeper mechanistic integration that becomes fully evident with the identification of NDC80 as a shared effector [36].

Our results identify a coordinated NEK2/ZWINT–NDC80 regulatory axis. While these components are highly correlated in patient samples, our functional assays primarily confirm that NEK2 and ZWINT promote tumor progression via the upregulation of NDC80. Future studies are needed to explore whether NDC80 provides reciprocal regulation to its upstream activators. Its association with gemcitabine resistance provides a plausible mechanistic explanation for therapeutic failure in high-grade disease and suggests NDC80 as a potential target to overcome chemoresistance [37].

Our findings also expand the conceptual landscape of bladder cancer biology by emphasizing kinetochore-associated regulatory networks, which have been comparatively overlooked relative to canonical pathways. By revealing how chromosome segregation machinery intersects with EMT programs and drug response, this study opens new avenues for therapeutic intervention [38–40].

Several limitations remain. Orthotopic or PDX models could provide deeper translational validation. The biochemical mechanisms linking NEK2–ZWINT activity to NDC80 regulation require further elucidation. Additionally, whether this axis modulates responses to other chemotherapeutics or immunotherapies merits investigation (Fig. 7).

Fig. 7.

Proposed working model of the NEK2–ZWINT–NDC80 regulatory axis in bladder cancer. High-grade bladder cancers exhibit coordinated upregulation of NEK2 and ZWINT, which identifies an aggressive tumor cell subpopulation and initiates downstream activation of NDC80. NEK2 and ZWINT cooperatively regulate NDC80 expression, where NEK2 and ZWINT cooperatively drive NDC80 expression, collectively sustaining the aggressive potential of bladder cancer cells. This regulatory axis promotes proliferation, migration, EMT-associated cellular states, clonogenicity, and in vivo tumorigenicity, ultimately contributing to tumor progression and gemcitabine resistance. Targeting components of this axis may offer therapeutic opportunities for patients with high-risk bladder cancer

In summary, this study demonstrates that the NEK2–ZWINT–NDC80 axis represents a central regulatory circuit driving bladder cancer progression and chemoresistance. By bridging transcriptional co-activation, cell-state identity, downstream effector biology, and clinical resistance, our findings provide a mechanistic foundation for developing targeted therapeutic strategies in high-risk bladder cancer [41].

Conclusions

This study identifies a cooperative NEK2–ZWINT–NDC80 axis that drives bladder cancer progression and chemoresistance. NEK2⁺ZWINT⁺ cells represent an aggressive subpopulation marked by enhanced proliferation, EMT activation, and migratory capacity. Dual Knockdown of NEK2 and ZWINT markedly suppresses tumor growth, while NDC80 functions downstream to reinforce malignant signaling and promote gemcitabine resistance. These findings highlight this axis as a promising therapeutic target for high-risk bladder cancer.

Electronic Supplementary Material

Below is the link to the electronic supplementary material.

Author contributions

All authors contributed substantially to the conception and execution of the study, encompassing clinical sample acquisition, specimen processing, data curation, bioinformatics analyses, and critical revision of the manuscript. All authors have reviewed and approved the final version of the manuscript and accept full accountability for the accuracy, rigor, and scientific integrity of their individual contributions. Xiaowei Hu and Yubo Zhao, Zhichao Tong conceived and designed the study. Zhichao Tong, Hongjian Song supervised the project. Zhe Wang, Qi Liu, Ziyi Liu, Haonan Li, Haiqiang Duan, were responsible for clinical sample collection. Yaowei Li, Shiyu Huang, Qing Shi, Tianxi Yu, Guangzheng Wu, Shuoyi Yan, Zhishuai Zhang, Di Wang and Zongzheng Yang curated clinical data, performed pathological classification, and verified data accuracy. Zhihao Yin, Jianwei Wang, and Peng Zhang, Peng Dai and Yaowei Li drafted the manuscript.

Funding

This work was supported by the National Natural Science Foundation of China (82573847; 82572347; 81902569), The Nn10 project at the Affiliated Cancer Hospital of Harbin Medical University (Grant No: Nn102024-01), Heilongjiang Province Postdoctoral Foundation (LBH-Z22030), Excellent Youth Project of Heilongjiang Provincial Natural Science Foundation (YQ2024H023), the Harbin Medical University Cancer Hospital Haiyan Foundation (JJZD2024-24; JJQN2022-07), the Heilongjiang Provincial Natural Science Foundation (PL2024H123), Collectively, these funding sources enabled the comprehensive execution of this study.

Data availability

The datasets analysed during the current study are available from the TCGA-BLCA cohort, the GEO Database (https://www.ncbi.nlm.nih.gov/geo/) including GSE135337 (n = 8), GSE169379 (n = 26) and GSE222315 (n = 13), and the IMvigor210 dataset (http://research-pub.gene.com/IMvigor210CoreBiologies/). The codes used during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

All animal experiments were performed in accordance with the guidelines approved by the Animal Care and Use Committee of Harbin Medical University and complied with all relevant ethical regulations for animal use (GJZDYF2024-001). The maximal tumor burden permitted by the ethics committee was a tumor volume not exceeding 1,500 mm³. During the course of the study, tumor growth was carefully monitored, and the maximal allowed tumor size was not exceeded in any experimental animal. All procedures involving human participants were conducted in accordance with the Declaration of Helsinki and were approved by the Ethics Committee of Harbin Medical University (KY2024-100). Informed Consent to Participate was obtained from the participate to conduct the study.

Consent for publication

Written informed consent for publication was obtained from all participants.

Competing interests

The authors declare no competing interests.