Co-expression of RON and MET is a prognostic indicator for patients with transitional-cell carcinoma of the bladder

Recepteur d'Origine Nantais (RON) is a distinct receptor tyrosine kinase in the c-met proto-oncogene family. We examined the mutational and expression patterns of RON in eight human uroepithelial cell lines. Biological effects of RON overexpression on cancer cells were investigated in vitro, and the prognostic significance of RON and/or c-met protein (MET) expression was analysed in a bladder cancer cohort (n=183). There was no evidence of mutation in the kinase domain of RON. Overexpression of RON using an inducible Tet-off system induced increased cell proliferation, motility, and antiapoptosis. Immunohistochemical analysis showed that RON was overexpressed in 60 cases (32.8%) of primary tumours, with 14 (23.3%) showing a high level of expression. Recepteur d'Origine Nantais expression was positively associated with histological grading, larger size, nonpapillary contour, and tumour stage (all P<0.01). In addition, MET was overexpressed in 82 cases (44.8%). Co-expressed RON and MET was significantly associated with decreased overall survival (P=0.005) or metastasis-free survival (P=0.01) in 35 cases (19.1%). Recepteur d'Origine Nantais-associated signalling may play an important role in the progression of human bladder cancer. Evaluation of RON and MET expression status may identify a subset of bladder-cancer patients who require more intensive treatment.

In our prior profiling experiment (Cheng et al, 2002), RON was one of the receptor-type PTKs expressed in bladder cancer cells. Interestingly, this receptor protein and its cognate ligand MSP are all located within chromosome 3p21.3, a region frequently amplified in human bladder cancer (Koo et al, 1999). As crosstalk between RON and MET was observed in epithelial cancer (Chen et al, 1997;Maggiora et al, 2003), we investigated the clinical significance of RON and MET overexpression in human bladder cancer.

Immunoprecipitation
Urine samples (n ¼ 32) were concentrated using a centrifugal filter device with a molecular weight cutoff of 50 kDa (UltraFree-4; Millipore, Bedford, MA, USA) as described previously (Yan et al, 2001). Protein concentrations of the concentrated urine samples were determined using the Micro BCA method (Pierce Chemical, Rockford, IL, USA). Equal amounts of proteins (20 mg) were loaded onto 4 -15% gradient gels and separated using sodium dodecyl sulphate (SDS) -polyacrylamide gel electrophoresis under nonreducing conditions. Resolved proteins were electrophoretically transferred to TransBlot nitrocellulose membranes (Bio-Rad, Richmond, CA, USA). The membranes were blocked with 5% low-fat dry milk in TBS-T (10 mM Tris, pH 7.2, 50 mM NaCl, 0.5% Tween 20) for 1 h at room temperature, followed by incubation with primary antibody (1 : 100 dilution) against human MSPa (R&D Systems, Minneapolis, MN, USA) at 41C for 18 h. Blots were washed eight times with TBS-T (5 min per wash) and incubated with a 1 : 5000 dilution of horseradish peroxidase-conjugated secondary antibody (Vector Laboratories, Inc., Burlingame, CA, USA) diluted in TBS-T containing 3% bovine serum albumin for 1 h at room temperature. Labelled proteins were visualised with enhanced chemiluminescence (Amersham Biosciences Corp., Piscataway, NJ, USA). Cell lysates from HepG2 and UB37 were used as positive controls. Informed consent was obtained from all patients before their urine was taken for analysis.

Chemicals and plasmids
Wortmannin was obtained from Sigma (Sigma-Aldrich, St Louis, MO, USA). Plasmid pLP-TRE2-RON was constructed in our laboratory, and co-transfected with pTet-Lac-Hygo in the subsequent experiments as described (Chow et al, 2003). In this study, the Tet-off system using doxycycline was used to regulate exogenous RON expression.

Preparation of total RNA and real-time polymerase chain reaction (RT -PCR) analysis
Total RNA was extracted using Trizol reagent (Invitrogen Corp., Carlsbad, CA, USA). First-strand cDNA was synthesised in a total volume of 20 ml using an M-MLV Reverse Transcriptase kit (Invitrogen). The Roche LightCycler Real-time PCR system (Roche Diagnostics Corporation, Roche Applied Science, Indianapolis, IN, USA) was used to quantify the expression of target genes with SYBR Green I labelling. Polymerase chain reactions were performed in a total volume of 10 ml containing 3 ml cDNA (1 : 100 dilution), 2.5 mM MgCl 2 , 10pM forward and reverse primers, 1 ml SYBR Green I, and 4.8 ml PCR-grade H 2 O. The reaction was performed for 7 min, to denature DNA and activate the Hot-Start polymerase, followed by 50 PCR cycles denatured at 951C for 10 s, annealed at 601C for 5 s, and elongated at 721C for 12 s in glass capillaries. The primers used were: RON receptor forward 5 0 -AGCCCACGCTCAGTGTCTAT-3 0 , RON receptor re-verse 5 0 -GGGCACTAGGATCATCTGTCA-3 0 ; PPIA forward 5 0 -GTTTGCAGACAAGGTCCCA-3 0 , and PPIA reverse 5 0 -ACCCGTAT GCTTTAGGATG-3 0 .

Western blot analysis
In all, 50 mg of total protein was denatured in SDS lysis buffer (Tris-HCl (50 mM), pH 6.8, SDS (2%), glycerol (10%), and dithiothreitol (100 mM)) and then loaded into duplicated 12% SDS -polyacrylamide gels. After electrophoresis, one of the gels was stained with 0.5% colorimetric Coomassie brilliant blue (Sigma) as a quantitative control. The other gel was transferred to a polyvinylidene difluoride membrane (Stratagene, La Jolla, CA, USA) and blocked with 5% skimmed milk in PBST (sodium chloride (100 mM), disodium hydrogen phosphate (80 mM), sodium dihydrogen phosphate (20 mM), Tween 20 (0.2%), pH 7.5) solution at 41C overnight. After being washed with PBST and phosphatebuffered saline, the membrane was hybridised with the monoclonal antibodies for MET, RON, or b-actin (Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA) at 371C for 1 h. The membrane was then incubated with anti-rabbit or anti-mouse IgG conjugated with horseradish peroxidase at 251C for 1 h. After being thoroughly washed, the membrane was exposed to radiographic film following a reaction with enhanced chemiluminescence detection reagents (Amersham). Human b-actin was used as an internal control.

Single-strand conformation polymorphism analysis and mutation analysis of the RON gene
Each exon of the kinase domain, plus a region containing the proteolytic cleavage site in exon 1, was examined for mutation using a specific PCR primer pair as described previously (Angeloni et al, 2000). First-strand cDNAs from primary uroepithelium, UB40, and UB47 cells were prepared using reverse transcription of 1 mg of RNA, according to the instructions for the SuperScript preamplification system (Gibco BRL). The primers used for amplification of RON for PCR were chosen from a prior report . To check for potential mutations of RON, genomic DNA was also amplified using two primers flanking this region: sense oligomer corresponding to nucleotides 2619 -2641 (5 0 -ATCCACCCAGTGCCAACCTAGTT-3 0 ) and antisense oligomer corresponding to nucleotides 2873 -2895 (5 0 -GGCCAGATGGGGT CCCACA GAG-3 0 ) . Each sequence was confirmed with two independent PCR reactions.

Cell viability staining
The cells were re-suspended in DMEM medium. Equal volumes of cell suspension and ethidium-bromide/acridine-orange solution were mixed and placed underneath the cover slides. The cells were observed under a microscope. The live cells labelled with acridine orange are green, and the dead cells labelled with ethidium bromide are orange. The cells were counted and their viability was calculated.

Wound-healing assay
The UB09 cells (3 Â 10 6 ) were seeded on a 30-mm dish overnight after transfection. A mid-line wound was made on the monolayer cells, and the healing process was recorded every 20 min until the wound was completely healed (Ho et al, 2005). The Image-Pro plus computer program (Media Cybernetics, San Diego, CA, USA) was used to calculate the distance between the wounded edges.

Clinicopathologic characteristics
For this study, archival blocks were collected from 183 patients (117 men and 66 women; age range: 23 -93 years old; mean age: 64.2710. 4 years) with primary TCC of the urinary bladder treated at our hospital between July 1, 1988, and July 31, 1994. Patients with primary carcinoma in situ were excluded from this study. The study protocol was approved by the Human Investigations Committee in our hospital. All cases were reviewed for histological grade according to the World Health Organization classification (Eble et al, 2004). Clinical staging was determined according to the Tumor-Node-Metastasis System of the Union Internationale Contre le Cancer (Sobin and Wittekind, 2002) using a survey of the clinical details, image studies, and pathologic data. But no definite tumour staging could be determined at diagnosis in 10 patients (5.5%). This cohort was different from our prior study on MET (Cheng et al, 2002).
The treatment and follow-up of patients were conducted according to the standard strategy previously described in detail (Cheng et al, 2002). Briefly, all patients with superficial bladder cancer (n ¼ 93) received transurethral resection and postoperative intravesical chemotherapeutic agent instillation with either thiotepa (30 mg in 30 ml of normal saline for 70 patients) or epirubicin (40 mg in 40 ml of normal saline for 34 patients) weekly for 8 consecutive weeks. Those patients who received intravesical bacillus Calmette-Guerin therapy or neoadjuvant chemotherapy were excluded from this study to preclude any potential bias in prognostic correlation (e.g. recurrence rate or tumour progression). The patients were followed up every 3 months for the first 2 years, every 6 months for another 2 years, and yearly thereafter. Recurrent tumours were confirmed by biopsy, and the patient was given a transurethral operation. This was followed by another 8-week course of intravesical chemotherapeutic instillation therapy or by a radical or partial cystectomy if there was disease progression. Disease progression at recurrence was defined as being at a higher stage than the previous result. In all, 77 patients with muscle-invasive tumours were given a radical cystectomy, and nine patients were given a partial cystectomy. The four patients with distant metastasis at diagnosis received a transurethral resection only. A total of 75 patients received systemic chemotherapy with methotrexate (MTX), cisplatin, doxorubicin, and vinblastine. Survival status was determined by checking outpatient clinic records, interviewing patients' families, or both. Clinical follow-up ranged from 24 to 95 months (median: 54 months).

Immunohistochemistry (IHC) of RON and MET expression
Immunostaining procedures were described previously (Cheng et al, 2002). Briefly, tissue sections were incubated at room temperature for 2 h with monoclonal anti-MET and anti-RON antibodies (Santa Cruz Biotechnology) raised against human MET or RON protein, respectively. The optimal dilution (1 : 100) was determined using human kidney tissue as a positive control (Rampino et al, 2002). The StrAviGen Super Sensitive MultiLink kit (BioGenex Laboratories, Inc., San Ramon, CA, USA) was used to detect the resulting immune complex. Peroxidase activity was visualised using an aminoethyl carbazole substrate kit (Zymed Laboratories, Inc., San Francisco, CA, USA). Finally, sections were counterstained with haematoxylin. For the negative control, nonimmune mouse immunoglobulin was substituted for the primary antibody in the incubation.
When they evaluated the staining results, the reviewers (N-HC and C-LH) were blinded to the clinical outcomes. Since there was no apparent difference in staining intensity, we used a threecategory scoring system for RON and MET based on the proportion of tumour cells stained, as described previously (Chow et al, 2001;Cheng et al, 2002). 'High level of expression' indicates that more than 50% of the tumour cells exhibited immunostaining; 'low level of expression', between 10% and 50% reactivity; and 'negative', less than 10% or zero staining for RON protein.

Statistics
Correlations between RON or MET expression and clinicopathologic indicators of bladder cancer, and the biological effects of RON expression on cancer cells were examined, where suitable, using analysis of variance, Fisher's exact test, or w 2 test. The relationship between IHC expression pattern or biological indicators and clinical outcome was analysed using a multiple logistic regression model. Overall survival was calculated using Kaplan -Meier analysis, and the Cochran -Mantel -Haenszel test (log-rank test) was used to assess the significance of RON and MET expression in relation to tumour recurrence or patient survival. The relative risk (RR) in relation to patient prognosis was assessed using a Cox proportional hazards model after adjustment for clinicopathologic parameters. Only those variables with a P-value o0.05 were considered significant.

Measurement of MSP in human urine
To examine the involvement of MSP/RON signalling in bladder carcinogenesis, we analysed the appearance of urinary MSP in a total of 17 cases of urothelial carcinomas (two grade 1, nine grade 2, and six grade 3) and 15 cases of non-neoplastic inflammatory urinary tract diseases. Immunoprecipitation showed that 10 of 17 with cancer and one with an inflammatory urinary tract disease had detectable urinary MSP (P ¼ 0.04). Representative results of cancer (n ¼ 9) and xanthogranulomatous pyelonephritis (n ¼ 1) cases are shown in Figure 1. The appearance of urinary MSP was basically independent of the RON expression status in primary tumours (data not shown). These results imply that the activation of MSP/RON may be involved in the development of human bladder cancer.

Expression of RON and MET receptors in uroepithelial cell lines
The expression of RON was examined at the mRNA level using RT -PCR in a primary culture of uroepithelium and eight cancer cell lines (Figure 2). TCCSUP had the highest level of RON expression in this panel, while the other seven cancer cell lines had variable level of RON expression, with RT4 being the lowest one. Unexpectedly, primary culture of uroepithelium was found to have high RON expression.
Then, expression of RON was assessed at protein level using Western blot (Figure 3). Primary uroepithelial cells and seven of

HepG2 E6
Urine 701 Urine 704 Urine 512 Urine 580 Urine 669 Urine 752 Urine 753 Urine 754 Urine 772 DMEM UB37 Figure 1 Immunoprecipitation of MSP in human urine. All nine patients with transitional-cell bladder cancer had detectable MSP in their urine, but MSP was not measurable in the urine of patients with xanthogranulomatous pyelonephritis (case 772). The column of DMEM was taken from medium used for culturing of HepG2 and UB37 cells.  (Rossi et al, 2001). Therefore, it is plausible to speculate that primary culture grown in serum-enriched media may exhibit a distinctly different phenotype.
To determine whether constitutive activation of RON is caused by mutations in the kinase domain of RON, genomic DNA was submitted for single-strand conformational polymorphism screening using a panel of intron-based primers covering exon 1 and exons 14 -20 of the RON gene. The amplified DNA fragments were also sequenced. No mutations were found in the kinase domain of the RON gene in any of the cell lines tested (Chow et al, 2003).
All of the uroepithelial cells expressed mature p140 MET and precursor p170 MET , except that a relatively lower level of p170 MET was observed in TCCSUP cells (Figure 2). Altogether, coexpression of RON and MET was a universal event in uroepithelial cells, highly suggestive of the potential clinical significance of crosstalk between RON and MET.  Growth curve of UB09 cells after transfection of RON. The cells (1 Â 10 4 per well for six wells) were cultured in 10% serum-containing medium with or without doxycycline (1 mg ml À1 ). The cell number was calculated from 3 to 9 days using a cell counter. Those cells expressing higher RON had a significantly faster growth rate, as indicated by asterisks (P ¼ 0.0028, t-test).

Effects of RON overexpression on the biological activities of cancer cells
As RT4 cells are resistant to transfection, UB09 cells expressing a relatively lower level of RON were transfected with plasmid DNA to express exogenous RON protein. The plasmids (80 mg) pLp-TRE2-RON and pTet-Lac-Hygo at a molar ratio of 1 : 1 were transfected into UB09 cells using electroporation (180 V, 1000 mF). The cells were subcultured into a six-well tray (1 Â 10 4 per well) 12 h later. The cell numbers were counted at a 3-day interval from day 0 through day 9. Overexpression of RON (UB09/RON) induced a higher growth rate of UB09 cells compared to the parental cells (P ¼ 0.0028) (Figure 4).
To evaluate the effect of RON upon cell motility, UB09 cells (1 Â 10 6 ) were transfected and cultured on a cover glass for 24 h. Time-lapse recording was then used to evaluate the progression of wound healing produced by a yellow tip (Ho et al, 2005). The images were recorded every 20 min for a total of 30 -48 h. It showed that UB09/RON cells migrated faster than control cells (P ¼ 0.04) ( Figure 5).
To test the effect of RON on apoptosis, UB09 cells were first treated with MTX for 48 h and stained with acridine orange for apoptotic cell measurement. After MTX treatment (50 mg ml À1 ), 60% of the UB09 cells were apoptotic, but only 25% of the UB09/ RON cells were apoptotic (P ¼ 0.024). The antiapoptotic effect, however, was blocked by wortmannin treatment (10 nM), a P13K inhibitor ( Figure 6). Altogether, overexpression of RON increased the growth rate and motility of the cells and conferred an antiapoptotic efficacy on cancer cells through a PI3K-related signaling pathway.

Immunohistochemical expression of RON or MET receptor in primary bladder cancer
To clarify the clinical significance of RON receptor expression, we studied the IHC expression in a transitional-cell bladder-cancer cohort (n ¼ 183). Membranous staining for RON was seen in only a few non-neoplastic urothelial cells (Figure 7a). Overexpression of RON was defined as membranous staining in more than 10% of tumour cells (Figure 7b). In all, 60 cases (32.8%) were classified as overexpression of RON, and 14 (23.3%) of those were high expression (summarised in Table 1). Overexpression of RON was positively associated with histological grading (P ¼ 0.003), larger tumours (P ¼ 0.003), nonpapillary contour (P ¼ 0.005), and tumour stage (P ¼ 0.01) using Fisher's exact test or the w 2 test (Table 1).
We also examined the expression of MET in this cohort (Table 1). In total, 82 cases (44.8%) overexpressed MET, and 27 (32.9%) of those were high expression. MET expression was positively associated with histological grading, nonpapillary contour, tumour size, and muscle invasion using Fisher's exact test or the w 2 test (all Po0.001). In all study cases, co-expression of both receptors was detected in 35 cases (19.1%), and expression of RON or MET was detected in 25 cases (13.7%), and 47 cases (25.7%), respectively (Table 2).

Prognostic significance of co-expression of RON and MET in a bladder cancer cohort
In univariate analysis, none of the biological indicators analysed were correlated with tumour recurrence using the log-rank test (P40.1, respectively) (data not shown), except that muscleinvasive tumours tended to have a higher risk of local recurrence (P ¼ 0.06). Factors associated with decreased patient survival were overexpression of RON (P ¼ 0.003), co-expression of RON/MET (P ¼ 0.005), and overexpression of MET (P ¼ 0.02). Patients who co-expressed RON/MET in their bladder tumours had a significantly worse overall survival rate (P ¼ 0.005) or metastasis-free survival rate (P ¼ 0.01) compared with the remaining expression status of the two receptors ( Figure 8).
Multivariate analysis using a logistic regression model revealed that neither biological indicator nor expression of RON or MET or both was associated with tumour recurrence (Table 2). For longterm survival, important prognostic indicators were multiplicity (P ¼ 0.009), co-expression of RON/MET (P ¼ 0.03), expression of RON (P ¼ 0.04), and stage classification (P ¼ 0.05) using the logrank test. Overexpression of MET by itself tended to correlate with poor patient survival in this cohort (P ¼ 0.06). We next used a Cox proportional hazards models to determine the RR of overall survival with 95% confidence interval (CI). The RR of poor longterm survival was 2.46 for multiple tumours, 2.22 for co-expression of RON and MET, 2.03 for RON expression alone, and 1.98 for tumour staging.

Significance of co-expression of RON and MET in superficial bladder cancer
In the subset of superficial bladder cancer (Stages O and A), the frequency of RON and MET overexpression was 19.4% (18 out of 93) and 32.3% (30 out of 93), respectively (Table 1). Indicators associated with poor overall survival were co-expression of RON and MET (P ¼ 0.03) and multiplicity (P ¼ 0.05), with RR estimated at 3.43 (95% CI 1.16 -10.12) and 2.22 (95% CI 1.01 -4.91), respectively (data not shown). But expression of RON or MET or both did not predict the chance of long-term survival for patients with muscle-invasive tumours.

DISCUSSION
In this study, we found that expression of RON or MET, or both, is positively associated with aggressive biological indicators and decreased patient survival, which supports the hypothesis that RON-related signalling events play an important role in the progression of bladder cancer. Our results are substantially in agreement with studies on primary carcinomas of the breast (Maggiora et al, 1998), colorectum (Zhou et al, 2003), liver (Chen et al, 1997), and ovary (Maggiora et al, 2003). Moreover, our in vitro experiments indicated that increased cell growth or motility and the antiapoptotic effect are the underlying mechanisms for RON in bladder carcinogenesis. These results are consistent with reports showing a mitogenic/invasive response induced by constitutive activation of RON  and activation   of signalling pathways, including PI3K, in regulating cell adhesion, motility, growth, and survival after stimulation with MSP Peace et al, 2001).
The most intriguing observation of this investigation, however, is the prognostic importance of the co-expression of RON and MET, especially for patients with superficial bladder cancer (P ¼ 0.03; RR ¼ 3.43). This observation has clinical implications because this group of patients represents the largest group of urological malignancy in common practice. Co-expression of both receptors thus appears to signify a distinctive 'invasive growth' programme for carcinoma cells. This conclusion essentially concurs with the current hypothesis of crosstalk between subfamily members of the ErbB receptors and their prognostic significance (Wada et al, 1990;Sliwkowski et al, 1994;Pinkas-Kramarski et al, 1996;Chow et al, 2001). Therefore, evaluation of the RON and MET expression status may identify a subset of bladder cancer patients who might require more intensive treatment.
Further support for clinical relevance of RON/MET interaction in human cancer comes from in vitro experiments by Follenzi et al (2000). First, MET and RON exist on the cell membrane surface as preformed dimers before ligand stimulation, and there is a bidirectional transphosphorylation between MET and RON after stimulation with either hepatocyte growth factor or MSP. Second, although RON is a less efficient kinase than MET, formation of MET/RON complexes induces a more efficient RON transphosphorylation by MET, leading to a more sustained signal than that induced by the RON/RON homodimer. Activation of both MET and RON may thus possibly initiate a cooperative or synergistic response to their ligands. Further investigation is mandatory to clarify the benefits of combined therapies against RON/MET receptors and their signal transducers.
In our series, the most important prognostic indicators for bladder cancer patients were, first, multiple tumours at diagnosis and, second, advanced tumour stage. A similar discovery was also reported in some earlier studies (Holmang et al, 1995;Crew et al, 1997;Cheng et al, 2002), although no complete rationale was proposed to explain for the occurrence. Since dysplastic change in the field mucosa correlates well with clinical outcome in cancer patients, multiple tumours at diagnosis may represent a sign of unstable urothelium. Alternatively, if multiple bladder cancers are monoclonal in origin (Sidransky et al, 1992), the occurrence might signify an intraluminal spreading and implantation of tumour cells. A prospective study is necessary to verify the consequences of such spreading and implantation before determining what might be appropriate treatment plans for bladder cancer patients.
In summary, the results of our study indicate that RON-related molecular events are important in the progression of bladder carcinogenesis. Evaluation of the expression pattern of MET and RON is of great help in selecting bladder cancer patients for more aggressive therapy.