Dependence of Wilms tumor cells on signaling throughinsulin-like growth factor 1 in an orthotopic xenograftmodel targetable by specific receptor inhibitionAleksandra Bielena, Gary Boxb, Lara Perrymana, Lynn Bjerkea, Sergey Popova,c, Yann Jamind, Alexa Jurya,Melanie Valentib, Alexis de Haven Brandonb, Vanessa Martinsb, Vincent Romanete, Sebastien Jeaye,Florence I. Raynaudb, Francesco Hofmanne, Simon P. Robinsond, Suzanne A. Ecclesb, and Chris Jonesa,c,1

aDepartment of Pediatric Oncology and bCancer Research United Kingdom Cancer Therapeutics Unit, Institute of Cancer Research, Sutton SM2 5NG, UnitedKingdom; cDepartment of Pediatric Oncology, Royal Marsden Hospital, Sutton SM2 5PT, United Kingdom; dCancer Research United Kingdom and Engineeringand Physical Sciences Research Council Cancer Imaging Centre, Institute of Cancer Research and Royal Marsden National Health Service Foundation Trust,Sutton SM2 5NG, United Kingdom; and eNovartis Pharma, CH-4002 Basel, Switzerland

Edited by Dennis A. Carson, University of California, San Diego, La Jolla, CA, and approved March 28, 2012 (received for review March 29, 2011)

We have previously demonstrated an increased DNA copy numberand expression of IGF1R to be associated with poor outcome inWilms tumors. We have now testedwhether inhibiting this receptormay be a useful therapeutic strategy by using a panel of Wilmstumor cell lines. Both genetic and pharmacological targeting resultedin inhibition of downstream signaling through PI3 and MAP kinases,G1 cell cycle arrest, and cell death, with drug efficacy dependent onthe levels of phosphorylated IGF1R. These effects were further asso-ciated with specific gene expression signatures reflecting pathwayinhibition, and conferred synergistic chemosensitisation to doxoru-bicin and topotecan. In the in vivo setting, s.c. xenografts of WiT49cells resembled malignant rhabdoid tumors rather than Wilmstumors. Treatment with an IGF1R inhibitor (NVP-AEW541) showedno discernable antitumor activity and no downstream pathway in-activation. By contrast, Wilms tumor cells established orthotopicallywithin the kidney were histologically accurate and exhibited signif-icantly elevated insulin-like growth factor–mediated signaling, andgrowthwas significantly reducedon treatmentwithNVP-AEW541 inparallel with signaling pathway ablation. As a result of the paracrineeffects of enhanced IGF2 expression in Wilms tumor, this diseasemay be acutely dependent on signaling through the IGF1 receptor,and thus treatment strategies aimed at its inhibitionmaybeuseful inthe clinic. Such efficacy may bemissed if only standard ectopic mod-els are considered as a result of an imperfect recapitulation of thespecific tumor microenvironment.

nephroblastoma | MAPK | PI3K

Wilms tumor is an embryonal renal tumor accounting fornearly 6% of all pediatric cancers and more than 90% of

kidney tumors in children (1). It is the fourthmost common type ofsolid tumor in children, usually presenting between the ages of 2 yand 4 y. During the past 40 y, improvement of treatment protocolshas raised survival rates from 30% to more than 90%, and treat-ment of this tumor has become a paradigm for successful cancertherapy (1). However, approximately 15% of patients with favor-able histology, and 50% of patients with anaplastic Wilms tumor,will experience recurrence, and, for these patients, the survivalrate is closer to 60% (2).Work in our laboratory using microarray-based comparative

genomic hybridization revealed a significant correlation betweenincreased copy number at the IGF1R locus at chromosome15q26.3 and Wilms tumor relapse (3). Approximately 10% ofWilms tumors exhibited a low-level gain corresponding to three orfour copies of the gene by microarray-based comparative genomichybridization analysis, with tumors exhibiting increased copies andmRNA/protein overexpression having a significantly worse out-come (3). This was independent of specific histological subtype,although an increased prevalence of IGF1R aberrations werenoted in tumors with anaplastic vs. favorable histology.

We hypothesized that paracrine activation of IGF1R by IGF2,produced in large amounts by the tumor as a result of loss ofheterozygosity or loss of imprinting (4, 5), may result in an in-creased mitogenic/antiapoptotic action through PI3-kinase/Akt/S6K and/or Ras/MAP kinase signaling pathways. Novel thera-peutic strategies targeting the IGF1 receptor might therefore bebeneficial in patients with anaplastic/relapsedWilms tumor, aloneor in combination with existing chemotherapeutic agents.Insulin-like growth factor (IGF) signaling has become an at-

tractive target for novel cancer therapeutic strategies because ofits crucial role in regulating cancer cell proliferation and survival(6). Numerous experimental approaches have been used to inhibitIGF1R signaling, including dominant-negative mutants, antisenseoligonucleotides, soluble IGFBPs, antagonistic and/or neutraliz-ing antibodies, and small-molecule kinase inhibitors (7). The mostadvanced strategies are those involving anti-IGF1R antibodiesand small-molecule inhibitors, with both in vitro and in vivo effi-cacy demonstrated in a range of tumor types, including numerouschildhood cancer models (8).A major hindrance to the development of novel therapeutic

strategies in Wilms tumor has been the lack of appropriate modelsystems, both in vitro and in vivo, for mechanistic and preclinicalstudies, with few well-characterized cell lines available. Most re-cently, the Pediatric Preclinical Testing Program has developeda small number of Wilms tumor cell lines (one anaplastic, threenonanaplastic) as s.c. xenografts for the screening of a variety ofchemotherapeutic and targeted agents in the wider context ofchildhood cancer (9).The present study sought to assess the efficacy of strategies

targeting the IGF1 receptor in a panel of Wilms tumor cell lines,and to determine the major determinants of response to geneticand pharmacologic inhibition. In particular, we have addressed thelack of clinically relevant models available by xenografting Wilmstumor cells in the kidney of immunocompromised mice, and in-vestigated the differential signaling activation, and thus thera-peutic response, comparedwith standard ectopic (i.e., s.c.)models.

Author contributions: A.B., S.J., F.H., S.A.E., and C.J. designed research; A.B., G.B., L.P.,L.B., S.P., Y.J., A.J., M.V., A.d.H.B., V.M., V.R., F.I.R., and S.P.R. performed research; G.B.,F.H., and S.A.E. contributed new reagents/analytic tools; A.B., G.B., L.P., L.B., S.P., Y.J., A.J.,M.V., A.d.H.B., V.M., V.R., S.J., F.I.R., S.P.R., S.A.E., and C.J. analyzed data; and A.B., F.H.,S.P.R., S.A.E., and C.J. wrote the paper.

Conflict of interest statement: V.R., S.J., and F.H. are employees of Novartis Pharma.

This article is a PNAS Direct Submission.

Data deposition: The sequence reported in this paper has been deposited in the ArrayEx-press database, www.ebi.ac.uk/arrayexpress/ (accession no. E-TABM-891).1To whom correspondence should be addressed. E-mail: chris.jones@icr.ac.uk.

See Author Summary on page 7604 (volume 109, number 20).

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1105034109/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1105034109 PNAS | Published online April 23, 2012 | E1267–E1276

MED

ICALSC

IENCE

SPN

ASPL

US

Dow

nloa

ded

by g

uest

on

Janu

ary

6, 2

020

ResultsIn Vitro Sensitivity of Wilms Tumor Cells to NVP-AEW541 Is IGF1R-Dependent. To screen the in vitro efficacy of the small moleculeIGF1R inhibitor NVP-AEW541 in Wilms tumor cells, we as-sembled a panel of cell lines (17.94, WT-CLS1, CCG 99–11,STA2A, and WiT49), as well as human embryonic kidneyHEK293 cells, and fibroblast lines engineered to have high (R+)or no (R−) IGF1R expression for comparison. DifferentialIGF1R protein expression was observed across the lines, withparticularly high levels noted in 17.94, and an absence of receptorinWT-CLS1 as determined byWestern blot (Fig. 1A). By contrast,WT-CLS1 cells expressed the closely related insulin receptor,common to all lines with the exception of 17.94 (Fig. 1A). Higher

levels of phosphorylated IGF1R were detected in WiT49 and17.94 cells compared with the very low levels found in the otherWilms tumor lines, as assessed by the electrochemiluminescentMeso Scale Discovery (MSD) System assay (Fig. 1B).In vitro sensitivity to NVP-AEW541 appeared to correlate with

the levels of IGF1R/phospho-IGF1R, with the most sensitive lines(WiT49 and 17.94) having the highest levels of constitutive ex-pression and concentration needed to reduce the growth of trea-ted cells to 50% of that of untreated cells (GI50) of 0.71 μM and0.75 μM, respectively, even lower than the R+ cells (1.10 μM; Fig.1C). The least sensitive line was WT-CLS1 (5.10 μM), in whichIGF1R expression was undetectable by Western blot (Fig. 1C).GI50s for CCG 99–11 (2.52 μM) and STA2A (1.41 μM) cells werebetween these values (Fig. 1D). Culturing WiT49 cells in the

Fig. 1. IGF1R as a therapeutic target in Wilms tumor cell lines. (A) Western blot for IGF1R protein expression in a panel of Wilms tumor cell lines in vitro. Alsoincluded for comparison are the IGF1R-null (R−) and -overexpressing (R+) fibroblast cell lines, as well as blots for the homologous insulin receptor (IR). β-Actinis used as a loading control. (B) Quantitative measure of phosphorylated IGF1R in Wilms tumor cell lines assessed by the MSD assay. (C) Effects on cell survivalof treatment with Wilms tumor cells with the IGF1R inhibitor NVP-AEW541, in comparison with the resistant R− and sensitive R+ cells. (D) GI50 values for celllines treated with NVP-AEW541 as assessed by the MTS assay. Mean and range (in μM) for triplicate experiments are given. (E) Increased downstream sig-naling although PI3 and MAP kinase pathways in WiT9 cells in the presence of IGF2 assessed by Western blot analysis of phospho-Akt and phospho-Erk1/2. (F)NVP-AEW541 induces a concentration-dependent reduction in cell survival in WiT49 cells in the absence and presence of IGF2. Cells were treated with 1× and3× GI50 concentrations of NVP-AEW541 in the absence of IGF2, and in serum-free medium including 50 ng/mL IGF2. (***P < 0.001 and **P < 0.01, t test).

E1268 | www.pnas.org/cgi/doi/10.1073/pnas.1105034109 Bielen et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

6, 2

020

presence of the IGF1R ligand IGF2 resulted in an increase ofphospho-Akt and phospho-Erk1/2 levels, indicating activation ofPI3 and MAP kinase pathways (Fig. 1E). Cell survival was di-minished in a concentration-dependentmanner byNVP-AEW541both in the absence and presence of IGF2 (Fig. 1F).

IGF1R Inhibition Results in Cell Cycle Arrest, Apoptosis, and Chemosensi-tization via Down-Regulation of PI3 and MAP Kinase Pathways in WilmsTumor Cells in Vitro. To assess the effects on downstream signalingof IGF1R abrogation in Wilms tumor cells, we compared geneticknockdown by specific siRNA with pharmacological inhibition byNVP-AEW541. siRNA directed against IGF1R in WiT49 cellsresulted in efficient reduction of IGF1R (Fig. 2A) along with

inhibition of phospho-Akt and induction of apoptosis as mea-sured by PARP cleavage (Fig. 2B); no effect on phospho-Erk1/2levels were observed. A profound cell cycle arrest at G1 was seenby FACS analysis after 48 h (Fig. 2C). These observations werereplicated in the absence and presence of IGF2 (Fig. 2D). Sucheffects on protein expression (Fig. 2E) and cell cycle (Fig. 2F)were mimicked by NVP-AEW541 in WiT49 cells in a concentra-tion- and time-dependent manner, with the additional inhibitionof phospho-Erk1/2 levels. NVP-AEW541 blocked IGF2-inducedPI3-kinase activation as detected by loss of phospho-Akt after 1 hof treatment and induced apoptosis as seen by PARP and caspase3 cleavage after 24 h in WiT49 cells. Levels of phospho-Erk1/2

Fig. 2. Effects of genetic and pharmacological targeting of IGF1R on downstream signaling in Wilms tumor cells. (A) Relative IGF1R mRNA expression inWiT49 cells transfected with siRNA targeting the gene, as determined by quantitative RT-PCR (***P < 0.001, t test). (B) Western blots demonstrate efficientknockdown of IGF1R protein in association with diminished phospho-Akt and induced PARP cleavage in WiT49 cells transfected with IGF1R siRNA. (C) FACSanalysis of siRNA-transfected WiT49 cells vs. scrambled control oligos demonstrate an accumulation of cells in G1 and sub-G1 phases in contrast to a reductionof S and G2 phases. (D) Western blots confirming the knockdown of IGF1R, reduction in phospho-Akt and phospho-Erk1/2, and induction of PARP cleavageafter treatment of WiT49 cells with IGF1R siRNA in the presence of the ligand IGF2. (E) Effects on downstream signaling in WiT49 cells after treatment withNVP-AEW541 in the presence of IGF2. Cells were treated for 1, 3, 6, 24, or 48 h with 1×, 3×, or 5× GI50 compound (increase in triangle). Treatment with IGF1Rinhibitor decreased phospho-Akt and phospho-Erk1/2 and induced PARP and caspase-3 cleavage in a time- and concentration-dependent manner. At 48 h, thehighest concentration of compound results in a significant cell death with little protein recoverable. β-Actin is used as a loading control. (F) Effects on cell cyclein WiT49 cells treated with NVP-AEW541. An accumulation of cells in G1 and sub-G1 phases is induced by NVP-AEW541 in a time- and concentration-de-pendent manner. At 48 h, the highest concentration of compound results in significant cell death.

Bielen et al. PNAS | Published online April 23, 2012 | E1269

MED

ICALSC

IENCE

SPN

ASPL

US

Dow

nloa

ded

by g

uest

on

Janu

ary

6, 2

020

were reduced after 3 h drug exposure. After 48 h, there was sig-nificant cell death and evidence of receptor degradation.To better understand the mechanism of action of NVP-

AEW541 in Wilms tumor cells, global gene expression profilingand pathway analysis on samples was performed after a time-course exposure to 5×GI50 concentration. Genes dysregulated byIGF1R inhibition in a time-dependent manner relative to vehiclecontrol included numerous genes associated with cell cycle pro-gression and DNA replication, including CCNA2, CCNB1, CKS2,

CDC20, E2F2, FEN1, MCM2, and CENPF (Fig. 3A). Examiningeffects on canonical pathway signaling using the Ingenuity Path-way Analysis tool, the two highest scoring pathways with dysre-gulated gene expression upon treatment with 5× GI50 NVP-AEW541 for 24 h were those associated with PI3K/AKT (P <0.0001) and ERK/MAPK signaling (P < 0.0001; Fig. 3B).Given the known chemosensitizing effects of IGF1R inhibition

in other cell systems showing similar effects on cell signaling andgene expression, we investigated the use of NVP-AEW541 in

Fig. 3. Treatment of WiT49 cells with NVP-AEW541 induces gene expression changes associated with cell cycle arrest and PI3K/MAPK down-regulation, andsensitizes the cells to chemotherapy. (A) Heat map demonstrates expression of genes associated with cell cycle progression in WiT49 cells treated with 5× GI50 ofNVP-AEW541 for 1, 6, and 24h. Genes are colored according to global, not relative, expression values (blue, down-regulated; red, up-regulated). Top-ranking geneset enrichment analysis scores for GNF2_CCNA2 [enrichment score = −0.89, P < 0.00001, false discovery rate (FDR) q < 0.00001] GNF2_CKS2 (enrichment score=−0.88,P< 0.00001, FDRq< 0.00001), andGNF2_CENPF (enrichment score =−0.85, P< 0.00001, FDRq< 0.00001) highlights coordinate down-regulation of genes inthe expression neighborhood of these cell cycle control genes. (B) Top-ranking canonical pathways identified by Ingenuity Pathway Analysis after treatment ofWiT49 cells with NVP-AEW541 were those associated with PI3K/AKT (P < 0.0001) and ERK/MAPK signaling (P < 0.0001). Pathway components are colored by theirglobal gene expression levels (green, down-regulated; red, up-regulated). (C) Median effects analysis of combining NVP-AEW541 with chemotherapeutic agentsin WiT49 cells. Highly synergistic interactions are seen for doxorubicin (combination index = 0.43) and topotecan (combination index = 0.39).

E1270 | www.pnas.org/cgi/doi/10.1073/pnas.1105034109 Bielen et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

6, 2

020

combination with clinically relevant chemotherapeutics in Wilmstumor cells. Combined treatment of NVP-AEW541 with doxo-rubicin or topotecan gave rise to highly synergistic interactions invitro, with combination indices of 0.43 and 0.39, respectively,using median effects analysis (Fig. 3C).

NVP-AEW541 Inhibits Tumor Growth in an Orthotopic, but Not s.c.,Wilms Tumor Xenograft Model. Initially, WiT49 cells were in-oculated s.c. in female athymic nude mice. Palpable tumors de-veloped within 7 d with a 100% incidence. At this point, animalswere treated with 25 mg/kg, 50 mg/kg, or 75 mg/kg NVP-AEW541,and their tumor growth rate compared with a vehicle-treatedcontrol group. No significant differences were observed in tumorvolume (Fig. 4A) or final tumor weight (Fig. 4B) upon treatmentwith the IGF1R inhibitor. This was despite pharmacokinetic as-sessment of drug concentration revealing the presence of 3.66 μM,16.79 μM, and 210.59 μM NVP-AEW541 within the tumors ofanimals treated with 25 mg/kg, 50 mg/kg, and 75 mg/kg, re-spectively, amounting to fivefold, 24-fold, and 30-fold higher thanthe values of the in vitro GI50 for WiT49 cells. Assessment ofpharmacodynamic endpoints by MSD demonstrated a significantinhibition of phospho-IGF1R (P< 0.001, t test) and phospho-IRS1

(P = 0.006) at the highest doses of compound, but no effect onphospho-Akt or phospho-Erk1/2 (Fig. 4C), the latter of which wasfound at particularly high constitutive levels. Histopathologicalassessment of the tumors revealed them to be composed of islandsand sheets of medium and large undifferentiated cells surroundedby delicate fibrovascular stroma. The cells had prominent nucleoliand abundant slightly eosinophilic cytoplasm, more typical of ma-lignant rhabdoid tumors of the kidney than classic Wilms tumors(Fig. 4D). A misidentification of the cell line was ruled out byscreening for SMARCB1 (INI1) mutation/deletion, which wasfound to be wild-type.We hypothesized that the specific microenvironment of the

tumor cells may be playing a significant role in the phenotypedemonstrated byWiT49 cells grown s.c., and also their response totargeted therapy.We therefore established an orthotopicmodel ofthese Wilms tumor cells by implantation into the renal subcapsuleof the left kidney in female athymic nude mice. Tumors weredetected by using ultrasound within 14 d, with a 100% take rate.Tumor cells grew deep into the renal parenchyma in a penetratingfashion reflecting a highly invasive phenotype (Fig. 5A). Largeareas with epithelial differentiation and prominent tubular for-mation were seen, and in some places a thin fibrotic pseudocap-

Fig. 4. Treatment with NVP-AEW541 shows no efficacy in an s.c. xenograft Wilms tumor model. (A) Tumor volumes in mice implanted with WiT49 cells s.c.and treated with 25 mg/kg, 50 mg/kg, or 75 mg/kg for 15 d. Tumor volume is plotted as percentage of day 0 for all groups, with no statistically significantdifferences compared with vehicle-treated controls. (B) Final tumor weights for the same experiment also show no differences between treated and un-treated groups. (C) Pharmacodynamic biomarkers assessed in the WiT49 s.c. model after treatment with NVP-EW541. Although a dose-dependent decrease inphospho-IGF1R and, to a lesser extent, phospho-IRS1 were observed, no significant inhibition of the downstream PI3 and MAP kinase pathways was seen. (D)Histology of the s.c. WiT49 model. The cells had prominent nucleoli and abundant slightly eosinophilic cytoplasm, with a morphological appearance moreclosely resembling malignant rhabdoid tumors of the kidney than classic Wilms tumors. [Original magnification of ×100 (inset, magnification of ×400).]

Bielen et al. PNAS | Published online April 23, 2012 | E1271

MED

ICALSC

IENCE

SPN

ASPL

US

Dow

nloa

ded

by g

uest

on

Janu

ary

6, 2

020

Fig. 5. Antitumor activity of NVP-AEW541 in an orthotopic Wilms tumor xenograft model. (A) Low-power histological section of WiT49 cells implanted in thekidney. Tumor cells demonstrated a highly invasive phenotype, growing deep into the renal parenchyma. (Original magnification of ×100.) (B) High-powerhistological section of the orthotopic WiT49 tumor. Large areas with epithelial differentiation and prominent tubular formation were seen, with some cellsexhibiting anaplastic features such as large cells, hyperchromatic nuclei, and atypical mitotic figures. (Original magnification of ×400.) (C) Significant re-duction in tumor volume of the orthotopic WiT49 model after treatment with NVP-AEW541 (**P < 0.01, t test). (D) Significant reduction in final kidneyweights of the orthotopic WiT49 model after treatment with NVP-AEW541. Tumor weights could not be directly assessed as a result of the infiltrative growthpatterns observed (**P < 0.01, t test). (E) Images of the kidney in vehicle-treated control and NVP-AEW541–treated mice with orthotopic WiT49 xenografts.Extensive growth of tumor cells is observed extrarenally and deep within the kidney in the control animals, with considerably less observed in mice treatedwith the IGF1R inhibitor. (F) Hydrogen 1 MRI of a representative vehicle control- and NVP-AEW541–treated mouse implanted orthotopically with WiT49 cells.Tumors appear hyperintense on T2-weighted images. Considerably less tumor burden is evident in animals treated with IGF1R inhibitor compared withvehicle controls, as indicated by the yellow arrow.

E1272 | www.pnas.org/cgi/doi/10.1073/pnas.1105034109 Bielen et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

6, 2

020

sule was detected. Cells mostly demonstrated a cohesive growthpattern, often arranged in trabeculae. Moreover, cells exhibitedanaplastic features such as large hyperchromatic nuclei andatypical mitotic figures (Fig. 5B). In total, WiT49 xenograftsestablished within the kidney closely recapitulated the histology ofanaplastic Wilms tumors. Similar morphological differences be-tween orthotopic and s.c. xenografts were observed for 17.94 andCCG 99–11 cells (Fig. S1); however, neither model was readilysuitable for in vivo efficacy assays. Orthotopically implanted CCG99–11 cells formed large, rapidly growing, and predominantlyextrarenal tumors, thus precluding study within the specific kidneymicroenvironment. Attempts to implant cells deeper within therenal parenchyma resulted in a significant lowering of the take rate(four of 24, 17%). The 17.94 cells were considerably slower-growing, and only four of eight (50%) s.c. implants formed tumors,even after enhancement, at more than 100 d. Orthotopically, smalltumors developed within the kidney after extensive optimizationof conditions with a similar efficiency (12 of 39, 31%) albeit withinthe shorter timeframe of 31 d. Because of the difficulty in estab-lishing these models efficiently in both sites, we thus focused ourefforts on WiT49 cells.Upon treatment with 50 mg/kg NVP-AEW541 for 14 d, ortho-

topic WiT49 tumors showed a significantly reduced final tumor

volume, as measured by MRI (P= 0.006, t test; Fig. 5C). Becauseof the infiltrative growth within the kidney, final organ weights ofvehicle- or drug-treated animals were compared with normal (i.e.,tumor-free) kidneys in age-matched mice. There was a significantreduction in final kidney weights with the NVP-AEW541–treatedmice compared with vehicle controls (P= 0.002, t test), with meankidney weights approaching those without established tumors (Fig.5D). In some cases, almost-complete regression of tumor wasnoted onmacroscopic examination (Fig. 5E) and byMRI (Fig. 5F).The pharmacokinetics of the orthotopic and s.c. models were

similar, with a drug concentration of 24.86 μM within the WiT49cells grown in the kidney. Pharmacodynamically, however, majordifferences were observed in response to equivalent doses of NVP-AEW541. In the kidney, tumor cells showed a reduction ofphosphorylated IGF1R levels of nearly 70% (Fig. 6A) comparedwith less than 30% when implanted s.c. (Fig. 4C), with attendantsignificant reductions in downstream signaling activation not ob-served in the heterotopic transplants. Significant reductions inphospho-IGF1R (P = 0.0059, t test), phospho-IRS1 (P = 0.045,t test), phospho-Akt (P= 0.031, t test), and phospho-Erk1/2 (P =0.0077, t test) were all seen on treatment with 50 mg/kg NVP-AEW541 in the orthotopic model. This is despite any confoundingcross-reactivity between mouse and human protein in the in-

Fig. 6. Pharmacodynamic biomarkers of response of an orthotopic Wilms tumor model treated with IGF1R inhibitor. (A) MSD assay analysis of relativephospho-IGF1R, phospho-IRS1, phospho-Akt, and phospho-Erk1/2 in WiT49 cells implanted in the kidney. All markers were significantly reduced on treatmentwith 50 mg/kg NVP-AEW541 (**P < 0.01 and *P < 0.05, t test). (B) Western blot of constitutive levels of phospho-/total protein in WiT49 cells grown in thekidney (“ortho”) compared with the equivalent s.c. (“subcut”) model and normal mouse kidneys. (C) Representative immunohistochemistry for phospho-IGF1R in the orthotopically grown WiT49 cells demonstrates high levels in the vehicle-treated controls and significant reduction on exposure to NVP-AEW541.

Bielen et al. PNAS | Published online April 23, 2012 | E1273

MED

ICALSC

IENCE

SPN

ASPL

US

Dow

nloa

ded

by g

uest

on

Janu

ary

6, 2

020

filtrative orthotopic models (Fig. 6B), as significant reduction ofactivated IGF1R on treatment with NVP-AEW541 could be vi-sualized by immunohistochemistry (Fig. 6C). Thus, the specificorthotopic microenvironment in which the cells were grownclearly exerts a significant effect on tumor response, mediated byenhanced abrogation of downstream signaling, rendering the cellsmore susceptible to targeted therapeutic inhibition.

DiscussionDespite a clear link between Wilms tumorigenesis and dysregu-lation of the IGF network, specific inhibition of this pathway hasnot been extensively assessed as a molecularly targeted thera-peutic strategy in this malignancy. This has in part been because ofa lack of well validated experimental Wilms tumor models. Forreasons that remain unclear, Wilms tumors appear very resistantto establishment as long-term cultures, and several early modelssuch as SK-NEP-1 and G401 have later been proven to be tumorsother than Wilms (10, 11). In the present study, we have useda small panel of more recently described Wilms tumor cell linesderived from primary in vitro culture [17.94 (12), WT-CLS1 (13),CCG 99–11 (14), STA2A (15)] or s.c. mouse xenografts of humanWilms tumor metastasis [WiT49 (16)].We addressed the lack of in vitro preclinical data by studying the

effect of targeting the IGF1R in these models by genetic andpharmacological means. We demonstrated that the small-mole-cule inhibitor NVP-AEW541 had a potent cytotoxic effect againstWilms tumor cells in vitro through down-regulation of PI3K andMAPK pathways, as well as cell cycle control genes such asCCNA2 and CCNB1, leading to a marked G1 arrest and sub-sequent induction of apoptosis in a caspase-dependent manner.These responses appeared well correlated with constitutiveIGF1R activation; although significant off-target effects cannot beruled out, they are not suggested by cell signaling and gene ex-pression assays. It thus seems likely that, in these relatively cyto-genetically stable tumors, a critical dependence is forced onIGF1R mediation of downstream signaling cascades, with theresult that monotherapy may effectively block multiple pathways.Given that we have previously shown copy number gain of

IGF1R to be associated with treatment failure and relapse inWilms tumor (3), coupled with paracrine activation of the re-ceptor by IGF2, also overexpressed via loss of heterozygosity orloss of imprinting (4, 5), IGF1R appears to represent a usefultherapeutic target whose inhibition may be beneficial, especially inpatients with relapsed and/or anaplastic Wilms tumor. Addition-ally, we provide evidence for a significant synergistic chemo-potentiation of the clinically relevant chemotherapeutic agentsdoxorubicin (2) and topotecan (17). Thus, for the broadest ap-plicability, IGF1R-targeting compounds may be combined withconventional chemotherapy and used in an upfront setting, po-tentially allowing for reduced long-term toxicity through lowercytotoxic drug doses (18).To assess whether the in vitro observations of IGF1R inhibition

could be replicated in the in vivo setting, we established Wilmstumor cells both s.c. as well as within the kidney. A previous studyusing the anti-IGF1Rmonoclonal antibody IMC-A12 showed littleefficacy against three s.c. Wilms tumor models (i.e., favorable his-tology) as part of the Pediatric Preclinical Testing Program initia-tive (19). No information regarding IGF1R status in these modelswas provided. Similarly a more recent study that used the smallmolecule BMS-754807 showedmodest activity in the samemodels,but again without correlativemolecular data (20). Our data provideclear evidence that cells grown in the appropriate kidney environ-ment not only more accurately recapitulate the histology of humanWilms tumors, but also have considerably enhanced response toabrogated IGF-mediated signaling than when implanted s.c.Ectopic models are widely used as they provide a simple, rapid

and reproducible means to evaluate emerging therapies. However,they do not imitate the natural tumor environment, omitting or-

gan-specific host–tumor interactions that may influence responseto therapy (21–24). This has previously been demonstrated in thekidney, in which levels of basic fibroblast growth factor were foundto be 10 to 20 times higher in renal cell carcinoma cells implantedorthotopically compared with s.c. (25). In the context of Wilmstumor, the predictive value of the IGF2-stimulated in vitromodels,in which critical dependence on IGF1R-mediated signaling wasalso evident, was only recapitulated in the orthotopic setting,calling into question the value of studies focusing only on ectopicmodels, in which IGF2 levels are low (26–28). Although our effi-cacy data were based, for practical reasons, on only a single cell linemodel, the phenotypic effects of the differing microenvironmentswere recapitulated in two additional cell lines, suggesting a gener-alizable conclusion. Of note, these cell lines were originally har-vested from extrarenal metastatic sites, which is clinically relevant,as recurrences tend not to occur in the kidney, as the patientshave undergone nephrectomy. Specific intracellular signaling de-pendencies associated with, e.g., cell growth in the lungmay also beconsidered as useful models better reflective of treatments aimedat the recurrent setting.A recently developed genetically engineered mouse model

combined loss of Wt1 function and Igf2 up-regulation to drivetumor initiation and proliferation through IGF1R signalingtransduced via phospho-IRS1 and phospho-ERK1/2, thus poten-tially providing a biologically relevant model for testing the effi-cacy of agents targeting the receptor (29). Although impressiveclinical results have been observed in pediatric sarcomas with anti-IGF1R monotherapy, in most tumor types, blockade of the re-ceptor is being considered in combination with additional ratio-nally designed agents (30). Although Wilms tumors may beparticularly dependent on IGF1R because of the relative lack ofgenomic complexity (31) and the high levels of constitutive MAPkinase activation observed in xenograft and transgenic models,and with a significant fraction of human Wilms tumors also dis-playing increased phospho-IRS1 and phospho-ERK1/2 (29),strategies aimed at coinhibition of MEK or mTOR may open upfurther yet-unexplored combination strategies in this disease.

Materials and MethodsCell Lines and Reagents. R− (i.e., IGF1R-null) andR+ (i.e., IGF1R-overexpressing)mouse fibroblasts were a gift from Renato Baserga (Thomas Jefferson Uni-versity, Philadelphia, PA). The human embryonic kidney HEK293 cell line wasobtained from the American Type Culture Collection, and the Wilms tumorWT-CLS1 cells were obtained from Cell Lines Service. TheWilms tumor cell lineWiT49was donated byHermanYeger (Hospital for Sick Children, Toronto,ON,Canada); STA2A by Peter Ambros (St. Anna Children’s Hospital, Vienna, Aus-tria); 17.94 by Keith W. Brown (University of Bristol, Bristol, United Kingdom);and CCG99-11 by Jonathan D. Licht (Robert H. Lurie Comprehensive CancerCenter, Chicago, IL). The IGF1R inhibitor NVP-AEW541 was provided byNovartis Pharma. Doxorubicin and topotecan were purchased from Sigma-Aldrich and Eurasia, respectively. IGF2 was purchased from Abcam.

Tumor Cell Growth Inhibition in Vitro. Growth inhibition following drug ex-posurewasdeterminedusingtheMTSassayaspreviouslydescribed (32). For theassessment of combination effects, cells were treated with increasing con-centrations of drugs alone or concurrently at their equipotentmolar ratio, andcombination indices were calculated by the method of Chou and Talaly (33).All values are given as mean ± SD of at least three independent experiments.

siRNA. Predesigned siRNA duplexes directed against IGF1R (catalog no.SI02624552 and SI00017521) were purchased from Qiagen. Cells were trans-fected with 5 nM IGF1R siRNAs as well as with a scrambled sequence controlduplex by using HiPerFect (Qiagen) transfection reagent.

Western Blot Analysis. Immunodetectionwasperformedaspreviouslydescribed(34) by using antibodies against IGF1R (no. 3027), phospho-Aktser473 (no. 9271),Akt (no. 9271), phospo-Erk1/2 (no. 9101), Erk1/2 (no. 9102), PARP (no. 9542),and caspase-3 (no. 9662), all at 1:1,000 dilution (Cell Signaling Technology).

E1274 | www.pnas.org/cgi/doi/10.1073/pnas.1105034109 Bielen et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

6, 2

020

Electrochemiluminescent Immunoassay. MSD 96-well multispot assays for total/phospho-IGF1R/IR/IRS-1, total/phospho-Akt, and total/phospho-Erk1/2 werecarriedoutper themanufacturer’s protocol. Briefly, plateswereblocked for 1h,and then 20 to 25μg of protein was added to the plate in duplicate wells andincubated overnight at 4 °C. Plates were washed and incubated with detectionantibody for 2 h with shaking. Plates were washed four times, read buffer wasadded, and the plates were analyzed on a SECTOR 6000 instrument (i.e., MSD).

mRNA Expression Profiling Analysis. Expression profiling after treatment ofWiT49 cells with NVP-AEW541 at 5× GI50 by Illumina HT-12 BeadChips wascarried out according to the manufacturer’s instructions and normalizedusing the lumi package in R software (version 2.11). All data has been de-posited in the Minimum Information About a Microarray Experiment-com-pliant ArrayExpress database (accession no. E-TABM-891). Supervisedanalysis was performed by using an absolute signal to noise metric ofgreater than 1.5 in GenePattern software. Time-course analysis was carriedout by using a Pearson correlation coefficient metric. To determine path-ways and networks that were significantly dysregulated upon treatmentwith NVP-AEW541, pathway analysis was performed by using the IngenuityPathway Analysis program, and coordinate gene regulation was identifiedby using Gene Set Enrichment Analysis.

In Vivo Antitumor Activity of NVP-AEW541. Six- to 8-week-old female athymicnude mice (CrTac:NCr-Foxn1nu; Taconic) were used to establish s.c. andorthotopic Wilms tumor xenograft models. Experiments were conducted inaccordance with United KingdomHomeOffice regulations under the Animals(Scientific Procedures) Act 1986 and National Council of Rural Institutesguidelines for the welfare and use of animals in cancer research (35). WiT49cells were implanted s.c., bilaterally, at 5 × 106 per site. Mice were randomlyassigned to three groups (n = 6), and treated with NVP-AEW541 (25, 50, or75 mg/kg orally twice per day) or vehicle comprising 25 mM L-(+)-tartaric acid.Bodyweight and tumor volumeweremeasured every 3 to 4d until the animalswere euthanized, withfinal tumorweightsmeasured followingdissection. Forthe orthotopic Wilms tumor model, mice were anesthetized with isoflurane,a small incision between the spleen and the spinal column was made to theskin and abdominalwall, and the left kidneywas carefully exteriorized. A totalof 5 × 105 WiT49 cells in 10 μL Matrigel (BD Biosciences) was injected into thesubcapsule of the kidney. The abdominal wall was closed with sterile 4–0Mersilk suture (Ethicon) and the skin closed with surgical clips. Postoperativeanalgesia was administered according to the manufacturer’s instructions.Animalswere treatedwith 50mg/kgNVP-AEW541 or vehicle control bymouthtwice per day for 15 d (n = 9). Tumor size was monitored by ultrasound (Vevo770; VisualSonics) and final tumor volume measured by MRI (as described

later). Tumor samples were homogenized and extracts used for theMSD assaydescribed earlier. Tumor drug levelsweremeasuredby LC/tandemMSby usinga Synergi Polar-RP analytical column (80 Å, 4 μeq, 50 × 2 mm; 00B-4336-B0;Phenomenex) with a gradient mobile phase of methanol and 0.1% formicacid, at a flow rate of 0.2 mL/min, in positive ionization mode by multiplereaction monitoring. Expression of phospho-IGF1R was assessed by immuno-histochemistry using a rabbit monoclonal antibody against IGF1R pY950(clone 7/4; Novartis) at 1:10,000 dilution using an epitope retrieval solution at100 °C (Vision Biosystems).

MRI. Tumor volume after NVP-AEW541 treatment of the orthotopic Wilmstumor model was measured by 1H MRI performed on a 7T Bruker horizontal-bore microimaging system by using a 3-cm birdcage coil. Anesthesia wasinduced by an i.p. injection of a combination of fentanyl citrate (0.315 mg/mL) plus fluanisone (10 mg/mL; Hypnorm; Janssen), midazolam (5 mg/mL;Roche), and water (1:1:2). Anatomical T2-weighted images were acquiredfrom 20 contiguous 1-mm-thick transverse slices through the mouse abdo-men with an in-plane resolution of 120 μm [rapid acquisition with relaxationenhancement (RARE) sequence; number of excitations (NEX), 4; effectiveecho time, 36 ms, repetition time, 4,500 ms; turbo factor, 8; 1-mm-thickcontiguous; matrix, 256 × 256]. Tumor volume was measured by using seg-mentation from regions of interest drawn on every slice containing tumorusing in-house software (ImageView; written under Interactive Data Lan-guage programming platform; ITT).

Statistical Analysis. All statistical tests were performed in GraphPad Prism 4(GraphPad). Associations between continuous and categorical variables wereanalyzed using ANOVA, with the posttest Bonferroni multiple comparisonmethod applied. Student t test was applied for the other statistical analyses.All tests were two-tailed, with a confidence interval of 95%. P values lessthan 0.05 were considered statistically significant.

ACKNOWLEDGMENTS. Expression profiling was carried out by UniversityCollege London Genomics, and we thank Dr. Anita Grigoriadis (Break-through Breast Cancer Unit, King’s College London) for bioinformatic assis-tance. SMARCB1 (INI1) mutation screening was carried out by the PaediatricMalignancy Unit, Great Ormond Street Hospital, London. We also thank Dr.David MacVicar (Royal Marsden Hospital) for the assessment of ultrasoundimages. This work was supported by Cancer Research United KingdomGrants C13468/A6718 and C309/A8274; Cancer Research United Kingdomand Engineering and Physical Sciences Research Council Cancer Imaging Cen-tre Grant C1060/A10334 in association with the Medical Research Counciland Department of Health (England); and National Health Service fundingto the National Institute for Health ResearchBiomedical Research Centre.

1. Pastore G, et al. (2006) Malignant renal tumours incidence and survival in European

children (1978-1997): Report from the Automated Childhood Cancer Information

System project. Eur J Cancer 42:2103–2114.2. Kalapurakal JA, et al. (2004) Management of Wilms’ tumour: Current practice and

future goals. Lancet Oncol 5:37–46.3. Natrajan R, et al. (2006) Blastemal expression of type I insulin-like growth factor

receptor in Wilms’ tumors is driven by increased copy number and correlates with

relapse. Cancer Res 66:11148–11155.4. Bjornsson HT, et al. (2007) Epigenetic specificity of loss of imprinting of the IGF2 gene

in Wilms tumors. J Natl Cancer Inst 99:1270–1273.5. Rainier S, et al. (1993) Relaxation of imprinted genes in human cancer. Nature 362:

747–749.6. Pollak MN, Schernhammer ES, Hankinson SE (2004) Insulin-like growth factors and

neoplasia. Nat Rev Cancer 4:505–518.7. Chitnis MM, Yuen JS, Protheroe AS, Pollak M, Macaulay VM (2008) The type 1 insulin-

like growth factor receptor pathway. Clin Cancer Res 14:6364–6370.8. Sachdev D, Yee D (2007) Disrupting insulin-like growth factor signaling as a potential

cancer therapy. Mol Cancer Ther 6:1–12.9. Houghton PJ, et al. (2007) The pediatric preclinical testing program: Description of

models and early testing results. Pediatr Blood Cancer 49:928–940.10. Smith MA, et al. (2008) SK-NEP-1 and Rh1 are Ewing family tumor lines. Pediatr Blood

Cancer 50:703–706.11. Garvin AJ, Re GG, Tarnowski BI, Hazen-Martin DJ, Sens DA (1993) The G401 cell line,

utilized for studies of chromosomal changes in Wilms’ tumor, is derived from

a rhabdoid tumor of the kidney. Am J Pathol 142:375–380.12. Dallosso AR, et al. (2007) Alternately spliced WT1 antisense transcripts interact with

WT1 sense RNA and show epigenetic and splicing defects in cancer. RNA 13:

2287–2299.13. Royer-Pokora B, et al. (2010) Wilms tumor cells with WT1 mutations have

characteristic features of mesenchymal stem cells and express molecular markers of

paraxial mesoderm. Hum Mol Genet 19:1651–1668.14. Kim MK, et al. (2008) A pathologic link between Wilms tumor suppressor gene, WT1,

and IFI16. Neoplasia 10:69–78.

15. Stock C, et al. (2002) Genetic changes of two Wilms tumors with anaplasia and

a review of the literature suggesting a marker profile for therapy resistance. Cancer

Genet Cytogenet 135:128–138.16. Alami J, Williams BR, Yeger H (2003) Derivation and characterization of a Wilms’

tumour cell line, WiT 49. Int J Cancer 107:365–374.17. Metzger ML, et al. (2007) Topotecan is active against Wilms’ tumor: Results of a multi-

institutional phase II study. J Clin Oncol 25:3130–3136.18. Casa AJ, Dearth RK, Litzenburger BC, Lee AV, Cui X (2008) The type I insulin-like

growth factor receptor pathway: A key player in cancer therapeutic resistance. Front

Biosci 13:3273–3287.19. Houghton PJ, et al. (2010) Initial testing of a monoclonal antibody (IMC-A12) against

IGF-1R by the Pediatric Preclinical Testing Program. Pediatr Blood Cancer 54:921–926.20. Kolb EA, et al. (2010) Initial testing (stage 1) of the IGF-1 receptor inhibitor BMS-

754807 by the pediatric preclinical testing program. Pediatr Blood Cancer, 10.1002/

pbc.22741.21. Li MH, Yamase H, Ferrer F (2010) Characterization of a WiT49 cell line derived

orthotopic model of Wilms tumor. Pediatr Blood Cancer 54:316–318.22. Box C, Rogers SJ, Mendiola M, Eccles SA (2010) Tumour-microenvironmental

interactions: Paths to progression and targets for treatment. Semin Cancer Biol 20:

128–138.23. Talmadge JE, Singh RK, Fidler IJ, Raz A (2007) Murine models to evaluate novel and

conventional therapeutic strategies for cancer. Am J Pathol 170:793–804.24. Kerbel RS (2003) Human tumor xenografts as predictive preclinical models for

anticancer drug activity in humans: better than commonly perceived-but they can be

improved. Cancer Biol Ther 2(4 suppl 1):S134–S139.25. Singh RK, et al. (1994) Organ site-dependent expression of basic fibroblast growth

factor in human renal cell carcinoma cells. Am J Pathol 145:365–374.26. Van Dyke T (2010) Finding the tumor copycat: Approximating a human cancer. Nat

Med 16:976–977.27. Ellis LM, Fidler IJ (2010) Finding the tumor copycat. Therapy fails, patients don’t. Nat

Med 16:974–975.28. Hazen-MartinDJ, ReGG,GarvinAJ, SensDA (1994)Distinctive properties of ananaplastic

Wilms’ tumor and its associated epithelial cell line. Am J Pathol 144:1023–1034.

Bielen et al. PNAS | Published online April 23, 2012 | E1275

MED

ICALSC

IENCE

SPN

ASPL

US

Dow

nloa

ded

by g

uest

on

Janu

ary

6, 2

020

29. Hu Q, et al. (2011) Wt1 ablation and Igf2 upregulation in mice result in Wilms tumorswith elevated ERK1/2 phosphorylation. J Clin Invest 121:174–183.

30. Olmos D, Basu B, de Bono JS (2010) Targeting insulin-like growth factor signaling:Rational combination strategies. Mol Cancer Ther 9:2447–2449.

31. Natrajan R, et al. (2006) Array CGH profiling of favourable histology Wilms tumoursreveals novel gains and losses associated with relapse. J Pathol 210:49–58.

32. Cory AH, Owen TC, Barltrop JA, Cory JG (1991) Use of an aqueous soluble tetrazolium/formazan assay for cell growth assays in culture. Cancer Commun 3:207–212.

33. Chou TC, Talaly P (1977) A simple generalized equation for the analysis ofmultiple inhibitions of Michaelis-Menten kinetic systems. J Biol Chem 252:6438–6442.

34. Bax DA, et al. (2009) Molecular and phenotypic characterisation of paediatric gliomacell lines as models for preclinical drug development. PLoS ONE 4:e5209.

35. Workman P, et al.; Committee of the National Cancer Research Institute (2010)Guidelines for the welfare and use of animals in cancer research. Br J Cancer 102:1555–1577.

E1276 | www.pnas.org/cgi/doi/10.1073/pnas.1105034109 Bielen et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

6, 2

020