Cancer Therapy: Preclinical

TargetingtheExtracellularDomainofFibroblastGrowthFactor Receptor 3 with Human Single-Chain Fv Antibodies Inhibits Bladder Carcinoma Cell Line Proliferation 1 1 1 2 Jorge Martlı´nez-Torrecuadrada,¤ Gabriela Cifuentes, Paula Lo´o¤ pez-Serra, Pilar Saenz, 2 1 Antonio Martlnez,¤ı´ and J. Ignacio Casal

Abstract Purpose: Previous expression studies have shown that fibroblast receptor 3 (FGFR3) is overexpressed in early stages of bladder cancer. To study the potential use of thera- peutic antibodies against FGFR3, we have produced a collection of human single-chain Fv (scFv) antibody fragments by using phage display libraries. Experimental Design:Two ‘‘naı¨l« ve’’semi-synthetic human scFv libraries were used to select anti- bodies against the extracellular domain of FGFR3a(IIIc).The reactivity of the selected scFvs with a recombinant FGFR3 was characterized by an enzyme immunoassay and surface plasmon reso- nance analysis and with RT112bladder carcinoma cells by a fluorescence-activated cell sorter. The capacity of the selected scFvs to block RT112cell proliferation was determined. Results: We have isolated six human scFv antibody fragments directed against FGFR3. These human scFvs specifically bound FGFR3, but not the homologous molecule FGFR1.Biacore anal- ysis was used to determine the affinity constants, which ranged from 12to 40 nmol/L. Competi- tion analysis showed that the FGF9ligand was able to block the binding of two scFvs, 3C and 7D, to FGFR3, whereas FGF1only blocked 7D. Immunoprecipitation and flow cytometric analysis confirmed the specificity of the antibodies to native membrane FGFR3. Two scFvs, 3C and 7D, gave an strong immunofluorescence staining of RT112cells. Moreover, they recognized equally well wild-type and mutant FGFR3 containing the activating mutation S249C. Furthermore, they blocked proliferation of RT112cells in a dose- and FGF-dependent manner. Conclusion: Our results suggest that these human anti-FGFR3 scFv antibodies may have poten- tial applications as antitumoral agents in bladder cancer.

The fibroblast growth factors (FGF) represent one of the largest homology at the level. The FGFR structure consists of an families of polypeptide growth and differentiation factors for extracellular ligand-binding domain, a transmembrane domain, mesodermal and epithelial cells (1). FGFs have been implicated in and an intracellular kinase domain. The ligand-binding domain multiple biological processes during embryo development, wound contains three different immunoglobulin-like domains (called healing, hematopoiesis, and angiogenesis. In addition, FGFs have immunoglobulin I, immunoglobulin II, and immunoglobulin been shown to increase the invasiveness of a variety of tumors III). The ligand domain is followed by a single transmembrane from prostate, bladder, kidney, breast, and pancreas (2–5). domain and a cytoplasmic split domain in the Although more than 20 FGFs with different effects on various intracellular domain. Alternative splicing of the mRNAs cells have been reported, only five FGF receptors (FGFR) have corresponding to FGFR1-3 gives place to two isoforms, a and been described (6, 7). These receptors share 55% to 72% h. Only isoform a, which contains the three immunoglobulin domains, has been found for FGFR3. Alternative splicing of FGFR3 transcripts involving the COOH-terminal domain of the immunoglobulin III domain yields two isoforms, IIIb and IIIc. 1 Authors’ Affiliations: ProteinTechnology Unit, Biotechnology Program, Centro The two variants show different binding affinities: the IIIc variant Nacional de Investigaciones Oncolo´o¤ gicas, Madrid, Spain and 2Progenika, Zamudio, Bilbao, Spain is more promiscuous and binds a wide range of FGFs, including Received 2/7/05; revised 6/6/05; accepted 6/10/05. FGF1, FGF2, FGF4, and FGF9; on the other hand, the IIIb variant Grant support: Spanish Ministry of Science and Technology grants BIO02003- preferentially binds FGF1 and, at a lower extent, FGF8 and FGF9 01481, FIT-010000-2003-86, and FIT-010000-2004-65. The Centro Nacional (8). After binding to FGFRs in the presence of heparin, FGFs de Investigaciones Oncolo´o¤ gicas is partially supported by the Red Temo´a¤ tica de induce receptor dimerization, resulting in autophosphorylation Investigacio´o¤ n Cooperativa de Centros de Co´a¤ ncer (FIS C03/10). The costs of publication of this article were defrayed in part by the payment of page of the intracellular kinase domain and a downstream activation charges. This article must therefore be hereby marked advertisement in accordance of intracellular signaling cascades (9). Following activation of with 18 U.S.C. Section 1734 solely to indicate this fact. the ligand receptor complex, various signal transduction path- Requests for reprints: Ignacio Casal, Biotechnology Program, Centro Nacional ways are initiated by FGFs: elevation of intracellular calcium de Investigaciones Oncoloo´¤ gicas, Melchor Ferno´a¤ ndez Almagro 3, 28029 Madrid, Spain. Phone: 34-91-224-69-20; Fax: 34-91-224-69-72; E-mail: i [email protected]. levels, induction of mitogen-activated protein kinase and protein F 2005 American Association for Cancer Research. kinase C pathways, stimulation of adenylate cyclase, and doi:10.1158/1078-0432.CCR-05-0282 induction of the proto-oncogenes c-myc and c-fos (6).

Clin Cancer Res 2005;11(17) September 1, 2005 6280 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 23, 2021. © 2005 American Association for Cancer Research. FGFR3 Blocking scFvs

Specific mutations have been found in FGFR3 that lead to the isoforms IIIc of the extracellular domains of FGFR3a or FGFR1a, activation of its tyrosine kinase activities and to different respectively, fused to the Fc region of human IgG1. FGF9, FGF1, and syndromes related with bone development, multiple myeloma (EGF) were provided by ReliaTech. (10, 11), cervical carcinomas (12, 13), and bladder carcinomas (12, Mouse anti-FGFR3 monoclonal antibody (mAb) was obtained from Santa Cruz Biotechnology (Santa Cruz, CA). Goat anti-human IgG (Fc (12, 14–17). There are recent observations that FGFR signaling specific), mouse anti–c-myc mAb (clone 9E10), and antitubulin were is absolutely necessary for the survival of prostate cancer cells from Sigma. Horseradish peroxidase–conjugated anti–c-myc (clone in vitro (18). Recently, FGFR3 has been proposed as a possible 9E10) was purchased from Roche (Mannheim, Germany). Anti-6xHis therapeutic target for multiple myeloma (10, 19). Although the was provided by NeoMarkers (Westinghouse Dr, Fremont, CA). Anti- presence of activating mutations is well documented, little was M13 and horseradish peroxidase–conjugated anti-M13 mAbs were known about the expression of FGFR3 in tumoral tissues. obtained from GE Healthcare. FITC-conjugated rabbit anti-mouse IgG Recently, after analysis by using gene chip was from DAKO (Glostrup, Denmark). R-phycoerythrin goat anti- technology, it was found that FGFR3 was overexpressed in mouse IgG was purchased from Molecular Probes (Eugene, OR). Mouse IgG TrueBlot was provided by eBioscience (San Diego, CA). tumoral samples corresponding to stages pTa,pT1 (fold change FGFR3 cDNA cloning and cell transfection. For cellular expression of >8), and pT2 (fold change >2) of bladder carcinoma when compared with the normal counterparts (20). The gene recombinant FGFR3s, the extracellular domain of the protein was fused to the COOH-terminal Fc region of human IgG1. Vector pcDNA3.1-Fc, expression levels were confirmed at the protein level by which contains the human Fc region, was used to clone the cDNA Western blot and immunohistochemical analysis. In fact, this fragment encoding the extracellular domain of FGFR3-IIIc (amino acids protein overproduction seems to occur more frequently in 1-370) to generate pcDNA3.1-FGFR3(IIIc)WT-Fc. This plasmid was used transitional carcinomas than activating mutations (20). All as template to prepare the mutant FGFR3 S249C by site-directed muta- these data together suggest that FGFR3 might constitute an genesis using the primer 5V-CCCGCCTGCAGGATGGGCCGGTGCGGGC- attractive target for therapy in urologic tumors. Bladder cancer AGCGC-3V. The resulting plasmid, pcDNA3.1-FGFR3(IIIc)S249C-Fc, is the second most common malignancy of the genitourinary was sequenced to confirm the presence of the S249C mutation. tract. About 40% to 50% of the bladder carcinoma showed pcDNA3.1-FGFR3(IIIc)WT-Fc and pcDNA3.1-FGFR3(IIIc)S249C-Fc FGFR3-activating gene mutations (17, 21); the incidence was were transfected into HEK293T cells with FuGENE 6 (Roche) and the significantly bigger (80%) in superficial tumors than in invasive expression and reactivity of the wild-type and mutant were analyzed 2 days after transfection by immunoblotting with anti-FGFR3 tumors. mAb and by FACS with the specific scFvs as described below. With the increasing interest in FGFR3 as a therapeutic target Selection of fibroblast 3–specific single-chain Fv in different neoplasias and the recent findings about its antibodies from phagemid libraries. The human scFv libraries Tomlin- overexpression in transitional cell carcinoma, we have started son I + J, helper phage KM13, and Escherichia coli TG1 and HB2151 were the development of human antibodies for therapeutic purposes obtained from MRC Geneservices. Both libraries were grown separately by phage display. Phage antibody display is one of the best and the phages mixed 1:1 for selection. Phage selection and scFv alternatives for the production of human antibodies for production were done on microtiter plates (Maxisorb, Nunc, Rockilde, research, clinical, and therapeutic applications (for reviews on Denmark) according to the scheme shown in Fig. 1. In the first round of this topic, see refs. 22–24). However, FGFR3 is a very elusive selection, wells of the microplates were coated with 1 Ag of FGFR3 or j molecule for antibody production given the high homology human IgG in PBS at 4 C overnight. Next, wells were washed with PBS and blocked with 2% MPBS (skimmed milk in PBS) for 2 hours. Then, between FGFR3 from mouse and human species (92%). Only 1011 phages in 2% MPBS were incubated first for 1 hour with human IgG recently, and by using a very large commercial Fab library with 10 to remove the reactivity against the Fc fragment of the recombinant a size of 2.1 Â 10 , it has been described the production of Fab antigen, and then for 2 hours with FGFR3-coated wells. Plates were fragments against a nondisclosed FGFR3 isoform (25). In our washed 10 times with PBS-0.1% Tween 20 (20 times for the subsequent experiments, we have used two available public libraries of rounds) and treated with 100 AL of trypsin (Sigma) for elution of the single-chain Fv (scFv) antibodies, Tomlinson I + J (MRC bound phages. Elution and production of phages for other rounds of Geneservices, Cambridge, United Kingdom), which have been panning was carried out essentially as described by Goletz et al. (26). previously described (26). The size of both libraries is around In other panning experiments and for a major Fc blocking efficiency, 1.4 Â 108. scFvs penetrate tumors better than immunoglobulin phage selection was carried out by mixing first the phage suspension A Gs (IgG) and Fabs, are cleared more rapidly, and have greater with 50 g of human IgG in 2% MPBS. After 1-hour incubation at 37jC, this mixture was directly added to FGFR3-coated wells for specificity. In this report, we have developed a collection of 2 hours at room temperature. human scFv antibodies specific for FGFR3a isoform IIIc. These Phage ELISA. Flexible microtiter plates (Falcon, BD Biosciences, antibodies reacted with the bladder carcinoma cell line RT112 Franklin Lakes, NJ) were coated overnight with 0.3 Ag of FGFR3, FGFR1, by fluorescence-activated cell sorting (FACS) analysis and or human IgG in PBS. After washing thrice with PBS, plates were blocked inhibited cellular proliferation, showing promise for further with 2% MPBS for 2 hours at room temperature. Following additional therapeutic application. washing, different dilutions of the purified phage suspension coming from the selection steps, in the presence of 2% MPBS, were incubated for Materials and Methods 1 hour at room temperature. After washing, a 1:5,000 dilution of anti- M13 labeled with peroxidase in 2% MPBS was added for 1 hour and the Cell lines, proteins, and antibodies. RT112 cells were obtained from color was developed with 3,3V,5,5V-tetramethylbenzidine substrate the German collection of Microorganisms and Cell Cultures (DSMZ; (Sigma). The reaction was stopped after 10 minutes by adding Braunschweig, Germany). Human embryonic kidney (HEK293T) and 1 mol/L sulfuric acid and the absorbance was read at 450 nm. RT112 cells were cultured in RPMI 1640 (Sigma, St. Louis, MO) Monoclonal phage ELISA. Usually groups of 96 colonies from the supplemented with 10% fetal bovine serum and antibiotics. second or third round of selection were picked either manually or by The recombinant human FGFR3a(IIIc)/Fc and FGFR1a(IIIc)/Fc using an automatic picker (Q-pix, Genetix, Hampshire, United chimera were purchased from R&D Systems (Minneapolis, MN) and Kingdom) and grown in 100 ALof2Â TY medium containing 100 ReliaTech (Braunschweig, Germany). These proteins consist of the Ag/mL ampicillin and 1% glucose in a 96-well plate (Sarstedt,

www.aacrjournals.org 6281 Clin Cancer Res 2005;11(17) September 1,2005 Downloaded from clincancerres.aacrjournals.org on September 23, 2021. © 2005 American Association for Cancer Research. Cancer Therapy: Preclinical

Fig. 1. Flowchart to obtain FGFR3-specific scFvs by using phage display antibody libraries.

Numbrecht, Germany) at 37jC. The next day, cultures were diluted purification and to separate monomers from dimers and larger 1:100 in the same medium and shaken for 2 hours at 37jC. Then, 25 AL multimers, scFv-containing fractions obtained from immobilized metal of 2Â TY medium containing 109 KM13 helper phages were added and affinity chromatography were pooled, dialyzed against PBS, applied to a the culture continued for 1 hour. Bacterial cells were pelleted, HiLoad 26/60 Superdex 75 pg column equilibrated with the same resuspended in 2Â TY with 100 Ag ampicillin and 50 Ag/mL buffer, and run in an A¨ KTA-FPLC (GE Healthcare) at a flow rate of 0.75 kanamycin, and grown overnight at 30jC. Finally, the culture was mL/min. Fractions were analyzed by SDS-PAGE and those containing centrifuged and 50 AL of the supernatant were used for the monoclonal the monomeric form were pooled and concentrated using Vivaspin phage ELISA in the same conditions detailed above. concentrators (Vivascience, Hannover, Germany). Sequence analysis. Phagemid DNAs from individual colonies were Immunoprecipitation-Western blot analysis. About 2 Â 108 RT112 amplified by PCR with the primers LMB3 5V-CAGGAAACAGCTATGAC- cells for each experiment were washed with ice-cold PBS and scraped in 3V and pHENseq 5V-CTATGCGGCCCCATTCA-3V. PCR products were ice-cold lysis buffer [50 mmol/L Tris-HCl (pH 7.4), 300 mmol/L NaCl, purified with the QIAquick PCR purification (Qiagen, Hilden, 5 mmol/L EDTA, 1% Triton X-100] supplemented with 1Â protease Germany) and directly sequenced by using fluorescent dideoxy chain inhibitor cocktail (Roche). The lysates were incubated for 30 minutes terminators and by automated sequencing (ABI7002, Applied Bio- on ice and centrifuged at 16,000 Â g for 15 minutes at 4jC. In parallel, systems, Foster City, CA). a precipitation complex of antibodies was made by incubating 300 AL Production and purification of soluble single-chain Fv antibody and of protein G-Sepharose (GE Healthcare) with 1 AL of anti–c-myc ELISA. The procedure for E. coli induction was identical to that (Sigma) for 90 minutes at 4jC. After three washes with lysis buffer, 50 described above except that at the exponential phase (A600 = 0.9), E. coli Ag of the purified scFvs were added and left to bind for 2 hours at 4jC HB2151 cells were induced with 2Â TY containing 100 Ag/mL followed by three washes with lysis buffer. For immunoprecipitation, ampicillin and 1 mmol/L isopropyl-L-thio-h-D-galactopyranoside and the cell lysates were added to the preformed scFv precipitation complex incubated at 30jC overnight. For the preparation of periplasmic and were incubated overnight at 4jC, pelleted by centrifugation, and extracts, bacteria were harvested and resuspended in 1:50 of the culture washed thrice with lysis buffer. Proteins were released with 0.1 mol/L volume of ice-cold TES buffer [0.5 mol/L sucrose in 200 mmol/L Tris- glycine-HCl (pH 2.7), precipitated with trichloroacetic acid (Sigma), HCl (pH 8.0), 0.5 mmol/L EDTA], containing 20 Ag/mL benzamidine and loaded onto 7.5 % SDS-PAGE for electrophoresis. They were then (Sigma) and 10 Ag/mL soybean trypsin inhibitor (Sigma), and rapidly transferred to Hybond-C membrane (GE Healthcare) and incubated diluted in 0.2Â TES. Cells were incubated on ice for 30 minutes and with anti-FGFR3 mAb overnight at 4jC. The filter was washed four centrifuged at 16,250 Â g for 10 minutes at 4jC. The supernatant times and bound antibodies were detected with Mouse IgG TrueBlot containing the periplasmic fraction was dialyzed overnight at 4jC and SuperSignal Substrate (Pierce, Rockford, IL). against 50 mmol/L Tris-HCl (pH 7.0) and 300 mmol/L NaCl and Surface plasmon resonance analyses. The binding kinetics of soluble subjected to affinity chromatography using TALON resin (BD Bio- scFvs to FGFR3 was measured at 25jC on a Biacore X (Biacore, Uppsala, sciences) following the instructions of the manufacturer. Fractions were Sweden). For immobilization of the FGFR3-Fc, goat anti-human IgG analyzed by SDS-PAGE and Coomassie blue staining. For further (Fc fragment) antibodies (Sigma) were immobilized on CM5 chips

Clin Cancer Res 2005;11(17) September 1, 2005 6282 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 23, 2021. © 2005 American Association for Cancer Research. FGFR3 Blocking scFvs

Ta b l e 1. Effect of the blocking procedures in the phage selection process

Selection Panning Picking Tested clones Positive, soluble scFv Unique sequences Selected clones A1 Third Manual 96 5 23A, 3B A2Second Picker 96 3 1 3C B Third Picker 96 7 3 1D, 2D,7D Total 288 15 6

(Biacore) by using the amino coupling kit (Biacore). For each native FGFR3 molecules, RT112 cells were gently detached from plastic experiment, 1 Ag of purified FGFR3-Fc was first captured and then the plates by using a nonenzymatic cell dissociation solution (Sigma) different purified scFvs were added at various concentrations ranging followed by washing with PBS. Approximately, 1 Â 106 cells and 1011 from 500 to 10 nmol/L. After each experiment, the surface was phage particles from fresh preparations were incubated in 2% MPBS for regenerated with 0.1 mol/L glycine-HCl (pH 2.5). Sensorgrams were 1 hour on ice. Bound phages were detected using anti-M13 diluted obtained at each concentration and evaluated using the BIA Evaluation 1:100 in 1% bovine serum albumin/PBS (1 hour on ice) and R- 3.0 program to determine the rate constants Kon and Koff to calculate phycoerythrin goat anti-mouse diluted 1:100 in 1% bovine serum the Kd. Experiments were made in duplicate. albumin/PBS (30 minutes on ice). After each incubation step, cells were To determine the binding sites of the different scFvs, we carried out a washed twice with 1% bovine serum albumin/PBS. Finally, cells were competitive binding assay; 0.5 Ag FGFR3-Fc was immobilized and then resuspended in 500 AL of PBS and immediately analyzed in a a1Amol/L solution of each scFv was injected followed by the injection FACScalibur with CellQuestPro software (BD Biosciences). of the same amount of the competitor antibody. If there was an increase For soluble scFvs, RT112 or transfected HEK293T cells (2 Â 105) were in the resonance units of the sensorgram, then the binding sites of both incubated with the corresponding scFv at a concentration of 50 Ag/mL scFvs were different. If there was no difference, then the binding sites in FACS buffer (PBS containing 3% fetal bovine serum and 0.1% were the same or overlapping. sodium azide) for 30 minutes at room temperature. After washing with The competitions for binding to FGFR3 between the ligand FGF9, FACS buffer, cells were fixed with 4% paraformaldehyde (Merck, FGF1, or control EGF and scFvs were also monitored by surface plasmon Darmstadt, Germany) in PBS for 10 minutes. Bound scFvs were resonance. After capturing FGFR3 on the sensor chip, 1 Amol/L FGF9, detected with anti-6xHis tag antibody diluted 1:50 in FACS buffer for FGF1, or EGF and 100 Ag/mL heparin were injected until saturation of 30 minutes at room temperature and then FITC-labeled anti-mouse binding. Then, 1 Amol/L of each scFv was subsequently injected and the antibody diluted 1:50 in FACS buffer. After each incubation step, cells binding curve of the antibody to the FGFR3-FGF ligand complex was were washed with FACS buffer. Cells were finally resuspended in 500 AL compared with the binding to FGFR3 alone obtained with each scFv at PBS and analyzed in a FACScalibur. the same concentration. Labeled RT112 cells were also examined with an ultraspectral Flow cytometry and confocal microscopy analyses. Phages and soluble confocal microscopy system (TCSSP2-AOBS-UV, Leica-Microsystems, scFvs were used for flow cytometry. Phages were prepared from 50 mL Mannheim, Germany) after nucleus counterstaining with 4V,6-diami- of culture, concentrated by polyethylene glycol precipitation, and dino-2-phenylindole (Molecular Probes). Images were acquired with a resuspended in 2 mL of PBS. To determine binding of phage scFv to 63Â oil immersion objective using the Leica Confocal Software.

Fig. 2. Sequence analysis of the selected FGFR3-specific scFv. Amino acid sequence alignments of theVH (A)andVL (B) regions of the anti-FGFR3 3A, 3B, 3C, 1D, 2D, and 7D scFv clones. CDR2and CDR3 are boxed. Dots, identical residues; asterisks, stop codons.

www.aacrjournals.org 6283 Clin Cancer Res 2005;11(17) September 1,2005 Downloaded from clincancerres.aacrjournals.org on September 23, 2021. © 2005 American Association for Cancer Research. Cancer Therapy: Preclinical

Fig. 3. Purification of soluble scFvs. A, 3B, 3C, 2D, and 7D scFvs with 6xHis and c-myc tags at the COOH terminus were expressed in E. coli HB2151 cells by induction with 1mmol/L isopropyl-L-thio-B- D-galactopyranoside overnight at 30jC. After osmotic shock with sucrose, His-tagged scFvs were purified from the periplasmic fraction by usingTalon resin. Elution fractions (1-7) were examined by 10% SDS-PAGE and Coomassie staining. Arrowheads, full-length scFvs. Molecular weight markers in kilodaltons are indicated on the left. B, size exclusion chromatography on a HiLoad 26/60 Superdex 75 pg column of immobilized metal affinity chromatography ^ purified 3C scFv in PBS (pH 7.4). Arrows, elution volumes of the molecular weight markers bovine serum albumin (67 kDa) and glutathione S-transferase (26 kDa). Inset, fractions corresponding to dimeric (1-4) and monomeric (1-5) forms were collected, loaded onto a 10% SDS-PAGE, and stained with Coomassie brilliant blue.

Cell proliferation assays. RT-112 cells were plated onto 96-well colonies were randomly picked and tested by ELISA after the 3 tissue culture plates at a density of 2 Â 10 cells per well in RPMI 1640 different selection procedures, yielding 15 FGFR3-specific supplemented with 10% fetal bovine serum and antibiotics in antibodies. The final selection efficiency for the blocking duplicate. Once attached, cells were washed with PBS and cultured process A was 1.5% and for process B was 3.1%, showing that with different concentrations of anti-FGFR3 scFvs (ranging from 0.02 to the inclusion of a high concentration of blocking molecule in 2 Amol/L) in RPMI 1640 containing 25 ng/mL FGF9, FGF1, EGF, or FGF9 plus 0.5 Ag/mL of recombinant FGFR3, together with 50 Ag/mL the phage suspension is more efficient than a previous heparin (Sigma) and antibiotics for 48 hours at 37jC. Cell proliferation absorption of the phage on IgG-coated wells. was scored by staining cells with the chromogenic dye 3-(4,5- Characterization of the selected single-chain Fv antibodies. dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide. After removal After DNA sequencing, we found that 6 of 15 scFv sequences of the medium, 100 AL of 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenylte- specific for FGFR3 were unique and different (Fig. 2). Two of trazolium bromide reagent (Sigma) were added to the culture wells at a the six clones (3A and 1D) contained a stop codon within the final concentration of 1 mg/mL in RPMI 1640 and the cells were further sequence; they were identical except for a mutation in the incubated for 1 hour at 37jC. Then, 100 AL of DMSO (Sigma) were complementary-determining region (CDR) 2 of the VL region added to each well. Absorbance was read at 570 nm. All the (DxA). The other four scFvs were different from each other. All experiments were done at least thrice. Cell viability is represented as the mutations were concentrated on the hypervariable regions the ratio of absorbance given by scFv-treated cells over absorbance of nontreated cells, expressed as a percentage. CDR2 and CDR3. CDR3 was highly variable in both VH and VL, but CDR2 was mainly variable in VH. All the scFvs were expressed at high levels as soluble molecules. Results For scFv purification, the periplasmic fraction was purified by standard immobilized metal affinity chromatography proce- Selection of receptor 3–specific single- dures (Fig. 3A) followed by gel filtration on a Superdex 75 chain Fv antibodies. Two different phage selection procedures column (Fig. 3B). This step allowed for a good resolution and were carried out by using different blocking conditions. separation of monomers, dimers, and larger-size molecules. The Selection process A was made by using a preblocking step with purification of monomers is strictly necessary for determination IgG-coated wells. In selection B, phages were mixed with the of the kinetic variables of binding to the antigen. IgG for blocking. Three rounds of panning were done on each To assess the specificity of the selected antibodies, the selection phase. Results are shown in Table 1. In total, 288 periplasmic scFvs were tested in ELISA with FGFR3 and the

Clin Cancer Res 2005;11(17) September 1, 2005 6284 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 23, 2021. © 2005 American Association for Cancer Research. FGFR3 Blocking scFvs highly homologous (>62%) FGFR1. The results of the ELISA analysis are shown in Fig. 4. Although different from clone to Ta b l e 2 . Affinity analyses of anti-FGFR3 scFvs by clone, there was a significant reactivity of all the scFvs against surface plasmon resonance FGFR3. In contrast, the reactivity against FGFR1 was marginal in all the cases and about the same magnitude as the background scFv Kon (1/Ms) Koff (1/s) Kd (nmol/L) response against human IgG. Given the lower reactivity of the 3A 2D 2.29e+5 5.65eÀ326.8 and 1D clones, besides the presence of stop codons in the scFv 7D 3.16e+5 1.22eÀ240.9 gene, they were discarded for further analysis. 3B 3.66e+5 4.66eÀ312.4 Fibroblast growth factor receptor-3 binding affinity of purified 3C 2.64e+6 6.14eÀ22 5.0 single-chain Fv antibodies. Monomeric forms of scFv were used for determination of the affinity constants by surface plasmon resonance on immobilized FGFR3 using a Biacore instrument. effective for blocking the binding of 7D, but not the binding of The results for K , K , and K determinations are shown in on off d 3C. No blocking effect was noticed for an irrelevant growth Table 2. The affinities oscillated between 40.9 and 12.4 nmol/L, factor such as EGF. indicating a moderate to high affinity, which is rather good if Reactivity of single-chain Fv antibodies to fibroblast growth we consider the medium size and complexity of the original factor receptor 3–expressing RT112 cells. The binding of scFvs libraries (1.6 Â 108). However, the K of two of the scFvs, 3C off to native FGFR3 expressed by RT112 bladder carcinoma cells and 7D, are relatively poor: 16.28 seconds and 1.36 minutes, was assessed by flow cytometry using freshly prepared phages respectively. or soluble scFv. Either six scFv phages or four soluble scFvs were Single-chain Fv antibody and fibroblast growth factor compe- used for the cytometry experiments. They stained RT112 cells tition analysis. To distinguish different binding sites, we did a albeit at various levels. Flow cytometric analyses with phages competitive binding analysis by using FGFR3 immobilized on a displaying the six different scFvs revealed that all of them sensorchip and consecutive binding of the two antibodies that reacted at a similar significant level with viable bladder we wanted to study by surface plasmon resonance. The first carcinoma RT112 cells (Fig. 6A). When purified scFv fragments antibody was passed in a saturating amount, in such a way that corresponding to 3B, 3C, 2D, and 7D were used, two of them, signal could increase only if the binding sites of the antibodies 3C and 7D, gave an intense immunofluorescence staining with were different. The results are shown in Fig. 5A. In summary, a significant fraction of viable RT112 cells. Some cells remained 2D and 3B should recognize similar or overlapping epitopes unstained, indicating some heterogeneity at the cellular level. as they inhibit binding with each other. 7D and 3C must 2D and 3B stained a lower percentage of cells, suggesting a recognize different epitopes. However, the 7D epitope accessi- different accessibility to the epitopes by the scFvs. The analysis bility can be partially blocked by 2D and by previous binding of the 3C- and 7D-stained cells by confocal microscopy of 3B, but not in the opposite way (data not shown). revealed the characteristic staining pattern of membrane The ability of the natural ligands FGF9 and FGF1 to block proteins (Fig. 6A). the binding of scFvs was also tested by Biacore. As a negative None of the selected scFvs was functional by Western blot control, we used EGF. The results for 3C and 7D are shown in assays (data not shown). Therefore, to confirm the specific Fig. 5B. Previous binding of FGF9 to FGFR3 significantly blocks binding to FGFR3, an immunoprecipitation in nondenaturing the union of the two scFvs, indicating that FGF9 and the scFvs conditions followed by Western blot assay was carried out with compete for the same binding site. In contrast, FGF1 was very two purified scFvs. Results are shown in Fig. 6B. Immunopre- cipitation of RT112 cell extracts with scFvs 3C and 7D yielded a prominent band with the correct size that was recognized by a FGFR3-specific mAb, confirming the ability of these two scFvs to recognize soluble and native FGFR3. Reactivity of single-chain Fv antibodies to fibroblast growth factor receptor 3 containing activating mutations. Because FGFR3 is frequently mutated in bladder cancer, we decided to test the binding ability of these two scFvs to a mutant form of FGFR3 containing a Cys in position 249 instead of a Ser. This activating mutation has been reported as the most frequent in bladder cancer (27). We prepared this mutant by site-directed mutagenesis using PCR with a modified oligonucleotide. The mutation was confirmed by DNA sequencing (Fig. 7A). The correct expression of the mutant form after transfection of HEK293T cells was assessed by Western blot analysis (Fig. 7B). The mutant displayed similar levels of expression when compared with the wild-type form. Tubulin expression was

Fig. 4. Specificity of soluble scFvs from selected clones by ELISA. scFv-pIII fusion used for normalization of the loaded protein. Flow cytometric proteins from supernatant culture of scFv-expressing E. coli TG1strain were analysis showed that both scFvs, 3C and 7D, were able to analyzed for binding to FGFR3-Fc (black columns), FGFR1-Fc (gray columns), and recognize HEK293T cells transfected with the FGFR3 S249C human IgG (white columns) immobilized on an ELISA plate. Bound scFvs were detected with horseradish peroxidase ^ labeled anti ^ c-myc using 3,3V,5,5V- mutant even more strongly than cells transfected with the wild- tetramethylbenzidine as substrate. Results are expressed as absorbance at 450 nm. type form (Fig. 7C).

www.aacrjournals.org 6285 Clin Cancer Res 2005;11(17) September 1,2005 Downloaded from clincancerres.aacrjournals.org on September 23, 2021. © 2005 American Association for Cancer Research. Cancer Therapy: Preclinical

Fig. 5. scFv competition analyses for FGFR3 binding. A, Biacore sensorgrams showing pairwise binding competition of scFvs on FGFR3. Purified monomers of a FGFR3- binding scFv were first injected on CM5 chips containing immobilized FGFR3 until saturation; then a second scFv was loaded. ScFvs recognizing different epitopes should give additional increases in response units (RU), whereas an unchanged signal suggests the recognition of the same or overlapping epitope. Dashed lines, 3B; dashed lines, 3C; dotted lines, 2D; dashed-dotted lines,7D. B, sensorgrams showing binding competition between FGF9, FGF1,or EGFand 3C or 7D scFvs to FGFR3. Purified 3C and 7D scFvs (1 Amol/L) were injected either to FGFR3 immobilized biosensor (dashed line) or to FGFR3 previously saturated with FGF9/heparin (solid line), FGF1/heparin (dotted line), or EGF/heparin (dashed-dotted line). Each sensorgram was overlaid and zeroed on the y-axis to the average baseline before scFv injection.The start injection time for each scFv was set to zero on the x-axis. Competition was estimated by subtracting the response units of a given scFv injected into a FGFR3-immobilized flow cell from the response units of the injected scFv into the FGFR3-Fc immobilized chip in the presence of bound FGF9, FGF1, or EGF.

Clin Cancer Res 2005;11(17) September 1, 2005 6286 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 23, 2021. © 2005 American Association for Cancer Research. FGFR3 Blocking scFvs

Blocking of RT112 proliferation. We examined the efficacy of significantly inhibited FGFR3-mediated proliferation of RT112 the scFvs in inhibiting tumor cell proliferation in vitro. The cells. 2D inhibition was at a much lower degree and 3B did not RT112 tumor cell line was used in this study because it cause any inhibitory effect. expresses significant levels of FGFR3 (20). They were cultured A further confirmation of the specificity of the blocking was in the presence of FGF9 and heparin and in the absence of FCS. made by the addition of different growth factors. The scFv Three different concentrations of scFv were added to the cells inhibition in the presence of different growth factors is shown in for 48 hours. In two cases, 3C and 7D, there was a significant Fig. 8B. The inhibitory effect was factor specific. Both 3C and 7D inhibition of the proliferation that was dependent on the scFv inhibited the proliferation induced by FGF9, but only 7D was concentration added to the cultures (Fig. 8A). The highest able to block FGF1. None of them was able to block the inhibition was achieved at 2 Amol/L, but it was also significant proliferation induced by EGF. Moreover, the addition of soluble at 0.2 Amol/L. Therefore, anti-FGFR3 scFvs 3C and 7D FGFR3 to the cultures abolished the scFv blocking capability.

Fig. 6. Binding of phage and soluble scFvs to RT112human bladder carcinoma epithelial cells by flow cytometry and confocal microscopy. A, Bound phages were detected with anti-M13 followed by R-phycoerythrin ^ conjugated goat anti-mouse IgG. For soluble scFvs, the bound scFv was fixed with paraformaldehyde and detected with sequential incubations with anti-6xHis mAb and FITC-labeled anti-mouse IgG. FACS histograms show the binding of each scFv clone (bold line) and the backgrounds of phycoerythrin- or FITC-conjugated secondary antibodies (gray). RT112cells labeled with 3C or 7D scFvs and FITC-labeled anti-mouse IgG were counterstained with 4V,6-diamidino-2- phenylindole and visualized with laser confocal microscopy. Representative micrographs show scFv membrane staining (green)and4V,6-diamidino-2-phenylindole nuclear staining (red). Bar, 8 Am. B, FGFR3 immunoprecipitation (IP) of RT112cellular extracts followed byWestern blot (WB) analysis with FGFR3-specific mAb. A, Tr i t o n X-100 extracts of RT112cells; B, scFv anti ^ glutathione S-transferase immunoprecipitation (negative control); C, scFv 3C immunoprecipitation; D, scFv 7D immunoprecipitation.

www.aacrjournals.org 6287 Clin Cancer Res 2005;11(17) September 1,2005 Downloaded from clincancerres.aacrjournals.org on September 23, 2021. © 2005 American Association for Cancer Research. Cancer Therapy: Preclinical

Fig. 7. Reactivity of the selected scFvs with FGFR3 mutated in the extracellular domain. A, electropherograms showing the mutation in the DNA sequence to change Ser (TCC) 249 to Cys (TGC). B, immunoblotting analysis. HEK293Tcells transfected with either wild-type or mutant (S249C) FGFR3 were resolved on SDS-PAGE, transferred to nitrocellulose filters, immunoblotted with mAb against FGFR3 or tubulin (bottom), and visualized by enhanced chemiluminiscence. C, FACS analysis of scFv binding to HEK293Tcells expressing either wild-type or mutant FGFR3 (S249C).Transfected HEK293Tcells were incubated with scFv 3C (left)and 7D (right). Bound antibodies were fixed with paraformaldehyde and detected with mAb against 6xHis and phycoerythrin- conjugated goat anti-mouse IgG. Overlay histograms show the binding of the corresponding scFv to mock-transfected (gray), FGFR3(IIIc)WT-transfected (black line), and FGFR3(IIIc)S249C- transfected (dashed line)cells.

These results correlate well with the competition analysis, flow to the Fc region of human IgG1 as antigen for antibody cytometry, and the accessibility of the epitopes on the cell selection. Therefore, the specific reactivity obtained against membrane. Finally, from the morphologic point of view, FGFR3 must be targeted to the extracellular domains. After treatment with 3C and 7D scFvs resulted in changes in cellular ELISA selection, we obtained six unique FGFR3-specific scFv morphology and slower growth (Fig. 8C), with a more sequences. The specificity of the antibodies was confirmed by prominent vacuolization in the case of 7D. immunoprecipitation, Biacore analysis, and flow cytometry. None of the selected scFvs was able to recognize FGFR3 by Discussion immunoblot analysis (data not shown). These results suggest that the recognized epitopes are conformational, nonlinear, FGFR3 is a implicated in cell growth and susceptible to denaturation, probably due to the use of and tumorigenesis through activation of different signal soluble antigen for the panning procedures. These findings transduction pathways such as extracellular signal-regulated make difficult the identification of the amino acid sequence kinases 1 and 2, phosphatidylinositol 3 kinase, and phospho- corresponding to these epitopes and its position into the FGFR3 lipase c-g pathways (9, 28, 29). Overexpression of receptor molecule. Competition experiments have shown that the tyrosine kinase in tumors is generally associated with poor antibodies recognize different but, in some cases, overlapping prognosis (30). Apart from bladder cancer, FGFR3 is mutated epitopes. 2D and 3B probably recognize the same epitope. and overexpressed in a variety of cancers and, as a result, Biacore analysis was used for the determination of the kinetic particular attention has been given to FGFR3 for therapeutic constants of the scFv binding to the antigen. One of the use. Blocking of the transduction pathways used by FGFR3 problems associated with phage display is related to the could be an appropriate way to inhibit cell growth and generally low affinity shown by these scFv. As previously differentiation induced by this molecule. Thus, FGFR3 con- described, for selection of antigens immobilized on polystyrene stitutes an excellent target for effective cancer intervention. In a plates, the resulting scFvs are a mixture of monomers and previous report (20), it has been shown that murine mAbs dimers (31). The resulting homodimers show a significant specific for FGFR3 provoked an antiproliferative effect in the increase in apparent affinity (avidity; ref. 32). Due to this RT112 cell line. However, the use of mouse mAbs for human increase in avidity, phages displaying mixtures of monomeric cancer therapy is not realistic due to the strong human anti- and dimeric forms will be preferentially selected. To determine mouse antibody response, which has made almost impossible the kinetic constants of the scFvs, it was necessary to separate any therapeutic approach based on mouse mAbs. Currently the monomers by gel filtration. In our case, the affinities of the available alternatives include ‘‘chimerization’’ and ‘‘humaniza- selected scFvs oscillated between 40.9 and 12.4 nmol/L, tion’’ of mouse antibodies. Both approaches are labor- and indicating moderately high affinities, which is rather good if time-intensive and are unpredictable procedures. we consider the low size and complexity of the original libraries We have directly prepared human antibodies from a human (1.6 Â 108) and compare very favorably with the affinities phage display library using a soluble chimeric molecule obtained by using a much larger library (2 Â 1010) of Fab composed of the extracellular domain FGFR3a(IIIc) coupled fragments, which were approximately in the same range or 1

Clin Cancer Res 2005;11(17) September 1, 2005 6288 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 23, 2021. © 2005 American Association for Cancer Research. FGFR3 Blocking scFvs order of magnitude better (25). In terms of application to cancer This result indicates the capacity of these antibodies to use the therapy, the Kd of these scFvs might prove appropriate taking same binding site as FGF9, blocking its natural effector function. into account that recent reports have shown that a very high Furthermore, 3C and 7D inhibit FGF9-induced proliferation in a affinity can restrict the tumor penetration of the scFvs (33). significant and concentration-dependent manner. Both 7D and À2 À2 Currently available antibodies for therapeutic use have a Kd 3C have a very rapid off rate (1.22 Â 10 and 6.14 Â 10 , range of 10 to 0.1 nmol/L (34). respectively), which might lead to a very short association with Flow cytometric analysis confirmed the specific binding of the the tumoral cells due to the rapid turnover and internalization. A 3C and 7D scFvs to the native membrane FGFR3. More reduction in Koff is typically the major kinetic mechanism interestingly, 3C and 7D were able to recognize even better a resulting in higher affinity. Thus, it would be convenient to mutant FGFR3 containing one of the most abundant activating improve this affinity to slow the off rate. Because of the relatively mutations present in bladder cancer, S249C, stressing the ability of poor Koff of 3C and 7D when compared with other scFvs, it is these two scFvs to bind tumoral tissues. This enhanced recognition very likely that an improvement in the affinity of these molecules could be explained by the natural tendency to dimerize the might suppose an improvement in their blocking ability. The 3C mutant forms, which might stabilize the antibody binding. and 7D scFv sequences will serve as a template for affinity Previous results have shown that secreted soluble, extracellular maturation studies according to well-established technologies ligand-binding domains of the native FGFRs play an inhibitory such as complementary-determining region mutagenesis (36) or effect on cell proliferation, mopping up the ligand binding to the somatic hypermutation mimicry (37). receptor (35). The antibody fragments may mimic this capability There are a number of important human or humanized of the secreted extracellular domains. In fact, several results antibodies in clinical trials against different molecules impli- presented in this study suggest that the mechanism of RT112 cell cated in cancer progression (38). Antibodies are showing a growth inhibition by 3C and 7D scFvs is probably due to a direct great promise in cancer treatment. Compared with small block of FGF-FGFR3 interaction, resulting in the inhibition of molecules, antibody therapy may provide higher specificity to FGFR3 activation. This is based on the observation that 3C and tumoral cells and better safety profile. The moderate affinity of 7D specifically bind to both soluble and cell-surface expressed the scFv fragments is sufficient to react only with a population FGFR3 on RT112 cells. Moreover, 3C and 7D were blocked by of cells expressing significant levels of FGFR3, but not towards FGF9 from binding to FGFR3, but only 7D competed with FGF1. cell populations that express only low levels of the receptor. The

Fig. 8. Inhibition of RT112cell proliferation by FGFR3-specific scFvs. A, inhibitory activity of 3B, 3C, 2D, and 7D scFvs on RT112growth was tested at various concentrations: 2 Amol/L (black columns), 0.2 Amol/L (gray columns), and 0.02 Amol/L (white columns). Af ter 4 8 hours of incubation, cell viability was determined by 3-(4,5-dimethylthiazol-2-yl)- 2,5-diphenyltetrazolium bromide assay at 450 nm. Absorbance of the untreated control cells was taken as 100% and absorbance of treated cells was taken as percentage of relative growth for each concentration used. Columns, mean of three independent cell proliferation experiments (each concentration tested in duplicate); bars, SD. B, the inhibitory effects of 2 Amol/L 3C (dashed columns)and7D (dotted columns) scFvs were assayed as described above in the presence of FGF9, FGF1,EGF,or0.5Ag/mL of recombinant FGFR3 in the media (FGF9 + FGFR3 columns). C, the number and morphology of cells, either untreated or treated with 2 Amol/L 3C and 7D scFvs, after 48 hours of incubation were visualized before the 3-(4,5-dimethylthiazol-2-yl)-2,5- diphenyltetrazolium bromide assay at 20Â and 40Â magnifications by using an inverted Leica DMIL microscope equipped with a Leica DFC350 FX digital camera.

www.aacrjournals.org 6289 Clin Cancer Res 2005;11(17) September 1,2005 Downloaded from clincancerres.aacrjournals.org on September 23, 2021. © 2005 American Association for Cancer Research. Cancer Therapy: Preclinical scFvs could be used as therapeutic agents either alone or a molecular target for therapy (19). Specific inhibitors of FGFR3 combined with other molecules such as immunotoxins or induced cell cycle arrest and apoptosis in human myeloma radionuclides that can selectively kill the tumoral cells cells. It will be interesting to test the antiproliferative effect of displaying FGFR3, as previously shown for angiogenic mole- these antibodies on multiple myeloma cell lines. Our data, cules and growth factor receptors, such as vascular endothelial together with previous results of gene expression analysis and growth factor or erbB2 (neu/HER2; ref. 24), or other overex- polyclonal anti-FGFR3 antibodies, support further evaluation pressed proteins (39). The small size of the scFvs considerably of 3C and 7D scFv antibodies as antitumor agents. Previous increases their tumor penetration capabilities (33, 40) and findings (20) support that FGFR3 inhibition should be pharmacokinetic properties (41). Moreover, the use of anti- introduced at early disease stages (when FGFR3 overexpression body fragments confers a high specificity to the inhibitory is more abundant) before progressive loss of expression. effect, which allows the discrimination between the different FGFRs. This is particularly relevant given the opposite effects Acknowledgments they play. Thus, whereas FGFR3 presents an oncogenic activity, FGFR2 shows a tumor suppressor activity (42). We thank Arancha Garcia, Diego Megı´as, and Maria Montoya of the Centro Nacional de Investigaciones Oncolo¤ gicas confocal and cytometry unit for their help Apart from its overexpression in bladder or cervical cancer, and collaboration in the flow cytometric and confocal microscopy analysis.We thank dysregulated FGFR3 has been show to play a major role in Silvia Romero for her excellent technical assistance and Giovanna Roncador for her multiple myeloma tumorigenesis and has already been tested as help in the transfection experiments.

References 1. Cronauer MV, Schulz WA, Seifert HH, Ackermann R, bladder and cervix carcinomas. Nat Genet 1999; 29. Hart KC, Robertson SC, Kanemitsu MY, Meyer AN, Burchardt M. Fibroblast growth factors and their 23:18^20. TynanJA, Donoghue DJ.Transformation and Stat acti- receptors in urological cancers: basic research and 15. Sibley K, Cuthbert-Heavens D, Knowles MA. Loss vation by derivatives of FGFR1, FGFR3, and FGFR4. clinical implications. Eur Urol 2003;43:309 ^ 19. of heterozygosity at 4p16.3 and mutation of FGFR3 Oncogene 2000;19:3309 ^20. 2. Cronauer MV, Hittmair A, Eder IE, et al. Basic fibro- in transitional cell carcinoma. Oncogene 2001;20: 30. Agazie YM, Movilla N, Ischenko I, Hayman MJ. blast growth factor levels in cancer cells and in sera of 686^91. The phosphotyrosine phosphatase SHP2is a critical patients suffering from proliferative disorders of the 16. Billerey C, Chopin D, Aubriot-Lorton MH, et al. Fre- mediator of transformation induced by the oncogenic prostate. Prostate 1997;31:223 ^ 33. quent FGFR3 mutations in papillary non-invasive blad- fibroblast growth factor receptor 3. Oncogene 2003; 3. Miyake H, Yoshimura K, Hara I, Eto H, Arakawa S, der (pTa) tumors. AmJPathol 2001;158:1955 ^ 9. 22:6909 ^ 18. Kamidono S. Basic fibroblast growth factor regulates 17. van Rhijn BW, van Tilborg AA, Lurkin I, et al. Novel 31. Schier R, Bye J, Apell G, et al. Isolation of high- matrix metalloproteinases production and in vitro inva- fibroblast growth factor receptor 3 (FGFR3) mutations affinity monomeric human anti-c-erbB-2single chain siveness in human bladder cancer cell lines. J Urol in bladder cancer previously identified in non-lethal Fv using affinity-driven selection. J Mol Biol 1996; 1997;157:2351 ^ 5. skeletal disorders. EurJHum Genet 2002;10:819^ 24. 255:28 ^ 43. 4. Ruohola JK, Viitanen TP,Valve EM, et al. Enhanced 18. Ozen M, Giri D, Ropiquet F, Mansukhani A, Ittmann 32. GriffithsAD,MalmqvistM,MarksJD,etal.Human invasion and tumor growth of fibroblast growth factor M. Role of fibroblast growth factor receptor signaling anti-self antibodies with high specificity from phage 8b-overexpressing MCF-7 human breast cancer cells. in prostate cancer cell survival. J Natl Cancer Inst display libraries. EMBO J1993;12:725 ^ 34. Cancer Res 2001;61:4229 ^ 37. 2001;93:1783 ^ 90. 33. Adams GP, Schier R, McCall AM, et al. High affinity 5. Yamanaka Y, Friess H, Buchler M, et al. Overexpres- 19. PatersonJL, Li Z,Wen XY,et al. Preclinical studies of restricts the localization and tumor penetration of sin- sion of acidic and basic fibroblast growth factors in fibroblast growth factor receptor 3 as a therapeutic gle-chain fv antibody molecules. Cancer Res 2001;61: human pancreatic cancer correlates with advanced target in multiple myeloma. Br J Haematol 2004;124: 4750 ^ 5. tumor stage. Cancer Res 1993;53:5289 ^ 96. 595^603. 34. Hoet RM, Cohen EH, Kent RB, et al. Generation 6. Powers CJ, McLeskey SW, Wellstein A. Fibroblast 20. Gomez-Roman JJ, Saenz P, Cuevas Gonzalez J, of high-affinity human antibodies by combining growth factors, their receptors and signaling. Endocr et al. Fibroblast growth factor receptor 3 is overex- donor-derived and synthetic complementarity- Relat Cancer 2000;7:165 ^ 97. pressed in urinary tract carcinomas and modulates determining-region diversity. Nat Biotechnol 2005; 7. Sleeman M, Fraser J, McDonald M, et al. Identifica- the neoplastic cell growth. Clin Cancer Res 2005;11: 23:344 ^ 8. tion of a new fibroblast growth factor receptor,FGFR5. 459^65. 35. Ogawa T, Takayama K,Takakura N, Kitano S, Ueno Gene 2001;271:171 ^ 82. 21. van Rhijn BW,Vis AN, van der Kwast TH, et al. Mo- H. Anti-tumor angiogenesis therapy using soluble 8. Chellaiah A, Yuan W, Chellaiah M, Ornitz DM. Map- lecular grading of urothelial cell carcinoma with fibro- receptors: enhanced inhibition of tumor growth when ping ligand binding domains in chimeric fibroblast blast growth factor receptor 3 and MIB-1is superior to soluble fibroblast growth factor receptor-1 is used growth factor receptor molecules. Multiple regions pathologic grade for the prediction of clinical out- with soluble vascular endothelial growth factor determine ligand binding specificity. J Biol Chem come. J Clin Oncol 2003;21:1912^ 21. receptor. Cancer GeneTher 2002;9:633 ^ 40. 1999;274:34785^94. 22. Kretzschmar T, von Ruden T. Antibody discovery: 36. Schier R, McCall A, Adams GP, et al. Isolation of 9. Kanai M, Goke M,Tsunekawa S, Podolsky DK. Signal phage display.Curr Opin Biotechnol 2002;13:598^602. picomolar affinity anti-c-erbB-2single-chain Fv by transduction pathway of human fibroblast growth fac- 23. Winter G, Griffiths AD, Hawkins RE, Hoogenboom molecular evolution of the complementarity determin- tor receptor 3. Identification of a novel 66-kDa phos- HR. Making antibodies by phage display technology. ing regions in the center of the antibody binding site. phoprotein. J Biol Chem 1997;272:6621 ^ 8. Annu Rev Immunol 1994;12:433 ^ 55. JMolBiol1996;263:551^67. 10. Plowright EE, Li Z, Bergsagel PL, et al. Ectopic 24. Hudson PJ, Souriau C. Engineered antibodies. Nat 37. Chowdhury PS, Pastan I. Improving antibody affini- expression of fibroblast growth factor receptor 3 pro- Med 2003;9:129 ^ 34. ty by mimicking somatic hypermutation in vitro.Nat motes myeloma cell proliferation and prevents apop- 25. Rauchenberger R, Borges E, Thomassen-Wolf E, Biotechnol 1999;17:568 ^72. tosis. Blood2000;95:992 ^ 8. et al. Human combinatorial Fab library yielding specific 38. Harris M. Monoclonal antibodies as therapeutic 11. PollettJB,Trudel S, Stern D, Li ZH, Stewart AK. Over- and functional antibodies against the human fibroblast agents for cancer. Lancet Oncol 2004;5:292 ^ 302. expression of the myeloma-associatedoncogenefibro- growth factor receptor 3. J Biol Chem 2003;278: 39. Niv R, AssarafYG, Segal D, Pirak E, ReiterY.Target- blast growth factor receptor 3 confers 38194^205. ing multidrug resistant tumor cells with a recombinant resistance. Blood 2002;100:3819^ 21. 26. Goletz S, Christensen PA, Kristensen P, et al. Selec- single-chain FV fragment directed to P-glycoprotein. 12. Sibley K, Stern P, Knowles MA. Frequency of fibro- tion of large diversities of antiidiotypic antibody frag- Int J Cancer 2001;94:864 ^ 72. blast growth factor receptor 3 mutations in sporadic ments by phage display. J Mol Biol 2002;315:1087 ^ 97. 40. Jain RK. Delivery of molecular medicine to solid tumours. Oncogene 2001;20:4416^ 8. 27. L’Hote CG, Knowles MA. Cell responses to FGFR3 tumors. Science1996;271:1079 ^80. 13. Dai H, Holm R, Kristensen GB, AbelerVM, Borresen- signalling: growth, differentiation and apoptosis. Exp 41. Pastan I, Kreitman RJ. Immunotoxins in cancer ther- Dale AL, Helland A. Fibroblast growth factor receptor Cell Res 2005;304:417^ 31. apy. Curr Opin Investig Drugs 2002;3:1089 ^ 91. 3 (FGFR3)analyses of the S249C mutation and 28. Hart KC, Robertson SC, Donoghue DJ. Identifica- 42. Bernard-Pierrot I, Ricol D, CassidyA, et al. Inhibition protein expression in primary cervical carcinomas. tion of tyrosine residues in constitutively activated of human bladder tumour cell growth by fibroblast Anal Cell Pathol 2001;23:45 ^ 9. fibroblast growth factor receptor 3 involved in mito- growth factor receptor 2b is independent of its kinase 14. Cappellen D, De Oliveira C, Ricol D, et al. Fre- genesis, Stat activation, and phosphatidylinositol activity. Involvement of the carboxy-terminal region of quent activating mutations of FGFR3 in human 3-kinase activation. Mol Biol Cell 2001;12:931 ^ 42. the receptor. Oncogene 2004;23:9201 ^ 11.

Clin Cancer Res 2005;11(17) September 1, 2005 6290 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 23, 2021. © 2005 American Association for Cancer Research. Targeting the Extracellular Domain of Fibroblast Growth Factor Receptor 3 with Human Single-Chain Fv Antibodies Inhibits Bladder Carcinoma Cell Line Proliferation

Jorge Martínez-Torrecuadrada, Gabriela Cifuentes, Paula López-Serra, et al.

Clin Cancer Res 2005;11:6280-6290.

Updated version Access the most recent version of this article at: http://clincancerres.aacrjournals.org/content/11/17/6280

Cited articles This article cites 42 articles, 12 of which you can access for free at: http://clincancerres.aacrjournals.org/content/11/17/6280.full#ref-list-1

Citing articles This article has been cited by 8 HighWire-hosted articles. Access the articles at: http://clincancerres.aacrjournals.org/content/11/17/6280.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 Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/11/17/6280. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.

Downloaded from clincancerres.aacrjournals.org on September 23, 2021. © 2005 American Association for Cancer Research.