Stat3 Activation in Urothelial Stem Cells Leads to Direct Progression to Invasive Bladder Cancer

Bladder cancer is the fifth most common cancer with 69,250 new cases annually in the United States. Urothelial carcinoma represents approximately 90% of bladder cancers, which arise from an epithelial origin. Two subtypes of bladder urothelial carcinomas exist: noninvasive papillary and muscle-invasive cancer. Evidence supports that these 2 subtypes develop through their own independent pathologic and molecular pathways, although certain overlap does exist (1–4). The vast majority of muscle-invasive cancers arise de novo from carcinoma in situ (CIS) without prior clinical progression through noninvasive papillary lesions (2, 4). Muscle-invasive bladder cancer is clinically unfavorable with only a 5-year overall survival of 48% to 67% even after radical cystectomy (removal of entire bladder) for localized disease (5).

Several signaling pathways, such as p53, pRB, PTEN, and their downstream interacting proteins, have been described in mediating the development of invasive bladder cancer (6–9). For instance, TP53 mutation and RB inactivation are common in human bladder CIS (7, 8) and invasive cancer (6) and were shown to be associated with poor prognosis (10, 11). However, mouse model carrying urothelial specific deletions of p53 and pRB only produced late-onset hyperplasia and low-grade noninvasive papillary bladder tumors (12). Exposure of these urothelial specific p53/pRB-deficient mice to subcarcinogenic dose of the carcinogen, N-butyl-N-(4-Hydroxybutyl)nitrosamine (BBN), led to 50% incidence of invasive bladder cancer, whereas single knock out mice of p53 or pRB did not develop invasive cancer in response to BBN (12). Independently, lack of PTEN protein in human bladder cancer is shown to associate with poor outcome, and mouse urothelial specific deletions of PTEN and p53 via an adeno-Cre delivery system to the bladder induces invasive cancer development (9). Despite the current understanding of signaling pathways and mouse models for invasive bladder cancer, a mouse model representing the unique clinical progression of invasive bladder cancer directly from CIS, without prior pathologic appearance as noninvasive papillary tumor, is lacking. In this study, we describe a novel mouse model that targets an active dimerized form of Stat3 to the basal cells of bladder epithelium (urothelium). When Stat3 is constitutively active, the carcinogen BBN induced rapid progression of urothelial progenitor cells to CIS formation and subsequent muscle-invasive cancer. This mouse model implicates a role for Stat3 in predisposing urothelial basal cells toward de novo CIS formation and invasive cancer development, which closely resembles the clinical pathogenesis of human invasive bladder cancer.

K5.Stat3-transgenic mice were characterized as previously described (13). Adult transgenic mice and wild-type littermates at 6 to 8 weeks of age were treated with 0.05% BBN in drinking water for 12 weeks, followed by regular drinking water. Mice were sacrificed at 1 week (n = 4), 2 weeks (n = 4), 4 weeks (n = 4), 6 weeks (n = 4), 13 weeks (n = 4), and 20 weeks (n = 42) after first BBN treatment. Mouse bladders were either fixed in 10% formalin and paraffin embedded for histologic analyses or freshly dissociated for in vitro tumor-sphere forming assay.

Tumor sections were analyzed following standard hematoxylin and eosin (H&E) procedures or immunohistochemical analysis protocols (Dako; ref. 14). Nikon microscopy system and NIS Elements software were used for imaging and semiautomated quantification of CK14⁺ and CK18⁺ cells. Primary antibodies used are listed as follows: Flag (Sigma F1804), Stat3 (Cell Signaling 9139), pTyrStat3 (Cell Signaling 4113), CK14 (Convance PRB-155P), CK5 (Abcam ab75869), CK18 (Abcam ab668), and cleaved caspase-3 (Cell Signaling 9661).

Bladder tumors were enzymatically dissociated into single-cell suspension as previously described (14), and their ability to generate sphere-forming stem cell colonies was analyzed in an in vitro assay as previously described (15). In brief, viable single-cell suspension of tumor cells were resuspended in 1:1 ratio of serum-free Keratinocyte Growth Media (Gibco/Invitrogen) and Growth Factor Reduced Matrigel (BD Biosciences, 356231). Tumor sphere formation was assayed 12 days after first plated.

All animal procedures were approved under protocol AN-5529, and all patient materials were approved under Institutional Review Board protocol H-26809.

Stat3 is a latent transcription factor that normally resides in the cytoplasm. Upon growth factor/cytokine receptor or nonreceptor tyrosine kinase-mediated activation, Stat3 rapidly translocates into the nucleus where it binds to consensus promoter region and activates target gene transcription (16). The association of Stat3 to human invasive bladder cancer and poor survival has been reported (14, 17). However, its biologic role in de novo bladder tumorigenesis has not been fully explored. We previously reported the generation of transgenic mice overexpressing a dimerized form of active Stat3 (Stat3C; ref. 18) by targeting its expression to epithelial basal cells via the keratin 5 promoter (ref. 13; Supplementary Fig. S1A). Here, we investigated the relevance of these transgenic mice in modeling the pathogenesis of human-invasive bladder cancer by exposing them to a well-established chemical carcinogen regimen BBN for inducing rodent bladder cancer (19). Transgene expression in the mouse bladder of Stat3-transgenic mice was first validated by Western blot analysis of Flag protein expression, a polypeptide that was tagged onto the Stat3 transgene (Supplementary Fig. S1B). The urothelial overexpression of Stat3 protein was 5.6-fold greater in Stat3-transgenic mice compared with wild-type littermates, as shown by densitometry analysis (Supplementary Fig. S1B). The restricted expression of Stat3 to the nucleus of urothelial basal cells was determined by immunohistochemistry (Supplementary Fig. S1C and D).

Although hyperplasia was evident in bladder urothelium of Stat3-transgenic mice, spontaneous bladder tumors were not observed after 14 to 16 months of age [Fig. 1A (TG, n = 14) and Fig. 1G (WT, n = 9)], suggesting active Stat3 signaling alone is not sufficient for driving bladder tumorigenesis. After short exposure to the carcinogen BBN, dysplastic lesions resembling CIS were evident in Stat3-transgenic mice as early as 1 week after carcinogen treatment (Fig. 1B). Early invasion into the lamina propria was observed at 2 weeks (Fig. 1C), and invasive cancer was evident at 4 weeks (Fig. 1D) after first carcinogen treatment. These CIS lesions are characterized by atypical cellular morphology from basal, intermediate, to umbrella cells, and almost the entire urothelium analyzed contained CIS in Stat3-transgenic mice (Fig. 1A–D). Careful analyses of these early lesions did not reveal evidence of noninvasive tumors. These results implicated a role for Stat3 in predisposing carcinogen-initiated urothelial cells to rapid progression into CIS, which bypassed noninvasive tumor stage. In contrast, in BBN-treated wild-type mice, both urothelial hyperplasia and focal CIS were observed as premalignant lesions in the bladder (Fig. 1H–J). These results suggested that in wild-type mice, carcinogen-induced bladder tumorigenesis could go through both pathologic pathways: (i) step-wise progression from urothelial hyperplasia into nonmuscle invasive papillary tumors and eventually muscle-invasive bladder cancer, or (ii) CIS into muscle-invasive bladder cancer. However, due to the limitation of the current model, it is impossible to trace the fate of these early lesions from wild-type mice and specifically pinpoint the pathologic pathway followed during tumor progression.

At 20 weeks after first BBN treatment, 100% of Stat3-transgenic mice developed grossly visible muscle-invasive bladder cancers (Supplementary Fig. S1E), which were confirmed by H&E staining in histologic sections showing microscopic invasion into muscle bundles (Fig. 1E and F). On the other hand, development of either hyperplasia or noninvasive cancer was seen in 41.38% of age-matched wild-type mice (Fig. 1K), whereas 58.62% developed invasive bladder cancer (Fig. 1L). Kaplan–Meier analysis revealed a better survival of wild-type mice than survival of Stat3-transgenic mice (Supplementary Fig. 1F). The expression of Stat3 has a reported association with invasive properties and poor clinical prognosis in human bladder cancer (14, 17). The distinct bladder tumor phenotype from Stat3-transgenic mice versus wild-type littermates in this carcinogenesis experiment clearly supports an involvement of active Stat3 in de novo development of early CIS and muscle-invasive cancer. Another phenotype worth highlighting in Stat3-transgenic mice is the rapid progression into CIS and invasive cancer, without prior appearance as noninvasive tumors. In human bladder cancer, 70% to 80% go through a stepwise progression from hyperplasia to noninvasive papillary tumor, but a majority of these tumors do not progress into invasive cancer (Supplementary Fig. S2A, D, and E). The cancer progression in Stat3-transgenic mice closely resembles the alternative pathway in humans from CIS to invasive bladder cancer (Supplementary Fig. S2A–C). However, it should be noted that Stat3 is unlikely the only pathway that can drive the development of BBN-induced invasive bladder cancer, as only approximately 30% of invasive bladder cancer from wild-type mice showed nuclear localization of Stat3 (data not shown).

We have previously reported bladder cancer patients with a higher frequency of CK14⁺ cancer stem cells associate with poor survival outcome in 2 independent patient cohorts (20). Because bladder cancer with poor survival outcome are primarily high-grade invasive cancer, we hypothesized that active Stat3 may regulate a possible expansion of CK14⁺ stem cell in this transgenic mouse model, which subsequently leads to early CIS and invasive cancer progression. To test our hypothesis, we set off to examine the relative expression of stem and differentiated cell markers in bladder tumors derived from Stat3-transgenic mice in comparison with that from wild-type littermates. We recently reported CK14 as a primitive stem cell marker precursor to CK5 in human bladder cancer (20). CK14 is an acidic type I cytokeratin that commonly heterodimerizes with the basic type II cytokeratin CK5, and it is perceived that they coexpress within the same urothelial cells. Immunohistochemical staining of urothelium from young adult wild-type mouse revealed that CK5 is expressed continuously throughout the basal cell layer (Fig. 2F). Interestingly, CK14 expressing cells is scattered and represents only a subpopulation of CK5⁺ urothelial cells (Fig. 2E, marked by arrows). These staining patterns agree with the concept that CK14⁺ cells are a subpopulation and may be precursor to CK5⁺ urothelial cells, although lineage-tracing is required for definitive proof. We therefore proceeded to use CK14 as a putative stem/progenitor cell marker and CK18 as a more differentiated cell marker to compare the relative frequency of these cellular compartments in premalignant lesions or bladder tumors derived from Stat3-transgenic mice and wild-type littermates. Premalignant CIS lesions induced by BBN in Stat3-transgenic mice showed a significant expansion of CK14⁺ cells into at least 3 to 6 cell layers (Fig. 2C and J) as compared with early hyperplastic lesions induced by BBN in wild-type mice, which had a smaller expansion of CK14⁺ cells to form a continuous layer in urothelial basal cells (P < 0.0001) and a greater single layer of CK18⁺ differentiated superficial cells (Fig. 2G and J, P = 0.0002). In noninvasive papillary tumor induced by BBN in wild-type littermates, a significant proportion of tumor cells retain a distribution of more differentiated tumor cells marked by CK18 (Fig. 2H, red), although a clonal outgrowth of CK14⁺ cells (20.6%) was also observed (Fig. 2H, green color). In invasive bladder tumors induced by BBN in wild-type mice, there is an increase of CK14⁺ tumor cells to 61.3% with the presence of CK18⁺ differentiated cells (Fig. 2I, higher magnification in Supplementary Fig. S3A–D), whereas in invasive bladder tumors induced by BBN in Stat3-transgenic mice, CK18⁺ differentiated cells are completely absent (0%) with predominantly CK14⁺ stem cells (77.0%, Fig. 2D, higher magnification in Supplementary Fig. S3E–H). Statistical analysis revealed a significant difference in the percentage of CK14⁺ stem cells in invasive cancer derived from Stat3-transgenic mice, in comparison with both noninvasive and invasive tumors derived from wild-type littermates (Fig. 2K, P < 0.0001 and P = 0.01, respectively). These results revealed that a shift of balance toward CK14⁺ stem cells is evident in invasive bladder tumors from both Stat3-transgenic mice and wild-type mice, although more significant in that from Stat3-transgenic mice. Therefore, Stat3 is unlikely the only pathway that can drive expansion of CK14⁺ stem cells in invasive bladder tumors. Nevertheless, active Stat3 clearly contributed to a significant expansion of CK14⁺ stem cells in early CIS that likely contributed to subsequent invasive tumor formation in Stat3-transgenic mice.

To examine the validity of CK14⁺ tumor cells as functional stem cells, we have successfully adapted an in vitro assay to analyze the frequencies of sphere-forming stem cells in corresponding bladder tumors derived from BBN-treated Stat3-transgenic mice and those from wild-type littermates (15, 21). Interestingly, when equal numbers of tumor cells were analyzed in this assay, Stat3-driven bladder tumor cells generated a significantly higher number of sphere-forming stem cells in comparison with wild-type littermates (Fig. 3A, P = 0.02). These results are consistent with earlier immunofluorescence staining, confirming the increase of CK14⁺ cells in BBN-induced bladder tumors in Stat3-transgenic mice (Fig. 2D and K) are indeed functional stem cells. Although the sphere size difference was not statistically significant (Fig. 3B), a small proportion of tumor cells from BBN-induced Stat3-transgenic mice generated spheres that are larger in size (>90 μm; Fig. 3B and D) than the regular size (35–90 μm) generated from BBN-induced wild-type tumors (Fig. 3E). We attempted to serially passage these spheres in vitro. Unfortunately, these primary tumor spheres seemed to secrete extracellular matrix that made it technically challenging to enzymatically retrieve viable cells for serial passaging (data not shown). As a complementary approach, we overexpressed Stat3C in mouse bladder carcinoma cell line MB49, and Stat3C MB49 cells showed higher sphere formation in primary and secondary passages compared with control (Supplementary Fig. S4A and B).

Next, we explored the potential mechanisms leading to this significant expansion of CK14⁺ stem cells in lesions from BBN-treated Stat3-transgenic mice, in comparison with that from BBN-treated wild-type littermates. Because Stat3 has a recognized antiapoptotic role by mediating downstream expression of Bcl-XL, we hypothesized that Stat3 may confer better survival of urothelial cells in response to carcinogen treatment. In early lesions treated by BBN, we examined the frequency of caspase-3–positive cells to quantify apoptotic cells. In wild-type and Stat3-transgenic premalignant lesions, we observed a portion of caspase-3–positive cells within the CK14⁻ superficial layers (Fig. 3F and G, marked by arrows), as well as some caspase-3–positive cells within CK14⁺ cell layers (Fig. 3F–I, marked by asterisk). There was not a significant difference in apoptotic cells within the CK14⁺ cells of wild-type and Stat3-transgenic mice (Fig. 3J). Interestingly, within the CK14⁻ cells, there was an unexpected increase in the percentage of apoptotic cells in the early lesions from Stat3-transgenic mice in comparison with that from wild-type littermates (Fig. 3J, P = 0.007). We reasoned that although Stat3 protects CK14⁺ cells from BBN-induced apoptosis, it is not targeted to CK14⁻ superficial cells (Supplementary Fig. S1D). Therefore, BBN might preferentially induce apoptosis of CK14⁻ cells in Stat3-transgenic mice. This might provide a possible mechanistic insight into the significant expansion of CK14⁺ cell compartment in early lesions from Stat3-transgenic mice, in comparison with wild-type control. In wild-type tumors, there was an overall and statistically significant increase in apoptotic index in invasive versus noninvasive tumors (Fig. 3K, P = 0.002), which is consistent with a previous report on BBN-induced rodent bladder tumors (22). Interestingly, invasive tumors from Stat3-transgenic mice exhibit a significant reduction in apoptotic index, in comparison with invasive tumors from wild-type littermates (Fig. 3K, P = 0.004). Antiapoptotic role of Stat3 at this later stage of tumorigenesis may be important for survival of cancer cells that have accumulated additional alterations for malignant progression.

To examine the human relevance of these bladder tumor results from mice, which implicate a role of Stat3 in invasive progression, we analyzed 8 early-stage (6 pTa LG, 2 Ta HG) and 8 advanced stage human bladder cancers (1 TIS, 2 pT2, 5 pT3; Supplementary Table S1). In early-stage bladder tumors, we observed that 37.5% (3 of 8) express nuclear active Stat3, whereas 100% (8 of 8) of higher stage bladder cancers expressed nuclear active Stat3 (Supplementary Table S1). Immunohistochemical analysis of representative sections from higher stage tumors with nuclear expression of active Stat3 (Fig. 4A and C) revealed that they also predominantly express CK14⁺ stem cells (Fig. 4B and D). In contrast, representative sections from pTa low-grade tumors revealed a lack of nuclear Stat3 staining (Fig. 4E and G) and CK14 staining (Fig. 4F and H).

Collectively, these results from Stat3-transgenic mice and human bladder cancer for the first time implicate a role for Stat3 in predisposing urothelial basal cells toward the CIS progression pathway into invasive bladder cancer. Because chronic inflammation induced by cigarette smoking and persistent urinary tract infections have been shown to be a promoting factor for bladder cancer (23), Stat3 activation may be an important downstream mediator to inflammatory cytokines such as interleukins (IL-17 and IL-6) during bladder tumorigenesis. Further studies to identify the upstream activators for Stat3 and to investigate its expression in a large human cohort of human CIS and invasive bladder cancer will likely reveal its significance in driving urothelial cells toward the pathway of CIS and invasive cancer progression. This unique mouse model reported in this study will provide a valuable platform to understand other interacting pathways in driving CIS and invasive bladder cancer progression.

No potential conflicts of interest were disclosed.

Conception and design: K.S. Chan

Development of methodology: P.L. Ho, E.J. Lay, K.S. Chan

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): P.L. Ho, E.J. Lay, W. Jian, D. Parra, K.S. Chan

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): P.L. Ho, E.J. Lay, K.S. Chan

Writing, review, and/or revision of the manuscript: P.L. Ho, E.J. Lay, K.S. Chan

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): P.L. Ho, E.J. Lay, K.S. Chan

Study supervision: K.S. Chan

The authors thank Dr. John DiGiovanni and Steve Carbajal in coordinating the transfer of Stat3-transgenic mice used for research in this study, Dr. Li Xin for his expertise and advice in the sphere-forming assay, Dr. Jonathan Levitt for providing MB49 cells, Antonina Kurtova for her technical assistance, Dr. Michael Ittmann, Dr. Sayeeduddin Mohammad, and Dan L. Duncan Cancer Center Human Tissue and Pathology (HTAP) Core Resource for their consultation and services, and Dr. Carolyn Smith and Dr. David Rowley for their editorial comments.

This work was supported by grant funding from NCI R00 CA129640-05 (to K.S. Chan), V Foundation for Cancer Research V Scholar Award (to K.S. Chan), and gift fund from Bladder Cancer Partnership (to K.S. Chan).