Written informed consent was obtained from all patients to use their sample and the Institutional Review Board at Yamaguchi University Hospital approved this study (approval no. 17). Tumor specimens were obtained from 36 patients with MIBC treated with radical cystectomy with standard lymphadenectomy and from 62 patients with NMIBC treated with transurethral resection (TUR-BT) and diagnosed as high-grade pT1 (high-risk NMIBC) in Yamaguchi University Hospital between January 2001 and December 2012. All specimens were diagnosed as UC by the pathologist. Details of patient backgrounds are listed in Table I.

Formalin-fixed and paraffin-embedded tissue specimens were used for IHC staining. For each sample, 5 µm-thick sections were deparaffinized in xylene, dehydrated in ethanol, and incubated in 0.3% hydrogen peroxide solution in methanol for 10 min. The sections were then microwaved in 0.01 M citrate-buffered solution (pH 6.0) for 15 min and covered in blocking solution (IMMUNO SHOT; Cosmo Bio Co., Ltd., Tokyo, Japan) for 30 min. After addition of a blocking solution, primary antibodies [anti-CD44v9 (1/500 dilution, kindly provided by Professor Saya, Keio University, Tokyo, Japan), anti-cytokeratin 5/6 (CK5/6) (1/200 dilution), or anti-cytokeratin 20 (CK20) (1/100 dilution) (both from Dako, Kyoto, Japan)] were added according to manufacturer's instructions, followed by incubation with the respective secondary antibody (Nichirei Bioscience, Tokyo, Japan) for 30 min at room temperature. Double IHC studies were performed on paraffin-embedded sections using double labeling with CD44v9 and CK5/6 or CK20 by diaminobenzidine and fast red. To evaluate the IHC, the H-score was used in this study. Briefly, >500 tumor cells were counted with different three fields of views in each case, and the H-score was subsequently calculated by multiplying the percentage of the positive cells by the intensity (strongly stained, 3×; moderately stained, 2×; and weakly stained, 1×), yielding a possible range of 0–300 (14,15). Paraffin-embedded cell pellets were obtained by freezing aliquots of two human bladder cancer cell lines (HT1376, 5637) at −80°C followed by fixation in 10% formalin and subsequent processing for use in IHC staining as described (16).

The human bladder cancer cell lines HT1376 and 5637 and the human breast cancer cell line MDA-MB-468 were obtained from the American Type Culture Collection (ATCC; Manassas, VA, USA). The MDA-MB-468 cell line was selected based on previous reports of elevated CD44v9 expression. Cells were routinely cultured in DMEM (Thermo Fisher Scientific, Inc., Waltham, MA, USA) supplemented with 10% FBS (Biological Industries, Cromwell, CT, USA) and antibiotic antimycotic solution (Sigma-Aldrich, Tokyo, Japan) in a 5% CO₂ atmosphere at 37°C.

CD44, CD44v9, and control siRNAs were obtained from Thermo Fisher Scientific, Inc. siRNA sequences were as follows: CD44 siRNA sense, 5′-UAU UCC ACG UGG AGA AAA Att-3′ and antisense, 5′-UUU UUC UCC ACG UGG AAU Aca-3′; CD44v9 siRNA sense, 5′-CUA CUU UAC UGG AAG GUU Att-3′ and antisense, 5′-UAA CCU UCC AGU AAA GUA Gtt-3′. High GC duplex was used as the negative control siRNA. Each cell line was transiently transfected with siRNA using Lipofectamine RNAi MAX (Thermo Fisher Scientific, Inc.) according to the manufacturer's instructions. After transfection, cells were incubated at 37°C in a CO₂ incubator for 48 h. Quantitative evaluation of mRNA and protein expression was verified by western blotting and RT-PCR.

Total cellular RNA was extracted using TRIzol reagent (Thermo Fisher Scientific, Inc.). A total of 1 mg RNA was subjected to cDNA synthesis using the Reverse Transcription Kit (Thermo Fisher Scientific, Inc.) according to the manufacturer's instructions. PCR reactions were carried out in the StepOnePlus Real-Time PCR System (Thermo Fisher Scientific, Inc.) using SYBR-Green Supermix (Takara Bio, Inc., Shiga, Japan) according to the manufacturer's instructions. The primer sequences were as follows: CD44 forward, 5′-CCG CTA TGT CCA GAA AGG A-3′ and reverse, 5′-CTG TCT GTG CTG TCG GTG AT-3′; CD44 standard isoform forward, 5′-AAA GGA GCA GCA CTT CAG GA-3′ and reverse, 5′-TGT GTC TTG GTC TCT GGT AGC-3′; CD44v9 forward, 5′-GGC TTG GAA GAA GAT AAA GAC C-3′ and reverse, 5′-TGC TTG ATG TCA GAG TAG AAG TTG-3′; and β-actin forward, 5′-GCA TCC TCA CCC TGA AGT A-3′ and reverse, 5′-TGT GGT GCC AGA TTT TCT CC-3′. The differences in gene expression levels between samples were measured using the 2^−ΔΔCt method.

Aliquots of cells (5×10⁵) were prepared in assay tubes. After the addition of 2% FBS and 0.02% sodium azide in PBS followed by a rinse and centrifugation, cells were incubated in diluted primary antibody (anti-CD44v9) for 30 min at 4°C. Cells were then washed twice with 2% FBS and 0.02% sodium azide in PBS, and resuspended cells were incubated in phycoerythrin (PE)-labeled secondary antibody for 30 min at 4°C in the dark. After being rinsed twice, the conjugated cells were analyzed according to the manufacturer's instructions (Cell Analyzer EC800; Sony, Tokyo, Japan).

Protein lysates from each tested cell sample were prepared in a radio-immunoprecipitation assay buffer with proteinase inhibitors. Each lysate sample (30 µg) was separated by SDS-PAGE, and electro-transferred to a PVDF membrane. Following blocking in 5% non-fat milk or 5% BSA, these membranes were incubated with each primary antibody overnight at 4°C. After being washed in TBS with 0.05% Tween-20 (TBST), the membranes were incubated with HRP-conjugated secondary antibodies for 1 h at room temperature. After subsequent washing with TBST, membrane signals were detected using an ECL detection system (ChemiDoc™ XRS+; Bio-Rad Laboratories, Inc., Berkeley, CA, USA). Primary antibodies were used at the following dilutions: anti-CD44 (1/1,000 dilution) and anti-β-actin (1/1,000 dilution), both from Abcam (Cambridge, UK); and anti-E-cadherin (1/1,000 dilution); anti-N-cadherin (1/500 dilution); anti-snail (1/500 dilution); and anti-slug (1/500 dilution), from Santa Cruz Biotechnology (Dallas, TX, USA). β-actin was used for protein band normalization.

Equal numbers of transfected and control cells (1×10⁵) in DMEM without FBS were added to the upper insert invasion chambers with Matrigel (BD BioCoat Matrigel Invasion Chamber, 12 wells, 8 µm; Becton-Dickinson, Bedford, MA, USA) and the control chambers without Matrigel. DMEM supplemented with 10% FBS was added to the lower compartment, and the chambers were incubated at 37°C for 48 h. After the incubation period, cells from the upper surface of the filter were wiped off with a cotton swab. The lower surface of the filter was stained with Diff-Quik (Dade Behring AG, Düdingen, Switzerland). The number of cells having migrated to the bottom of the chamber was counted from five randomly selected fields using a light microscope. The mean number of cells was calculated per field. We evaluated the invasion capacity using the ratio between the number of cells in the invasion chamber and the number of cells in the control chamber. Two sets of experiments were carried out, each in triplicate.

The effect of lithium on cell migratory activity was examined using a scratch-wound assay. Each set of transfected and control cells was seeded in a 12-well chamber in DMEM with 10% FBS and grown until confluency. One linear scar was drawn in the monolayer with a 200-µl tip (width, 500 µm). Wound closure (cell migration) was observed under a phase-contrast microscope (×100 magnification; Olympus, Tokyo, Japan), and photographed when the wound was made and at designated times after wounding. Cells were counted within the initial and the remaining wound area and the wound closure was expressed as the percentage of migrating cells compared to the control. Two sets of experiments were carried out, each in duplicate.

Statistical analysis was performed using JMP version 10 (SAS, Cary, NC, USA). Differences between the two groups were analyzed using Student's t-test and the Cox proportional hazard model was applied to determine prognostic factors for progression. Progression-free survival and cancer specific-free survival were calculated using the Kaplan-Meier method and compared by a log-rank test. The Pearson's product-moment correlation coefficient was applied to study the correlation of the H-score in IHC. P<0.05 was regarded as statistically significant. To investigate the relation of IHC CD44v9 expression status to patient outcome, a receiver operating characteristic (ROC) curve was applied to determine the optimal cut-off value to discriminate cancer-specific death. H-scores of 153 and 170 were set as the cut-offs in MIBC [area under the curve (AUC) =0.6428] and high-risk NMIBC (AUC = 0.6354), respectively.

The CD44v9 expression level was evaluated by IHC in 36 MIBC and 62 high-risk NMIBC tissues. Fig. 1A shows the representative pictures of higher and lower expression staining in MIBC and high-risk NMIBC specimens, respectively. Table I shows the clinicopathological characteristics of MIBC and high-risk NMIBC patients with a median follow-up period of 39 months (range, 3–104 months) and 50 months (range, 18–125 months), respectively. In MIBC, 25 specimens (69.4%) showed lower CD44v9 expression (H-score <153), whereas 11 specimens (30.6%) showed higher expression. In high-risk NMIBC, 38 specimens (61.2%) showed lower CD44v9 expression (H-score <170), whereas 24 specimens (38.8%) showed higher expression. Table II shows the prognosis factors in CD44v9-positive and -negative groups. In MIBC, the positive rate of lymphovascular invasion and lymph node metastasis were higher in CD44v9-positive group than in negative group (64 vs. 36%, 18 vs. 8%) (Table IIA). In univariate and multivariate analysis, the expression of CD44v9 and recurrent tumor were significant predictors for progression-free survival in high-risk NMIBC (Table III; both P<0.01). Patients with higher CD44v9 expression exhibited significantly lower progression-free survival and cancer-specific survival than those with lower CD44v9 expression in MIBC (Fig. 2A and B; both P<0.05) as well as in high-risk NMIBC (Fig. 2C and D; P<0.01, P<0.05).

We compared the expression site of CK5/6, a basal-type marker, and CK20, a luminal-type marker, to that of CD44v9 using IHC. Fig. 1B depicts a representative MIBC sample showing the corresponding sites of expression of CD44v9 and CK5/6. In MIBC, the H-score of CD44v9 expression was significantly correlated with that of CK5/6 (Fig. 1C; R2= 0.6329, P<0.001). In high-risk NMIBC, the expression sites of CD44v9 and CK5/6 were colocalized in the invasive front of the tumor, which is defined as the deepest invasion site in this cancer. In contrast, the expression sites of CK20 were localized to the superficial tissue, and distinctly separated from the expression site of CD44v9 in high-risk NMIBC (Fig. 1D).

CD44v9 expression was evaluated by IHC (Fig. 3A) and flow cytometry (Fig. 3B) and western blotting (Fig. 3C) using the 5637 and HT1376 cell lines. The two cell lines were positive for both CK5/6 and CD44v9/CD44 expressions, while being negative for CK20 in IHC (data not shown). Fig. 3D shows the inhibitory effect of CD44 and CD44v9 siRNA on protein and mRNA expression by western blotting and RT-PCR, respectively.

We examined the cell migration and invasion ability associated with the suppression of CD44v9 expression. Initially, we measured the invasion ability of HT1376 and 5637 cells after transient transfection with scrambled or with CD44 and CD44v9 siRNA using a Matrigel invasion assay. The invasion ability of the cells transfected with CD44 and CD44v9 siRNA was significantly lower than that of cells transfected with scrambled siRNA (Fig. 4A and B). Secondly, we measured the migration ability of CD44 and CD44v9 knockdown cells using a wound healing assay. CD44/CD44v9 knockdown cells showed significantly decreased migration ability as compared to scrambled siRNA transfected cells (Fig. 4C and D). No significant difference in the proliferation ability or resistance to cisplatin was observed between the cells transfected with CD44/CD44v9 siRNA and those transfected with control siRNA (data not shown).

We evaluated the correlation of CD44v9 to the EMT markers E-cadherin, N-cadherin, snail, and slug by western blotting (Fig. 5). The cells transfected with CD44 and CD44v9 siRNA showed of N-cadherin, snail, and slug compared with those transfected increased expression of E-cadherin and decreased expression with the scrambled siRNA.