The Splicing Factor PTBP1 Promotes Expression of Oncogenic Splice Variants and Predicts Poor Prognosis in Patients with Non–muscle-Invasive Bladder Cancer Free

Bladder cancer is the sixth most common cancer in men, with an estimated 429,800 cases diagnosed in 2012 and 165,100 deaths for the disease worldwide (1). It is typically subdivided in three main categories on the basis of management goals and prognosis: non—muscle-invasive bladder cancer (NMIBC), muscle-invasive bladder cancer, and metastatic bladder cancer. NMIBC represents the most frequent form and includes carcinoma in situ (stage Tis), papillary lesions confined to the urothelium (stage Ta), or invading the lamina propria (stage T1; ref. 2). Complete transurethral resection of bladder tumors (TURBT) is the routine initial diagnostic and therapeutic step in management. However, over 50% NMIBCs will recur and up to 25% will progress to muscle-invasive disease (3). Because of the probability of recurrence and progression, NMIBC requires repeated surveillance and intervention and is among the most expensive cancers to treat from diagnosis to death (4).

Criteria assessing patient and tumor characteristics provide valuable information about disease recurrence, progression, and proposed treatments. The European Organisation for Research and Treatment of Cancer (EORTC) electronic risk calculator (http://www.eortc.be/tools/bladdercalculator) is commonly used to assess recurrence and progression potential of newly diagnosed cancers. The parameters used are tumor size and number, pathologic stage and grade, presence of carcinoma in situ (CIS), and prior recurrence rate (5). Ease of use and absence of expensive molecular tests represent the main advantages of this and other scoring models (6, 7). However, reproducibility of pathologic stage and grade is modest and remains a critical clinical concern.

Strong efforts are currently aimed at identifying molecular markers with robust diagnostic and prognostic value for NMIBC (8). Although several molecular markers are currently approved by the FDA and its European counterpart (9), their value in predicting NMIBC recurrence and progression is limited and none of them is recommended by clinical guidelines (10). Thus, identification of valuable markers or therapeutic targets that reduce the likelihood of recurrence and/or progression is a clinical priority for NMIBC management.

Dysregulation of splicing is emerging as a key feature of human cancers with therapeutic perspective (11). Splicing leads to excision of introns and ligation of exons from the precursor mRNA (pre-mRNA) and is orchestrated by the spliceosome, a macromolecular machinery composed of five small nuclear ribonucleoprotein particles and hundreds of auxiliary proteins (12). When the exon–intron boundaries (splice sites) display high levels of conservation, exons are almost always included in the mRNA (constitutive exons), whereas exons lacking strong consensus sequences are subjected to regulation (alternative exons). In this case, exon recognition is tuned by trans-acting splicing factors (SF) that bind to cis-regulatory sequences. The interplay between antagonistic SFs determines whether a target exon is included or skipped through a process named alternative splicing (AS; ref. 12). Because most splice isoforms engage specific interactions and behave as distinct proteins (13), AS increases the coding potential of genomes and represents an evolutionary advantage (14). However, its flexible regulation is prone to errors and defective splicing contributes to neoplastic transformation (11, 13). In this regard, splicing inhibition is envisioned as an effective anticancer therapy, as many tumors are very sensitive to this approach (11).

The main classes of SFs are the serine-arginine (SR) proteins and the heterogeneous nuclear ribonucleoproteins (hnRNP), which often act antagonistically in AS regulation (12). Members of both classes, like SRSF1 (15, 16) and hnRNPA1, A2 and I (also known as PTBP1; refs. 17, 18), were shown to play oncogenic functions. Notably, although AS dysregulation has been reported also in bladder cancer (19), limited information is available regarding the SFs responsible for this process and their possible association with prognosis. Herein, we have performed bioinformatics search for SR and hnRNP proteins associated with bladder cancer progression. Our study identified hnRNPI/PTBP1 as a factor linked to disease progression and suggest PTBP1 as a new prognostic factor and possible therapeutic target for NMIBC progression.

R2 genomics platform (http://r2.amc.nl) and Oncomine database (www.oncomine.org) were used to evaluate the association of splicing factors with bladder cancer grade using Hoglund, Dyrskjot, and Sanchez-Carbayo datasets (20–22).

Patients with histologically proven primary pTa and pT1 bladder urothelial carcinoma were enrolled in the study. Inclusion criteria were as follows: age ≥18 years, adequate bone marrow reserve, normal renal and liver function and Eastern Cooperative Oncology Group performance status between 0 and 2 (23). Exclusion criteria were as follows: previous bladder cancer; nonurothelial carcinomas; previous or concomitant urinary tract carcinoma in situ and urothelial carcinoma of the upper urinary tract and urethra; bladder capacity less than 200 mL; untreated urinary tract infection; severe systemic infection; urethral strictures preventing endoscopic catheterization; previous radiotherapy to the pelvis; other concurrent chemotherapy, radiotherapy, treatment with biological response modifiers; other malignant diseases within 5 years of trial registration (except for adequately treated basal cell or squamous cell skin cancer, and in situ cervical cancer); pregnancy; other factors precluding study participation.

All patients underwent TURBT of endoscopically detected tumors, ensuring that muscle was included in resected samples, as specified by European Association of Urology guidelines (2). Before TURBT, a cold-cup biopsy of apparently nontumor-bearing and tumor were taken and stored at −80°C until analysis. Tumors were staged in accordance with the TNM classification (24) and biopsies were graded according to WHO classifications (25). Risk categories for recurrence and progression were assessed in accordance with EORTC risk tables for NMIBC (5). Within 6 hours of TURBT, patients received a single intravesical instillation of 40 mg mitomycin. Four to 6 weeks after TURBT, patients without muscle in resected samples, positive or suspect cytology, carcinoma in situ, stage T1, or grade 3 tumors underwent restaging transurethral resection, random cold-cup bladder and prostatic urethra biopsies, and upper urinary tract imaging. No adjuvant intravesical therapy was given to patients with low-risk NMIBC. Patients with intermediate-risk and high-risk NMIBC were scheduled to receive adjuvant long-term intravesical chemotherapy with mitomycin, or immunotherapy with bacillus Calmette-Guerin, starting approximately 3 weeks after TURBT procedures. Patients were followed up with abdominal ultrasonography, urinary cytology, and cystoscopy every 3 months for 2 years, twice during the third year, and yearly thereafter.

All procedures performed in studies involving human participants were in accordance with the ethical standards and with the Declaration of Helsinki. The Institutional Review Board approved the study design, and all enrolled patients signed an informed consent form providing details of treatments.

PTBP1 IHC was performed on formalin-fixed, paraffin-embedded tissues obtained from 152 patients with primary pTa-pT1 NMIBC. Antigen was retrieved by microwave treatment at 750 W for 10 minutes in 10 mmol/L sodium citrate buffer (pH 6.0). Sections were incubated for 60 minutes at room temperature with 1:500 anti-PTBP1 antibody (sc-16547, Santa Cruz Biotechnology), or nonimmune serum as control, using the LSAB Signal Amplification Kit (K0690, Dako). PTBP1 expression was defined by the presence of nuclear staining in tumor cells. IHC analysis was done in blind by two pathologists (M. Pieraccioli and R. Lattanzio) unaware of clinicopathologic information, with 80% concordance in evaluation. Thirty cases (19.7%) in which evaluation of PTBP1 expression differed by >10% between the two observers were reevaluated. After consensus was achieved, the absolute interobserver variability (mean difference % ± SD) recorded was 3.95% ± 3.14%.

The relationship between PTBP1 expression and clinicopathologic parameters was investigated by Pearson χ² test or Fisher exact test. All clinical analyses were performed on an intent-to-treat basis. Primary endpoint was disease-free interval (DFS), defined as time from initial TURBT randomization to first cystoscopy noting recurrence. Secondary endpoints were progression-free (PFS), overall (OS), and disease-specific survival (DSS). Starting from initial TURBT, PFS was defined as time until onset of muscle-invasive disease as recorded by pathologic assessment of transurethral resection samples or biopsy samples, OS as time until death from any cause, and DSS as time until death from bladder cancer. Patients without recurrence or progression were censored at the last cystoscopy, and those lost to follow-up were censored at the last known day of survival. Endpoints were studied by the Kaplan–Meier method and comparisons were made by the log-rank test. Association of PTBP1 expression with outcome, adjusted for other prognostic factors, was tested by Cox proportional hazards model. Appropriateness of the proportional hazard assumption was assessed by plotting the log cumulative hazard functions over time and checking for parallelism. SPSS Version 15.0 (SPSS) was used throughout and P < 0.05 was considered statistically significant.

Bladder cancer cell lines, RT4, RT112, and EJ (26), were purchased and kindly provided by Dr. Francesca Velotti (La Tuscia University, Viterbo, Italy), and further authentication has not been performed. Bladder cancer cell lines were cultured in RPMI1640 medium (LONZA) supplemented with 10% FBS (Gibco) at 37°C with 5% CO₂. Mycoplasma contamination of cell cultures was routinely tested by Hoechst 33258 stain (27), every 2 to 3 weeks. Upon thawing, bladder cancer cell lines have been subcultured for not more than 15 passages. RNAi experiments, protein extractions, and immunofluorescence analyses were performed as described previously (28, 29).

For clonogenic, cell cycle, proliferation, and adhesion assays, cells were silenced twice with CTRL or PTBP1 siRNAi (80 nmol/L). Colony-forming assay was performed as described previously (30). Briefly, 24 hours after transfection, 400 cells were plated in 60-mm dishes and cultured for one week. After fixation-staining for 30 minutes with glutaraldehyde 6.0% (v/v)/crystal violet 0.5% (w/v) solution, colonies with n >50 cells were counted.

For cell cycle/sub-G₁ analyses, cells were pulse-labeled with 10 μmol/L BrdU for 45 minutes before harvesting and fixed with a 30% PBS/70% ethanol solution for 30 minutes in ice. Hypodiploid events and cell cycle were evaluated by flow cytometry using propidium iodide (PI) staining (20 μg/mL) and anti-BrdU antibody in the presence of 13 Kunitz U/mL ribonuclease A as described previously (31). A total of 15 × 10³ events were counted with FACSCalibur flow cytometer (Becton Dickinson) and analyzed using FlowJo program (Becton Dickinson).

Cell death was evaluated by staining for Annexin V (1 μg/mL) and PI (1 μg/mL; eBioscience) and analyzed by flow cytometry (FACSCanto; BD Biosciences). The combination of Annexin V and PI staining discriminates live cells (Annexin V/PI double negative) from apoptotic cells (Annexin V/PI double positive).

RNA was extracted from cells using TRIzol reagent (Invitrogen), and PCR analyses were performed as described previously (28, 29). Primers are listed in Supplementary Information. Quantitative expression level of CD44 variable exons was calculated by ΔC_t method relative to total CD44. CLIP assays were performed as extensively described and RNA associated with PTBP1 was represented as percentage of input (32).

Public datasets (R2 genomics analysis and visualization platform, http://r2.amc.nl) were queried for correlation between expression of SR proteins and hnRNPs and bladder cancer progression. Expression of PTBP1 (i.e., hnRNPI) was found to significantly correlate with disease progression in the Hoglund cohort including 308 patients (Fig. 1A; Supplementary Fig. S1A). PTBP1 expression also resulted significantly higher in invasive T1 stage compared with noninvasive Ta stage cancers (Fig. 1B). Analysis of two other datasets (https://www.oncomine.org: Dyrskjot and Sanchez-Carbayo datasets) confirmed that PTBP1 expression is increased in infiltrating and superficial bladder cancers compared with normal tissue (Supplementary Fig. S1B–S1D). More importantly, Kaplan–Meier analysis highlighted significant association between high PTBP1 mRNA expression and low OS probability (P < 0.001; Fig. 1C). In contrast, expression of other SFs was either unchanged or slightly affected, or associated only with G3 tumors (i.e., hnRNPA2B1, hnRNPC, hnRNPL, SRSF7, and SRSF10; Fig. 1A; Supplementary Fig. S1A).

To determine whether the bioinformatics analysis of mRNA expression levels in public databases represents a valuable tool to identify proteins with oncogenic potential, PTBP1 upregulation was also tested at the protein level by Western blot analysis using a cohort of surgical tumor (T) and nontumor (NT) specimens from 50 patients with NMIBC. PTBP1 protein was significantly increased in the neoplastic lesions compared with the adjacent nonneoplastic urothelium (P = 0.003; Fig. 2A). To extend this analysis, we enrolled a cohort of 178 patients. After initial TURBT, we excluded 13 patients because of concomitant urothelial carcinoma of the upper urinary tract (n = 3) and prostatic urethra (n = 4), histology other than pure urothelial carcinoma (n = 4), and consent refusal (n = 2). Restaging TURBT was performed in 69 patients (54 with high-risk disease and 15 with multifocal intermediate-risk disease) and histologic findings showed that 10 patients with residual stage pTa or pT1 tumor were eligible, whereas 9 with concomitant carcinoma in situ (pTis) and 4 with muscle-invasive disease (stage pT2) were ineligible. Of the remaining 152 patients (see clinical features and tumor characteristics in Table 1), 74 cases (48.7%) were noninvasive papillary carcinoma (pTa) and 78 (51.3%) were tumors invading subepithelial connective tissue (pT1). According to EORTC risk tables classification for NMIBC (5), the cancer progression risk was low in 27 cases (17.7%), intermediate in 70 cases (46.1%), and high in 54 cases (36.2%; Table 1).

Percentage of PTBP1-positive cells was significantly higher in tumor specimens with respect to the nonneoplastic urothelium (Supplementary Fig. S2A), and it ranged from 3% to 98% (mean ± SE = 40.7 ± 2.9) in tumor tissues. Examples of PTBP1 expression in patients with low-, intermediate-, and high-risk of cancer progression are shown in Fig. 2B. Notably, significant increase in PTBP1 expression was associated with risk of cancer progression (Fig. 2C).

To dichotomize PTBP1 expression, a cutoff of 6% positive cells was chosen according to the ROC curve analysis (AUC = 0.779, P < 0.001; Supplementary Fig. S2B and S2C). Tumors with >6% positive cells (n = 105) were considered PTBP1^High and those with ≤6% positive cells (n = 47) were PTBP1^Low. PTBP1 expression directly correlated with pT1 tumors (P < 0.001), grade 3 tumors (P < 0.001), and tumors with high risk of progression (P < 0.001; Table 1). We also found direct correlation between PTBP1^High and the number of patients experiencing tumor recurrence (P = 0.001), cancer progression (P = 0.021), overall mortality (P = 0.047), and cancer mortality (P = 0.029; Table 1). Furthermore, Kaplan–Meier analysis indicated that PTBP1^High expression was significantly associated with lower DFS, PFS, OS, and DSS rates (Fig. 2D). In particular, survival rates for patients with PTBP1^High and PTBP1^Low tumors were 39.0% versus 68.1% (DFS; P < 0.001), 89.5% versus 100.0% (PFS; P = 0.023), 57.1% versus 74.5% (OS; P = 0.006), and 90.5% versus 100.0% (DSS; P = 0.024). Finally, univariate analysis showed that PTBP1^High was significantly associated with poor DFS [HR = 2.6; 95% confidence interval (CI), 1.5–4.6; P = 0.001) and OS (HR = 2.4; 95% CI, 1.3–4.5; P = 0.008; Supplementary Table S1). Multivariate analysis demonstrated that the only factors negatively impacting on survival were tumor grade (DFS; P = 0.018), and age of patients at diagnosis (OS; P < 0.001; Supplementary Table S2).

These observations confirm the predictive value of our bioinformatics analysis of deposited mRNA expression levels and highlight PTBP1 as a novel marker of poor prognosis and disease progression, thus suggesting its oncogenic role in bladder cancer.

To investigate whether PTBP1 was functionally relevant for bladder cancer cells, we knocked down its expression by two rounds of transfection with siRNAs (siPTBP1) and obtained an almost complete depletion in three bladder cancer cell lines (RT4, RT112, and EJ; Supplementary Fig. S3A). Silencing of PTBP1 expression significantly reduced growth of all cell lines as measured by clonogenic assays (Fig. 3A; Supplementary Fig. S3B). Cytofluorimetric analysis of the cell cycle indicated that PTBP1 knockdown resulted in the reduction of cells engaged in active S-phase (BrdU-positive cells) and in the concomitant increase of cells in inactive S-phase (not incorporating BrdU but stalled with DNA content between 2N and 4N; Fig. 3B; Supplementary Fig. S3C). Moreover, depletion of PTBP1 dramatically increased the population of cells with a sub-G₁ DNA content (Fig. 3C; Supplementary Fig. S3D), suggesting that stalling in the S-phase was followed by cell death. Indeed, double staining with Annexin V and PI (Fig. 3D; Supplementary Fig. S3E) and immunofluorescence analysis of caspase-3 cleavage (Supplementary Fig. S3F) confirmed the significant increase in apoptotic cells after depletion of PTBP1 in all three cell lines.

To test whether depletion of PTBP1 affects the response of bladder cancer cells to chemotherapeutic treatment with mitomycin C, we used a suboptimal concentration of the drug (0.03 μmol/L) and set out conditions to reduce PTBP1 without strongly affecting cell survival (lower siRNA concentration and single round of transfection). Under these conditions, silencing of PTBP1 enhanced the cytotoxic effect of mitomycin C on bladder cancer cells death, as indicated by double staining with Annexin V and PI (Fig. 3E; Supplementary Fig. S3G) and analysis of caspase-3 cleavage (Supplementary Fig. S3H).

These results demonstrate that PTBP1 favors bladder cancer cell proliferation and survival and that it protects them from chemotherapeutic treatment.

PTBP1 is best known for its role in AS (33) and expression of oncogenic splice variants has been positively associated with disease progression in patients with bladder cancer (19, 34). We noted that some of these AS events are potential PTBP1 targets (Supplementary Table S3; refs. 35, 36). To test whether PTBP1 promotes the expression of these prooncogenic variants in bladder cancer cells, we examined its impact on the splicing pattern of eight genes. The oncogenic splice variants of seven of these genes were expressed in all three cell lines, whereas CD44 alternative variants were detected only in RT112 cells (Supplementary Fig. S4A and S4B). Notably, knockdown of PTBP1 reverted splicing of the prooncogenic variant of all its potential bladder cancer–related target genes (Fig. 4A and B; refs. 35, 36). On the other hand, exon 3 in PIK4CB and exon 5–6 in LRRFIP2, which are not its predicted target exons (Supplementary Table S3), were unaffected by PTBP1 depletion.

In addition to being involved in cell proliferation (PKM and NUMB) or cell death (FAS), which are affected by PTBP1 depletion (Fig. 3), several of the PTBP1 target genes encode for proteins that regulate the cell cytoskeleton and adhesion (ACTN1, MACF1, TPM1, CD44, and CTNND1). Accordingly, we found that knockdown of PTBP1 impaired cell adhesion in bladder cancer cell lines (Supplementary Fig. S4C), indicating that changes in the splicing pattern of these genes may be functionally relevant. In this regard, CD44 is a prototypic example of AS-regulated gene. It contains 19 exons, nine of which (V2-V10) are alternatively spliced to yield multiple variants (CD44v) that encode for variable extracellular ligand-binding domains of this transmembrane glycoprotein and whose inclusion correlates with tumor progression and metastasis (37). To investigate which of the variable CD44 exons is regulated by PTBP1, we performed semiquantitative PCR (sqPCR) analyses in RT112 cells. Interestingly, inclusion of variable exons between v2 and v7 was reduced in cells depleted of PTBP1, whereas that of other variable exons (v8 to v10) was not (Supplementary Fig. S5A). Quantitative PCR (qPCR) confirmed this exon-specific regulation in PTBP1-depleted cell (Supplementary Fig. S5B).

To test whether PTBP1 plays a direct role in the regulation of its target exons, we investigated PTBP1 recruitment on pre-mRNAs in vivo by performing UV-crosslink and RNA immunoprecipitation (CLIP) assays (32). PTBP1 exerts a position-dependent effect on AS by promoting exon inclusion when it binds in the downstream intron and exon skipping when it binds in the upstream intron (Supplementary Fig. S5C and S5D, top; refs. 35, 36). Thus, CLIP experiments were performed in the presence of RNase I (fragment size ∼200 bp; ref. 32) and we analyzed PTBP1 recruitment near the downstream intron of the positively regulated exon v7 of CD44 and exon 99 of MACF1 (Supplementary Fig. S5C) and the upstream intron of the negatively regulated exon 6a of TPM1 and exon 2 of CTNND1 (Supplementary Fig. S5D). In both cases, we detected a significant enrichment of PTBP1 binding in the target regions with respect to nonregulated exons (Supplementary Fig. S5C and S5D). These results demonstrate that PTBP1 specifically binds target exons in vivo and highlight its direct role in AS regulation of genes with strong relevance for bladder cancer.

To evaluate whether regulation of splicing by PTBP1 was also detected in NMIBC specimens, we performed qPCR analysis of CD44 variable exons that are regulated (v5 and v7) or not (v9) by PTBP1. We selected 18 patients with variable PTBP1 levels (0%–95% positive cells in IHC) from the cohort (Fig. 5A). Strikingly, PTBP1 expression was positively correlated with inclusion of its target exons v5 and v7, whereas the PTBP1-insensitive v9 exon was not significantly associated (Fig. 5B). These results strongly indicate that PTBP1 expression influences splicing outcome in patients with bladder cancer.

NMIBC mortality rate and management costs are strikingly high, underlining the need for valuable prognostic markers and therapeutic targets. In this regard, the splicing signature of human cancers is recently emerging as a sophisticated marker to distinguish tumor subtypes and precisely stratify patients (38). Herein, an unbiased approach to identify SFs with prognostic value in NMIBC undertaken by querying public databases has highlighted PTBP1 as a key predictive factor. We report that PTBP1 expression levels positively associate with disease progression and worse prognosis in patients. Accordingly, PTBP1 promotes prosurvival features of bladder cancer cells and favors splicing of oncogenic variants in both cell lines and patients. Thus, our study shows the predictive value of bioinformatics analysis of deposited gene expression datasets. Furthermore, it uncovers an oncogenic role for PTBP1 in NMIBC and identifies new potential prognostic and therapeutic targets for this disease.

Aberrant splicing regulation often confers selective advantage to tumor cells by favoring oncogenic splice variants of cancer-related genes (13, 39, 40). For instance, upregulation of PTBP1 in human cancer cells affected glycolytic metabolism by promoting splicing of the PKM2 variant (17), leading to acquisition of chemotherapy resistance in pancreatic cancer (18). Likewise, we found that knockdown of PTBP1 in bladder cancer cells augmented the cytotoxic effects of mitomycin C. More importantly, PTBP1 expression level exhibits a strong prognostic value in bladder cancer, as we found a significant correlation between PTBP1 expression, disease progression and worse outcome in a relatively large cohort of patients. Indeed, although, multivariate analysis did not identify PTBP1 expression as an independent predictor of outcome, its expression level is included among the variables that impact on disease progression and outcome, while Kaplan–Meier curves revealed that high expression of PTBP1 correlates with poor DFS, PFS, OS, and DSS. These observations reveal that PTBP1 can be considered a valuable marker to stratify patients with NMIBC in term of risk prediction and evaluation of its expression levels could support the current histologic classification and improve clinical evaluation of patients. This information may also help guiding clinical decisions regarding the follow-up, such as choosing between conservative adjuvant therapy or more aggressive treatment strategies.

We observed a 2- to 3-fold increase of PTBP1 expression in bladder cancer through public datasets analysis. Although this change in expression might appear limited, comparable small changes in expression of other splicing factors were previously shown to trigger oncogenic transformation, like in the case of SRSF1 (15, 16). On the contrary, strong upregulation of splicing factors can be toxic for the cell (15, 41). Thus, the reported increase in expression of PTBP1 in bladder cancer might be sufficient for its oncogenic function. Nevertheless, we cannot rule out the possibility that posttranscriptional regulation of PTBP1 also contributes to its oncogenic potential. For instance, alternative usage of exon 9 generates PTBP1 isoforms showing different target specificity and splicing activity (42). This regulation might impact on the oncogenic role of PTBP1 in bladder cancer and might also influence the multivariate analysis. However, because antibodies specific for these PTBP1 isoforms are not available and no data regarding isoform-specific expression were deposited in the public datasets, we could not investigate this aspect in our cohort of patients.

Expression of prooncogenic splice variants is associated with bladder cancer progression (19). We noted that several of these bladder cancer–regulated genes were potential target of PTBP1. Accordingly, knockdown of PTBP1 expression readily reverted splicing of these genes to the less oncogenic variant. Moreover, we report a positive correlation between the expression level of PTBP1 and the inclusion of variable exons v5 and v7 of CD44 in NMIBC patients' specimens. On the basis of these observations, we hypothesize that the relationship between AS changes in these genes and bladder cancer progression might rely on PTBP1 expression. PTBP1 likely exerts a direct effect on these targets, as we detected its recruitment in proximity of the regulated exons in bladder cancer cells. Notably, PTBP1-regulated genes encode for proteins related to cell survival (FAS), proliferation (NUMB, PKM), cytoskeleton organization (ACTN1, MACF1, TPM1, and CTNND1), and interaction with the extracellular matrix (CD44). Furthermore, depletion of PTBP1 affected proliferation, survival, and adhesion of bladder cancer cells, suggesting a causative role for this SF in oncogenic features of NMIBC cells. For instance, the increase in apoptosis observed in absence of PTBP1 may rely on a splicing switch in favor of the proapoptotic and membrane-associated FAS variant (+E6). In contrast, repression of exon 6 inclusion by PTBP1 generates a soluble prosurvival isoform of FAS (sFAS; ref. 43). Likewise, PTBP1 promotes inclusion of variable exon v6 in CD44 and this isoform promotes stemness and metastasis in colorectal cancer cells (44). Increased inclusion of CD44 variable exons has been reported in bladder cancer (34, 45, 46). Our study now correlates inclusion of variable exons v5 and v7 with PTBP1 expression in 19 patients with NMIBC, suggesting that deregulation of CD44 AS in bladder cancer may be directly caused by altered expression of PTBP1. Noteworthy, although we focused on PTBP1 splicing targets with direct relevance for bladder cancer, our findings do not rule out possible effects of PTBP1 on other variants of the same or other genes in this disease.

In conclusion, our study shows that upregulation of PTBP1 in bladder cancer cells alters their transcriptome in favor of prooncogenic splice variants and correlates with worse outcome in patients. Thus, we suggest that PTBP1 expression and its splicing signature represent novel outcome-predictor markers for NMIBC.

A. Giannantoni is a consultant/advisory board member for Ipsen. C. Sette is an employee of Fondazione Santa Lucia. No potential conflicts of interest were disclosed by the other authors.

Conception and design: P. Bielli, S.M. Di Stasi, C. Sette

Development of methodology: P. Bielli, V. Panzeri, R. Lattanzio, S.M. Di Stasi

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): V. Panzeri, M. Pieraccioli, E. Volpe, A. Giannantoni, S.M. Di Stasi, C. Sette, S. Mutascio

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): V. Panzeri, M. Pieraccioli, E. Volpe, A. Giannantoni, S.M. Di Stasi, C. Sette

Writing, review, and/or revision of the manuscript: P. Bielli, V. Panzeri, R. Lattanzio, V. Pagliarulo, M. Piantelli, S.M. Di Stasi, C. Sette

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): P. Bielli, R. Lattanzio, A. Giannantoni, S.M. Di Stasi

Study supervision: V. Pagliarulo, S.M. Di Stasi, C. Sette

Other: S. Mutascio

We wish to thank Dr. Christian Verri (University of Rome Tor Vergata, Italy) for help in the initial steps of this work and Dr. Francesca Velotti (La Tuscia University, Viterbo, Italy) who kindly provided bladder cancer cell lines. The research was supported by the Associazione Italiana Ricerca sul Cancro (AIRC; IG18790) and by Italian Ministry of Health “Ricerca Finalizzata 2011” (GR-2011-02348423).

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.