An Oncogenic Mechanism for Mutant FGFR3 Erica Di Martino1, Gavin Kelly2, Jo-An Roulson3, and Margaret A

Published OnlineFirst September 15, 2014; DOI: 10.1158/1541-7786.MCR-14-0022

Genomics Molecular Cancer Research Alteration of Cell–Cell and Cell–Matrix Adhesion in Urothelial Cells: An Oncogenic Mechanism for Mutant FGFR3 Erica di Martino1, Gavin Kelly2, Jo-An Roulson3, and Margaret A. Knowles1

Abstract

Activating mutations of FGFR3 are a common and early event and MMP10 transcripts was found in a large fraction of primary in bladder cancer. Ectopic expression of mutant FGFR3 in bladder tumors analyzed, supporting their key role in bladder normal urothelial cells has both pro-proliferative and antia- tumorigenesis in vivo. However, no correlation was found poptotic effects at confluence, suggesting that mutant cells are between their protein and/or mRNA expression and FGFR3 insensitive to cell–cell contact inhibition. Herein, detailed mutation status in tumor specimens, indicating that these genes analysis revealed that these cells have reduced cell–cell adhe- may be targeted by several converging oncogenic pathways. sion, with large intercellular spaces observable at confluence, Overall, these results indicate that mutant FGFR3 favors the and diminished cell–substrate adhesion to collagen IV, colla- development and progression of premalignant bladder lesions gen I, and fibronectin. These phenotypic alterations are accom- by altering key genes regulating the cell–cell and cell–matrix panied by changes in the expression of genes involved in cell adhesive properties of urothelial cells. adhesion and extracellular matrix remodeling. Silencing of endogenous mutant FGFR3 in bladder cancer cells induced Implications: The ability of mutant FGFR3 to drive transcription- converse changes in transcript levels of CDH16, PLAU, MMP10, al expression profiles involved in tumor cell adhesion suggests a EPCAM, TNC, and HAS3, confirming them as downstream gene mechanism for expansion of premalignant urothelial lesions. Mol targets of mutant FGFR3. Overexpression of EPCAM, HAS3, Cancer Res; 13(1); 138–48. 2014 AACR.

Introduction domain, inducing the formation of disulfide bonds between adjacent monomers, and therefore receptor dimerization and FGFR3 is one of four transmembrane tyrosine-kinase receptors activation in the absence of ligand (4, 5). These are the most mediating the cellular effects of FGFs. Binding of FGFs to the frequent of all FGFR3 mutations observed in bladder tumors extracellular portion of FGFR3 induces its dimerization, followed and include S249C (61%), Y375C (19%), R248C (8%), by phosphorylation of its cytoplasmic tyrosine kinase domain and G372C (6%) (3). Mutations in exon 15 cause a confor- and activation of a number of downstream signaling pathways mational change in the regulatory region of the receptor, regulating a range of cellular functions (1). leading to constitutive phosphorylation and activation (6). Activating mutations of FGFR3 are a common finding in These mutations involve codon 652 and are rarer in bladder urothelial carcinoma of the bladder, particularly in the subset tumors (<2% of mutant tumors). Other mutations of the of low-grade and low-stage tumors, where their frequency hotspot regions are only found in a very small proportion of reaches 70% (2). They map to three mutation hotspots in cases (<5%) (3). Besides bladder cancer, activating mutations exons 7 (codons 248 and 249), 10 (codons 372, 373, 375, 382, of FGFR3 are also detected in benign skin tumors (7) and, at and 393) and 15 (codon 653) (3). Mutations in exons 7 and lower frequency, in multiple myeloma (8) and cervical cancer 10 produce an unpaired cysteine residue in the extracellular (9). FGFR3 mutations are thought to be an early change during 1Section of Experimental Oncology, Leeds Institute of Cancer and urothelial transformation, as they are often detected in flat urothe- Pathology, St. James's University Hospital, Leeds, United Kingdom. lial hyperplasia, a proposed precursor lesion for urothelial carci- 2 Bioinformatics and Biostatistics Service, Cancer Research UK, Lon- noma (10). Interestingly, a number of bladder tumors without don Research Institute, Lincoln's Inn Fields Laboratories, London, United Kingdom. 3Section of Pathology and Tumour Biology, Leeds FGFR3 mutation display overexpression of wild-type FGFR3 as an Institute of Cancer and Pathology, St. James's University Hospital, alternative mechanism of FGFR3 upregulation (11). Therefore, Leeds, United Kingdom. overall a very high proportion of bladder tumors are characterized Note: Supplementary data for this article are available at Molecular Cancer by FGFR3 dysregulation. In vitro studies have shown that silencing Research Online (http://mcr.aacrjournals.org/). or inhibition of FGFR3 in bladder cancer cells is associated with Corresponding Author: Margaret A. Knowles, Leeds Institute of Cancer and decreased proliferation, reduced anchorage-independent growth, Pathology, Cancer Research UK Clinical Centre, St. James's University Hospital, and enhanced apoptosis (12–14), and FGFR inhibitors have Beckett Street, Leeds, LS9 7TF, UK. Phone: 441132064913; Fax: 441132429886; therefore been proposed as novel therapeutic agents in the treat- E-mail: [email protected] ment of bladder tumors (15). However, little is known about the doi: 10.1158/1541-7786.MCR-14-0022 detailed molecular mechanism by which FGFR3 signaling con- 2014 American Association for Cancer Research. tributes to the malignant transformation of urothelial cells.

138 Mol Cancer Res; 13(1) January 2015

Mutant FGFR3 Alters Adhesive Properties of Urothelial Cells

We have shown that expression of ectopic mutant FGFR3 in Aldrich Company Ltd.) for 1 hour at room temperature. E-cad- normal urothelial cells (NHUC) induces aberrant activation of the herin and b-catenin localization was detected by fluorescence MAPK and PLCg1 signaling pathways and increases cell prolifer- microscopy using 1:300 Alexa Fluor 488 goat anti-mouse and ation and viability at confluence (16). As a result, NHUC expres- Alexa Fluor 594 goat anti-rabbit IgGs (Life Technologies Ltd). sing mutant FGFR3 have an increased saturation density, reaching higher cell numbers in confluent conditions compared with Cell–substrate detachment assays control cells. The intensity of the phenotypic effect is mutation To test cell–substrate detachment, cells were grown to high dependent and reflects the relative mutation frequencies observed confluence in 6-well plates coated with different substrates (BD in bladder tumors (16). On the basis of these results, we hypoth- Biosciences). Cells were then subjected to trypsinization by incu- esized that mutant FGFR3 may allow urothelial cells to escape bating for 12 minutes with Trypsin-EDTA solution (0.5 g/L cell–cell contact inhibition and therefore contribute to the devel- porcine trypsin and 0.2 g/L EDTA; Sigma-Aldrich Company Ltd.). opment of hyperplastic lesions in the bladder. For qualitative results, attached cells were fixed and stained with In this study, we investigate the consequences of FGFR3 muta- 1% methylene blue in 50% ethanol; for quantitative results, both tion on the adhesive properties of urothelial cells. We show that detached and attached cells were counted with a Z2 Coulter normal urothelial cells expressing ectopic mutant FGFR3 display analyzer (Beckman Coulter Ltd.). reduced cell–cell and cell–substrate adhesion and identify a number of FGFR3 target genes that could be mediating this RNA extraction and microarray processing phenotype in vitro. Overall, our data provide a deeper under- RNA from cell lines was extracted using TRIzol reagent (Life standing of the oncogenic effects of mutant FGFR3 by indicating Technologies Ltd), following the manufacturer's instructions. molecular mechanisms by which it may drive the development of RNA was DNAse treated and repurified using the RNeasy mini bladder tumors. Kit (Qiagen Ltd.). RNA from snap-frozen tumor tissue was extracted with the RNeasy mini Kit (Qiagen Ltd.), including the Materials and Methods optional DNAse treatment step. For expression array analysis, RNA labeling and hybridization were performed by the Cancer Cell lines Research UK Manchester Institute Microarray Service. Briefly, 2 mg Normal urothelial cells immortalized with telomerase (TERT- of total RNA was used to prepare biotinylated target cRNA, NHUC) and expressing mutant FGFR3 were obtained as described according to the Affymetrix One Cycle Target Preparation Proto- previously (16). Expression of endogenous S249C FGFR3 was col, which was then hybridized on Affymetrix HG U133 Plus 2.0 silenced in the bladder cancer cell line 97-7 using shRNA, as oligonucleotide arrays (Affymetrix UK Ltd.). These data are avail- described previously (13). Cell lines were cultured in standard able in the GEO NCBI database (accession no. GSE61352). growth media at 37 Cin5%CO2. Cell line identity was verified by short tandem repeat DNA typing using a Powerplex 16 kit (Pro- Expression array analysis fi mega). Pro les were compared with matched initial patient Expression array data were analyzed using R (http://www.R- fi samples (TERT-NHUC) and pro les obtained from early passage project.org) and Bioconductor (20). They were processed at the cells from the originating laboratory (97-7). probe level using the RMA algorithm (21), and the difference between NHUC expressing ectopic S249C FGFR3 and control Tumor samples NHUC containing the empty vector was tested using moderated t fi Bladder tumor samples classi ed according to World Health tests, where the variances were estimated using an empirical Bayes – – Organization and Tumor Node Metastasis guidelines were col- method to correct for underestimation of noise (22). To control lected in Leeds (UK) between 2001 and 2004, with ethical approval for false discovery rate (FDR) due to multiple testing effects, the P and written informed consent (Leeds East 99/156). RNA, DNA, and values were adjusted using the Benjamini–Hochberg method. A fi fi formalin- xed and paraf n-embedded (FFPE) tissue were threshold of 1% FDR was employed to determine differential obtained as described previously (17). Representative fresh-frozen expression. Genes involved in cell–cell and cell–substrate adhe- sections were stained with hematoxylin and examined to deter- sion were highlighted within the differentially expressed genes mine tumor content. Only samples containing >70% tumor mate- using the Ingenuity Pathway analysis 7.6 software (Ingenuity FGFR3 rial were included in the study. mutation status was assessed Systems Inc.), the publicly available DAVID Bioinformatics data- by SNaPshot or high-resolution melting curve analysis followed by base (23), and a literature search. sequencing, as described previously (18, 19). The mRNA analysis included a total of 44 tumors of all grades and stages. Information Real-time RT-PCR about grade, stage, and FGFR3 status is summarized in Supple- cDNA synthesis and Taqman RealTime PCR were performed mentary Table S1. Protein analysis included a total of 48 pTaG2 as described previously (17). The following TaqMan Gene tumors,of which 34 were mutant for FGFR3 (26 S249C, 3 G372C, 2 Expression Assays (Applied Biosystems) were used: Y375C, 2 R248C, 1 K652E) and 14 were wild type. Hs00188166_m1 (SDHA, internal control), Hs01013953_ m1 (CDH1), Hs00901888_g1 (EPCAM), Hs00355084_ Immunofluorescence m1 (DSG1), Hs00951428_m1 (DSC2), Hs00187880_m1 Cells were seeded on glass coverslips and grown to confluence. (CDH16), Hs00193436_m1 (HAS3), Hs01547054_m1 Cells were then fixed with 4% paraformaldehyde for 10 minutes at (PLAU), Hs00938315_m1 (PLAT), Hs00169627_m1 (CD36), 4 C, permeabilized with 0.1% Triton X-100 (Sigma-Aldrich Com- Hs00233987_m1 (MMP10), and Hs01115665_m1 (TNC). pany Ltd.), and incubated with 1 mg/mL DAPI, 1:500 mouse Gene expression was normalized to SDHA using the DCt meth- monoclonal anti–E-cadherin antibody (HECD-1; Abcam Plc.), od and quantified relative to a positive control sample. Non- and 1:500 rabbit polyclonal anti–b-catenin antibody (Sigma- template negative controls were included in each plate.

www.aacrjournals.org Mol Cancer Res; 13(1) January 2015 139

di Martino et al.

Immunohistochemistry trolled trypsinization to promote their detachment from standard Immunohistochemistry was performed as described previously culture dishes and culture dishes coated with various extracellular (17). Samples were incubated with 1:400 mouse anti-EPCAM (C- matrix (ECM) proteins. For qualitative results, detached cells were 10; Santa Cruz Biotechnology Inc.) or 1:400 goat anti-HAS3 (E15; gently washed away and attached cells were visualized by staining Santa Cruz Biotechnology Inc.). Incubation with primary anti- with methylene blue; for quantitative results, both detached and body was performed either for 1 hour at room temperature attached cells were counted. A higher number of control cells (EPCAM), or 4C overnight (HAS3). Scoring was performed and cells expressing wild-type FGFR3 remained attached to stan- independently by three investigators (J.-A. Roulson, M.A. dard culture dishes compared with cells expressing mutant FGFR3 Knowles, and E. di Martino). Samples stained for EPCAM were (P ¼ 0.037; Fig. 1B and C). The effect was greater in cells expres- scored for percentage of positive cells, and for staining intensity sing the S249C- and Y375C-mutant proteins compared with across the whole tumor, quantified using the following arbitrary K652E-FGFR3, in accordance with our previous results that indi- units: 0 ¼ negative, 1 ¼ weak, 1.5 ¼ weak/moderate, 2 ¼ cated a stronger phenotypic effect for these two mutations (16). moderate, 2.5 ¼ moderate/strong, 3 ¼ strong. For each sample, Furthermore, cells expressing S249C-FGFR3 detached more easily a staining score was calculated as staining intensity % positive from collagen IV, collagen I, and fibronectin, compared with cells, and an average staining score calculated for each sample. control cells (P ¼ 0.05), whereas no difference was observed in Because of generally weak staining and heterogeneity across the detachment from laminin (Fig. 1D). Overall, these results suggest tissue, tumors stained for HAS3 were only scored for overall that FGFR3 mutation may disrupt the cell–substrate adhesive expression in relation to normal bladder tissue (positive ¼ properties of urothelial cells. expression above control, negative ¼ expression equal or below control). Mutant FGFR3 dysregulates genes involved in cell adhesion and ECM remodeling in normal urothelial cells Statistical analysis To identify molecular targets of mutant FGFR3 which could be Significant differences in gene expression between tumors modulating the observed changes in cell–cell and cell–substrate stratified according to stage, grade, and FGFR3 mutation status adhesion, the gene expression profiles of confluent TERT-NHUC– were assessed using the Mann–Whitney test, with the SPSS 12.0 expressing S249C FGFR3 and control cells containing the empty statistical analysis software (SPSS Inc.). P 0.05 was accepted as vector were compared using Affymetrix Human Genome U133 significant. Plus 2.0 arrays. In total, 750 probe sets corresponding to 576 specific genes were significantly up- or downregulated above a 2- Results fold threshold in mutant cells. Of these, 72 (12.5%) were linked to Mutant FGFR3 alters cell–cell adhesion of normal urothelial cell–cell or cell–matrix adhesion (Table 1), including genes cells encoding structural proteins of desmosomes (DSC2, DSC3, Our previous results suggested that telomerase-immortalized DSG1, PKP1, and PKP3), adherens junctions (CDH1, CDH16, normal human urothelial cells (TERT-NHUC) expressing ectopic and EPCAM), and focal adhesions (PXN, ZYX). Mutant FGFR3- mutant FGFR3 may lack cell–cell contact inhibition. Thus, we expressing cells also exhibited changes in their integrin expression investigated the cell–cell and cell–substrate adhesive properties of profiles (ITGAV, ITGA2, ITGB5, and ITGB6) and in genes mod- FGFR3-mutant TERT-NHUC in more detail. We stained normal ulating cell–substrate adhesion and ECM remodeling (TNC, and mutant FGFR3-expressing cells for E-cadherin and b-catenin. CD36, PLAU, PLAT, MMP10). On the basis of magnitude of This highlighted some morphologic changes in cells expressing change in gene expression level, biologic function, evidence of S249C-mutant FGFR3. Although control cells and cells expressing involvement in bladder or other cancers, and availability of wild-type FGFR3 formed a tight monolayer, cells expressing validated antibodies for immunohistochemistry, 11 of these S249C-FGFR3 displayed looser cell–cell junctions and increased genes were chosen for validation and further investigation: DSG1, intercellular spaces (Fig. 1A). It has been shown that FGFR1 can DSC2, EPCAM, CDH1, CDH16, CD36, MMP10, HAS3, PLAU, bind to E-cadherin, and this interaction is followed by their PLAT, and TNC. co-endocytosis and E-cadherin delocalization to the cytoplasm Downregulation of DSG1, DSC2, CDH16, and CD36 and (24). However, no differences in the localization or level of E- upregulation of EPCAM, PLAU, PLAT, MMP10, and HAS3 were cadherin or b-catenin were observed between control and S249C- confirmed by Taqman RealTime RT-PCR in TERT-NHUC expres- FGFR3–expressing TERT-NHUC. Western blotting of fractioned sing mutant FGFR3 compared with control cells (Fig. 2), whereas membrane and cytoplasmic protein extract confirmed that in no significant downregulation of CDH1 was detected (data not both control and S249C-FGFR3 cells E-cadherin was localized shown). Expression of TNC was higher in cells expressing mutant mainly in the membrane compartment (data not shown). Fur- FGFR3 although some variability was observed between experi- thermore, we could not demonstrate any interaction between E- mental replicates (Fig. 2). Changes in expression were generally cadherin and mutant FGFR3 in TERT-NHUC by protein coimmu- stronger in cells expressing S249C-FGFR3 than in cells expressing noprecipitation and Western Blotting (data not shown). Taken K652E-FGFR3, consistent with the previously observed mutation- together, these results suggest that some other molecular mechan- dependent intensity of the phenotypic effects of mutant FGFR3. isms may mediate the effects of mutant FGFR3 on cell–cell contact. Mutant FGFR3 regulates EPCAM, PLAU, TNC, MMP10, HAS3, and CDH16 expression in the bladder cancer line, 97-7 Mutant FGFR3 alters cell–substrate adhesion of normal We then investigated whether converse changes in the expres- urothelial cells sion of these genes could be induced by shRNA silencing of FGFR3 The ability of cells expressing mutant FGFR3 to detach from the in the bladder cancer cell line 97-7, which contains an endoge- substrate was also tested. Confluent cells were subjected to con- nous S249C mutation. Indeed, silencing of mutant FGFR3 was

140 Mol Cancer Res; 13(1) January 2015 Molecular Cancer Research

Mutant FGFR3 Alters Adhesive Properties of Urothelial Cells

Figure 1. Cell–cell and cell–matrix adhesive properties of TERT-NHUC expressing wild-type (WT) or mutant FGFR3 (S249C, Y375C, K652E) compared with control cells (transduced with the empty vector). A, confluent control cells and cells expressing S249C-mutant FGFR3 were stained with anti E-cadherin (green) and anti b-catenin (red) antibodies, and DAPI (blue). Control cells show tight monolayers, whereas mutant cells display large intercellular spaces, indicated by the arrows. B and C, control, WT, and mutant FGFR3-expressing cells grown to confluence on standard tissue culture plates were subjected to controlled trypsinization. Detached cells were removed and remaining attached cells were either stained (B) or counted (C). A significantly higher proportion of mutant cells detached from the plate compared with WT or control cells (P ¼ 0.037). D, control, WT, and S249C FGFR3-expressing cells were grown to confluence on plates coated with various ECM proteins, subjected to controlled trypsinization, and the number of attached and detached cells quantified. A significantly higher number of mutant FGFR3-expressing cells detached from collagen IV, collagen I, and fibronectin compared with WT and control cells (P < 0.05). , significant (P 0.05) differences from control.

accompanied by downregulation of EPCAM, PLAU, TNC, normal urothelium and it was strongly upregulated in 30 of 44 MMP10, HAS3 and upregulation of CDH16, and these changes (68%) tumors (Fig. 4C). TNC expression was not detectable in were reversed upon reexpression of S249C-FGFR3 (Fig. 3). How- normal urothelium and therefore a tumor/normal ratio could not ever, DSG1, DSC2, CD36, and PLAT did not show the expected be calculated. However, expression of TNC was detected in 18 of changes in expression after modulation of FGFR3 in these cells 40 (45%) tumors tested, showing that TNC is often upregulated in (data not shown). Overall, these results confirm that mutant bladder tumors. No HAS3, MMP10,orTNC expression was FGFR3 modulates the expression of EPCAM, PLAU, TNC, MMP10, detected in 1, 14, and 22 tumors, respectively. However, this HAS3, and CDH16 in bladder cancer cells. could have been the result of low RNA concentration; therefore, expression in these patients was considered inconclusive. CDH16 EPCAM, MMP10, HAS3, TNC, and CDH16 mRNA expression in mRNA was not detected in uncultured normal urothelial cells, bladder tumors in relation to FGFR3 mutation normal bladder tissue, or tumors with the assay used. To assess whether mutant FGFR3 regulates EPCAM, HAS3, No statistical difference in expression of HAS3, TNC,orMMP10 MMP10, TNC, and CDH16 in vivo, we examined the mRNA was observed when samples were stratified for tumor grade, stage, expression of these genes in a panel of 44 bladder tumors with or FGFR3 mutation status (P > 0.05). No difference in expression known FGFR3 mutation status and in normal uncultured urothe- of EPCAM was observed in relation to tumor grade and FGFR3 lium (Fig. 4). Of 44 tumors, 23 (52%) showed upregulation and 6 mutation, but a significant difference was observed in relation to (14%) downregulation of EPCAM (>2-fold; Fig. 4A). HAS3 was tumor stage, with nonmuscle invasive (Ta-T1) tumors showing overexpressed in 15 of 44 tumors (34%) and downregulated in 7 an average expression significantly lower than muscle invasive of 44 (16%; Fig. 4B). MMP10 was expressed at very low levels in tumors (T2 or higher; P ¼ 0.033), consistent with previous reports

www.aacrjournals.org Mol Cancer Res; 13(1) January 2015 141

di Martino et al.

Table 1. Adhesion-related genes showing differential expression (>2-fold) in NHUC expressing mutant FGFR3 Affymetrix ID Gene symbol Fold change Gene product 200974_at ACTA2 þ2.14 Actin, alpha 2 1569362_at ALCAM 2.74 Activated leukocyte cell adhesion molecule 228520_s_at APLP2 þ2.71 Amyloid beta precursor-like protein 2 213606_s_at ARHGDIA 5.39 Rho GDP dissociation inhibitor alpha 201167_x_at ARHGDIA 3.86 Rho GDP dissociation inhibitor alpha 1555812_a_at ARHGDIB þ2.02 Rho GDP dissociation inhibitor beta 201288_at ARHGDIB þ2.02 Rho GDP dissociation inhibitor beta 216627_s_at B4GALT1 4.44 Beta-1,4-galactosyltransferase, polypeptide 1 238987_at B4GALT1 2.65 Beta-1,4-galactosyltransferase, polypeptide 1 211631_x_at B4GALT1 2.42 Beta-1,4-galactosyltransferase, polypeptide 1 235333_at B4GALT6 þ2.71 Beta-1,4-galactosyltransferase, polypeptide 6 202265_at BMI1 þ2.42 BMI1 polycomb ring ﬁnger oncogene 200935_at CALR 5.25 Calreticulin 218309_at CAMK2N1 þ2.35 Calcium/calmodulin-dependent protein kinase II inhibitor 1 229163_at CAMK2N1 þ3.12 Calcium/calmodulin-dependent protein kinase II inhibitor 1 208712_at CCND1 þ2.48 Cyclin D1 209555_s_at CD36 5.83 CD36 molecule 228766_at CD36 5.94 CD36 molecule 206488_s_at CD36 5.01 CD36 molecule 208727_s_at CDC42 2.47 Cell division cycle 42 201130_s_at CDH1 2.02 Cadherin 1, type 1 206517_at CDH16 2.60 Cadherin 16 235287_at CDK6 2.34 Cyclin-dependent kinase 6 1555756_a_at CLEC7A 4.31 C-type lectin domain family 7, member A 208791_at CLU þ2.09 Clusterin 215145_s_at CNTNAP2 2.39 Contactin associated protein-like 2 219300_s_at CNTNAP2 2.19 Contactin associated protein-like 2 201681_s_at DLG5 þ2.30 Discs, large homolog 5 204750_s_at DSC2 4.89 Desmocollin 2 204751_x_at DSC2 2.46 Desmocollin 2 206033_s_at DSC3 2.06 Desmocollin 3 206642_at DSG1 3.49 Desmoglein 1 218995_s_at EDN1 2.10 Endothelin 1 200879_s_at EPAS1 2.46 Endothelial PAS domain protein 1 228948_at EPHA4 2.86 EPH receptor A4 227449_at EPHA4 3.23 EPH receptor A4 216252_x_at FAS þ2.48 Fas (TNF receptor superfamily, member 6) 215719_x_at FAS þ2.70 Fas (TNF receptor superfamily, member 6) 204780_s_at FAS þ2.71 Fas (TNF receptor superfamily, member 6) 204781_s_at FAS þ3.14 Fas (TNF receptor superfamily, member 6) 203562_at FEZ1 þ2.56 Fasciculation and elongation protein zeta 1 219250_s_at FLRT3 2.60 Fibronectin leucine-rich transmembrane protein 3 206582_s_at GPR56 2.21 G protein–coupled receptor 56 229397_s_at GRLF1 þ2.21 Glucocorticoid receptor DNA binding factor 1 223541_at HAS3 þ4.31 Hyaluronan synthase 3 201392_s_at IGF2R 2.18 Insulin-like growth factor 2 receptor 202859_x_at IL8 þ5.67 Interleukin 8 211506_s_at IL8 þ2.48 Interleukin 8 204686_at IRS1 þ2.50 Insulin receptor substrate 1 227314_at ITGA2 þ3.11 Integrin, alpha 2 241769_at ITGAV 2.47 Integrin, alpha V 201125_s_at ITGB5 þ2.18 Integrin, beta 5 201124_at ITGB5 þ2.19 Integrin, beta 5 208084_at ITGB6 þ2.17 Integrin, beta 6 208083_s_at ITGB6 þ2.52 Integrin, beta 6 226535_at ITGB6 þ2.82 Integrin, beta 6 231183_s_at JAG1 þ2.02 Jagged 1 222468_at KIAA0319L 2.16 KIAA0319-like 226534_at KITLG þ2.57 KIT ligand 205680_at MMP10 þ 5.15 Matrix metallopeptidase 10 (stromelysin 2) 210360_s_at MTSS1 2.30 Metastasis suppressor 1 202149_at NEDD9 þ3.45 Neural precursor cell expressed, developmentally downregulated 9 217999_s_at PHLDA1 þ2.85 Pleckstrin homology-like domain, family A, member 1 217996_at PHLDA1 þ2.69 Pleckstrin homology-like domain, family A, member 1 217997_at PHLDA1 þ3.70 Pleckstrin homology-like domain, family A, member 1 225842_at PHLDA1 þ5.15 Pleckstrin homology-like domain, family A, member 1 212629_s_at PKN2 2.31 Protein kinase N2 (Continued on the following page)

142 Mol Cancer Res; 13(1) January 2015 Molecular Cancer Research

Mutant FGFR3 Alters Adhesive Properties of Urothelial Cells

Table 1. Adhesion-related genes showing differential expression (>2-fold) in NHUC expressing mutant FGFR3 (Cont'd ) Affymetrix ID Gene symbol Fold change Gene product 221854_at PKP1 2.14 Plakophilin 1 209872_s_at PKP3 2.03 Plakophilin 3 201860_s_at PLAT þ3.03 Plasminogen activator, tissue 211668_s_at PLAU þ2.24 Plasminogen activator, urokinase 205479_s_at PLAU þ3.06 Plasminogen activator, urokinase 210845_s_at PLAUR þ2.09 Plasminogen activator, urokinase receptor 1567213_at PNN 2.76 Pinin, desmosome associated protein 204469_at PTPRZ1 þ2.37 Protein tyrosine phosphatase, receptor-type, Z polypeptide 1 211823_s_at PXN 3.58 Paxillin 211823_s_at PXN 3.58 Paxillin 213603_s_at RAC2 þ2.05 Ras-related C3 botulinum toxin substrate 2 1553962_s_at RHOB 2.59 Ras homolog gene family, member B 212099_at RHOB 2.28 Ras homolog gene family, member B 201647_s_at SCARB2 2.06 Scavenger receptor class B, member 2 202936_s_at SOX9 þ2.82 SRY-box 9 202935_s_at SOX9 þ2.05 SRY-box 9 204963_at SSPN þ2.28 Sarcospan 211085_s_at STK4 2.27 Serine/threonine kinase 4 201839_s_at EPCAM þ2.59 Tumor-associated calcium signal transducer 1 201109_s_at THBS1 2.41 Thrombospondin 1 201108_s_at THBS1 2.05 Thrombospondin 1 201110_s_at THBS1 2.41 Thrombospondin 1 203083_at THBS2 þ3.34 Thrombospondin 2 201149_s_at TIMP3 2.41 TIMP metallopeptidase inhibitor 3 201645_at TNC þ2.46 Tenascin C 209114_at TSPAN1 þ2.81 Tetraspanin 1 216609_at TXN þ2.90 Thioredoxin 226029_at VANGL2 2.08 Vang-like 2 202205_at VASP 2.13 Vasodilator-stimulated phosphoprotein 215646_s_at VCAN þ3.16 Versican 204620_s_at VCAN þ3.94 Versican 221731_x_at VCAN þ4.70 Versican 200808_s_at ZYX 2.40 Zyxin 215706_x_at ZYX 2.20 Zyxin NOTE: Fold change is the ratio between expression level in conﬂuent NHUC expressing S249C FGFR3 and control cells; highlighted genes were further investigated by RealTime RT-PCR.

(25). Expression of TNC could not be examined in relation to of around 50%. The mean average score was higher in wild-type FGFR3 mutation, tumor stage, and tumor grade due to the small compared with mutant tumors, but the result did not reach number of positive tumors (n ¼ 18). statistical signiﬁcance (P ¼ 0.106). Interestingly, a small number of samples known to be HRAS or KRAS mutant showed a trend EPCAM and HAS3 protein expression in low-stage bladder toward highest EPCAM staining, although differences were not tumors in relation to FGFR3 mutation status signiﬁcant (P ¼ 0.156). We hypothesized that the lack of correlation between In normal bladder, we detected a spotty membranous HAS3 FGFR3 mutation status and the mRNA level of its target genes staining (Supplementary Fig. S1D), consistent with previous could be due to the clinical heterogeneity of the group of results in epithelial cells (26). The signal was mainly localized tumors analyzed, the presence of contaminating tissue, and/or to the apical cells, and was of moderate intensity. Only 1% to the presence of different subpopulations of cells within the 10% of other cells showed weak to moderate staining in normal same tumor. Thus, we used immunohistochemistry to gain a bladder. Staining intensity in tumors varied from weak to more detailed picture of protein expression across different strong, and it was very heterogeneous across individual tumors. cell types and different areas in 48 tumors of similar grade Only 7 of 48 tumors (14%) showed increased expression com- and stage (pTaG2; n ¼ 34 FGFR3 mutant, and n ¼ 14 FGFR3 pared with normal bladder, and due to the small numbers, cor- wild type) to assess whether a correlation could be found relation between HAS3 upregulation and FGFR3 mutation between mutation status and HAS3, MMP10, and EPCAM at could not be assessed. However, in contrast with normal the protein level. bladder, a high proportion of tumors tended to show expres- EPCAM staining in normal bladder was membranous, of weak sion throughout, and not limited to apical cells (Supplemen- to moderate intensity, and mainly localized in lower layers of the tary Fig. S1E and S1F). epithelium (Supplementary Fig. S1A). Surface cells were mostly MMP10 immunohistochemistry was performed as described negative, and only occasionally was surface positivity observed. previously (27). However, staining appeared artifactual in our Most tumors showed weak to strong membrane staining, stron- hands. Attempts to optimize staining by altering antigen retrieval gest within the lower third of the epithelium but focally extending and detection methods, primary antibody concentration, and to the surface (Supplementary Fig. S1B). Some tumors, however, temperature and length of primary antibody incubation failed showed staining throughout (Supplementary Fig. S1C). The to produce satisfactory staining. Therefore, MMP10 immunohis- number of positive cells varied from 10% to 90% with a mean tochemistry had to be abandoned.

www.aacrjournals.org Mol Cancer Res; 13(1) January 2015 143

di Martino et al.

Figure 2. Relative mRNA expression of DSG1, DSC2, CDH16, CD36, EPCAM, PLAU, PLAT, MMP10, HAS3,andTNC mRNA in TERT-NHUC overexpressing wild-type (WT), and mutant FGFR3 (S249C, Y375C, and K652E) compared with control cells (transduced with the empty vector).

Discussion changes suggestive of decreased cell–cell adhesion. Disruption of cell–cell adhesion in bladder and other cancers is often due to Although mutant FGFR3 plays an undisputed role in bladder alteration in the expression or localization of E-cadherin and carcinogenesis, little is known of the molecular mechanisms b-catenin (28), but no such changes were observed in our through which FGFR3 signaling contributes to urothelial trans- model. However, we detected dysregulation of genes encoding formation. Our previous results suggested a possible role in other proteins involved in the maintenance of adherens junc- disrupting contact-mediated inhibition of urothelial cell prolif- tions, such as EPCAM and CDH16. CDH16 encodes a kidney- eration, thereby favoring the development of hyperplastic lesions speciﬁc cadherin downregulated in renal neoplasms (29). Its in the bladder (16). expression was detected at relatively high levels in normal uro- In this study, we used isogenic cell lines as simple in vitro thelial cells in vitro and was downregulated upon expression of models to unravel phenotypic outcomes of mutant FGFR3 expres- ectopic mutant FGFR3. However, this gene was not expressed in sion and identify downstream effectors and targets. Comparison uncultured urothelial cells or normal bladder tissue, suggesting of normal urothelial cells with and without ectopic expression of that although mutant FGFR3 can downregulate this gene under mutant FGFR3 highlighted morphologic and gene expression cell culture conditions, this event is unlikely to play a crucial role

144 Mol Cancer Res; 13(1) January 2015 Molecular Cancer Research

Mutant FGFR3 Alters Adhesive Properties of Urothelial Cells

Figure 3. Relative mRNA expression of FGFR3, CDH16, PLAU, MMP10, EPCAM, TNC, and HAS3 mRNA in 97-7 cells after FGFR3 silencing (-FGFR3) and after re-expression of S249C-FGFR3 (þFGFR3), compared with control cells (transduced with the empty vector) or cells expressing a scrambled nonspeciﬁc shRNA (NS).

during bladder cancer development. Our data also indicate dis- luronan (HA), a glycosaminoglycan component of the ECM ruption of expression of desmosomal junction proteins, such as that promotes tumor growth and spreading (34). HA and HAS1 those encoded by DSG1 and DSC2, as an additional mechanism are upregulated in bladder cancer tissue (35), and HAS2 of cell–cell adhesion dysregulation in FGFR3-mutant urothelial knockdown in a bladder cancer cell line was shown to affect cells. proliferation and invasion (36). Interestingly, HAS3 overex- Reduction of cell–cell adhesion following expression of pression in rat fibroblasts was shown to halt cell–cell contact mutant FGFR3 was also accompanied by decreased cell adhe- inhibition of proliferation and increase their saturation density sion to ECM ligands, such as collagen IV, collagen I, and (37), similar to the phenotype mediated by mutant FGFR3 in fibronectin, which are normally found in the urothelial base- urothelial cells (16). ment membrane and connective tissue of the bladder (30), and Taken together, the results in our in vitro model system by consistent changes in the expression of genes involved in suggest that mutation of FGFR3 in normal urothelial cells can cell–substrate adhesion and ECM remodeling, such as down- favor some of the key steps necessary for tumors to acquire regulation of CD36, and upregulation of TNC, MMP10, PLAU, malignant potential and spread. In nonmuscle invasive bladder and PLAT. Interaction between ECM components and the tumors where FGFR3 mutation is most common, this may integrins expressed on the surface of urothelial cells is essential allow intraepithelial spread and development of exophytic for preserving bladder wall integrity. Indeed, degradation of tumors. Accumulation of other molecular changes is likely to laminin and collagen IV is associated with the disruption of be required for these tumors to acquire a muscle-invasive and basement membrane continuity observed in transitional cell metastatic phenotype. carcinomas and is linked to poorer prognosis (31). Many of the genes highlighted in this study were previously An interesting observation was the upregulation of HAS3 in found to be upregulated in a number of malignancies including mutant FGFR3-expressing cells. This is consistent with previous bladder tumors (25, 27, 38–40), but their relation with mutated reports that FGFs may induce HAS3 expression in some cell FGFR3 had not been tested. Interestingly, however, higher levels types (32, 33). HAS3 encodes hyaluronan synthase 3, one of of MMP10 have been detected in superficial bladder tumors (27), three transmembrane enzymes that mediate synthesis of hya- which are the subgroup with most frequent FGFR3 mutation (2).

www.aacrjournals.org Mol Cancer Res; 13(1) January 2015 145

di Martino et al.

Figure 4. Expression of EPCAM (A), HAS3 (B), and MMP10 (C) mRNA in wild-type and FGFR3-mutant bladder tumors, relative to levels in uncultured normal urothelial cells.

We tested whether a correlation between mutant FGFR3 and the silencing of endogenous mutant FGFR3 in the tumor cell line expression of some of these genes could also be observed in tumor 97-7 caused converse changes in their expression. MMP10, TNC, tissues in vivo. To narrow down the number of genes to investigate HAS3, and EPCAM were conﬁrmed as FGFR3 targets in this in tumor tissues, we further validated their role as downstream second in vitro model. However, although their mRNA and/or transcriptional targets of FGFR3 signaling by testing whether protein levels were upregulated in a high proportion of

146 Mol Cancer Res; 13(1) January 2015 Molecular Cancer Research

Mutant FGFR3 Alters Adhesive Properties of Urothelial Cells

specimens, confirming an important role in bladder malignant clarify the role of mutant FGFR3 in the development of bladder transformation, we found no correlation with FGFR3 mutation cancer. status in tumor samples. Although we cannot exclude that changes in these genes in response to FGFR3 mutation only occur fl under cell culture conditions, this lack of correlation may be due Disclosure of Potential Con icts of Interest fl to the fact that several distinct carcinogenic pathways may lead to No potential con icts of interest were disclosed. the dysregulation of these genes in tumor tissue, and their expres- Authors' Contributions sion may be influenced by other genetic and epigenetic events in Conception and design: E. di Martino, M.A. Knowles bladder tumors. All of these may have acted as confounding Development of methodology: E. di Martino factors preventing us from detecting a direct correlation between Acquisition of data (provided animals, acquired and managed patients, the expression of FGFR3 target genes and FGFR3 mutation status provided facilities, etc.): E. di Martino in vivo. Furthermore, FGFR3 overexpression or FGFR3 gene Analysis and interpretation of data (e.g., statistical analysis, biostatistics, fusions are alternative mechanisms of FGFR3 dysregulation in computational analysis): E. di Martino, G. Kelly, J.-A. Roulson, M.A. Knowles bladder tumors (11, 41). Consequently, some of the tumors Writing, review, and/or revision of the manuscript: E. di Martino, fi J.-A. Roulson, M.A. Knowles classi ed as wild type in this study may have displayed high Study supervision: M.A. Knowles levels of FGFR3 target genes due to FGFR3 dysregulation through other mechanisms. These results suggest that our in vitro model Acknowledgments based on isogenic cell lines is a simpler and more controlled The authors thank the Cancer Research UK Manchester Institute Microarray environment allowing the unraveling of molecular changes asso- Service for assistance with array experiments, Fiona Platt for FGFR3 genotyping, ciated with a single oncogenic event such as FGFR3 mutation. and Filomena Esteves for technical support with the immunohistochemistry. Using this model, it will now be possible to interrogate the role of specific FGFR3-modulated genes in determining phenotype. Of Grant Support particular interest will be whether there is a regulatory hierarchy This work was supported by Cancer Research UK (grant C6228/A5433 to within the cell adhesion–related genes identified or whether all M.A. Knowles). The costs of publication of this article were defrayed in part by the payment of are co-ordinately regulated. page charges. This article must therefore be hereby marked advertisement in In conclusion, we have shown that mutant FGFR3 alters the accordance with 18 U.S.C. Section 1734 solely to indicate this fact. expression of cell–cell and cell–substrate adhesion genes in two distinct in vitro models. Further studies are required to confirm this Received January 14, 2014; revised July 29, 2014; accepted August 9, 2014; oncogenic mechanism in vivo. Overall, this investigation helps to published OnlineFirst September 15, 2014.

References 1. Beenken A, Mohammadi M. The FGF family: biology, pathophysiology and 12. Bernard-Pierrot I, Brams A, Dunois-Larde C, Caillault A, Diez de Medina therapy. Nat Rev Drug Discov 2009;8:235–53. SG, Cappellen D, et al. Oncogenic properties of the mutated forms 2. Billerey C, Chopin D, Aubriot-Lorton MH, Ricol D, Gil Diez de Medina S, of fibroblast growth factor receptor 3b. Carcinogenesis 2006;27: Van Rhijn B, et al. Frequent FGFR3 mutations in papillary non-invasive 740–7. bladder (pTa) tumors. Am J Pathol 2001;158:1955–9. 13. Tomlinson DC, Hurst CD, Knowles MA. Knockdown by shRNA identifies 3. di Martino E, Tomlinson DC, Knowles MA. A decade of FGF receptor S249C mutant FGFR3 as a potential therapeutic target in bladder cancer. research in bladder cancer: past, present, and future challenges. Adv Urol Oncogene 2007;26:5889–99. 2012;2012:429213. 14. Lamont FR, Tomlinson DC, Cooper PA, Shnyder SD, Chester JD, Knowles 4. Adar R, Monsonego-Ornan E, David P, Yayon A. Differential activation of MA. Small molecule FGF receptor inhibitors block FGFR-dependent cysteine-substitution mutants of fibroblast growth factor receptor 3 is urothelial carcinoma growth in vitro and in vivo. Br J Cancer 2011; determined by cysteine localization. J Bone Miner Res 2002;17:860–8. 104:75–82. 5. d'Avis PY, Robertson SC, Meyer AN, Bardwell WM, Webster MK, Donoghue 15. Knowles MA. Novel therapeutic targets in bladder cancer: mutation and DJ. Constitutive activation of fibroblast growth factor receptor 3 by muta- expression of FGF receptors. Future Oncol 2008;4:71–83. tions responsible for the lethal skeletal dysplasia thanatophoric dysplasia 16. di Martino E, L'Hote CG, Kennedy W, Tomlinson DC, Knowles MA. Mutant type I. Cell Growth Differ 1998;9:71–8. fibroblast growth factor receptor 3 induces intracellular signaling and 6. Webster MK, D'Avis PY, Robertson SC, Donoghue DJ. Profound ligand- cellular transformation in a cell type- and mutation-specific manner. independent kinase activation of fibroblast growth factor receptor 3 by the Oncogene 2009;28:4306–16. activation loop mutation responsible for a lethal skeletal dysplasia, tha- 17. di Martino E, Taylor CF, Roulson JA, Knowles MA. An integrated genomic, natophoric dysplasia type II. Mol Cell Biol 1996;16:4081–7. transcriptional and protein investigation of FGFRL1 as a putative 4p16.3 7. Hafner C, Vogt T, Hartmann A. FGFR3 mutations in benign skin tumors. deletion target in bladder cancer. Genes Chromosomes Cancer 2013;52: Cell Cycle 2006;5:2723–8. 860–71. 8. Chesi M, Brents LA, Ely SA, Bais C, Robbiani DF, Mesri EA, et al. Activated 18. van Oers JM, Lurkin I, van Exsel AJ, Nijsen Y, van Rhijn BW, van der Aa MN, fibroblast growth factor receptor 3 is an oncogene that contributes to tumor et al. A simple and fast method for the simultaneous detection of nine progression in multiple myeloma. Blood 2001;97:729–36. fibroblast growth factor receptor 3 mutations in bladder cancer and voided 9. Cappellen D, De Oliveira C, Ricol D, de Medina S, Bourdin J, Sastre-Garau urine. Clin Cancer Res 2005;11:7743–8. X, et al. Frequent activating mutations of FGFR3 in human bladder and 19. Platt FM, Hurst CD, Taylor CF, Gregory WM, Harnden P, Knowles MA. cervix carcinomas. Nat Genet 1999;23:18–20. Spectrum of phosphatidylinositol 3-kinase pathway gene alterations in 10. van Oers JM, Adam C, Denzinger S, Stoehr R, Bertz S, Zaak D, et al. bladder cancer. Clin Cancer Res 2009;15:6008–17. Chromosome 9 deletions are more frequent than FGFR3 mutations in flat 20. Gentleman RC, Carey VJ, Bates DM, Bolstad B, Dettling M, Dudoit S, et al. urothelial hyperplasias of the bladder. Int J Cancer 2006;119:1212–5. Bioconductor: open software development for computational biology and 11. Tomlinson DC, Baldo O, Harnden P, Knowles MA. FGFR3 protein expres- bioinformatics. Genome Biol 2004;5:R80. sion and its relationship to mutation status and prognostic variables in 21. Irizarry RA, Bolstad BM, Collin F, Cope LM, Hobbs B, Speed TP. Summaries bladder cancer. J Pathol 2007;213:91–8. of Affymetrix GeneChip probe level data. Nucleic Acids Res 2003;31:e15.

www.aacrjournals.org Mol Cancer Res; 13(1) January 2015 147

di Martino et al.

22. Smyth GK. Linear models and empirical bayes methods for assessing hyaluronan synthesis in the epidermis by activation of keratinocyte hya- differential expression in microarray experiments. Stat Appl Genet Mol luronan synthases 2 and 3. J Biol Chem 2003;278:49495–504. Biol 2004;3:Article3. 33. Ohno T, Yoo MJ, Swanson ER, Hirano S, Ossoff RH, Rousseau B. Regen- 23. Dennis G Jr., Sherman BT, Hosack DA, Yang J, Gao W, Lane HC, et al. erative effects of basic fibroblast growth factor on extracellular matrix DAVID: database for annotation, visualization, and integrated discovery. production in aged rat vocal folds. Ann Otol Rhinol Laryngol 2009; Genome Biol 2003;4:P3. 118:559–64. 24. Bryant DM, Wylie FG, Stow JL. Regulation of endocytosis, nuclear trans- 34. Sironen RK, Tammi M, Tammi R, Auvinen PK, Anttila M, Kosma VM. location, and signaling of fibroblast growth factor receptor 1 by E-cadherin. Hyaluronan in human malignancies. Exp Cell Res 2011;317:383–91. Mol Biol Cell 2005;16:14–23. 35. Golshani R, Hautmann SH, Estrella V, Cohen BL, Kyle CC, Manoharan M, 25. Brunner A, Prelog M, Verdorfer I, Tzankov A, Mikuz G, Ensinger C. EpCAM et al. HAS1 expression in bladder cancer and its relation to urinary HA test. is predominantly expressed in high grade and advanced stage urothelial Int J Cancer 2007;120:1712–20. carcinoma of the bladder. J Clin Pathol 2008;61:307–10. 36. Okegawa T, Ushio K, Imai M, Morimoto M, Hara T. Orphan nuclear 26. Barnes L, Carraux P, Saurat JH, Kaya G. Increased expression of CD44 and receptor HNF4G promotes bladder cancer growth and invasion through hyaluronate synthase 3 is associated with accumulation of hyaluronate in the regulation of the hyaluronan synthase 2 gene. Oncogenesis 2013;2: spongiotic epidermis. J Invest Dermatol 2012;132:736–8. e58. 27. Seargent JM, Loadman PM, Martin SW, Naylor B, Bibby MC, Gill JH. 37. Itano N, Atsumi F, Sawai T, Yamada Y, Miyaishi O, Senga T, et al. Abnormal Expression of matrix metalloproteinase-10 in human bladder transitional accumulation of hyaluronan matrix diminishes contact inhibition of cell cell carcinoma. Urology 2005;65:815–20. growth and promotes cell migration. Proc Natl Acad Sci U S A 2002; 28. Dusek RL, Attardi LD. Desmosomes: new perpetrators in tumour suppres- 99:3609–14. sion. Nat Rev Cancer 2011;11:317–23. 38. Kramer MW, Escudero DO, Lokeshwar SD, Golshani R, Ekwenna OO, 29. Thedieck C, Kuczyk M, Klingel K, Steiert I, Muller CA, Klein G. Expression of Acosta K, et al. Association of hyaluronic acid family members (HAS1, Ksp-cadherin during kidney development and in renal cell carcinoma. Br J HAS2, and HYAL-1) with bladder cancer diagnosis and prognosis. Cancer Cancer 2005;92:2010–7. 2011;117:1197–209. 30. Wilson CB, Leopard J, Cheresh DA, Nakamura RM. Extracellular matrix and 39. El-Kott AF, Khalil AM, El-Kenawy Ael M. Immunohistochemical expres- integrin composition of the normal bladder wall. World J Urol 1996;14 sions of uPA and its receptor uPAR and their prognostic significant in Suppl 1:S30–7. urinary bladder carcinoma. Int Urol Nephrol 2004;36:417–23. 31. Gontero P, Banisadr S, Frea B, Brausi M. Metastasis markers in bladder 40. Booth C, Harnden P, Selby PJ, Southgate J. Towards defining roles and cancer: a review of the literature and clinical considerations. Eur Urol relationships for tenascin-C and TGFbeta-1 in the normal and neoplastic 2004;46:296–311. urinary bladder. J Pathol 2002;198:359–68. 32. Karvinen S, Pasonen-Seppanen S, Hyttinen JM, Pienimaki JP, Torronen K, 41. Williams SV, Hurst CD, Knowles MA. Oncogenic FGFR3 gene fusions in Jokela TA, et al. Keratinocyte growth factor stimulates migration and bladder cancer. Hum Mol Genet 2013;22:795–803.

148 Mol Cancer Res; 13(1) January 2015 Molecular Cancer Research

Alteration of Cell−Cell and Cell−Matrix Adhesion in Urothelial Cells: An Oncogenic Mechanism for Mutant FGFR3

Erica di Martino, Gavin Kelly, Jo-An Roulson, et al.

Mol Cancer Res 2015;13:138-148. Published OnlineFirst September 15, 2014.

Updated version Access the most recent version of this article at: doi:10.1158/1541-7786.MCR-14-0022

Supplementary Access the most recent supplemental material at: Material http://mcr.aacrjournals.org/content/suppl/2014/09/16/1541-7786.MCR-14-0022.DC1

Cited articles This article cites 40 articles, 9 of which you can access for free at: http://mcr.aacrjournals.org/content/13/1/138.full#ref-list-1

Citing articles This article has been cited by 2 HighWire-hosted articles. Access the articles at: http://mcr.aacrjournals.org/content/13/1/138.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at Subscriptions [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://mcr.aacrjournals.org/content/13/1/138. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.