Oct-3/4 Expression Reflects Tumor Progression and Regulates Motility of Bladder Cancer Cells

Cancer and embryonic stem cells exhibit similar behavior, including immortal, undifferentiated, and invasive activities. Here, we show that in clinical samples bladder tumors with intense expression of stem cell marker Oct-3/4 (also known as POU5F1) are associated with further disease progression, greater metastasis, and shorter cancer-related survival compared with those with moderate and low expressions. Expression of Oct-3/4 is detected in human bladder transitional cell carcinoma samples and cell lines. Overexpression of Oct-3/4 enhances, whereas knockdown of Oct-3/4 expression by RNA interference reduces, migration and invasion of bladder cancer cells. Oct-3/4 can up-regulate fibroblast growth factor-4 and matrix metalloproteinase-2 (MMP-2), MMP-9, and MMP-13 production, which may contribute to tumor metastasis. Finally, we show that Ad5WS4, an E1B-55 kD–deleted adenovirus driven by the Oct-3/4 promoter , exerts potent antitumor activity against bladder cancer in a syngeneic murine tumor model. Therefore, our results implicate that Oct-3/4 may be useful as a novel tumor biological and prognostic marker and probably as a potential therapeutic target for bladder cancer.


Introduction
In cancer biology, the relationship between embryogenesis and oncogenesis has long been a prevailing theme.Broadly speaking, cancer cells are similar to very early embryonic cells, which are immortal, undifferentiated, and invasive.It is important to study genes associated with embryogenesis and tumorigenesis.One good candidate is the homeobox gene family, as its members mediate a plethora of embryonic functions, such as cell type determination, organ development, and embryogenesis.Deregulated expression of certain homeodomain proteins that are encoded by homeobox genes may lead to the development of lymphomas, leukemias, and other solid tumors (1).
The POU homeobox gene family transcription factor Oct-3/4 (also known as POU5F1) acts as a key regulator of pluripotency in early stages of mammalian development (2,3).It has been shown that a critical amount of Oct-3/4 is required to sustain self-renewal of embryonic stem (ES) cells, and any up-regulation or downregulation induces divergent cell fates.In normal circumstances, Oct-3/4 is specifically expressed in both murine and human ES and germ cells, but not in cells from differentiated tissues.Deregulated expression of Oct-3/4 can be found by immunohistochemical staining in various human solid tumors (4)(5)(6), as well as in testicular germ cell tumors and their premalignant components (7).It is possible that aberrant expression of Oct-3/4 may contribute to the neoplastic process and play a role in cancer stem cell theory (8,9).In the genitourinary system, muscle-invasive transitional cell carcinoma (TCC) of the bladder is a common cause of death.The generally poor prognosis of advanced bladder cancer calls for new therapeutic modalities.Oncolytic adenoviruses, such as E1B-55 kD-deleted adenovirus, are promising anticancer agents, which replicate selectively in cancer cells, resulting in cancer-specific cytolysis (10).To improve the safety of oncolytic viral replication, tumor-specific promoter to control viral replication may play a key role.
In this study, we investigated the significance of Oct-3/4 expression in bladder cancer and its potential as a therapeutic target, as well as novel tumor biological and prognostic marker.

Materials and Methods
Clinical samples, cell lines, and mice.Clinical specimens were collected after clinically indicated transurethral resections obtained from 57 patients with superficial high-grade (grades 2 and 3) bladder TCC.Details of their clinicopathologic characteristics can be found in Supplementary Material.Human and murine bladder cancer cell lines (TCCSUP, J82, T24, TSGH-8301, and MBT-2), immortalized normal human urothelial cell line (SV-HUC-1), normal murine mammary gland epithelial cell line (NMuMG), murine ES cell line (HM-1), murine fibroblasts (NIH3T3 and primary fibroblasts), and C3H/HeN mice used in this study have been previously described (11,12).
Quantification of mRNA expressions.Semiquantitative reverse transcription-PCR (RT-PCR) was done as described previously (11).Expression levels of Oct-3/4, fibroblast growth factor-4 (FGF-4), and matrix metalloproteinase-13 (MMP-13) mRNA in cell lines were determined by real-time quantitative RT-PCR using LightCycler FastStart DNA Master SYBR Green I kit (Roche) and LightCycler 1.5 Real-Time PCR System (Roche).Annealing temperature for PCR was 55jC.Gene expression was normalized to h-actin before the fold change in gene expression was calculated using LightCycler Software 4.0 (Roche).The sequences of PCR primers are described in the supplementary material.
Histochemical, immunohistochemical, and immunoblot analyses.Tissue sections of the mouse lungs were prepared and stained with H&E as previously described (11).Detections of Oct-3/4 and adenovirus hexon and fiber proteins in frozen tissues or cell lines were done by immunohistochemical staining or immunoblotting (11).
Microarray hybridization and data analysis.Total RNA was extracted from TCCSUP/Oct-3/4 stable clone 4 and vector control clone (supplementary material), and their cDNA was synthesized and labeled with Cy3 or Cy5.Microarray hybridizations were carried out using Agilent human 1A Oligo Microarray kit (Agilent Technologies) according to the manufacturer's instructions.Fluorescent images were obtained using a GenePix Scanner 4000B (Axon Instruments).Raw data were normalized and analyzed using Genefilter software version 2.0 (Research Genetics;ref. 13).The microarray data have been submitted to the ArrayExpress database under accession number E-MEXP-1378.
In vitro wound healing, migration, and invasion assays.Confluent cells were wounded using a 200-AL pipette tip in six-well culture plates and incubated in DMEM with 5% or 10% of fetal bovine serum in the presence or absence of mitomycin C (5 or 10 Ag/mL).The wound width was photographed at 0 and 12 h.Quantitative analysis of the wound closure was calculated by counting the number of cells per 10 6 Am 2 wound area at 12 h.To trace single-cell migration, cells (1.5 Â 10 4 ) grown in 35-mm culture dishes overnight were monitored for their migration using a CCM-cultured cell monitoring system (Astec Co.) for 4 h.The images were obtained with CCD camera in every 10 min.Cell migration was analyzed using Metamorph 2.0 image analysis software (Universal Imaging Corp.), and the migration distances of 10 to 15 cells were measured, summed, averaged, and represented as pixels.
Migration of cancer cells (5 Â 10 4 ) through gelatin-coated, 8-Am-pore polycarbonate membranes (Nucleopore) was assessed using a 48-well modified Boyden chemotaxis chamber (NeuroProbe), as previously described (14).After incubation for 18 h for TCCSUP cells and 14 h for MBT-2 cells, the filters were fixed with methanol and stained with Giemsa solution.The number of migratory cells was counted under a microscope.In vitro invasion assay was done using modified Boyden chamber consisting of 8-Am membrane filter inserts (BD Biosciences) coated with Matrigel (Collaborative Research) in 24-well plates as previously described (15).Cancer cells (5 Â 10 4 ) were added to the upper chamber, whereas lower chambers were filled with conditioned medium collected from the individual cancer cells.After being incubated for 16 h, cells that migrated through the membrane to the lower surface were stained with Giemsa solution and counted.
Gelatin zymography.The enzymatic activities of MMP-2 and MMP-9 present in the cell supernatant were determined by gelatin zymography as previously described (16).
Analysis of Oct-3/4 and E1A promoter activities in murine cells.Cells were cotransfected with pGL-mOct-3/4-Luc or pGL-E1A-Luc reporter plasmid (supplementary material), as well as pTCY-LacZ driven by the hactin promoter, and relative luciferase activity was measured as luciferase activity divided by h-galactosidase (h-gal) activity to normalize transfection efficiency (17).
Cell proliferation and viability assays.Cell proliferation was determined by BrdUrd incorporation during DNA synthesis using the Cell Proliferation ELISA BrdUrd (colorimetric) kit (Roche) according to the manufacturer's instructions.To assess cell viability, confluent cells were infected with adenoviral vectors at various multiplicities of infection (MOI) and stained with crystal violet 7 d after infection.A colorimetric WST-1 assay (TaKaRa) was also used to assess cell viability 5 d after viral infection.
Animal studies.In the animal model of experimental pulmonary metastasis, groups of eight mice were injected with MBT-2/Oct-3/4 or MBT-2/LacZ cells (5 Â 10 5 ) in 0.1 mL PBS via the tail vein and killed 30 d after tumor cell inoculation.The lungs were immediately excised, and the wet lung weight and the number of tumor nodules on the lung surface were recorded.Paraffin-embedded, H&E-stained lung sections were analyzed for tumor nodules microscopically, as previously described (11).In the spontaneous metastatic model, groups of eight mice were inoculated s.c. with MBT-2/Oct-3/4 or MBT-2/LacZ cells (2 Â 10 6 ) and killed 60 d after tumor cell inoculation.Metastatic nodules on the lung surface were detected grossly and microscopically.
To evaluate the antitumor efficacy of oncolytic adenoviruses, MBT-2 cells (10 6 ) were inoculated s.c.into the right flank of C3H/HeN mice at day 0. At day 12, visible and palpable nodules developed in all mice ranging from 56 to 178 mm 3 (mean F SD 105.5 F 36.3 mm 3 , n = 40).Four groups of 10 mice were treated intratumorally (i.t.) with 10 9 plaque-forming units (pfu) of Ad5WS4, Ad5WS1, or heat-inactivated (56jC for 2 h) Ad5WS4 in 100 AL of saline, or with saline at days 12, 14, and 16.All mice were monitored for tumor growth and survival.Tumor volumes were measured as described previously (11).The recorded day of death or sacrifice, when the primary tumor reached 3,500 mm 3 , was used to calculate survival time.
Statistical analysis.Statistical significance between groups, unless otherwise stated, was assessed with Student's t test.Correlation between Oct-3/4 expression and clinical variables was analyzed by Fisher's exact test.The significance of Oct-3/4 expression in relation to tumor metastasis and progression was calculated by Kaplan-Meier analysis and log-rank test.The survival analysis in mice was performed using the Kaplan-Meier survival curve and log-rank test.

Results
Oct-3/4 may be a novel tumor biological and prognostic marker of superficial bladder TCC.We used RT-PCR and immunohistochemical approaches to examine Oct-3/4 expression in a panel of clinical bladder TCC samples.Our data illustrate a portion of RT-PCR results in which various levels of Oct-3/4 mRNA were detected in four human primary bladder carcinoma specimens (PT1 to PT4), but not in normal human bladder tissue (NT; Fig. 1A).Further immunohistochemical staining of frozen sections confirms Oct-3/4 expression in superficial bladder tumor tissues, but not in normal tissues (Fig. 1B).H&E-stained bladder tumor with moderate Oct-3/4 expression served as a comparison (Fig. 1B, bottom right).Vast majority of Oct-3/4-positive tumor tissues suggest that Oct-3/4 expression in human primary bladder carcinomas might induce a cascade of gene expression to promote tumorigenesis.
We then systemically studied 57 patients with histologically confirmed superficial bladder TCC and 10 cases of nonneoplastic bladder tissues obtained by cystourethroscopic biopsies.Paraffinembedded sections from tumor specimens were stained with H&E and reviewed by a pathologist (C.C.L.) to confirm the presence of at least 70% tumor cells (Supplementary Fig. S1).Overall, Oct-3/4 expression was seen in all tumor tissues but not in nonneoplastic bladder tissues.Of these tumor tissues, there were 25 (43.9%) with intense Oct-3/4 expression, 8 (14.0%) with moderate expression, and 24 (42.1%) with low expression.However, no significant association was found between Oct-3/4 expression and tumor grade, stage, morphology, and multiplicity (Supplementary Table S1).As for prognostic implication, disease progression, metastasis, and cancer death were observed in 15, 13, and 12 in high Oct-3/4expressing patients, as well as in 4, 4, and 3 in moderate-expressing to low-expressing groups, respectively.Univariate analysis revealed that multiplicity at diagnosis was the risk factor for tumor recurrence (data not shown).Most importantly, tumors with intense Oct-3/4 expression were associated with further disease progression (P = 0.0002) and greater metastasis (P = 0.0011) compared with those with moderate or low expression (Fig. 1C).Furthermore, intense Oct-3/4 expression was also a risk factor for disease progression in bladder cancer subgroups, including grade 2 (P = 0.047) or grade 3 (P < 0.0001), as well as stage Ta-T1 (P = 0.017) or stage T2 (P = 0.0028; data not shown).Patients with low Oct-3/4 expression had longer cancer-related survival (P = 0.0005) than those with high expression, as assessed by Kaplan-Meier survival analysis and log-rank test (Fig. 1C).The poor patient survival rate in Oct-3/4 high-expression group may be attributed to greater disease progression and higher metastasis.Therefore, intensity of Oct-3/4 expression provides the prognostic information for patients with superficial bladder TCC.
RT-PCR and immunoblot studies also show Oct-3/4 expression in human and murine bladder cancer cell lines, as well as murine HM-1 ES cells that served as the positive control for Oct-3/4 expression (Fig. 1D).In contrast, Oct-3/4 expression was not detectable in normal murine NMuMG epithelial cells.Of note, immortalized human epithelial (SV-HUC-1) and murine fibroblast (NIH3T3) cells expressed much smaller amounts of Oct-3/4 compared with bladder cancer cell lines.We also confirmed immunohistochemically Oct-3/4 expression in MBT-2 and HM-1 cells, but not in NIH3T3 cells (Supplementary Fig. S2).Taken together, these results implicate Oct-3/4 as a novel tumor biological and prognostic marker for bladder cancer.
Oct-3/4 overexpression enhances, whereas knockdown of Oct-3/4 expression reduces, migration and invasion of bladder cancer cells.To study the biological role of Oct-3/4 in bladder cancer, we established human TCCSUP and murine MBT-2 stable clones overexpressing Oct-3/4, as well as vector control clones.TCCSUP/Oct-3/4 clones 4 and 5, as well as MBT-2/Oct-3/4 clones 9 and 12, were chosen for further functional studies (supplementary material).Moreover, endogenous Oct-3/4 expression in parental TCCSUP and MBT-2 cells was also knocked down by transient transfection of Oct-3/4-specific short hairpin RNA (shRNA; supplementary material).Real-time quantitative RT-PCR and immunoblot analyses verified different expression levels of Oct-3/4 in these TCCSUP and MBT-2 derivatives (Supplementary Fig. S3).Densitometric analysis of the immunoblots showed that, compared with stable clones transfected with empty control plasmid (control vector clones), expression of Oct-3/4 was 2.6-fold to 3.3-fold higher in Oct-3/4-overexpressing clones, whereas it was knocked down by f75-80% in Oct-3/4 shRNA-transfected parental cells.To unravel the role of Oct-3/4 overexpression in the malignant phenotype of bladder cells, Affymetrix oligonucleotide microarray analysis was performed to determine whether any genes were differentially expressed between the TCCSUP/Oct-3/4 and control vector clones.A total of 2,073 genes showed at least 1.5-fold changes in expression levels.They involve different biological processes, in which 16 of the up-regulated and 8 of the down-regulated genes with well-characterized gene functions were grouped ontologically (Supplementary Tables S2 and S3).A number of these Oct-3/4regulated genes have been identified as regulators of cell motility, such as MMP-13, fibronectin 1, and transforming growth factor-h3.Collectively, these results suggest that Oct-3/4 overexpression might be involved in the control of bladder cancer cell migration in vitro and metastasis in vivo.
We next examined whether overexpression of Oct-3/4 was sufficient to confer these characteristics in bladder cancer cells.In an in vitro wound healing assay, TCCSUP/Oct-3/4 and MBT-2/Oct-3/4 cells migrated into the wound area more rapidly than control cells (Fig. 2A).Similar results were obtained in Boyden chamber assays (Supplementary Fig. S4).Giving that overexpression of Oct-3/4 also up-regulates genes involved in transcription/translation (Supplementary Table S2), we examined whether Oct-3/4 overexpression affected cell proliferation in bladder cancer cells.In a BrdUrd incorporation assay, proliferative potential was higher in TCCSUP/Oct-3/4 cells than in control cells (Fig. 2B).To further exclude the possibility that the effect of Oct-3/4 overexpression on cell migration could be associated exclusively to cell proliferation rather than to cell migration, we used mitomycin C to inhibit cell proliferation in the wound healing assay.Because mitomycin C at 5 and 10 Ag/mL completely inhibited proliferation of TCCSUP and MBT-2 cells with no cytotoxicity, respectively, these doses were chosen for subsequent experiments (Supplementary Fig. S5).Although mitomycin C reduced the number of cells migrating to the wound area by f25% in TCCSUP and f30% in MBT-2 clones, mitomycin C treatment did not abrogate Oct-3/4-mediated enhancement of cell migration (Fig. 2B).To further confirm these results, we compared the migration distance of cancer cells in which Oct-3/4 was overexpressed or knocked down.Migration of single cells was traced for 4 hours with visual data acquired every 10 minutes.Whereas bladder cancer cells overexpressing Oct-3/4 exhibited higher migratory activity, Oct-3/4 knockdown cells migrated slower than their control counterparts (Fig. 2C).In TCCSUP and MBT-2 cells transiently transfected with Oct-3/4 expression vectors or Oct-3/4 shRNA, cell migration and invasion through Matrigel were greatly enhanced in Oct-3/4-overexpressing cells but reduced in Oct-3/4 knockdown cells in Boyden chamber assays (Fig. 2D).Taken together, the migratory and invasive abilities of bladder cancer cells were correlated with Oct-3/4 expression.
Oct-3/4 overexpression promotes metastasis in animal tumor models.Giving that Oct-3/4 overexpression enhanced migration and invasion of bladder cancer cells in vitro, we next investigated in animal models whether the metastatic potential of bladder cancer cells correlated with Oct-3/4 expression.In the animal model of experimental pulmonary metastasis in which cancer cells were injected into mice via the tail vein, tumor nodules in the lung were detected more frequently in those inoculated with MBT-2/Oct-3/4 cells compared with those injected with the control cells, as revealed by gross examination and H&E staining (Fig. 3A).In particular, there was a significant increase in pulmonary metastatic colonization, both in lung weight and in the number and size of the tumor nodules (Fig. 3B).In the spontaneous metastatic model in which tumor cells were inoculated s.c.into the mice, the number of visible metastatic nodules in the lung was greater in mice inoculated with MBT-2/ Oct-3/4 cells than in those receiving control cells (Fig. 3C).Histologic sections also revealed that mice inoculated with MBT-2/ Oct-3/4 had more tumor lesions (Fig. 3D).Taken together, these results indicate that Oct-3/4 overexpression promoted the metastatic behavior of bladder cancer cells in vivo.BrdUrd incorporation assay shows that cell proliferation of TCCSUP/Oct-3/4 cells was increased compared with that of the control cells (left).Columns, mean of six determinations, which were consistent in two independent experiments; bars, SE.In the wound healing assay, numbers of the migratory cells at a defined wound area at 12 h were greater in Oct-3/4-overexpressing TCCSUP (middle ) and MBT-2 (right ) cells than those in their control cells, regardless of whether the cells were treated with mitomycin C.During the wound healing assay, cells were treated with or without mitomycin C (5 Ag/mL for TCCSUP and 10 Ag/mL for MBT-2 cells).Columns, mean of three determinations, which were consistent in three independent experiments; bars, SE.C, single-cell tracking shows that expression of Oct-3/4 affected cell migration.Tracking of single-cell migration of TCCSUP (left) and MBT-2 (right ) derivatives by video time-lapse microscopy for 4 h shows that Oct-3/4-overexpressing cells migrated faster and Oct-3/4 knockdown cells migrated slower than their control counterparts.Data are presented as bar and whisker graphs, showing the median and the distribution of 50% (bar) and 90% (whisker) of all tracked cells (n = 10-15), and are representative of two independent experiments.D, overexpression of Oct-3/4 enhanced, whereas Oct-3/4 knockdown reduced cell migration (left ) and invasion through Matrigel (right ).Cell migration was determined in modified Boyden chambers using a filter coated with Matrigel in the invasion assay.After incubation (18 h for TCCSUP cells and 14 h for MBT-2 cells in the migration assay and 16 h in the invasion assay), cells that migrated through the membrane to the lower surface were stained with Giemsa solution and three fields (at 40Â magnification) in each section were counted.Columns, mean of three determinations, which were consistent in three independent experiments; bars, SE.  (20).Overexpression of FGF-4 enhances the metastatic activity of MCF-7 cells (21).In our microarray data, MMP was up-regulated in TCCSUP/Oct-3/4 cells, in particular MMP-13, which was increased by 2.4-fold compared with the control cells (Supplementary Table S2).Previous studies also showed MMP expression in human bladder carcinomas (22,23).We, therefore, examined FGF-4 and MMP-13 expressions in TCCSUP and MBT-2 cells with varying degrees of Oct-3/4 expression.RT-PCR (Fig. 4A and B, left) and real-time quantitative RT-PCR (Fig. 4A and B, right) studies reveal that expressions of FGF-4 (Fig. 4A) and MMP-13 (Fig. 4B) transcripts were increased in TCCSUP/Oct-3/4 and MBT-2/Oct-3/4 cells, but decreased in their parental cells after their endogenous Oct-3/4 expression was knocked down by shRNA.Because MMP-13 was reported to play a central role in the activation cascades of MMP (24), we further investigated the role of Oct-3/4 in the activity of MMP-2 and MMP-9 in bladder cancer cells by gelatin zymography (Fig. 4C).TCCSUP/ Oct-3/4 and MBT-2/Oct-3/4 cells produced higher levels of MMP-2 compared with control vector cells.Up-regulation of MMP-9 was also observed in MBT-2/Oct-3/4 cells, whereas it was hardly detectable in all the TCCSUP derivatives.In the two parental cells, shRNA-mediated knockdown of Oct-3/4 resulted in 30% to 60% reduction of MMP-2 production.Furthermore, knockdown of Oct-3/4 expression also slightly reduced the endogenous level of MMP-9 in MBT-2 cells.However, MMP-9 was still undetectable even in Oct-3/4-overexpressing TCCSUP cells.Collectively, it is tempting to speculate that Oct-3/4 overexpression in bladder cancer cells led to enhanced FGF-4 expression, which may have contributed to the up-regulation of MMP and, thereby, to the metastatic phenotype.
Ad5WS4 exerts cytolytic effect in vitro against bladder cancer cells.As Oct-3/4 is specifically expressed in bladder TCC samples and cell lines but not in normal cells, it would be of interest to use Oct-3/4 as a potential therapeutic target for bladder cancer.Towards this end, we first assessed the Oct-3/4 promoter activity in different murine cells using a luciferase reporter assay.A much more pronounced transcriptional activity of the Oct-3/4 promoter was observed in MBT-2 cells than in NIH3T3 cells and normal mouse fibroblasts (Fig. 5A, top).This result led to further interest in constructing a conditionally replicating adenovirus with its E1B-55 kD gene-deleted and E1A gene driven by the mouse Oct-3/4 promoter for murine bladder cancer therapy.The transcriptional activity of mouse Oct-3/4 promoter was significantly higher than that of adenoviral E1A promoter in MBT-2 cells, but was down-regulated in NIH3T3 cells (Fig. 5A, bottom).In contrast, the E1A promoter activity in MBT-2 cells was similar to that in NIH3T3 cells.Thus, these data suggest that the Oct-3/4 promoter may control transgene expression in a cancer-specific manner in murine bladder tumors.We then constructed an E1B-55 kD-deleted adenovirus driven by the Oct-3/4 promoter, designated Ad5WS4.The cytolytic effects of Ad5WS4 and Ad5WS1 (an E1B-55 kD-deleted adenovirus driven by its original E1A promoter) were compared in cancer and normal cells.Two replication-defective adenoviral vectors, AdCMV carrying no transgene and AdLacZ carrying b-gal gene, were also tested in parallel, which were expected to be nonreplicative and nonlytic in all the cell lines tested.Crystal violet staining (Fig. 5B) and WST-1 assay (Fig. 5C) show that Ad5WS4 and Ad5WS1 induced significant cytopathic effect (CPE) in MBT-2 cells.In marked contrast, both oncolytic adenoviruses were not cytolytic to normal murine NMuMG Figure 4. Overexpression of Oct-3/4 stimulates FGF-4 expression and up-regulates MMP expression in bladder cancer cells.A and B, quantification of the mRNA levels of FGF-4 (A ) and MMP-13 (B ) in TCCSUP and MBT-2 derivatives.In semiquantitative RT-PCR analysis, relative density of signals for FGF-4 (A, left ) and MMP-13 (B, left) was determined by densitometric scanning and expressed as a ratio relative to the density of signal for h-actin for control vector-transfected cells.In real-time quantitative RT-PCR, the mRNA level of FGF-4 (A, right) and MMP-13 (B, right ) was normalized to that of h-actin.Relative mRNA level was expressed as a fold difference compared with control vector.C, representative gelatin zymography showing the activities of MMP-2 and MMP-9 in the conditioned medium of cells of various TCCSUP and MBT-2 derivatives.Relative density of signals for MMP-2 was determined by densitometric scanning and expressed as a ratio relative to the density of signal for control vector-transfected cells.The results shown are representative of three independent experiments.epithelial cells, primary fibroblasts, and NIH3T3 fibroblasts, as well as human SV-HUC-1 normal epithelial cells.However, in NIH3T3 cells, Ad5WS1 seemed to induce some degrees of CPE, whereas Ad5WS4 did not cause noticeable cytolytic effects at high viral doses (Fig. 5B).As expected, AdCMV and AdLacZ failed to induce CPE in normal and cancer cells.
We also examined the effect of Oct-3/4 on the expression of coxsackie and adenovirus receptor (CAR), the receptor for the commonly used adenovirus serotype 5, and the efficiency of adenovirus-mediated gene transfer in bladder cancer cells.Flow cytometric analysis shows that Oct-3/4 overexpression did not significantly change the expression level of CAR on TCCSUP cells (Supplementary Fig. S6).Accordingly, the efficiency of adenovirusmediated gene transfer was similar in TCCSUP/Oct-3/4 and TCCSUP cells, as determined by infection with Ad.Luc, a replication-defective adenoviral vector expressing luciferase (Supplementary Fig. S6).
Ad5WS4 exerts antitumor effects against MBT-2 bladder cancer.To evaluate the antitumor effects of Ad5WS4, mice bearing subcutaneous MBT-2 tumors with an approximate volume of 100 mm 3 were randomly allocated into four groups of 10 mice each and injected i.t. with Ad5WS4, Ad5WS1, heat-inactivated Ad5WS4, or saline for a total of three doses.Ad5WS4 or Ad5WS1 treatment significantly retarded tumor growth compared with heat-inactivated Ad5WS4 or saline treatment (Fig. 6A).Ad5WS4 was superior to Ad5WS1 in suppressing tumor growth at day 30 (Fig. 6B).Furthermore, Ad5WS4 treatment also significantly prolonged the survival time of the tumor-bearing mice compared with Ad5WS1, heat-inactivated Ad5WS4, or saline treatment (Fig. 6C).In mice treated with Ad5WS4, 90% survived by 45 days, whereas the survival rate was 40% and 0% in mice treated with Ad5WS1 and saline, respectively.To assess productive replication of Ad5WS4 in MBT-2 tumors, adenoviral fiber and hexon proteins, which are two viral late proteins, were detected immunohistochemically 3 days after virus injection.Numerous fiber/hexon protein-positive tumor cells were detected in the tumors injected with Ad5WS4 or Ad5WS1, but not in those treated with saline, suggesting active viral replication in the tumors after Ad5WS4 or Ad5WS1 treatment (Fig. 6D).Taken together, these results show that Ad5WS4 exerted higher antitumor efficacy than Ad5WS1 in an immunocompetent bladder tumor model.

Discussion
Although Oct-3/4 is expressed in some solid tumors and may have a role in tumor progression, the significance of this potential has not been elucidated thus far (4-6, 20, 25).In this study, we first confirm a previous finding showing that Oct-3/4 is highly expressed in bladder cancer (6).Most importantly, our clinical data indicate that higher Oct-3/4 expression correlates with tumor progression and metastasis, as well as contributes to patient mortality.It can be speculated that deregulation of Oct-3/4 gene may perturb normal differentiation program and predispose to tumor formation.In normal undifferentiated circumstances, Oct-3/4 is found in pluripotent embryonic cells and germ cells (2,3).Oct-3/4 has since been identified as a highly sensitive marker for seminoma, embryonal carcinoma, and mixed germ cell tumors (7,26).Moreover, its expression has also been addressed in metastatic embryonal carcinomas where it marks the embryonal origin of metastases (27).Our results indicate that Oct-3/4 might also correlate with stem-like cell properties of cancer cells apart from being a marker for ES cells.
Cancer stem cells have recently been reported in various cancers (28)(29)(30).To date, little is known regarding the phenotype of stem cells in healthy or diseased tissues and organs.Whereas some stem cell makers, such as Oct-3/4, may be widely distributed throughout the body, other stem cell antigens may be restricted to a certain organ system.Like classic stem cells, cancer stem cells exhibit plasticity for differentiation, which is associated with the ability to transcribe genes normally expressed in different tissues, including sites other than the primary site of cancer.This broad transcription accessibility can also contribute to the behavior of cancer cells by overexpressing genes that promote cell viability, growth, and metastasis.The CD44-positive prostate cancer cells expressed higher mRNA levels of several ''stemness'' genes including Oct-3/4 and were more proliferative, clonogenic, tumorigenic, and metastatic than the isogenic CD44-negative cells (31).It was shown that hypoxia mediates microenvironment effects on stem cell function by altering Oct-3/4 expression.Hypoxia-inducible factor-2a induces Oct-3/4 expression and promotes tumor growth, as well as collagen degradation and invasion (32,33).
By microarray analysis for comparison of gene expression profiles, we found that many of the Oct-3/4-regulated target genes have previously been described as regulators of cell motility in various tumor cell models (34)(35)(36)(37).Therefore, Oct-3/4 dysregulation may be a prognostic factor in superficial bladder TCC.Our microarray data also show that overexpression of Oct-3/4 upregulates genes involved in transcription/translation. Microarray analysis of mouse ES cells by overexpression (38) and downregulation (39) of Oct-3/4 has identified Tcl1, which is known to regulate cell proliferation, as one of the downstream targets of Oct-3/4.In bladder cancer cells, not only did TCCSUP/Oct-3/4 cells migrate more rapidly, they also proliferated faster than their control cells.Our results show that the effect of Oct-3/4 on cell migration could be distinguished from its effect on cell proliferation.It was shown previously that Oct-3/4 expression in breast cancer cells could be down-regulated by all-trans-retinoic acid and that this effect was associated with decreased cell proliferation and downregulation of FGF-4 expression (20).FGF-4 and osteopontin have been identified as two downstream target genes for Oct-3/4 (40,41).Furthermore, overexpression of FGF-4 in NIH3T3 cells results in increased MMP activity, which correlates with malignant progression and cell invasion (42).In breast, gastric, lung, prostate, liver, and colon cancers, increased osteopontin expression is associated with tumor invasion, progression, and metastasis (43).Therefore, high expression of Oct-3/4 in human bladder TCC may induce some gene deregulation to promote tumor progression.In this study, higher expression levels of MMP-13, MMP-2, and MMP-9 were noted in bladder cancer cells overexpressing Oct-3/4.Overexpression of MMP is associated with cancer promotion and invasion.MMP-13 can degrade type I collagen, whereas MMP-2 and MMP-9 degrade type IV collagen of basement membrane to facilitate tumor invasion (44).Disruption of the balance between the expressions of MMP and MMP inhibitors may facilitate tumor progression and metastasis (45).Therefore, it is unsurprising that overexpression of Oct-3/4 in cancer cells leads to inappropriate activation of MMP-13, MMP-2, and MMP-9 secretion, which promotes cell migration and progression.Moreover, overexpressions of MMP-13, MMP-2, and MMP-9 in our in vitro study with Oct-3/4 stable clones provide the evidence showing that human superficial bladder TCC with high Oct-3/4 expression confers greater tumor metastasis and progression than that with low Oct-3/4 expression.Although the precise role of Oct-3/4 in MMP activation within cancer cells remains to be elucidated, we show for the first time that Oct-3/4 up-regulates MMP-13, MMP-2, and MMP-9 in relation to tumor progression.
Because intact urothelium can function as an effective barrier, bladder is suitable for intravesical instillation of adenoviral vectors.However, low efficiency of gene transfer with adenoviral vectors in cancer gene therapy has been attributable to low expression on cancer cells of CAR.The expression of CAR is heterogeneous among human bladder cancer cell lines (46).It has been shown that the mRNA levels of CAR are reduced in invasive bladder cancer specimens compared with superficial bladder cancer specimens (47).We have previously reported that chemotherapeutic agent etoposide can enhance CAR expression on bladder cancer cells, which can lead to an increase in adenovirus-mediated gene transfer (12).This suggests that oncolytic adenovirus and etoposide can be used in combination for the treatment of bladder cancer.To this end, histone deacetylase inhibitors have been shown to upregulate CAR expression in bladder cancer cells and enhance adenovirus-mediated gene transfer (48).Furthermore, various strategies have been evaluated to circumvent low expression of CAR on target cells by retargeting adenovirus to receptors other than CAR for cell binding (49).
In the animal study, we used mouse MBT-2 bladder tumor model to evaluate the antitumor effect of Ad5WS4.It is generally believed that human adenovirus serotype 5 does not efficiently infect mouse cells nor does it replicate well in these cells.However, we have shown previously that the susceptibility of MBT-2 cells and human bladder cancer cells to adenoviral vector was similar, especially when higher viral dose was applied (12).We also showed recently that adenoviral late proteins ( fiber and hexon) were produced in MBT-2 cells after infection with oncolytic adenovirus Ad.9OC driven by nine copies of the Oct-3/4 response element (11).In this study, we also detected viral late protein production in MBT-2 cells after Ad5WS4 or Ad5WS1 infection.Although oncolytic adenovirus may more efficiently infect and lyse human bladder cancer cells than MBT-2 cells, the syngeneic MBT-2 tumor model allowed us to study the antitumor efficacy of oncolytic adenovirus in immunocompetent mice.Some other murine tumor models have been reported for the same purpose (50).
In this study, we show that expression of Oct-3/4 in bladder cancer promoted tumor progression and metastasis, which may be accounted for by the activation of MMP-13, MMP-2, and MMP-9 expressions by Oct-3/4.Therefore, the expression of Oct-3/4 may contribute to the group of bladder cancer patients with poorer survival, and Oct-3/4 may act as a novel target for cancer therapy especially in cancers with high propensities for metastasis.In the syngeneic MBT-2 tumor model in immunocompetent mice, we conclude that Ad5WS4, an oncolytic adenovirus driven by the Oct-3/ 4 promoter, may serve as a possible therapeutic strategy for bladder cancer with greater effectiveness and cancer-specific potentials.

Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.

Figure 1 .
Figure 1.Oct-3/4 expression detected in human superficial bladder TCC predicts tumor progression.A, examination of Oct-3/4 expression in four human bladder tumor tissues (PT1-PT4) and one normal bladder tissue (NT) by RT-PCR.B, immunohistochemical detection of Oct-3/4 expression in three representative tumor sections (intense, moderate, and low) of human superficial bladder TCC, but not in normal bladder tissue (200Â magnification).Top right, magnified image (320Â magnification) of the area indicated by the box (top middle), which reveals intranuclear staining of Oct-3/4 in tumor tissues.An Oct-3/4 moderate-stained section was also subjected to H&E-staining (bottom right ).Bars, 100 Am.C, correlation between Oct-3/4 expression and tumor progression (top left ), metastasis (top right ), and cancer-related survival (bottom left).D, detection of Oct-3/4 expression in bladder cancer cell lines, as well as murine ES cells, as determined by RT-PCR and immunoblot analysis.Note that either none or a very small amount of Oct-3/4 was detected in normal epithelial cells (SV-HUC-1 and NMuMG) and fibroblasts (NIH3T3).Expression of h-actin served as the loading control.

Figure 2 .
Figure2.Overexpression of Oct-3/4 enhances migration and invasion of bladder cancer cells.A, cell migration was increased in Oct-3/4-overexpressing TCCSUP and MBT-2 cells compared with their control cells.Wound healing assay was done, and wound closure was monitored 12 h later.Cell migration was visualized by light microscopy and photographed.The scale bars shown on 100Â images correspond to 200 Am.B, inhibition of cell proliferation did not abrogate Oct-3/4-mediated enhancement of cell migration.BrdUrd incorporation assay shows that cell proliferation of TCCSUP/Oct-3/4 cells was increased compared with that of the control cells (left).Columns, mean of six determinations, which were consistent in two independent experiments; bars, SE.In the wound healing assay, numbers of the migratory cells at a defined wound area at 12 h were greater in Oct-3/4-overexpressing TCCSUP (middle ) and MBT-2 (right ) cells than those in their control cells, regardless of whether the cells were treated with mitomycin C.During the wound healing assay, cells were treated with or without mitomycin C (5 Ag/mL for TCCSUP and 10 Ag/mL for MBT-2 cells).Columns, mean of three determinations, which were consistent in three independent experiments; bars, SE.C, single-cell tracking shows that expression of Oct-3/4 affected cell migration.Tracking of single-cell migration of TCCSUP (left) and MBT-2 (right ) derivatives by video time-lapse microscopy for 4 h shows that Oct-3/4-overexpressing cells migrated faster and Oct-3/4 knockdown cells migrated slower than their control counterparts.Data are presented as bar and whisker graphs, showing the median and the distribution of 50% (bar) and 90% (whisker) of all tracked cells (n = 10-15), and are representative of two independent experiments.D, overexpression of Oct-3/4 enhanced, whereas Oct-3/4 knockdown reduced cell migration (left ) and invasion through Matrigel (right ).Cell migration was determined in modified Boyden chambers using a filter coated with Matrigel in the invasion assay.After incubation (18 h for TCCSUP cells and 14 h for MBT-2 cells in the migration assay and 16 h in the invasion assay), cells that migrated through the membrane to the lower surface were stained with Giemsa solution and three fields (at 40Â magnification) in each section were counted.Columns, mean of three determinations, which were consistent in three independent experiments; bars, SE.

Figure 3 .
Figure 3. Overexpression of Oct-3/4 promotes metastasis in animal models of experimental pulmonary and spontaneous metastases.A and B, in the mouse model of experimental pulmonary metastasis, C3H/HeN mice (n = 8) were inoculated with MBT-2/Oct-3/4 or control cells via the tail vein and killed 30 d after tumor cell inoculation.A, gross appearance of two representative lungs from each group of mice (left ).The length of the small squire corresponds to 1 cm.Representative H&E staining of the lungs of two mice from each group (right ).The scale bars shown on 40Â images correspond to 1 mm.Note that the numbers and sizes of tumor nodules indicated by arrows were increased in mice inoculated with MBT-2/Oct-3/4 cells.B, wet lung weight of each group of mice (left ).Number of tumor nodules on the surface of the lung from each group of mice (right ).Columns, mean of eight determinations; bars , SE. C and D, in the model of spontaneous metastasis, C3H/HeN mice (n = 8) were inoculated s.c. with MBT-2/ Oct-3/4 or control cells and killed 60 d after tumor cell inoculation.C, number of metastatic nodules on the surface of the lung from each group of mice.The mean values are indicated by horizontal bars.D, representative lung histology stained with H&E of each group of mice.The arrows denote tumor nodules, and the scale bars shown on 40Â and 100Â images correspond to 500 Am.

Figure 5 .
Figure 5. Ad5WS4 exhibits tumor-specific replication and cytolysis in vitro.A, detection of Oct-3/4 (top and bottom ) and E1A (bottom ) promoter activities in various murine cells by luciferase reporter assays.Cells were cotransfected with pGL-mOct-3/4-Luc or pGL-E1A-Luc and pTCY-LacZ, and their relative luciferase activities were measured as luciferase activity divided by h-gal activity to normalize transfection efficiency.Columns, mean of three determinations, which were consistent in three independent experiments; bars , SE. B, CPE in MBT-2 and NIH3T3 cells after infection with various adenoviruses.Cells were infected with varying doses of Ad5WS4, Ad5WS1, AdCMV, or AdLacZ and monitored for CPE by crystal violet staining at 7 d postinfection.C, tumor-selective cytolytic effect of Ad5WS4 and Ad5WS1.Murine MBT-2 (n = 8), NMuMG (n = 6), mouse fibroblast (n = 3), and NIH3T3 (n = 8) cells, as well as human SV-HUC-1 cells (n = 6) were infected with Ad5WS4, Ad5WS1, or AdCMV at indicated MOI.Cell survival was determined after 5 d by WST-1 assay and expressed as the percentage of surviving cells relative to that in the mock-infected cells.Columns, mean; bars, SE.The results shown are representative of two independent experiments.

Figure 6 .
Figure 6.Ad5WS4 exerts antitumor effects in the subcutaneous MBT-2 tumor model.A-C, tumor-bearing mice (n = 10) were treated i.t. with 10 9 pfu of various adenoviruses or with saline at days 12, 14, and 16.A, mice were monitored twice weekly to measure tumor growth.Columns, mean; bars, SE.B, tumor volume of each mouse at day 30.C, Kaplan-Meier survival curves at day 60 for each group of the treated mice.P < 0.0001 for Ad5WS4 versus inactivated Ad5WS4 or saline; P < 0.01 for Ad5WS1 versus inactivated Ad5WS4; P < 0.05 for Ad5WS4 versus Ad5WS1.The results shown are representative of two independent experiments.D, detection of adenoviral fiber and hexon proteins in MBT-2 tumor-bearing mice treated with Ad5WS4 or Ad5WS1, but not in those treated with saline.Groups of three C3H/HeN mice bearing MBT-2 tumors were inoculated i.t. with 10 9 pfu of Ad5WS4 or Ad5WS1or with saline at day 15.Tumors were excised from the treated mice at day 18, and the cryosections were detected immunohistochemically for adenoviral fiber and hexon proteins.The scale bar shown on 320Â images corresponds to 100 Am.