Urologic Oncology: Seminars and Original Investigations 32 (2014) 877–884

http://dx.doi.org/10.1016/j1078-1439/r 2014 Elsev

1Present address: DeparGermany.

2Equal contributions.

* Corresponding author7659.E-mail address: stefan.

Original article

Phenotypic drug screening and target validation for improvedpersonalized therapy reveal the complexity of phenotype-genotype

correlations in clear cell renal cell carcinoma

Meike Schneidera,b,1,2, Julia Schülerc,2, Rouven Höfflinb,2, Nina Korzeniewskib,2,Carsten Grüllichd,e, Wilfried Rothe,f, Dogu Tebera,e, Boris Hadaschika,e, Sascha Pahernika,e,

Markus Hohenfellnera,e, Stefan Duensinga,b,e,*a Department of Urology, University of Heidelberg School of Medicine, Heidelberg, Germany

b Molecular Urooncology, Department of Urology, University of Heidelberg School of Medicine, Heidelberg, Germanyc Oncotest GmbH, Freiburg, Germany

d Department of Medical Oncology, National Center for Tumor Diseases, Heidelberg, Germanye Center for Kidney Tumors, National Center for Tumor Disease and University of Heidelberg School of Medicine, Heidelberg, Germany

f Department of Pathology, University of Heidelberg School of Medicine, Heidelberg, Germany

Received 8 January 2014; received in revised form 11 March 2014; accepted 11 March 2014

Abstract

Objectives: Novel personalized therapeutic approaches are urgently needed for patients with metastatic clear cell renal cell carcinoma(ccRCC).Methods and materials: We combined the development of a primary patient-derived ccRCC cell line with a phenotypic drug screen

consisting of 101 approved anticancer compounds.Results: We identified the MNNG HOS transforming gene (MET)-anaplastic lymphoma receptor tyrosine kinase (ALK) inhibitor crizotinib as

the top hit of our drug screen, whereas compounds targeting the vascular endothelial growth factor (VEGF) or mammalian target of rapamycin(mTOR) pathway showed no or only minor in vitro activity. Among the known major crizotinib targets MET, ALK, and ROS-1, only MET wasexpressed in our ccRCC cell line. Subsequent sequence analysis revealed a heterozygous R988C mutation of the MET gene and a VHL deletion inboth the primary tumor and the tumor-derived ccRCC cell line. However, we were unable to show an activation of MET and, further, METknockdown did not result in increased apoptosis or cytotoxicity. Therefore, our results suggest that MET R988C does not function as a majoroncogenic driver mutation but rather represents a sequence variant. However, we provide evidence that the cytotoxic effect of crizotinib in our cellline model correlates with its ability to inhibit P-glycoprotein (ABCB1)-associated transport functions.Conclusions: Our study shows that a phenotypic screen of a patient-derived tumor cell line can identify compounds with antitumor

activity but with an unexpected mode of action. Our results underscore that target validation and phenotype-genotype correlations remain amajor experimental challenge. The implications of our findings for a personalized management of patients with cancer are discussed. r 2014Elsevier Inc. All rights reserved.

Keywords: Renal cancer; Personalized therapy; Drug screen; MET; Crizotinib

.urolonc.2014.03.011ier Inc. All rights reserved.

tment of Urology, University Hospital Frankfurt,

. Tel.: þ49-6221-56-6255; fax: þ49-6-221-56-

duensing@med.uni-heidelberg.de (S. Duensing).

1. Introduction

Patients with metastatic renal cell carcinoma (RCC) have,with a few rare exceptions, no curative options. At the sametime, clear cell RCC (ccRCC) is a prime example of hownovel biological insights into the molecular pathogenesis candrive rational drug development [1–3]. Clear cell RCCs

M. Schneider et al. / Urologic Oncology: Seminars and Original Investigations 32 (2014) 877–884878

commonly harbor genetic alterations that inactivate the VHLtumor-suppressor gene and activate the phosphoinositide3-kinase/mammalian target of rapamycin (PI3K/mTOR) path-way. Targeted agents, including tyrosine kinase inhibitors(TKIs), that inhibit signaling cascades downstream of VHL orblock mTOR activation have therefore resulted in improvedpatient survival [4].

Renal cancers are genetically diverse and other examplesof recurrent mutations besides VHL include the METoncogene, which is commonly mutated in hereditary aswell as in a subset of sporadic RCCs of papillary morphol-ogy [5].

Recent evidence shows that ccRCCs not only frequentlyharbor VHL mutations but are also characterized byextensive intratumoral genomic heterogeneity [6]. Genomicinstability and tumor heterogeneity are important sources ofdrug-resistant tumor cell variants. This needs to be takeninto consideration for strategies to individualize the onco-logical management of patients with ccRCC.

In an effort to develop improved personalized therapeu-tic strategies for patients with metastatic ccRCC, wecombined the development of a primary, patient-derivedtumor cell line with a phenotypic drug screen. Results fromour study highlight the intricacies of phenotype-genotypecorrelations and have important implications for futureanticancer drug development efforts.

2. Materials and methods

2.1. Patient, primary tumor, and tumor-derived cell line

A primary, tumor-derived RCC cell line (RXF2209L)was obtained from a 49-year-old male patient with meta-static RCC (pT3aNxM1; Fuhrman Grade 3; Fig. 1A) whounderwent tumor nephrectomy at the Department of Urol-ogy of the University Hospital Heidelberg (Ethics Commit-tee votum S-453/2010). The histopathological analysis ofthe tumor showed a typical clear cell type RCC morphology(ccRCC; Fig. 1A). Additional characterization of the tumorshowed an average proliferative index of 25.1 cells perhigh-power field (�40) as measured by the proliferationmarker, Ki-67 (Fig. 1A). It is noteworthy that the prolifer-ation index showed a wide range between 2 and 121 Ki-67–positive cells per high-power field. The tumor was stronglypositive for nuclear HIF-1α (Fig. 1A) and showed anactivation of mTOR that was preferentially found at theperipheral zone/invasion front (Fig. 1A). HIF-2α wasexpressed only weakly and focally (not shown).

At the time of diagnosis, the patient showed signs oftumor dissemination to the bone, lung, and adrenal gland.The patient succumbed to progressive disease 10 monthsafter initial diagnosis following 7 months of stable diseaseunder the TKI, sunitinib.

Fresh tumor tissue was mechanically minced and incu-bated in an enzyme cocktail consisting of 41 U/ml

collagenase (Sigma), 125 U/ml DNAse (Roche), and100 U/ml hyaluronidase (Roche) at 371C for approximately45 minutes. The cells were passed through stainless steelsieves and washed. The tumor cell suspension was incu-bated in RPMI 1640 media supplemented with 5% fetalbovine serum, 100 U/ml of penicillin, 100 U/ml of strepto-mycin, and 1% L-glutamine (all from Gibco). The patient-derived primary tumor cell line was passaged 420 timesand cell line identity was confirmed by short tandem repeatanalysis (DSMZ Braunschweig, Germany). The mutationsstatus of VHL and MET in the primary tumor and on thepatient-derived cell line was determined using Sangersequencing (GATC Biotech, Konstanz, Germany). Alltumor-derived cell lines were generated by Oncotest (Freiburg,Germany). Normal renal epithelial cells used as controls wereobtained commercially (Lonza).

2.2. Phenotypic drug screening

The RXF2209L cells and normal renal epithelial cellsthat were used as control were subjected to a high-throughput drug screen using the National Cancer Insti-tute–Approved Oncology Drug Panel Set III (Fig. 1B) [7].A full list of compounds can be found under http://dtp.nci.nih.gov. The 101 compounds of the library included currentchemotherapeutic agents, small molecule inhibitors, andantihormonal substances. Briefly, both tumor cells andnormal renal epithelial cells were subcultured, counted,and inoculated into 96-well plates with 10,000 cells per wellallowed to adhere overnight. Drugs were then added to thetumor cells and the normal renal epithelial cells cultureplates at a final concentration of 1 and 10 mM, respectively.The plates were incubated at 371C for 48 hours, andapoptotic cell death was measured using a luminometricassay (Caspase-Glo 3/7 Assay; Promega). Results werevalidated using the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphe-nyltratrazolium bromide (MTT) assay (Molecular Probes).All experiments were performed in triplicates.

2.3. Immunological methods, antibodies, and reagents

Immunoprecipitation and immunoblot analyses wereperformed as previously described [8]. Immunohistochem-ical analysis of formalin-fixed, paraffin-embedded tumortissue has previously been described. Briefly, sections weredeparaffinized in xylene, rehydrated in a graded ethanolseries, and boiled in a microwave oven for 30 minutes in acitrate buffer (pH 6.0) followed by blocking and incubationwith primary antibodies against Ki-67 (Dako), HIF-1α,HIF-2α (both Novus), phospho-mTOR S2448 (cell signal-ing), and P-glycoprotein (P-gp) (C219, Calbiochem).Immunoperoxidase detection of primary antibodies wasperformed using the Histostain-Plus kit (Invitrogen) accord-ing to manufacturer's recommendations. Antibodies againstMET, phospho-MET Y1234/1235, phosphotyrosine (p-Tyr-102), phospho-mTOR S2448, ALK, ROS-1, glyceraldehyde

Fig. 1. Drug-screening procedure for improved personalized therapy of ccRCC. (A) Hematoxylin and eosin (H&E) stain of the primary tumor shows acharacteristic clear cell RCC morphology. Additional markers shown are Ki-67, HIF-1α, and phospho-mTOR S2448. The scale bar indicates 50 μm. (B) Anoverview of the drug screen protocol using a primary, tumor-derived cell line (RXF2209L) generated following standard mechanical and enzymaticdisintegration procedures and 101 compounds of the NCI-Approved Oncology Drug Set at a concentration of 1 or 10 μM. All experiments were done intriplicates and the response was measured using the Caspase-Glo 3/7 activity assay (Promega). Induction of apoptosis equal to or more than 3-fold greater thanthat of DMSO-treated controls was considered a hit if there was no toxicity (o3-fold apoptosis induction) in normal renal epithelial cells are used as controls.DMSO ¼ dimethyl sulfoxide; NCI ¼ National Cancer Institute. (Color version of figure is available online.)

M. Schneider et al. / Urologic Oncology: Seminars and Original Investigations 32 (2014) 877–884 879

3-phosphate dehydrogenase (GAPDH) (all Cell Signaling),and poly(ADP-ribose) polymerase (PARP) (both Calbio-chem) were used for immunoblot analyses. Rhodamine 123and verapamil were obtained commercially (Sigma). Intra-cellular accumulation of rhodamine 123 was analyzed byflow cytometry. Briefly, cells were treated with crizotinib orverapamil for 3 hours at concentrations indicated, and250 ng/ml of rhodamine 123 was added for an additional2 hours followed by analysis with flow cytometry. Crizo-tinib was obtained commercially (Selleck), dissolved indimethyl sulfoxide (DMSO), and used at concentrations andtime intervals as indicated.

2.4. Small interfering RNA

Synthetic RNA duplexes were obtained commer-cially (Qiagen), and cells cultured in 60-mm dishes with2 ml of antibiotic-free media were transfected with 12 μl of

20 μM annealed RNA duplexes. For MET knockdown,the following sequences were used: sense 50-GCCA-AUUUAUCAGGAGGUGTT-30 and antisense 50-CACCU-CCUGAUAAAUUGGCTT-30. The HiPerFect transfectionreagent (Qiagen) was used as recommended by the manu-facturer. Knockdown efficiency was monitored by immuno-blotting.

2.5. Quantitative PCR

Detection of P-gp messenger RNA (mRNA) expressionwas performed as previously described [9]. The followingprimers were used: forward—50-TTGCTGCTTACATT-CAGGTTTCA-30 and reverse—50-AGCCTATCTCCTGT-CGCATTA-30 (Integrated DNA Technologies). Caki-2cells (American Type Culture Collection) were used aspositive control and maintained as recommended by thedistributor.

M. Schneider et al. / Urologic Oncology: Seminars and Original Investigations 32 (2014) 877–884880

2.6. Statistical analysis

We used the Student 2-tailed t test for independentsamples wherever applicable (GraphPad Prism). P r 0.05was considered significant.

3. Results

3.1. Identification of crizotinib as an active compound in apatient-derived primary ccRCC cell line

We identified the MET-ALK inhibitor crizotinib [10] asthe top hit from 101 compounds used at 2 differentconcentrations in our screening protocol (Figs. 1 and 2).Crizotinib was found to induce a 7.7-fold increase ofapoptosis in RXF2209L cells when compared with controls(Fig. 2A). There was no significant increase in apoptosis innormal renal epithelial cells. The next most effective drugswere bortezomib (1 μM) with a 3.3-fold induction ofapoptosis and plicamycin (1 μM) with a 3.1-fold induction

Fig. 2. Identification of crizotinib as a hit in a phenotypic drug screen. (A) Fold apcell line RXF2209L treated with various compounds selected from the drug-scCaspase-Glo 3/7 assay was used as a readout. (B) Absence of phospho-mTOR Scancer cells as shown by immunoblot analysis. (C) Measurement of cell viabilityassay and normalized to DMSO-treated controls. Asterisks indicate a significan***P o 0.0001; the Student 2-tailed t test for independent samples). (D) ImmunRXF2209L cells treated with crizotinib at concentrations and time intervals as ind48 hours' treatment can be noted. DMSO ¼ dimethyl sulfoxide.

of apoptosis. Both the drugs were excluded because ofsignificant toxicity in normal renal epithelial cells. Surpris-ingly, established first- or second-line systemic RCC ther-apeutics that were part of the drug library did not inducesignificant cell death (Fig. 2A). The lack of activity ofvascular endothelial growth factor–targeted agents, despitethe fact that some RCCs are known to express vascularendothelial growth factor receptors, is most likely explainedby the absence of an endothelial cell compartment in vitro[11] and does not rule out in vivo activity as the patientachieved temporary disease stabilization when treated withsunitinib. The inefficiency of everolimus can be explained bythe fact that activated mTOR was not detectable in the tumor-derived cell line despite focal mTOR activation in theprimary tumor (Figs. 2B and 1A). Apoptosis induction bycrizotinib was confirmed by an MTT cell viability assay andimmunoblot analysis for PARP cleavage (Fig. 2C and D).Induction of PARP cleavage was seen at a 1 μM concen-tration of crizotinib, which is in the higher range of clinicallyachievable plasma concentrations of the drug [12]. However,intratumoral crizotinib concentrations have been reported to

optosis induction normalized to DMSO in the patient-derived primary RCCreening panel. Compounds were used at a 10 μM concentration and the2448 expression in the primary cell line when compared with HCC78 lungof the RXF2209L line treated with crizotinib as indicated using the MTT

t reduction when compared with controls (*P o 0.05, **P o 0.0005, andoblot analyses of apoptosis induction as measured by PARP cleavage inicated. The onset of apoptosis between 0.1 and 1 μM of crizotinib and 24 to

M. Schneider et al. / Urologic Oncology: Seminars and Original Investigations 32 (2014) 877–884 881

be significantly higher (up to 100-fold of the in vitro IC50)[13], which suggest that concentrations in the 10 μM rangecan realistically be achieved in the actual tumor tissue.

3.2. Identification of a MET sequence variant in RXF2209Lcells

Crizotinib is approved for the treatment of ALK-rearranged non–small cell lung cancer [14]. We thereforesought next to identify the crizotinib target in the patient-derived ccRCC cell line. The cell line showed an expressionof MET protein (Fig. 3A) but not of ALK or ROS-1, 2additional crizotinib targets, by immunoblot analysis(Fig. 3A). To rationalize the significant cytotoxicity inducedby crizotinib, a Sanger sequencing analysis was performed.A heterozygous R988C missense mutation of the MET gene(also designated R970C) was detected in both the primarytumor and the tumor-derived cell line (Fig. 3B) [15]. Inaddition, a deletion in exon 2 of the VHL tumor-suppressorgene was detected in both the primary tumor and thecorresponding cell line. The R988C mutation affects thejuxtamembrane domain of the MET receptor tyrosine kinaseand has been shown to function as an activating mutationwith oncogenic properties in certain tumor entities, includ-ing lung cancer [15,16]. However, MET R988C failed toshow kinase-activating functions and transformation in adifferent study [17], suggesting that it can also represent asequence variant.

Therefore, we next asked whether the R988C mutationled to MET activation in our cell line by measuring METphosphorylation [18]. Phosphorylated MET was undetect-able in the patient-derived cell lines when a phosphospecific

Fig. 3. Expression of crizotinib targets and MET mutational status. (A) Immuncompared with that of MET-positive H4441 lung cancer cells, ALK-rearranged H2presence of MET protein expression when ALK and ROS-1 expression was nottumor and the corresponding cell line RXF2209L shows a MET R988C mutatioavailable online.)

antibody against MET Y1234/1235 and a phosphotyrosineantibody were used (Fig. 4A and B). We next tested theeffect of small interfering RNA-mediated MET knockdownon apoptosis and cell viability. Depletion of MET did notcause increased apoptotic cell death as measured by PARPcleavage (Fig. 4C), and no reduction of cell viability wasdetected as measured by the MTT assay (Fig. 4D). Knock-down of the related RON kinase, which is also a crizotinibtarget [19], also had only a modest effect on cell viability(data not shown).

Taken together, we were not able to corroborate a role ofone of the known crizotinib targets MET, ALK, ROS-1, orRON in the drug effects observed. These findings make itlikely that MET R988C represents a sequence variant ratherthan an oncogenic driver mutation in RXF2209L cells, anotion that is supported by previous findings [17].

3.3. Inhibition of P-gp-mediated efflux correlates with thecytotoxic activity of crizotinib

Crizotinib has previously been shown to block thefunction of the multidrug-resistance transporter P-gp(ABCB1) [12]. This function was found to be independentfrom its function as a kinase inhibitor but was shared withother TKIs [12].

We tested whether crizotinib can block the efflux of thefluorescent dye rhodamine 123 in RXF2209L cells by flowcytometry. As shown in Fig. 5A, crizotinib was found tocause an intracellular rhodamine 123 accumulation at boththe concentrations—1 and 10 μM. However, when we usedthe calcium channel blocker verapamil to block P-gp, wedid not detect an increase of the intracellular accumulation

oblot analysis of MET, ALK, and ROS-1 expression in RXF2209L cells228 lung cancer cells, and ROS-1-rearranged HCC78 lung cancer cells. Thedetected can be noted. (B) Sanger sequencing analysis of both the primaryn corresponding to the juxtamembrane domain. (Color version of figure is

Fig. 4. MET R988C is not an oncogenic driver mutation in RXF2209L cells. (A) Immunoblot analysis for phospho-MET Y1234/1235 using a panel of 17primary tumor-derived renal cancer cell lines (Oncotest, Freiburg, Germany) all in comparison with RXF2209L cells (red). Immunoprecipitation (IP) of METwas performed in addition for RXF2209L cells and RXF1114 cells. The absence of detectable MET phosphorylation in RXF2209L cells can be noted.(B) MET immunoprecipitation followed by immunoblot analysis of RXF2209L cells for MET and phosphotyrosine expression. The absence of detectableMET phosphorylation can be noted. (C) Immunoblot analysis of apoptosis induction in RXF2209L cells after siRNA-mediated knockdown of MET (siMET).The absence of enhanced PARP cleavage can be noted. (D) Measurement of cell viability of the RXF2209L line treated with crizotinib as indicated using theMTT assay and normalized to control siRNA-transfected cells. siRNA ¼ small interfering RNA. (Color version of figure is available online.)

M. Schneider et al. / Urologic Oncology: Seminars and Original Investigations 32 (2014) 877–884882

of rhodamine 123 at a concentration of 10 μM in compar-ison with that of control cells (Fig. 5A). Although thisverapamil concentration may be too low for effectivelyblocking P-gp-mediated efflux, our cell viability assaysclearly show that the inhibition of P-gp function correlates

Fig. 5. Crizotinib-associated blockade of P-glycoprotein (P-gp)–mediated transpoassays show intracellular accumulation of the fluorescent dye (arrows) in crizotinibat a 10-μM concentration. (B) Measurement of cell viability of the RXF2209L celindicated concentrations. Asterisk indicates a significant reduction when comparedexpression in the primary tumor. The scale bar indicates 50 μm. DMSO ¼ dim

with cytotoxicity in RFX2209L cells (Fig. 5B). Expressionof P-gp in RXF2209L cells was confirmed by qPCR andwas found to be in the same range as in highly multidrug-resistant and P-gp-overexpressing Caki-1 renal cancer cells(not shown) [20]. We also analyzed P-gp expression in the

rt correlates with cytotoxicity. (A) Flow cytometric rhodamine 123 efflux-treated RXF2209L cells but not in controls and cells treated with verapamill line treated with DMSO as solvent control or crizotinib or verapamil at thewith controls (***P o 0.0001). (C) Immunohistochemical analysis of P-gp

ethyl sulfoxide. (Color version of figure is available online.)

M. Schneider et al. / Urologic Oncology: Seminars and Original Investigations 32 (2014) 877–884 883

primary tumor and found a positive staining result byimmunohistochemistry, which was most pronounced inthe peripheral zone (Fig. 5C).

4. Discussion

Our report highlights the complexities of phenotype-genotype correlations in personalized cancer medicine asexemplified by a patient with advanced ccRCC.

Our results have a number of ramifications and limita-tions. First, phenotypic drug screening can result in anidentification of unexpected active compounds (here: activityof a MET-ALK inhibitor in a nonpapillary RCC cell line) forwhich none of the known major tyrosine kinase targets(MET, ALK, ROS-1, and RON) can be corroborated. In thecase of our primary tumor-derived cell line, this scenario waseven more complex as a sequence alteration of the crizotinibtarget MET was found, which has previously been suggestedas an oncogenic driver mutation in certain tumor entities[15,16]. However, we were unable to show MET activationor increased apoptosis/cytotoxicity of our cell line followingMET knockdown. We conclude that the MET R988Cmutation likely represents a sequence variant as previouslysuggested by others and does not function as a majoroncogenic driver in our tumor-derived primary cell line[17]. We cannot rule of the possibility that the R988C allelecontributes to tumor development in an indirect way. None-theless, our findings underscore that a careful functionalvalidation of somatic mutations with potential oncogenicdriver activity is indispensable. Like other TKIs, crizotinibhas been reported to inhibit the multidrug-resistance trans-porter P-gp [12]. Our results confirm this finding in ourpatient-derived cell line and furthermore demonstrate acorrelation between P-gp inhibition and cytotoxicity. In theabsence of a chemotherapeutic drug, P-gp inhibition bycrizotinib could either lead to an intracellular accumulationof, for example, toxic metabolites or enhance the intracellularcrizotinib concentration as this drug is, like many others,both a substrate and a blocker of P-gp [21].

Second, the establishment of a primary, tumor-derivedcell line is unlikely to be achievable for most patients. In ourstudy, the patient succumbed to metastatic disease only 10months after diagnosis. It is likely that this highly aggressivetumor behavior was instrumental for the successful establish-ment of a patient-derived cell line. Moreover, results fromour study were obtained from a single patient and may notbe generalizable for most patients with ccRCC.

On a general note, more advanced disease yields highersuccess rates of cell line establishment. At the same time,timing is crucial for patients with rapidly progressingtumors as serial passages, drug-screening procedures, andvalidation experiments typically take several months. Thisraises the question of the level of scientific stringency forphenotypic drug screens, i.e., should such a screen be usedfor clinical decision making without the long and complex

process of target identification and validation, whichcommonly requires chemical proteomics. There are also anumber of regulatory concerns as drug screens can obvi-ously show activities of compounds that are not approvedfor the use in the respective tumor entity. The samedilemma applies to combination therapies that are likelyto become more relevant in the future.

There is an ongoing debate whether and to what extent atumor-derived cell line is able to reflect intratumoralgenomic heterogeneity. We concur that secondary in vitroselection processes need to be taken into consideration. Inour cell line, no expression of activated mTOR wasdetectable although the primary tumor showed focal pos-itivity, thus highlighting such selection processes. It isnonetheless conceivable that in vitro culturing promotesthe outgrowth of highly proliferative and hence clinicallyrelevant, tumor-propagating cells.

Taken together, phenotypic drug screening can be usefulfor the rapid identification of active antitumor compoundsin particular when combined with a patient-derived cell lineor a short-term culture. Given that only a single patient wasexamined, however, we believe that a clinical trial would bepremature at this point of time. Nevertheless, mousexenograft experiments have shown antitumor activity ofcrizotinib in an RCC model [22]. How to approach asituation in which a drug target is uncommon or remainselusive remains a major challenge and could involve deepsequencing analyses. In the context of genomic cancermedicine, our results underscore that proper validationexperiments to differentiate between oncogenic drivermutations and sequence variants with little or no role intumor growth are mandatory.

5. Conflict of interest

J.S. is a salaried employee of Oncotest GmbH, Freiburg,Germany. This employment does not alter our adherence toall the journal policies on sharing data and materials. Allexperiments were approved by the local ethics committeewith written informed consent. The authors declare nocompeting financial interests.

Acknowledgments

This work was supported by the Medical FacultyHeidelberg of the University of Heidelberg School ofMedicine. M.S. was supported by a short-term fellowshipfor clinician-scientists of the Medical Faculty Heidelberg.N.K. is supported by a postdoctoral fellowship from theMedical Faculty Heidelberg. R.H. is supported by adissertation stipend from the Medical Faculty Heidelberg.We would like to thank Anne Löhr for the excellenttechnical help and support and the flow cytometry corefacility of the German Cancer Research Center for instru-ment use and technical support.

M. Schneider et al. / Urologic Oncology: Seminars and Original Investigations 32 (2014) 877–884884

References

[1] Motzer RJ, Hutson TE, Tomczak P, Michaelson MD, Bukowski RM,Rixe O, et al. Sunitinib versus interferon alfa in metastatic renal-cellcarcinoma. N Engl J Med 2007;356:115–24.

[2] Haddad H, Rini BI. Current treatment considerations in metastaticrenal cell carcinoma. Curr Treat Options Oncol 2012;13:212–29.

[3] Figlin R, Sternberg C, Wood CG. Novel agents and approaches foradvanced renal cell carcinoma. J Urol 2012;188:707–15.

[4] Molina AM, Motzer RJ, Heng DY. Systemic treatment options foruntreated patients with metastatic clear cell renal cancer. Semin Oncol2013;40:436–43.

[5] Schmidt L, Duh FM, Chen F, Kishida T, Glenn G, Choyke P, et al.Germline and somatic mutations in the tyrosine kinase domain of theMET proto-oncogene in papillary renal carcinomas. Nat Genet1997;16:68–73.

[6] Gerlinger M, Rowan AJ, Horswell S, Larkin J, Endesfelder D,Gronroos E, et al. Intratumor heterogeneity and branched evolutionrevealed by multiregion sequencing. N Engl J Med 2012;366:883–92.

[7] Boichuk S, Lee DJ, Mehalek KR, Makielski KR, Wozniak A,Seneviratne DS, et al. Unbiased compound screening identifiesunexpected drug sensitivities and novel treatment options for gastro-intestinal stromal tumors. Cancer Res 2014;74:1200–13.

[8] Korzeniewski N, Zheng L, Cuevas R, Parry J, Chatterjee P, AndertonB, et al. Cullin 1 functions as a centrosomal suppressor of centriolemultiplication by regulating polo-like kinase 4 protein levels. CancerRes 2009;69:6668–75.

[9] Korzeniewski N, Hohenfellner M, Duensing S. CAND1 promotesPLK4-mediated centriole overduplication and is frequently disruptedin prostate cancer. Neoplasia 2012;14:799–806.

[10] Cui JJ, Tran-Dube M, Shen H, Nambu M, Kung PP, Pairish M, et al.Structure based drug design of crizotinib (PF-02341066), a potent andselective dual inhibitor of mesenchymal-epithelial transition factor(c-MET) kinase and anaplastic lymphoma kinase (ALK). J MedChem 2011;54:6342–63.

[11] Huang D, Ding Y, Li Y, Luo WM, Zhang ZF, Snider J, et al. Sunitinibacts primarily on tumor endothelium rather than tumor cells to inhibit thegrowth of renal cell carcinoma. Cancer Res 2010;70:1053–62.

[12] Zhou WJ, Zhang X, Cheng C, Wang F, Wang XK, Liang YJ, et al.Crizotinib (PF-02341066) reverses multidrug resistance in cancer

cells by inhibiting the function of P-glycoprotein. Br J Pharmacol2012;166:1669–83.

[13] Avan A, Caretti V, Funel N, Galvani E, Maftouh M, Honeywell RJ,et al. Crizotinib inhibits metabolic inactivation of gemcitabinein c-Met-driven pancreatic carcinoma. Cancer Res 2013;73:6745–56.

[14] Kwak EL, Bang YJ, Camidge DR, Shaw AT, Solomon B, Maki RB,et al. Anaplastic lymphoma kinase inhibition in non-small-cell lungcancer. N Engl J Med 2010;363:1693–703.

[15] Ma PC, Kijima T, Maulik G, Fox EA, Sattler M, Griffin JD, et al.c-MET mutational analysis in small cell lung cancer: novel juxta-membrane domain mutations regulating cytoskeletal functions. Can-cer Res 2003;63:6272–81.

[16] Ma PC, Jagadeeswaran R, Jagadeesh S, Tretiakova MS, Nallasura V,Fox EA, et al. Functional expression and mutations of c-Met and itstherapeutic inhibition with SU11274 and small interfering RNA innon-small cell lung cancer. Cancer Res 2005;65:1479–88.

[17] Tyner JW, Fletcher LB, Wang EQ, Yang WF, Rutenberg-SchoenbergML, Beadling C, et al. MET receptor sequence variants R970C andT992I lack transforming capacity. Cancer Res 2010;70:6233–7.

[18] Jeffers M, Schmidt L, Nakaigawa N, Webb CP, Weirich G, KishidaT, et al. Activating mutations for the met tyrosine kinase receptor inhuman cancer. Proc Natl Acad Sci U S A 1997;94:11445–50.

[19] Benvenuti S, Lazzari L, Arnesano A, Li Chiavi G, Gentile A,Comoglio PM, et al. Ron kinase transphosphorylation sustains METoncogene addiction. Cancer Res 2011;71:1945–55.

[20] Yu DS, Chang SY, Ma CP. The expression of mdr-1-related gp-170and its correlation with anthracycline resistance in renal cell carci-noma cell lines and multidrug-resistant sublines. Br J Urol 1998;82:544–7.

[21] Chuan Tang S, Nguyen LN, Sparidans RW, Wagenaar E, Beijnen JH,Schinkel AH, et al. Increased oral availability and brain accumulationof the ALK inhibitor crizotinib by coadministration of theP-glycoprotein (ABCB1) and breast cancer resistance protein(ABCG2) inhibitor elacridar. Int J Cancer 2014;134:1484–94.

[22] Zou HY, Li Q, Lee JH, Arango ME, McDonnell SR, Yamazaki S,et al. An orally available small-molecule inhibitor of c-Met, PF-2341066, exhibits cytoreductive antitumor efficacy through antipro-liferative and antiangiogenic mechanisms. Cancer Res 2007;67:4408–17.