Targeting an Autocrine Regulatory Loop in Cancer Stem-like Cells Impairs the Progression and Chemotherapy Resistance of Bladder Cancer

Bladder cancer is one of the most frequently diagnosed and lethal cancers worldwide, with an estimated 429,800 new cases and 165,100 deaths in 2012 (1). Although surgical resection is the preferred treatment for patients with bladder cancer, most tumors recur and progress to muscle-invasive bladder cancers (MIBC) and metastatic bladder cancer, which are prone to develop chemotherapy resistance after treatment and are associated with poor prognosis (2, 3). Cancer stem-like cells (CSCs) are a small subpopulation of tumor cells that are functionally defined by their strong stem-like properties, including self-renewal, drug resistance, and tumor initiation capacity upon serial passage, and the heterogeneity, progression and chemotherapy resistance of many cancers, including bladder cancer, have been attributed to CSCs (4–6). Therefore, unveiling the molecular mechanisms by which CSCs are regulated will likely facilitate the development of novel and efficacious therapies for advanced and chemoresistant bladder cancer.

Bladder CSCs can be isolated using several markers, such as CD44, SOX2, CD133, and aldehyde dehydrogenase 1 A1 (ALDH1A1; refs. 7, 8). In addition, many signaling pathways have been reported to regulate the self-renewal, chemotherapy resistance, and tumor initiation properties of bladder CSCs, including hedgehog signaling, the Wnt/β-catenin pathway, and the KMT1A-GATA3-STAT3 and E2F1-EZH2-SUZ12 cascades (9–12). A recent study reveals that variants of ARID1A, GPRC5A, and MLL2 drive the self-renewal of bladder CSCs, as indicated by single-cell sequencing (13). However, the mechanisms underlying the maintenance of the stem-like properties of bladder CSCs remain insufficiently understood.

Hippo signaling was initially identified in Drosophila and has been more recently examined for its regulatory role in mammalian cells; the Hippo pathway was found to play crucial roles in mediating carcinogenesis and self-renewal in stem cells and CSCs (14). The Hippo pathway effector Yes-associated protein (YAP) is negatively regulated by the mammalian Ste20-like kinases 1/2 (MST1/2) and large tumor suppressor 1/2 (LATS1/2), and this negative regulation results in retention of phosphorylated YAP in the cytoplasm (15). Conversely, YAP that is not phosphorylated enters the nucleus and functions as a transcription coactivator for transcription factors of the TEA domain transcription factor (TEAD) family to induce the expression of downstream genes that promote cell proliferation and differentiation (15). Recent studies have demonstrated that YAP is required for CSC self-renewal and expansion in various cancer types, such as breast cancer, prostate cancer, glioblastoma, and lung cancer (16–18). However, the role of YAP in bladder CSCs and the related mechanisms have yet to be established.

The epithelial marker OV6 serves as a putative marker for hepatic oval cells, bile duct epithelial cells, and also a CSC marker in epithelium-derived malignant tumors; it is associated with patient prognosis as shown in many studies (19–24). In this study, we found that high OV6 expression in specimens positively associated with disease progression and poor prognosis of patients with bladder cancer. Moreover, OV6⁺ bladder cancer cells harbored strong stem-like properties, and the pattern of OV6 expression in bladder CSCs was very similar to those of CD44 and CD133. In addition, we demonstrated that YAP drives the self-renewal of OV6⁺ CSCs to facilitate the invasion, migration and drug resistance of bladder cancer. Furthermore, YAP triggers PDGFB transcription via TEAD1, and reciprocally, PDGF-BB binds to its receptor PDGFR and stabilizes YAP by preventing its phosphorylation by LATS1/2, thus forming an autocrine-regulatory loop in OV6⁺ CSCs. Finally, an orthotopic bladder cancer mouse model was employed to show that blocking the autocrine regulatory loop in OV6⁺ CSCs using a YAP or PDGFR inhibitor overcame the resistance to cis-platinum in advanced bladder cancer.

Two cohorts, cohort 1 (n = 130, January 2009 to December 2013) and cohort 2 (n = 95, January 2004 to December 2008), of patients with bladder cancer from Changhai Hospital (Shanghai, China) were recruited in this study. The patients were merged and randomly divided by 1:1 ratio into a training (n = 113) and a validation set (n = 112). In addition to the cohorts above, this study also included a patient with recurrent bladder cancer (n = 1), patients with metastatic bladder cancer (n = 6), and patients with bladder cancer treated with cis-platinum (n = 5). Histologic grade and tumor stage were assessed according to the American Joint Committee on Cancer (2010) and the World Health Organization classification (2004; ref. 25).

Clinical follow-up data were available for all patients. The median follow-up period was 59.17 months (mean 50.3 months) for cohort 1 and 122.1 months (mean 91.5 months) for cohort 2. All patients were investigated according to a uniform method of the Department of Urology, Changhai Hospital (Shanghai, China) based on the guidelines (26, 27). Cystoscopy was applied for patients treated with TURBT at the first 3 months after surgery. For patients with low-risk tumors, cystoscopy was performed 1 year after surgery if the first cystoscopy was negative, and then yearly for 5 years. For patients with high-risk tumors, cystoscopy was done every 3 months for 2 years, then every 6 months for 2 years, and then yearly. Patients with intermediate-risk tumors had an in-between follow-up scheme using cystoscopy, which is adapted according to personal and subjective factors. For patients treated with radical cystectomy, Return visits were generally performed postoperatively at least every 3 to 4 months for the first year, semiannually for the second year, and annually thereafter. Follow-up visits consisted of a physical examination, urine cytology, serum chemistry evaluation, chest radiography, ultrasound examination (including liver, kidney and retroperitoneum), CT/MRI of the abdomen and pelvis.

The samples were obtained after written informed consent was provided by the patients according to an established protocol approved by the Ethics Committee of Changhai Hospital (Shanghai, China). The study was conducted in accordance with the International Ethical Guidelines for Biomedical Research Involving Human Subjects (CIOMS).

IHC was performed as described previously (28), and the following primary antibodies were used: mouse anti-OV6 (MAB2020, R&D Systems), rabbit anti-YAP (ab52771), mouse anti-Vimentin (ab8978), and rabbit anti-PDGF receptor beta (ab32570; Abcam). The percentage of positive cells (% of PPs) and the staining intensity (SI value) were determined and multiplied to obtain the immunoreactive score (IRS value; ref. 29), which ranged from a minimum score of 0 to a maximum score of 12. Time-dependent receiver-operating characteristic (ROC) analysis was performed to determine the cut-off value and AUC of OV6 expression in predicting 5-year cancer-specific survival (30) using R software (4.3.3). Patients with bladder cancer were divided into 2 groups: a low expression (IRS values of 0–3) group and a high expression (IRS values of 4–12) group.

The cell lines used in this study were obtained from the Cell Bank of the Type Culture Collection of the Chinese Academy of Sciences (Shanghai, China) in June 2016. UMUC3 and J82 cells were cultured in minimum essential medium (1871544, Gibco) supplemented with FBS (10%, 10099-141, Gibco). RT4 and T24 cells were maintained in McCoy 5A medium (1849736, Gibco) modified with FBS (10%, 10099-141, Gibco). SW780 cells were cultured in Leibovitz L-15 medium (1828421, Gibco) containing 10% FBS (Gibco). All cell lines were supplemented with 1% penicillin/streptomycin (15140122, Gibco) and cultured at 37°C in 5% CO₂.

The cell lines in this study were authenticated by short tandem repeat (STR) profiling and tested for Mycoplasma contamination with a Mycoplasma Detection Kit (B39032, Selleck Chemicals). The most recent tests were performed in December 2017. The cell lines used in this study were within 40 passages.

Flow cytometric assays were performed with a Cyan ADP Sorter (Beckman Coulter). Bladder cancer cell lines were magnetically labeled with APC-conjugated-OV6 antibody (FAB2020A, R&D Systems), FITC-conjugated–CD44 antibody (130-095-195, Miltenyi Biotec) or PE-conjugated-CD133 antibody (130-113-670, Miltenyi Biotec). The magnetic-activated cell sorting (MACS) assay was performed with a MiniMACS Cell Sorter (Miltenyi Biotec). Bladder cancer cell lines were magnetically labeled with OV6 antibody (MAB2020, R&D Systems), CD44 antibody (130-110-082, Miltenyi Biotec), or CD133 antibody (130-090-423, Miltenyi Biotec) and then subsequently incubated with rat anti-mouse IgG1 microbeads and separated on a MACS MS column (Miltenyi Biotec). All the procedures were carried out according to the manufacturer's instructions.

Briefly, after magnetic sorting, single-cell suspensions with 3,000 cells were seeded in 6-well ultra-low attachment culture plates (Corning) and cultured in serum-free DMEM/F12 (Gibco) supplemented with B27 (1:50, 17504-044, Thermo Fisher Scientific), N2 (1:100, 17502048, Thermo Fisher Scientific), 20 ng/mL EGF (Gibco), 10 ng/mL bFGF (Gibco), and ITS (1:100, 13146, Sigma) for 5 days. The number of spheroids formed was determined via microscopy, and representative images were obtained.

The spheroid formation assay was carried out in semisolid medium. In brief, single-cell suspensions with 3,000 cells were resuspended in 1:1 growth factor–reduced Matrigel (BD Biosciences, 356231)/serum-free DMEM/F12 (supplemented with B27, N2, EGF, bFGF, and ITS) in a total volume of 200 μL. Samples were plated around the rims of wells in a 12-well plate and allowed to solidify at 37°C for 10 minutes before 1 mL of serum-free DMEM/F12 (supplemented with B27, N2, EGF, bFGF, and ITS) was added. Medium was replenished every 3 days. Ten days after plating, spheres with a diameter over 50 μm were counted via microscopy, and representative images were obtained.

For the single-cell spheroid formation assay, bladder cancer cells were deposited by FACS in wells from the ultra-low-adhesion 96-well plates containing 200 μL stem cell medium (serum-free DMEM/F12 supplemented with B27, N2, EGF, bFGF, and ITS) at a concentration of a single cell per well, which was confirmed visually by microscope. Wells containing either none or more than 1 cell were excluded for further analysis. The single-cell wells were checked daily and a further 100 μL of stem cell medium was added after 5 days. Ten days after plating, wells with sphere were counted under a microscope, and representative images were obtained.

The proliferation of bladder cancer cells under the indicated conditions was evaluated using a CCK-8 kit (CK-04, Dojindo), which was described in our previous study (31). The proliferation rates are presented as a proportion of the control value, which was obtained from the treatment-free groups. Invasion and migration assays were carried out in transwell chambers (Millipore) with or without Matrigel (BD Biosciences) according to the manufacturer's instructions, which was explained in our previous study (22).

Apoptotic cells were evaluated with ANXA5 and propidium iodide (PI) staining (Invitrogen, A13201) according to the manufacturer's instructions and analyzed by flow cytometry with a Cyan ADP Sorter (Beckman Coulter).

The short hairpin RNA (shRNA) interference vector pLKO.1-GFP, containing a U6 promoter upstream of the shRNA, and the lentivirus packaging vectors pVSVG-I and pCMV-GAG-POL were obtained from Shanghai Integrated Biotech Solutions Co., Ltd. (Shanghai, China). The UMUC3/J82 cell line was transduced with the shRNA-expressing lentivirus (sh-YAP or sh-TEAD1) or control lentivirus. After 72 hours of transduction, the cells were observed and photographed under microscope. Stable UMUC3/J82 knockdown of YAP and TEAD1 were also generated using lentiviral constructs. The shRNA sequences are presented in Supplementary Table S18. Cell transfection with YAP and TEAD1 plasmid was carried out using Lipofectamine 3000 reagent (L3000015, Invitrogen) according to the manufacturer's protocol, and the sequence of the YAP and TEAD1 plasmid is shown in Supplementary Table S18.

Total RNA was extracted with RNAiso Plus (9109, Takara), and cDNAs were synthesized using a PrimeScript One Step RT reagent Kit (RR037B, Takara). Real-time PCR was performed with SYBR Green Real-Time PCR Master Mix (QPK201, Toyobo) on an ABI PRISM 7300HT Sequence Detection System. The results were normalized to the expression of β-actin. Fold change relative to the mean value was determined by 2^−ΔΔC_t. The primer sequences are presented in Supplementary Table S18.

The Western blot analysis was performed as we reported previously (24). Nuclear and cytoplasmic proteins from bladder cancer cells were extracted using an NE-PER Nuclear and Cytoplasmic Extraction Kit (#78899, Thermo Fisher Scientific). The following primary antibodies were used: rabbit anti-β-tubulin (#2128) and rabbit anti-Histone H3 (#4499), rabbit anti-phospho-p44/42 MAPK (Erk1/2; Thr202/Tyr204; #4370), rabbit anti-phospho-MEK1/2 (Ser217/221; #9154) from Cell Signaling Technology; rabbit anti-LATS1+LATS2 (ab70565), rabbit anti-LATS1+LATS2 (phospho-T1079+T1041; ab111344), rabbit anti-PDGF receptor beta (ab32570), rabbit anti-YAP (ab52771), and rabbit anti-YAP (phospho-S127; ab76252), rabbit anti-PDGF receptor beta (phospho Y1021) (ab134048), rabbit anti-ERK1 + ERK2 (ab17942), and rabbit anti-MEK1+MEK2 (ab178876) from Abcam; and mouse anti-TEF-1 (610922) from BD Biosciences. The secondary antibodies were horse anti-mouse IgG-HRP–linked antibody (#7076S) or goat anti-rabbit IgG-HRP–linked antibody (#7074S) from Cell Signaling Technology. Co-immunoprecipitation (co-IP) experiments were performed according to previously published protocols (24). The antibodies used are listed above.

The immunofluorescence analysis was performed as we reported previously (24). The primary antibodies rabbit anti-YAP (ab52771, Abcam) and mouse anti-OV6 (MAB2020, R&D Systems) were incubated with samples at 4°C overnight. Nuclei were stained with DAPI (E607303, Sangon). Fluorescence images were observed and collected under a Leica DM5000B fluorescence microscope (Leica).

The antibody–microarray experiment was performed as reported previously (31). Cytokine profiles were examined by Quantibody Human Inflammatory Array 3 (QAH-INF-G3, RayBiotech) containing 40 inflammation-associated cytokines. The PDGF-BB and ICAM-1 levels in cell culture medium were measured using ELISA Kits for PDGF-BB (DBB00, R&D Systems) and ICAM-1 (DCD540, R&D Systems) according to the manufacturer's instructions.

Chromatin immunoprecipitation (ChIP) assays were performed according to our previously published study (31). Primers complementary to the promoter region of PDGFB (forward: 5′- TGGCAGAGCAGGTTCCCACATA; reverse: 5′-TGCTGAGACCACCGTGCTGT-3′) were used to detect PDGFB genomic DNA, and primers specific to the human GAPDH promoter were used as the control (kit supplied). Enrichment of the targets was calculated as follows: fold enrichment = 2^{[C_t (PDGFB-ChIP) − C_t (IgG)]}.

The TEAD1-binding sites of the PDGFB promoter (NC_000022.11: −3000 to +100 relative to the PDGFB transcription site) or its mutant sequence (−2732 to −2723, CTCATTCCAT) were cloned into a pGL3-basic luciferase reporter vector (Promega). UMUC3 cells were cotransfected with 10 ng pTK-RL reporter control plasmid and 200 ng pGL3-basic-PDGFB-WT or pGL3-basic-PDGFB-Mut using Lipofectamine 3000 reagents (L3000015, Invitrogen) according to the manufacturer's protocols. Cells were collected 48 hours after transfection, and PDGFB transcription activity was evaluated by measuring luminescence with a Dual-Luciferase Assay Kit (E1910, Promega). Fold induction was derived relative to normalized reporter activity.

RNA was isolated from OV6⁻ and OV6⁺ UMUC3 cells using TRIzol reagent (Invitrogen). The total RNA was purified with a Qiagen RNeasy Mini Kit (Qiagen), and then, the purified RNA was checked to determine the quantity. Single- and double-stranded cDNA was synthesized from mRNA samples. The double-stranded cDNA was then purified for end repair, dA tailing, adaptor ligation, and DNA fragment enrichment. The libraries were quantified using Qubit (Invitrogen) according to the Qubit user guide. The constructed library was sequenced on an Illumina Hiseq 4000 sequencer. The paired end raw reads were aligned using TopHat version 1.2.0, which allowed 2 mismatches in the alignment. The aligned reads were assembled into transcripts using cufflinks version 2.0.0. The alignment quality and distribution of the reads were estimated using SAM tools. From the aligned reads, the gene and transcript expression was assessed using cufflinks version 1.3.036. The differential transcripts were analyzed via cuffdiff. Finally, GO and pathway functional analyses were performed for the differentially expressed transcripts.

All experimental animal procedures were approved by the Animal Care and Use Committee of the Second Military Medical University (Shanghai, China). UMUC3 cells were transfected with the luciferase reporter gene. For the in vivo limiting dilution assay, sorted tumor cells were diluted at an appropriate cell dose and injected into NOD/SCID mice (Shanghai Laboratory Animal Center, China); the number of tumors formed from each cell dose injected was then scored. The frequency of CSCs was calculated using the ELDA software (http://bioinf.wehi.edu.au/software/elda/index.html; The Walter and Eliza Hall Institute, Parkville, Victoria, Australia). In the animal experiments, human recombinant PDFG-BB (10 ng/mL), PDGF-BB–neutralizing antibody (100 ng/mL), or CP-673451 (500 nmol/mL) was added to the culture medium of sorted cells for 4 days before subcutaneously injecting into the mice.

For the in vivo metastasis assay, 5 × 10⁵ cells in 200 μL of PBS were intravenously injected through the tail vein of 6-week-old male NOD/SCID mice. Mice were sacrificed and examined 5 weeks after tumor cell injection. For mouse orthotopic bladder cancer, 6-week-old female NOD/SCID mice were used in this study. Briefly, 5 × 10⁵ UMUC3-Luc-OV6⁻ cells or UMUC3-Luc-OV6⁺ cells in 50 μL of PBS were respectively installed into the bladder via a 24-gauge catheter and maintained for 3 hours according to a previously described protocol (32). Tumor growth was monitored weekly by live-animal bioluminescence optical imaging using an IVIS Lumina II imaging system (PerkinElmer) after intraperitoneal injection of d-luciferin (150 mg/kg; Gold Biotech) in 100 μL of DPBS. Mice were sacrificed 4 weeks after tumor implantation.

For the in vivo treatment model, orthotopic tumor-bearing mice, generated by installation of 1 × 10⁶ UMUC3-Luc-OV6⁺ cells or T24-Luc-OV6⁺ cells in 50 μL of PBS, were divided into 4 groups on day 5 after tumor implantation. Cis-platinum (S1166, Selleck Chemicals), verteporfin (5305, Tocris Bioscience), and CP-673451 (HY-12050, MedChemExpress) were stocked in DMSO and prepared in PBS. Mice were treated by intraperitoneal injection of cis-platinum (3 mg/kg) alone or combined with verteporfin (100 mg/kg) or CP-673451 (30 mg/kg) every 2 days. Necropsies were performed after 6 weeks.

Numerical data are expressed as the mean ± SD. Statistical differences between variables were analyzed with a 2-tailed Student t test, χ² test, or Fisher exact test for categorical/binary measures or ANOVA for continuous measures. Correlation studies were analyzed by Pearson. Survival curves were plotted using the Kaplan–Meier method and compared via log-rank analysis. Variables with P < 0.1 on the univariate analysis were included in multivariate Cox proportional hazards analysis. Differences were considered significant at P < 0.05. Time-dependent ROC analysis was used to determine the cut-off value of censored data with “survival ROC” package, R software 3.4.4. Time-dependent area under the receiver-operating characteristic curve (AUC) was computed with the “time ROC” package. All the analyses were performed using the GraphPad Prism 5 (GraphPad Software, Inc.), SPSS 21.0 (IBM Corporation) software and R Project for Statistical Computing (version 3.4.4).

On the basis of the previous studies that OV6 closely associates with progression and prognosis of patients with tumors (22–24), we also examined whether the epithelial marker OV6 is associated with the clinicopathologic characteristics and prognosis of patients with bladder cancer. First, we analyzed OV6 expression in specimens via IHC. The positive and selective staining of OV6 in normal bladder epithelial cells staining of OV6 in normal bladder epithelial cells (Fig. 1A) suggested that it could serve as an epithelial marker. In addition, OV6 expression in bladder cancer specimens was determined by immunoreactive scores (IRS; ranging from a minimum of 0 to a maximum of 12; the rules are described in the “Materials and Methods”). Interestingly, the OV6 expression was higher in the muscle-invasive bladder cancer (MIBC) than in the nonmuscle-invasive bladder cancer (NMIBC) samples (Fig. 1A). In addition, OV6 expression was higher in the muscular layer than in the mucosa of the same specimen from patients with MIBC (Supplementary Fig. S1A). Of note, higher OV6 expression was observed in the invasive fronts of bladder cancer than in the core (Supplementary Fig. S1A). Metastatic lesions or postchemotherapy samples exhibited higher OV6 expression than primary tumors or prechemotherapy samples (Fig. 1B and C). These data indicate that OV6 is positively correlated with progression of bladder cancer.

Then, we determined whether OV6 expression in specimens was associated with disease progression and the prognosis of patients with bladder cancer. First, the OV6 expression level in specimens was assessed by IHC from patients with bladder cancer (Fig. 1D), and then, a time-dependent ROC curve analysis was performed to determine the cutoff between low and high OV6 expression in cohort 1; this analysis demonstrated that the best cut-off value was 4 (IRS score; Supplementary Fig. S1B). Thus, patients with bladder cancer from cohort 1 were divided into 2 groups: a low OV6 expression group (IRS values of 0–3) or a high OV6 expression (IRS values of 4–12) group, and OV6 expression in bladder cancer samples was significantly associated with multiple malignant clinicopathologic features of bladder cancer tumors, such as tumor stage, tumor grade, lymph node status, and TNM stage (Supplementary Table S1). More importantly, Kaplan–Meier analysis revealed that patients with higher OV6 expression exhibited markedly worse cancer-specific survival (CSS; P < 0.001), progression-free survival (PFS; P < 0.001), and overall survival (OS; P < 0.001) than their counterparts (Supplementary Fig. S1C–S1E). Furthermore, univariable and multivariable Cox regression analyses indicated that OV6 was an independent risk factor for prognosis of patients with bladder cancer (Supplementary Table S2). In addition, another cohort (cohort 2; Supplementary Table S3), comprising 95 consecutive bladder cancer patients, was employed as a validation set using the cut-off value (IRS = 4) derived from cohort 1. Kaplan–Meier analysis revealed that patients with higher OV6 expression exhibited markedly worse CSS (P = 0.018), PFS (P = 0.033), and OS (P < 0.001) than their counterparts (Supplementary Fig. S1F–S1H). Moreover, incorporating OV6 expression to TNM stage improved accuracy in predicting CSS of patients with bladder cancer as shown by time-dependent AUC analysis both in cohort 1 and cohort 2 (Supplementary Fig. S1I and S1J).

Furthermore, patients with bladder cancer in cohort 1 were merged with cohort 2, and all 225 patients with bladder cancer were randomly split into a training set (n = 113) and a validation set (n = 112) at 1:1 ratio. A time-dependent ROC curve analysis was performed to determine the cutoff between low and high OV6 expression. Consistent with the best cutoff derived from cohort 1, the best cut-off value was also 4 (IRS score; Fig. 1E). Using the cutoff 4 for high and low OV6 expression, high OV6 expression in bladder cancer samples was also significantly associated with high tumor stage, tumor grade, positive lymph node status, and high TNM stage in both the training set and the validation set (Supplementary Table S4). More importantly, Kaplan–Meier analysis revealed that patients with higher OV6 expression exhibited markedly worse CSS (P < 0.001, training set; P = 0.007, validation set), PFS (P < 0.001, training set; P = 0.043, validation set), and OS (P < 0.001, training set; P = 0.003, validation set) than their counterparts (Fig. 1F–K). In addition, univariate and multivariate Cox regression analyses indicated that OV6 was an independent risk factor for the CSS of patients with bladder cancer in both the training and validation sets (Supplementary Table S5). Moreover, improved accuracy in predicting CSS of patients with bladder cancer was also observed by combining OV6 expression with the TNM stage in both sets in time-dependent AUC analysis (Supplementary Fig.S1K and S1L). These results demonstrate that OV6 could serve as a helpful indicator of disease progression and prognosis for patients with bladder cancer.

Given that OV6 is associated with disease progression and prognosis of patients with bladder cancer and serves as a CSC maker for other tumors, we speculated that OV6 was a putative CSC marker for bladder cancer. First, we detected the percentage of OV6 in bladder cancer cell lines and determined whether the pattern of OV6 expression in bladder cancer CSCs was similar to other CSC markers. The flow cytometry analysis demonstrated that higher OV6 expression was observed in sphere-forming bladder cancer cells (which were reported to possess CSC properties; ref. 33) than in adherent cells from various bladder cancer cell lines (Fig. 2A). Second, immunofluorescence confirmed colocalization of OV6 with other CSC markers, such as CD44 or CD133, in bladder cancer specimens or spheres from the bladder cancer cell line UMUC3 (Fig. 2B and C; Supplementary Fig. S2A and S2B). Third, multimarker analyses indicated that OV6⁺ UMUC3 or J82 cells also expressed CD44 or CD133 and that the percentage of OV6⁺CD44⁺ or OV6⁺CD133⁺ cells was, respectively, increased in spheres (Supplementary Fig. S2C and S2D). Fourth, flow cytometry showed that OV6 was increased in CD133⁺ UMUC3 cells, which were sorted by magnetic-activated cell sorting (MACS), and CD133 was also upregulated in OV6⁺ UMUC3 cells (Supplementary Fig. S2E and S2F). These data suggest that the pattern of OV6 expression in bladder cancer CSCs may be similar to that of CD44 or CD133.

We then isolated OV6⁺ and OV6⁻ UMUC3 or J82 cells via MACS to compare their stem-like properties (Supplementary Fig. S2G). First, higher expression levels of stem-associated genes were observed in OV6⁺ cells than in control OV6⁻ cells via real-time PCR (Fig. 2D). Second, a spheroid assay in serum-free medium or in semisolid medium revealed that significantly more spheres were formed by OV6⁺ bladder cancer cells than by OV6⁻ cells (Fig. 2E; Supplementary Fig. S2H). In addition, single-cell spheroid formation assay revealed that OV6⁺ bladder cancer cells had significantly stronger capacity to generate spheroids than OV6⁻ cells (Supplementary Fig. S2I), indicating that OV6⁺ bladder cancer cells harbored strong self-renewal. Third, a tumor limited dilution assay to analyze OV6⁺ bladder cancer cells stably labeled with luciferase that were subcutaneously injected of into nonobese, diabetic, severe combined immunodeficient (NOD/SCID) mice was employed to determine whether OV6⁺ bladder cancer cells were more tumorigenic than control OV6⁻ cells in vivo. As expected, OV6⁺ bladder cancer cells initiated tumors with markedly higher frequency than OV6⁻ cells in 2 serial generations. (Fig. 2F; Supplementary Fig. S2J and S2K; Supplementary Table S6). Furthermore, we showed that OV6⁺CD44⁺ UMUC3 cells exhibited higher tumor initiation incidence and frequency than OV6⁻CD44⁺ cells in NOD/SCID mice, but there was not a statistical difference in tumorigenicity between OV6⁺CD44⁺ and OV6⁺CD44⁻ cells (Supplementary Fig. S2L; Supplementary Table S7), indicating that OV6 plays a stronger role than CD44 in the tumor initiation capacity of CSCs. Moreover, given that CSCs were resistant to chemotherapeutic drugs, we also examined whether OV6⁺ bladder cancer cells were more chemoresistant compared with OV6⁻ cells. Because chemotherapeutic drugs can enrich CSCs in tumors, a flow cytometric analysis was performed to detect the percentage of OV6⁺ cells in bladder cancer cells after treatment with chemotherapeutic drugs. As expected, cis-platinum enriched the ratio of OV6⁺ cells in UMUC3 or J82 cell populations (Fig. 2G). A Cell Counting Kit-8 (CCK-8) assay demonstrated that OV6⁺ CSCs exhibited higher cell survival during cis-platinum treatment than OV6⁻ cells (Fig. 2H). In addition, annexin V/ propidium iodide (PI) double staining revealed that cis-platinum treatment induced fewer apoptotic events in OV6⁺ CSCs (UMUC3, 11.1% ± 1.1%; J82, 12.89% ± 2.1%) but considerably more apoptotic events in OV6⁻ cells (UMUC3, 47.5% ± 5.1%; J82, 85.4% ± 9.45%; Fig. 2I). These results indicate that OV6⁺ bladder cancer cells possess strong stem-like properties.

Given that CSCs possess stronger tumorigenicity and facilitate tumor metastasis (28), we assessed the in situ tumorigenicity and metastatic abilities of OV6⁺ CSCs in bladder cancer. First, OV6⁺ and OV6⁻ cells obtained from UMUC3 cell populations stably labeled with luciferase were perfused into murine bladders through a urethral catheter, and the resulting bioluminescence was detected using an in vivo imaging system to monitor tumor growth (Supplementary Fig. S2M). OV6⁺ CSCs exhibited a higher rate of tumor formation in bladders than control OV6⁻ cells (Supplementary Fig. S2M-O). Third, Matrigel invasion assays and transwell migration assays respectively showed that OV6⁺ CSCs exhibited a higher rate of cell invasion and migration than OV6⁻ UMUC3 or J82 cells (Supplementary Fig. S2P and S2Q). In addition, OV6⁺ and OV6⁻ UMUC3 cells were injected into the caudal vein of mice, and the injection of OV6⁺ cells yielded increases in the incidence, number and size of lung metastases than OV6⁻ cells (Supplementary Fig. S2R–S2T). Epithelial-to-mesenchymal transition (EMT) is an essential step in tumor metastasis (34). We thus assessed the expression of the EMT marker Vimentin in OV6⁺ UMUC3 cells-derived metastatic lesion by IHC. As expected, higher Vimentin expression was observed in specimens from OV6⁺ cell–generated lung metastasis (Supplementary Fig. S2R–S2T). These results demonstrate that OV6⁺ CSCs can facilitate in situ tumorigenicity and metastasis of bladder cancer cells.

To identify strategies for targeting OV6⁺ CSCs and hereby reversing bladder cancer progression and chemotherapy resistance, we next explored the mechanisms that promote the expansion and self-renewal of OV6⁺ CSCs. First, OV6⁺ CSCs and OV6⁻ UMUC3 cells were analyzed by RNA-Seq to search for the critical genes or pathways involved in maintaining the stem-like properties of OV6⁺ CSCs. (Fig. 3A; Supplementary Fig. S3A–SD; Supplementary Table S8–S10). The results revealed that OV6⁺ CSCs exhibited differential expression of Hippo pathway–related genes, such as the higher expression of YAP (Fig. 3A). Real-time PCR confirmed that YAP mRNA was upregulated in OV6⁺ CSCs from UMUC3 and J82 cultures, and Western blotting also corroborated the increased YAP protein level in the nucleus of OV6⁺ CSCs (Fig. 3B and C; Supplementary Fig. S3E), suggesting that YAP was activated in OV6⁺ bladder cancer CSCs. In addition, the RNA-Seq results indicated that OV6⁺ CSCs presented elevated expression of YAP target genes (Fig. 3D), which were validated for connective tissue growth factor (CTGF) and cysteine-rich angiogenic inducer 61 (CYR61; Fig. 3B). In addition, p-YAP phosphorylated by p-LATS1/2 is well known to be retained in the cytoplasm, but only activated YAP (no phosphorylation) enters the nucleus to exert its biological function. Accordingly, Western blotting demonstrated that reduced YAP and LATS1/2 phosphorylation was observed in the cytoplasm of OV6⁺ CSCs (Fig. 3C; Supplementary Fig. S3E). In addition, relative to YAP expression in OV6⁻ cells, Western blotting and immunofluorescence assays both revealed that YAP was mainly located in the nucleus of OV6⁺ CSCs (Fig. 3C and E; Supplementary Fig. S3E and S3F). Taken together, these data suggest that YAP signaling is activated in OV6⁺ bladder CSCs.

We subsequently examined whether YAP was required for maintaining the stem-like properties of OV6⁺ CSCs. After the successful knockdown of YAP in bladder cancer cells was confirmed by Western blot analysis (Supplementary Fig. S3G), lower expression of stem-associated genes and fewer spheres were found in YAP-silenced OV6⁺ CSCs than in control OV6⁺ cells (Fig. 3F and G; Supplementary Fig. S3H and S3I). Furthermore, compared with control OV6⁺ CSCs, YAP-silenced OV6⁺ CSCs showed decreased tumor initiation in 2 serial generations, and enhanced apoptosis and decreased cell survival were observed in OV6⁺ CSCs under cis-platinum treatment (Fig. 3H and I; Supplementary Fig. S3J–S3L). In addition, to test whether upregulation of YAP could rescue these inhibitory effects of YAP knockdown on the stem-like properties of OV6⁺ CSCs, we reexpressed either wild-type (WT) or mutant (MT) YAP in YAP-knockdown OV6⁺ CSCs from UMUC3 and J82 cells. Western blotting demonstrated that bladder cancer cells transfected with MT-YAP exhibited a better overexpression effect (Supplementary Fig. S3M). Subsequently, exogenous expression of MT-YAP enhanced the expression of stem-associated genes and the sphere-forming capacity of YAP shRNA-transfected-OV6⁺ CSCs (Supplementary Fig. S3N and S3O). In addition, YAP overexpression increased the expression of stem-associated genes, the number of spheres, and the tumorigenicity and chemoresistance of OV6⁺ CSCs in bladder cancer (Fig. 3J–M; Supplementary Fig. S3P–S3T; Supplementary Table S11). Thus, YAP is crucial for maintaining the stem-like properties of OV6⁺ CSCs.

In addition, IHC analyses of specimens from patients with bladder cancer were performed (Supplementary Fig. S3U), and the results revealed a positive correlation between OV6 expression and nuclear YAP expression (Supplementary Table S12; P < 0.001). Furthermore, all patients were classified into 4 groups according to OV6 and nuclear YAP expression in specimens (Supplementary Table S13), and concomitantly elevated expression of OV6 and nuclear YAP in bladder cancer specimens was found to be associated with the poorest CSS (P < 0.001), PFS (P < 0.001), and OS (P < 0.001; Fig. 3N–P).

Given that autocrine signaling maintains the stem-like properties of CSCs (35), we investigated the crucial inflammatory factors in mediating OV6⁺ bladder cancer CSCs using a RayBio Human Cytokine Antibody Array (Fig. 4A–C; Supplementary Fig. S4A-C; Supplementary Table S14). As expected, the level of many cytokines was significantly upregulated in the conditioned medium (CM) from OV6⁺ CSCs from UMUC3 or J82 cell cultures relative to those in CM from OV6⁻ cells (Fig. 4A and B; Supplementary Fig. S4A-C; Supplementary Table S14). We then compared the significantly differentially expressed cytokines using a Venn plot and found that PDGF-BB and ICAM-1 were consistently increased in both cell types (Fig. 4C; Supplementary Table S14). On the basis of validation through ELISA analysis, the secretion of PDGF-BB but not ICAM-1 was significantly higher in the CM from OV6⁺ bladder CSCs than in the CM from OV6⁻ cells (Fig. 4D), which prompted us to investigate whether PDGF-BB was responsible for maintaining the stemness of OV6⁺ CSCs. First, human recombinant PDGF-BB protein was added to OV6⁺ CSC cultures, which resulted in higher expression of stem-related genes and higher sphere formation than that detected in naïve OV6⁺ CSC cultures (Fig. 4E and F; Supplementary Fig. S4D and S4E). Second, PDGF-BB–treated OV6⁺ CSCs presented a higher tumor incidence than control OV6⁺ cells (Fig. 4G). Conversely, the addition of a neutralizing antibody against PDGF-BB to the CM of OV6⁺ CSCs suppressed the stem-like properties (Fig. 4E–G; Supplementary Fig. S4D and S4E). In addition, PDGF-BB facilitated the stem-like properties of OV6⁻ bladder cancer cells (Supplementary Fig. S4F–S4H; Supplementary Table S15). Thus, PDGF-BB is required for the stem-like features of OV6⁺ bladder CSCs.

Given that PDGF-BB triggers the intrinsic signaling of tumor cells through its receptor, PDGFR (36), we also found that recombinant PDGF-BB protein activated the phosphorylation of PDGFR and its downstream kinases (MEK and ERK; ref. 37) in OV6⁺ CSCs (Supplementary Fig. S4I). We then determined whether the PDGFR inhibitor CP-673451 (38) suppressed the stem-like characteristics of OV6⁺ CSCs. As expected, CP-673451 inhibited the stem-like characteristics of OV6⁺ CSCs, including stem-associated gene expression, self-renewal, and tumorigenicity (Fig. 4E–G; Supplementary Fig. S4D and S4E). Thus, autocrine PDGF-BB/PDGFR signaling is required for maintenance of the stem-like characteristics of OV6⁺ CSCs in bladder cancer.

Because YAP was required for the stem-like properties of OV6⁺ CSCs, we next investigated whether autocrine PDGF-BB/PDGFR–mediated YAP activity. First, recombinant PDGF-BB protein inhibited YAP phosphorylation in the cytoplasm while increasing YAP expression in the nucleus of OV6⁺ CSCs, whereas CP-673451 had the opposite effects (Fig. 4H; Supplementary Fig. S4K). Second, YAP knockdown alleviated the promoting role of PDGF-BB in YAP stabilization and the stem-like characteristics of OV6⁺ CSCs, including stem-associated gene expression, self-renewal, and tumorigenicity (Fig. 4I-K; Supplementary Fig. S4L and S4M). Therefore, PDGF-BB/PDGFR facilitated OV6⁺ CSCs through stabilizing YAP and promoting YAP entry into the nucleus.

Moreover, we investigated how PDGF-BB/PDGFR mediated YAP activity in OV6⁺ CSCs. First, we tested whether PDGFR could directly interact with YAP to form a complex and prevent LATS1/2-dependent YAP phosphorylation. A co-IP analysis demonstrated that endogenous PDGFR directly interacted with YAP in OV6⁺ UMUC3 CSCs (Fig. 4L). Second, PDGF-BB enhanced the interaction between PDGFR and YAP, while CP-673451 abated the effects in OV6⁺ CSCs from UMUC3 or J82 cultures (Fig. 4M). In the same manner, PDGF-BB decreased the interaction between p-LATS1/2 and YAP, while CP-673451 abrogated the promoting effects of PDGF-BB (Fig. 4M). Third, PDGF-BB decreased the phosphorylation of YAP and LATS1/2 in the cytoplasm of OV6⁺ CSCs (Fig. 4H). In addition, PDGF-BB–treated OV6⁺ CSC–derived xenografts presented activated PDGFR and YAP, which were inhibited in CP-673451-treated OV6⁺ CSC–derived xenografts (Supplementary Fig. S4N-Q). Therefore, PDGFR activated by PDGF-BB can directly interact with YAP, which prevents p-LAST1/2 interaction with YAP and inhibits YAP phosphorylation in OV6⁺ CSCs.

Given that PDGF-BB/PDGFR upregulated and activated YAP in OV6⁺ bladder CSCs, we next examined whether YAP could reciprocally promote PDGFB expression and secretion. As shown in Fig. 5A, higher PDGFB expression was observed in OV6⁺ than in OV6⁻ CSCs from UMUC3 or J82 cells. However, YAP knockdown decreased the PDGFB levels in OV6⁺ CSCs (Fig. 5A), while YAP overexpression caused the opposite effects (Fig. 5B). Second, the concentration of PDGF-BB in CM from OV6⁺ CSCs was higher than that in CM from OV6⁻ cells (Fig. 5C). Silencing of YAP decreased the PDGF-BB concentration in CM from OV6⁺ CSCs (Fig. 5C), whereas YAP overexpression increased the PDGF-BB concentration (Fig. 5D). Third, we reexpressed YAP in YAP-knockdown OV6⁺ CSCs, which reversed the inhibitory effects of YAP knockdown, especially the decreases in PDGFB expression and PDGF-BB concentration (Supplementary Fig. S5A and S5B). The data indicate that YAP promotes PDGFB expression in OV6⁺ bladder CSCs.

Then, we examined the mechanisms underlying YAP regulation of PDGFB in OV6⁺ bladder CSCs. YAP triggers the transcription of downstream genes by recruiting TEAD in tumors (39), which prompted us to determine whether YAP upregulated PDGFB transcription via TEAD1 in OV6⁺ CSCs. We observed an interaction between YAP and TEAD1 in OV6⁺ UMUC3 or J82 CSCs, which was further enhanced by TEAD1 overexpression or recombinant human PDGF-BB treatment (Fig. 5E). Verteporfin, an inhibitor that blocks the interaction between YAP and TEAD (40), decreased PDGFB expression and PDGF-BB secretion by OV6⁺ CSCs (Fig. 5A and C). Third, TEAD1-knockdown attenuated the YAP-induced increase in PDGFB mRNA expression and PDGF-BB concentration in CM of OV6⁺ CSCs (Fig. 5B and D; Supplementary Fig. S5C). Furthermore, TEAD1 overexpression increased the levels of stem-associated genes and the number of spheres formed by OV6⁺ CSCs (Fig. 5F-G; Supplementary Fig. S5D-F), and TEAD1 knockdown alleviated the promoting effects of YAP or PDGF-BB on the stem-like properties of OV6⁺ CSCs from UMUC3 or J82 cells (Fig. 5F-G; Supplementary Fig. S5E-F). Therefore, YAP facilitates PDGFB expression in a TEAD1-dependent manner, contributing to the stem-like characteristics of OV6⁺ CSCs.

In addition, online JASPAR software (http://jaspar.genereg.net) was employed to predict putative transcription factor–binding sites of TEAD1 on the PDGFB promoter (Supplementary Fig. S5G). As expected, TEAD1 bound to the PDGFB promoter in OV6⁺ CSCs, as demonstrated through a chromatin immunoprecipitation (ChIP) assay (Fig. 5H). In addition, though TEAD1 or YAP overexpression or PDGF-BB treatment enhanced PDGFB transcriptional activity in OV6⁺ UMUC3 CSCs, mutated variants of the TEAD1-binding sites on PDGFB abolished the effects, which was validated by luciferase assays (Fig. 5I). Thus, PDGF-BB/PDGFR–induced YAP upregulation reciprocally promotes the PDGFB transcription through TEAD1 in OV6⁺ bladder CSCs.

As the autocrine regulatory loop YAP/TEAD1/PDGF-BB/PDGFR is required in OV6⁺ CSCs, we further investigated whether blocking the loop using verteporfin or CP-673451 could improve the therapeutic effect of cis-platinum in an orthotopic bladder cancer mouse model. OV6⁺ CSCs from UMUC3 or T24 cultures stably expressing a luciferase reporter were perfused into murine bladders through a urethral catheter, and tumor growth was monitored using an in vivo imaging system (Fig. 6A; Supplementary Fig. S6A). After the perfusion, mice were treated with cis-platinum or cis-platinum (Cis) combined with verteporfin (VP), or with CP-673451 (CP). As shown in Fig. 6A and B and Supplementary Fig. S6A and S6B, there was no significant difference in tumor growth between OV6⁺ CSC–derived xenografts treated with cis-platinum and naïve xenografts, indicating that OV6⁺ CSCs were resistant to cis-platinum in vivo, which is consistent with the in vitro experimental results. However, cis-platinum combined with verteporfin or CP-673451 inhibited tumor growth and reduced the expression levels of OV6 and YAP in tissues from OV6⁺ CSC–derived xenografts (Fig. 6A–D; Supplementary Fig. S6A–S6D). These results indicate that blocking the YAP/TEAD1/PDGF-BB/PDGFR autocrine regulatory loop in OV6⁺ CSCs with verteporfin or CP-673451 can reduce cis-platinum resistance in bladder cancer.

Furthermore, we examined the clinical significance of OV6 and PDGFR in patients with bladder cancer. We found a positive correlation between OV6 and PDGFR expression in bladder cancer specimens (Supplementary Fig. S6E; Supplementary Table S16). On the basis of intratumor OV6 and PDGFR expression, patients with bladder cancer were classified into 4 groups (Supplementary Table S17), and concomitantly elevated OV6 and PDGFR expression in patients with bladder cancer resulted in the poorest CSS (P < 0.001), PFS (P < 0.001), and OS (P < 0.001; Supplementary Fig. S6F–S6H).

Approximately 75% of newly diagnosed patients are diagnosed with NMIBC, and 25% are diagnosed with MIBC or metastatic disease (41). Despite receiving radical cystectomy and pelvic lymph node dissection for MIBC, more than half of patients will eventually develop tumors at distant sites (42). Bladder cancer relapse and treatment failure in most patients have been attributed to the existence of CSCs, which survive many commonly employed therapeutics, including chemotherapy. Although many studies have focused on bladder CSCs, the specific markers and underlying mechanisms of chemotherapy resistance have not been elucidated. However, our study identified a new subset of CSCs with OV6 expression that is closely associated with the progression and prognosis of patients with bladder cancer. Furthermore, the results showed that targeting OV6⁺ CSCs inhibited chemotherapy resistance and improved the therapeutic effects of cis-platinum on MIBC (Fig. 6E). These data suggest that OV6 can also serve as a potential therapeutic target. However, because OV6 itself is a poorly understood molecular entity, the practicality of using it to mark CSCs awaits further delineation of its constituent complexity.

YAP is a transcription cofactor suppressed by the Hippo pathway, and this cofactor exerts its protumorigenic function and drives CSC self-renewal in various malignant tumors. In bladder cancer, YAP has been reported to be an independent biomarker for poor prognosis of patients and to promote cell growth and migration (43, 44). A recent study indicated that YAP activation is a pharmacological target for enhancing the antitumor effects of DNA-damaging modalities in chemotherapy (45). However, the role and mechanism of YAP in bladder CSCs have not been clearly elucidated. In this study, the differentially expressed genes regulating bladder CSCs were screened through transcriptome sequencing, and YAP was found to be a crucial factor. In addition, we found that YAP is necessary for maintenance of the stem-like properties of OV6⁺ bladder CSCs and that the combined expression of YAP and OV6 predicts the prognosis of patients.

Many studies have demonstrated that autocrine signaling loops continuously activate intratumor pathways and regulate tumor growth and CSCs (10, 46, 47). To elucidate the mechanisms underlying the sustained activation of YAP and its role in the self-renewal of OV6⁺ CSCs, we also identified a positive autocrine loop, namely, the YAP/TEAD1/PDGF-BB/PDGFR loop. PDGF-BB has been reported to mediate mesenchymal stem cells and chemotherapy-resistant cancer cells (48, 49), but the biological function and mechanisms of PDGF-BB in CSCs are not fully understood. In addition, although YAP is upregulated by stimulation with PDGF-BB in vascular smooth muscle cells (50), the mechanisms underlying PDGF-BB regulation of YAP have not been elucidated. In our study, we found that PDGF-BB facilitates the stem-like characteristics of OV6⁺ CSCs by enhancing YAP stability in the cytoplasm and promoting increased YAP entry into the nucleus. A recent study revealed that a PDGFR-Src family kinase (SFK) cascade regulates YAP activation via tyrosine phosphorylation in cholangiocarcinoma (51). However, whether PDGFR directly mediates YAP stabilization in tumors has not been reported. Our study demonstrated that PDGFR interacts with YAP to prevent YAP phosphorylation by LATS1/2, which facilitates YAP activation in OV6⁺ CSCs.

Given that OV6 is associated with bladder cancer progression and the prognosis of patients with bladder cancer and the autocrine YAP/TEAD1/PDGF-BB/PDGFR signaling loop is required in OV6⁺ CSCs, we established a treatment model of orthotopic bladder cancer to examine whether blocking the autocrine loop inhibited the resistance of MIBC and augmented the therapeutic effect of cis-platinum. As expected, PDGFR or YAP suppression alleviated the chemotherapy resistance of OV6⁺ CSCs to cis-platinum and achieved a favorable therapeutic effect in vivo. We do, however, recognize that the YAP/TEAD1 inhibitor verteporfin used in our study has been reported to exhibit YAP-independent antiproliferative and cytotoxic effects in endometrial cancer cells (52). On the basis of the results of this study, our future studies will further elucidate the interaction between CSCs and the microenvironment, which will aid in the discovery of more effective therapeutic targets for drug resistance and recurrence in patients with bladder cancer.

No potential conflicts of interest were disclosed.

Conception and design: K.-J. Wang, C. Wang, Y.-H. Sun, C.-L. Xu

Development of methodology: C. Wang, S. Zeng, C.-L. Xu

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): K.-J. Wang, L.-H. Dai, J. Yang, Q.-Q. Tian, C.-L. Xu

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): K.-J. Wang, C. Wang, L.-H. Dai, J. Yang, H. Huang, X. Ma, X. Lu, S. Zeng, C.-L. Xu

Writing, review, and/or revision of the manuscript: C. Wang, X. Ma, S. Zeng, C.-L. Xu

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): C. Wang, J. Yang, Z. Zhou, Z.-Y. Yang, W. Xu, M.-M. Hua, H.-Q. Wang, Y.-Q. Cheng, D. Liu, Y.-H. Sun

Study supervision: Z. Zhang, Y.-H. Sun, C.-L. Xu

This work was supported by Innovation Program of Shanghai Municipal Education Commission (no. 2017-01-07-00-07-E00014); the National Natural Science Foundation of China (no. 81772720, 81773154, 81572509, and 81301861); Shanghai Natural Science Foundation of China (no. 13ZR1450700); Special Fund for Major Projects of Zhangjiang National Innovation Demonstration Zone; the National New Drug Innovation Program (2017ZX09304030); Shanghai Clinical Medical Center For Urinary System Diseases (2017ZZ01005); Shanghai Key Laboratory of Cell Engineering (14DZ2272300). Medical Discipline Construction Project of the Pudong New District (PWYgf2018-03). We also thank Dr. Zi-Wei Wang and Yu-xin Tan (Department of Urology, Changhai Hospital, Second Military Medical University, Shanghai, China), Jian Lu, and Qi Chen (Department of Health Statistics, Second Military Medical University, Shanghai, China) for their contributions in this study.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.