Redundant Signaling as the Predominant mechanism for Resistance to Antibodies Targeting the Type-I Insulin-like Growth Factor Receptor in Cells Derived from Childhood Sarcoma

Abstract

Antibodies targeting IGF-1R induce objective responses in only 5–15% of children with sarcoma. Understanding mechanisms of resistance may identify combination therapies that optimize efficacy of IGF-1R-targeted antibodies. Sensitivity to the IGF1R-targeting antibody TZ-1 was determined in rhabdomyosarcoma (RMS) and Ewing sarcoma (EWS) cell lines. Acquired resistance to TZ-1 was developed and characterized in sensitive Rh41 cells. BRD4 inhibitor, JQ1 was evaluated as an agent to prevent acquired TZ-1 resistance in Rh41 cells. The phosphorylation status of RTKs was assessed. Sensitivity to TZ-1 in vivo was determined in Rh41 parental and TZ-1-resistant xenografts. Of 20 sarcoma cell lines, only Rh41 was sensitive to TZ-1. Cells intrinsically resistant to TZ-1 expressed multiple (>10) activated RTKs or a relatively less complex set of activated RTKs (~5). TZ-1 decreased the phosphorylation of IGF-1R, but had little effect on other pRTKs in all resistant lines. TZ-1 rapidly induced activation of RTKs in Rh41 that was partially abrogated by knockdown of SOX18 and JQ1. Rh41/TZ-1 cells selected for acquired resistance to TZ-1 constitutively expressed multiple activated RTKs. TZ-1 treatment caused complete regressions in Rh41 xenografts, and was significantly less effective against the Rh41/TZ-1 xenograft. Intrinsic resistance is a consequence of redundant signaling in pediatric sarcoma cell lines. Acquired resistance in Rh41 cells, is associated with rapid induction of multiple RTKs indicating a dynamic response to IGF-1R blockade and rapid development of resistance. The TZ-1 antibody had greater antitumor activity against Rh41 xenografts compared to other IGF-1R-targeted antibodies tested against this model.

Introduction

The insulin-like growth factors (IGFs) have been implicated in the pathogenesis of numerous pediatric malignancies, including Ewing Sarcoma (EWS), rhabdomyosarcoma (RMS) and osteosarcoma (OS). A role for IGF-1R signaling has been proposed in the pathogenesis of RMS and EWS. For example, IGF-2 is highly expressed in RMS (1) and is reported to support proliferation and motility of RMS cells by binding to IGF-1R. RMS cell lines secrete IGF-2, which then binds and activates IGF-1R, resulting in autocrine proliferation and increased cell motility (2). EWS, peripheral primitive neuroectodermal tumor, and Askin tumor cells secrete IGF-1 and express high levels of IGF-1R, (3), and IGF-1R–blocking antibodies interrupt this autocrine loop (4). Furthermore, high levels of IGF-1 may play an important role in the pathogenesis of OS as the majority of OS patient express IGF ligands and 45% express IGF-1R (5). Moreover, exogenous IGF-1 promotes proliferation of OS cells, and monoclonal antibodies or antisense oligonucelotides against IGF-1R is reported to inhibit OS growth (6). In addition, activated IGF-1R signals to both PI3K and MAPK pathways via insulin-receptor substrates (IRS-1–4), which are all implicated in the pathogenesis of childhood sarcomas (7). The loss of imprinting at the IGF-2 locus may be a primary genetic event in development of embryonal RMS (8), and translocations associated with alveolar RMS (t(2;23)(q35;q14) and t(2;130(q36;q14) that generate the PAX3FOXO1 and PAX7-FOXO1 chimeric transcription factors, enhance transcription of IGF-1R (9). In Ewing sarcoma, the EWS-FLI1 chimeric transcription factor, generated from the t(11;22) chromosomal translocation may suppress IGFBP3, a negative regulator of IGF-1 (10).These studies indicate that IGF-1R signaling plays a critical role in the pathogenesis of RMS, EWS and OS, therefore targeting this pathway may have considerable utility in the therapy of sarcomas. Consistent with that, preclinical studies from the Pediatric Preclinical Testing Program showed that treatment of mice with human anti-IGF-1R antibodies resulted in tumor regression in OS, EWS and RMS xenograft models (11, 12).

Antibodies that prevent ligand binding to the IGF-1R may have therapeutic utility in childhood sarcomas. In preclinical models of childhood cancers, SCH717454, an IgG1 human antibody that binds IGF-1R and prevents IGF-1R-ligand binding, significantly inhibits growth of some RMS xenografts and induces regressions in several sarcoma histotypes, notably OS and EWS (11). Similarly, another IGF-1R-targeted antibody, R1507, was found to inhibit growth of OS xenografts (13). MK-0646, a humanized IGF-1R-targeting antibody, retarded growth of several sarcoma models and in vivo xenografts, with activity proportional to the level of IGF-1R expression (9). In contrast, other studies did not support a correlation between the level of IGF-1R expression and antitumor activity of CP751871 (human IgG2 antibody) (14, 15).

Phase-1 or −2 clinical trials with eight fully human or humanized antibodies that target IGF-1R have been reported (16). Data have not emerged suggesting that one antibody is more effective than another. In an extended phase 1 trial of patients with EWS, the objective response rate to figitumumab (CP751871) treatment was 12% with 36% stable disease for 4 months (17). The response to R1507 in EWS patients was 9.7%, and durable stable disease was again observed in some patients (18). A subsequent phase 2 trial in children showed a lower objective response rate with objective responses only in osteosarcoma and rhabdomyosarcoma patients (19).

The low objective response rate suggests that intrinsic resistance is the predominant limitation of targeting IGF-1R, and in the subset of responsive tumors, acquired resistance almost invariably develops. Intrinsic resistance to IGF-1R-targeted antibodies has been proposed to be through alternative signaling through the insulin receptor (IN-R), and resistance to antibody targeting IGF-1R may differ from that induced by small molecule inhibitors (20–23). Alternatively, tumor autonomous mechanisms have been proposed to explain intrinsic resistance. For example, inhibition of the IGF-1R leads to a compensatory increase in serum growth hormone levels that may circumvent the block on IGF-1R (1, 24) and antibodies against IGF-1R increase both IGF-1R and IN-R levels and may lead to increased circulating levels of IGF-1 (24, 25). Intrinsic mechanisms that drive resistance to IGF-1R inhibition appear variable and complex, in part due to the complexity of IGF-1R/IN-R signaling in tumor cells, and dynamic homeostatic mechanisms in intact organisms (26). Here, we have examined tumor-intrinsic mechanisms that account for responses to IGF-1R inhibition in panels of pediatric sarcoma cell lines.

Materials and Methods

Reagents, drugs and antibodies

LY2874455, NPS-1034, ceritinib, crenolinib and JQ1 were purchased from Selleck Chemicals (Houston, TX). All drugs were prepared as stock solutions in DMSO. In all experimental conditions, the final concentration was less than 0.1% DMSO. TZ-1, an anti-IGF1R antibody, was provided by MedImmune LLC (subsidiary of AstraZeneca). TZ-1, is a fully germlined human IgG1, and binds to human IGF1R with an affinity of 40 pM, as determined by SPR (BiaCore). The antibody is species cross-reactive to mouse, rat and monkey. TZ-1 is specific and does not bind to other members of the insulin receptor family. TZ-1 is a functional cell antagonist of IGF-1, inhibits the human IGF ligand binding to NIH 3T3 IGF1R cells at 0.3nM, and inhibits IGF1-mediated BxPC3 survival at 75nM. In addition, upon binding to the IGF1R receptor, TZ1 triggers internalization and down regulation of the receptor. The TZ-1 antibody was produced recombinantly using the CHO-G22 cell-line. The transient expression level is ~540 mg/L after 10 days of culturing using AstraZeneca proprietary tissue culture media and feeding supplements (Dr. R. Fleming, Astrazeneca, pers comm). The TZ-1 antibody was reconstituted in 10mM His-HCl (pH6). The TZ-1 Fab fragment has been used to engineer a bispecific antibody (iMab-EI) that binds EGFR and IGF1R. Further details regarding the characteristics of iMab-EI and parental antibodies are given in Dimasi et al (27).

SCH717454 was provided by Schering-Plough Research Institute and was diluted in 20 mmol/L sodium acetate buffer (pH 5) containing 150 mmol/L sodium chloride. Primary antibodies for Western blot were purchased from Cell Signaling and include GAPDH (cat. no 2118) and PARP (cat. no 9542). The SOX18 antibody was purchased from Santa Cruz (cat. no sc-166025). Secondary antibodies for Western blot: anti-mouse IgG, HRP-linked (Cell Signaling, cat. no 7076) and anti-rabbit IgG, HRP-linked (Cell Signaling, cat. no 7074). Reagents used: puromycin (Sigma-Aldrich, cat. no P9620), human immunoglobulin (Sigma-Aldrich, cat. no I4506), and Lipofectamine RNAiMAX (ThermoFisher Scientific, cat. no 13778150).

Cell Lines and Cell Culture

Ten RMS cell lines (Rh4, Rh5, Rh18, Rh28, Rh36, Rh41, CT-TC, JR-1, and RD) and 10 EWS cell lines (ES1, ES2, ES3, ES4, ES6, ES7, ES8, EW8, CHLA-258 and TC-71) were tested for sensitivity to TZ-1. Authentication of lines used was confirmed by routine comparison of short tandem repeat assays and our original short tandem repeat profiles established on early passage lines stored in a Master Bank. All sarcoma cells were maintained in antibiotic-free RPMI high glucose supplemented growth medium (Invitrogen) with 10% FBS (Sigma-Aldrich) and 2mmol/L glutamine. They were grown at 37^oC with 5% CO₂. In order to make a resistant cell line, Rh41 cells were exposed to TZ-1 and SCH717454 at half of their IC₅₀ concentration. The concentration was then increased with every other passage until the resistant cells remained viable in the presence of the antibody at 10μg/mL.

Plasmids and viral infections

pCMV-SOX18 was purchased from Dharmacon (MHS6278–202759994, Clone Id:6183010). Lentivirus was produced by transfecting 293T cells with shRNA expression plasmids (Mission shRNA, Sigma) along with the packaging plasmid (psPAX2), an envelope plasmid (pMD2.G) and the lentiviral backbone plasmid pLKO1. Sequences for the shRNA and siRNA are in Supplemental Table S1. Virus particles were collected 72 Hr post transfection, and filtered through a 0.45 μm membrane. Cell lines were transduced with the virus in the presence of polybrene (10 μg/mL) for 48 Hr at which point the media was replaced with new media containing puromycin (1 ug/mL). A stable pool of cells was obtained by prolonged exposure to puromycin.

Cell viability/proliferation assay

Cell viability assays were used to determine the concentration at which 50% of cell growth is inhibited by drug treatment (IC₅₀ concentration). Cells were seeded in 96-well plates at a density of 2–5×10³ cells per well and allowed to adhere overnight. The following day, cells were treated with drug at increasing concentrations. After 96 Hr, cell viability was quantified using Alamar blue by adding the reagents directly to the culture medium at a final concentration of 10% v/v. Cells were further incubated for 4 Hr and fluorescence was measured (excitation 530nm, emission 590 nm). Alamar blue was added to medium without cells was used as a negative control. The results were expressed as mean viable cells relative to cells treated with DMSO alone ± SD. The IC₅₀ concentration was calculated by XLFit software. Combination studies were performed using the same protocols, the results were expressed as mean viable cells relative to the condition of the fixed drug alone (considered 100% viability). Assays were done in triplicate and repeated three times.

Reverse Transcription Quantitative Real-Time PCR (RT-qPCR)

Expression levels of genes were analyzed using RT-qPCR. Total RNA was extracted after a 24 Hr treatment with TZ-1 (10μg/mL) and JQ1 (500nM). RNA was extracted using TRIzol (Life Technologies) following the manufacturer’s protocol and further purified by RNeasy Mini Kit (Qiagen). RNA (1μg) was used for cDNA Synthesis using the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems, cat no 4375222). The cDNA samples were diluted to 20ng/μl. Gene-specific primers are shown in Supplemental Table S2. RT-qPCR was performed using the SYBR Green PCR Master Mix (Qiagen) and amplification was performed on the QuantStudio 7 Flex Real-Time PCR System. The thermal cycling conditions included 2 min at 50^oC followed by an initial denaturation step of 10 min at 95^oC, and 40 cycles consisting of an annealing step at 95^oC for 15 seconds and an extension step at 60^oC for 1 minute. Each sample was analyzed in triplicate. Relative gene expression was determined using the ΔΔCt method. Fold changes in gene expression were normalized to an internal control gene, UBB and HPRT. The ΔC_T values between conditions were compared.

Phospho-RTK profiling

Changes in RTK signaling were assessed using proteomic profiler human phospho RTK arrays (R&D Systems, cat no. ARY001B). Cells were plated at 1×10⁶ cells per 100mm dish and treated with TZ-1 (10μg/mL) or JQ1 (500nM) or the combination for 48 Hr. Cell lysates were prepared following manufacturer’s instructions. For each sample, 400 μg total protein were diluted and incubated with the RTK array membranes overnight at 4⁰C. The arrays were developed on X-ray films following exposure to chemiluminescent reagents. The procedures were performed according to the manufacturer’s protocol. Quantitation of the arrays was performed using ImageJ and the raw values of each spot of the arrays are shown.

Reverse Phase Protein Array

Cells were plated in 6-well plates at 5×10³ cells/well followed by a 48 Hr treatment with human IgG or TZ1 (10 μg/mL). Cells were washed with cold PBS, lysed in the following buffer: (1% Triton X-100, 50 mm HEPES, pH 7.4, 150 mM NaCl, 1.5 mM MgCl2, 1 mM EGTA, 100 mM NaF, 10 mM Na pyrophosphate, 1 mM Na3VO4, 10% glycerol) supplemented with protease and phosphatase inhibitors (Thermo Fisher Scientific). RPPA was performed in the MD Anderson Cancer Center RPPA Core Facility as described previously (28).

Transcription factor activity profiling analysis

Transcription factor (TF) profiling plate arrays were purchased from Signosis, Inc. and each assay performed according to the manufacturer’s protocol. Rh41 cells were exposed to TZ-1 (10 μg/mL) and/or JQ1 (500 nM) for 24 Hr. Six micrograms of nuclear protein extracts were incubated with biotin-labeled probe for 30 min at room temperature. TF-probe complexes were separated from free probes using spin-column purification and detected with streptavidin-HRP and measured with the Spectramax M5 microplate reader (Molecular Devices). TF activities in control treated cells were compared with the activities in treated cells, and selected based on fold-change method (≤−1.5 or ≥1.5).

RNA Interference

For small interfering RNA (siRNA)-mediated knockdown of SOX18 or BRD4, cells were plated at 2–5×10⁵ cells/well in a 6-well dish to 30–50% confluence and the following day were transfected with 100 nM of either the control or targeting siRNA (Dharmacon) using RNAiMAX (Invitrogen). Media was replaced 24 Hr after transfection and cells were collected 48–72 Hr posttreatment depending on the assay. For SOX18 knockdown, four independent duplexes of siRNAs were tested and results from highest SOX18 gene knockdown (Dharmacon, J-019035–07) are used throughout. For BRD4 knockdown, a pool of four siRNAs was used to achieve knockdown (Dharmacon, L-004937–00-0005). The sequences are provided in Supplemental Table S1.

Caspase 3/7 mediated apoptosis

Kinetic quantification of apoptosis was performed by measuring the number of cells positive for activated caspase 3/7 over a 24 Hr period using live-cell imaging. The cells were plated at 15,000 cells/well in a 96-well plate and the following day were treated with the IGF-1R targeting monoclonal antibody, TZ1 (20 μg/mL) in the presence of 5 μM Caspase 3–7 reagent (Essen BioSciences, cat no. 4440). The apoptosis activity was monitored using an Incucyte ZOOM (Essen BioSience) which gathered images every 2 Hr using green fluorescence and is plotted as number of caspase-3/7 positive cells, n=3 wells per data point.

In vivo studies

C.B.17SC scid−/− (C.B-Igh-1b/IcrTac-Prkdcscid) female mice (Envigo, Indianapolis, IN) were used to propagate subcutaneously implanted tumors as previously described [2]. All mice were maintained under barrier conditions and experiments were carried out using protocols and conditions approved by the institutional animal care and use committee of UTHSA. Rh41 and Rh41/TZ-1 resistant cells were harvested in logarithmic growth, and 10⁶ cells inoculated into the flank. Treatment was initiated when tumors were 200–400 mm³ (Mean±SD: Rh41: 240 ± 72 mm³; Rh41/TZ1R: 280 ± 73 mm³). Methods similar to those developed in the Pediatric Preclinical Testing Program (PPTP) were used [2], only for these studies tumor volumes were measured until tumors reached 125% of their volume at start of treatment following tumor regression (defined as an ‘Event’). Endpoints were Event-free survival (EFS), and percent tumor regression. Complete Regression (CR; defined as tumor volume <40mm³, the level of detection). TZ-1 antibody was reconstituted in 10mM His-HCl (pH6) and administered intravenously twice weekly for 3 consecutive weeks at a dose of 33 mg/kg. When dosed at 10 mg/kg in normal, non-tumor bearing CD1 mice, TZ-1 has a half-life of 68.05 Hr and a clearance rate of 9.19 mL/kg/Hr (Dr. R. Fleming, Astrazeneca, pers comm).

Statistical Analysis

Experiments were performed in triplicate and the results were presented as mean ± SD. Statistical significance between multiple groups was determined by one-way ANOVA (single variable) or two way ANOVA (two variables) and Tukey’s multiple-comparison posttest. GraphPad Prism 6.0 software (GraphPad Software Inc.) was utilized for all statistical analyses. P < 0.05 was considered to be statistically significant.

Results

Sensitivity of pediatric sarcoma cell lines to TZ-1.

We investigated whether TZ-1, a monoclonal antibody that specifically blocks ligand binding to IGF-1R, would be effective at reducing cellular proliferation in RMS and EWS cell lines. Previous studies show that TZ-1 is a functional cell antagonist that inhibits human IGF ligand binding in NIH 3T3 IGF1R overexpressing cells. A panel of 10 RMS and 10 EWS cell lines were incubated with or without antibody at increasing concentrations for 96 Hr. The only RMS cell line out of 10 that responded to TZ-1 treatment was Rh41 with an IC₅₀ of 1.02 μg/mL (~7nM), Figure 1A. None of the EWS cell lines were sensitive to TZ-1 and all had an IC₅₀ of greater than 10 μg/mL (Figure 1B). TZ1 inhibition of IGF-1R phosphorylation was confirmed in the sensitive Rh41 as well as resistant cells (Rh18, ES2 and EW8) although the effect of TZ-1 in Rh18 and Rh36 was minimal (Supplemental Figure S1). Both TZ-1 and control IgG transiently stimulated pAkt and pERK1/2 in Rh18 and ES2 cells.

Figure 1. Sensitivity of RMS and EWS cells to the IGF-1R inhibitor, TZ-1 is dependent on RTK activation status.

A panel of RMS (A) and EWS (B) cell lines were treated with TZ-1 and dose response curves are shown. Cells were treated with TZ-1 for 96 Hr and cell viability was measured by Alamar Blue assay. Data is representative of three independent experiments in triplicate and presented as the mean ± SD. Phospho-RTK array showing phosphorylation status of 49 RTKs in RMS (C) and EWS (D) cell lines treated with IgG control antibody or TZ-1 (10μg/mL) for 48 Hr. Four phospho-tyrosine positive controls are on the corners of each array and each RTK is represented in duplicate with each pair of horizontal spots representing one receptor. Upregulated RTKs are indicated.

Phospho - RTK profiling reveals activation of multiple RTKs.

Profiling was used to determine if there was a change in RTK-activated cell signaling that was driving the response to TZ-1 in the Rh41 cell line. Phosphorylation of 49 receptor tyrosine kinases (RTKs) using a phospho-RTK array was used to observe changes in activation of RTKs following treatment with TZ-1. A subset of RMS and EWS cell lines was treated with or without the antibody TZ-1 (10μg/mL) for 48 Hr. Cell lysates were collected and incubated with the phospho-RTK array membranes. The phosphorylation of growth factor RTKs is shown for RMS (Figure 1C) and EWS cells (Figure 1D). Results from quantification of the RTK arrays is shown in Supplementary Figures S2 and S3. In all of the cell lines, treatment with TZ-1 reduced the phosphorylation of IGF-1R as expected. The untreated RMS cell lines and all cell lines intrinsically resistant to TZ-1, had multiple activated RTKs at baseline. Upon treatment with TZ-1, there was no noticeable change in the phosphorylation pattern for each of the cell lines, other than the reduction in phospho-IGF-1R. Activated RTKs detected in the control (i.e. untreated) RMS cells included the EGFR family of receptors, IGF-1R, IR, PDGFR, FGFR4, HGFR, RYK, ALK, DDR1, AXL, and DTK. Three of the EWS cells (ES2, ES4 and EW8) had a complex basal phosphorylation pattern consisting largely of the same activated RTKs as the RMS cells. In contrast, two EWS cell lines (ES3 and ES7) had a less complex basal phosphorylation pattern, which included IGF-1R, IR, HGFR, ERBB4, AXL and DTK. TZ-1 treatment of these EWS cell lines decreased phosphorylation of IGF-1R, but had little effect on other RTKs.

Development of acquired resistance.

Changes in RTK activation and development of acquired resistance were examined in Rh41 cells. Acquired resistance to the IGF-1R-targeted antibodies, SCH717454 and TZ-1, designated as Rh41-AB1R and Rh41-AB2R, respectively, was achieved by growing cells in increasing concentration of the respective antibody. Because Rh41 cells do not grow at very low densities, development of resistant cell lines from a single cell was not possible. The resistant line, Rh41-AB1R was initially developed in the presence of the SCH717454 antibody. However, during the study, the development of SCH717454 was halted and was no longer available for our studies. We evaluated whether the Rh41-AB1R cell line would also be resistant to the TZ-1 antibody and found that it was just as resistant to TZ-1 as to SCH717454 (Supplemental Figure S4). From this point forward, Rh41-AB1R was grown and treated in the presence of TZ-1.

TZ-1 activates multiple receptor tyrosine kinases in Rh41 parental and Rh41 cells with acquired resistance to TZ-1.

Rh41 was the only cell line sensitive to TZ-1 (Figure 1A) and in contrast to the cell lines intrinsically resistant to TZ-1, under control conditions the predominant activated RTK was IGF-1R with low level expression of EGFR, EPHA7, and VEGFR1 (Figure 2A). TZ-1 exposure (10 μg/mL, 48 Hr), caused a marked decrease in IGF-1R phosphorylation and increase in phosphorylation of numerous other RTKs including ErbB3, FGFR4, AXL, RYK, VEGFR1 and ALK. To quantify the degree of differential expression, the dot blots were quantified by densitometry; measurements and results of selected RTKs are shown in Figure 2B.

Figure 2. Evidence for phosphorylation of multiple RTKs induced in Rh41 IGF1-R inhibitor-resistant cells.

A) Rh41 parent and derivative cells resistant to TZ1 (Rh41-AB1R and Rh41-AB2R) were treated with TZ-1 or isotype matched IgG (control) (10 μg/mL) or left untreated for 48 Hr and cell lysates were assessed for levels of phosphorylated RTKs using phospho-RTK arrays. Four phospho-tyrosine positive controls are on the corners of each array and allow for comparison between arrays. Each RTK is represented in duplicate with each pair of horizontal spots representing one receptor. Upregulated RTKs are indicated (arrows). B) Dots from the phospho-RTK arrays were quantified by densitometric analysis using ImageJ software. C) Relative mRNA levels of multiple RTKs were determined by RT-qPCR in Rh41 parental cells treated with IgG or TZ-1 (10μg/mL) for 24 Hr, fold change expression are shown. D) Rh41 parent and resistant cells were treated with increasing concentrations of the panFGFR inhibitor (LY2874455), the MET/AXL inhibitor (NPS-1034), the ALK inhibitor (ceritinib) and the PDGFRα/β inhibitor (crenolinib) for 96 Hr. Cell viability was determined by Alamar Blue Assay at 96 Hr following treatment and IC₅₀ values were calculated. Data represents three independent experiments in triplicate and presented as the mean ±SD.

Sensitivity to TZ-1 in Rh41 and Rh41/TZ-1 resistant cells in vivo.

To determine the sensitivity to TZ-1 treatment in vivo, Rh41 and Rh41/TZ-1R resistant cells were established as subcutaneous xenografts, and mice received TZ-1 when tumors were 200–400 mm³ (Rh41 240± 72 mm³, Rh41TZ1R 280± 73 mm³). Rh41 xenografts were highly sensitive to TZ-1 treatment with all tumors regressing completely, with subsequent regrowth, and Event-Free Survival (EFS) of 60.9 ± 11.2 days (mean ± SD). In contrast, TZ-1 induced partial regressions of Rh41/TZ-1 xenografts, but with rapid regrowth (EFS 28.2 ± 20.2 days [mean ± SD]) significantly shorter than in Rh41 (P=0.0002). Western blot analysis of Rh41 parental tumor tissue showed a decrease in both total and phospho-IGF-1Rβ along with a reduction in pAKT and pS6. (Supplemental Figure S5).

Characterization of Rh41 cells with acquired resistance to TZ-1 in vitro.

In contrast to parental Rh41 cells, those selected for resistance to SCH717454 (Rh41-AB1R) had high expression of phosphorylated EGFR, FGFR4, AXL, and DTK. The Rh41-AB2R line, selected for TZ-1 resistance, had high expression of phosphorylated EGFR, FGFR4, AXL, PDGFRβ, VEGFR1 and DTK. Interestingly, the Rh41 parent line treated with TZ-1 and the resistant lines shared high overall levels of phosphorylated EGFR, FGFR4, AXL and VEGFR1 (Figure 2B). This suggests that in Rh41 cells, the rapid activation of these receptors is initiated following inhibition of IGF-1R, is sustained in resistant cell lines, and potentially an important mechanism of acquired resistance.

Next, we performed Reverse Phase Protein Array (RPPA) to identify changes in protein expression in response to IGF-1R inhibition in the Rh41 parental line compared to the resistant lines Rh41-AB1R and Rh41-AB2R. Of the 296 proteins from the RPPA dataset, TZ-1 treatment changed 80 significantly (P<0.05). Common to all three cell lines, TZ-1 treatment induced an increase in insulin receptor substrate (IRS-1) consistent with decreased endocytic recycling of IGF-1R, antibody treatment also decreased poly (ADP-ribose; PAR) and p70S6 phosphorylation (consistent with deceased IGF-1R/TORC1 signaling). The decrease in fatty acid synthetase (FASN) detected in all lines is consistent with IGF-1 regulation of FASN induction (Supplemental Figure S6). In the two lines selected for acquired resistance (Rh41-AB1R, Rh41-AB2R), an increase in the levels of both anti-apoptotic Mcl-1 and the tumor suppressor programmed cell death 4 (Pdcd4) was common suggesting that high levels of these proteins may be associated with resistance to TZ-1 through suppressing antibody-induced apoptosis.

IGF-1R blockade leads to transcriptional upregulation of RTKs.

To determine if increased RTK was due to a transcriptional response in the presence of IGF-1R blockade, RTK transcript levels were analyzed by quantitative real-time PCR in Rh41 cells. Following 24 Hr treatment with TZ-1, a moderate increase (1 to 8-fold increase) in FGFR2, FGFR3, FGFR4, IGF-1R, ALK, PDGFRα and VEGFR1 was detected. A more robust increase (13–26-fold) was seen in IRβ, AXL and PDGFRα/β.These changes, induced by short exposure to TZ-1, were similar to the two resistant lines grown for several months in TZ-1 (Figure 2C).

We examined whether inhibition of these activated RTKs would confer a vulnerability to cells with acquired resistance to IGF-1R blockade. In our previous results, we saw increased activity of FGFR, AXL, ALK and PDGFR. We treated the Rh41 parent and resistant cells with drugs targeting these RTKs and compared the half-maximal inhibitory concentration (IC₅₀) of the drugs (Figure 2D). The antibody resistant lines Rh41-AB1R and Rh41-ABR2 were more sensitive to the pan-FGFR inhibitor LY2874455, and the ALK inhibitor ceritinib. LY2874455 had an IC₅₀ value of 2.18 μM in the Rh41 parental line while the resistant lines were more sensitive to the drug, Rh41-AB1R (0.73 uM) and Rh41-AB2R (0.14 μM). A similar trend occurred for the ALK inhibitor between Rh41 parental (IC₅₀ = 1.45 μM) and the resistant lines Rh41-AB1R (IC₅₀ = 0.76 μM) and Rh41-AB2R (IC₅₀ = 0.006 μM). The IC₅₀ values for both drugs were much lower in the resistant cell lines, which suggests that sustained inactivation of IGF-1R signaling resulted in increased dependency on the ALK and FGFR signaling pathways. Rh41-ABR2 was also more sensitive to the AXL/MET inhibitor NPS-1034.

Inhibition of Brd4 prevents adaptive kinome reprogramming and sensitizes Rh41 cells to TZ-1.

The activation of kinase signaling networks following kinase-directed inhibitor treatment has been termed “adaptive kinome reprogramming” and can contribute to targeted therapy resistance (29–34). It has been proposed that chromatin readers are involved in adaptive reprogramming. The chromatin readers include a class of bromodomain and extraterminal domain (BET) proteins that recognize histone acetylated lysine residues and recruit transcriptional machinery to facilitate transcriptional activation (35). The BRD4 inhibitor, JQ1, has shown utility in the treatment of cancer and has been proposed as a treatment strategy to overcome resistance (36). We assessed whether inhibition of BRD4 would attenuate the increases in RTKs following inhibition of IGF-1R. Rh41 parental cells were treated with TZ-1, JQ1 alone and in combination for 24 Hr. Changes in expression of the RTKs by RT-qPCR was assessed. Treatment with JQ1 alone had only slight effect on RTK transcript levels, slightly increasing IRβ, and PDGFRα/β and slightly suppressing IGF-1R transcripts. However, JQ1 blocked TZ-1-mediated increase in IGF-1R, IR, AXL, PDGFRα, PDGFRβ and VEGFR1 expression (Figure 3A). JQ1 did not suppress the TZ-1 mediated increase in FGFR2, and slightly potentiated the increase in FGFR3 (Supplemental Figure S7). To validate these findings, we knocked down BRD4 gene expression using siRNA in Rh41 parental cells and detected changes in gene expression. Knockdown of BRD4 resulted in a greater than 50% reduction in many RTKs including IGF-1R, INSRβ, AXL, ErbB3, PDGFRα and PDGFRβ (Figure 3B).

Figure 3. BRD4 inhibition blocks TZ-1 mediated increased expression of multiple RTKs.

A) Relative mRNA levels of multiple RTKs were determined by RT-qPCR in Rh41 parental cells treated with DMSO and IgG (Control), TZ-1 (10μg/mL) JQ1 (500nM) or a combination of TZ1 and JQ1 for 24 Hr. B) Relative mRNA levels of BRD4 and RTKs following knockdown of BRD4 by siRNA and TZ1 (10μg/mL) treatment for 48 Hr were determined by RT-qPCR. C) The combination of TZ-1 and JQ-1 reduces cell proliferation as detected by Alamar Blue assay in Rh41-parental (C) and Rh41-AB2R resistant cells (D). Rh41 cells treated with TZ-1 (10μg/mL), JQ1 (100nM) alone and in combination for 96 Hr. E) The combination of TZ-1 and JQ-1 increased apoptosis as detected by Annexin V in cells treated for 18 Hr and imaged on the Celigo Imager (**p<.05, ***p < 0.001, ****p < 0.0001, n = 3 by one-way ANOVA. Bars are means +/− SD).

Because JQ1 inhibited TZ-1-induced increased expression of most RTKs, we investigated its ability to suppress cell viability. Rh41 parental and TZ-1-resistant cell lines were treated with JQ1 alone or JQ1 in the presence of TZ-1. The relative sensitivities of Rh41 and Rh41-AB2R are shown (Figure 3C, D). JQ1 alone reduced proliferation in Rh41 parental cells or Rh41-AB2R resistant cells, TZ-1 in combination with JQ1 further reduced proliferation (Figure 3C,D) and caused increased apoptosis (Figure 3E). Of note, TZ-1 induced apoptosis in Rh41 cells whereas JQ1 did not, however the combination resulted in enhanced apoptosis. Additional drug combinations were also investigated and notably, the combination of either JQ1 or a pan FGFR inhibitor (LY2874455) with TZ-1 resulted in substantial reduction in cell viability and increased apoptosis (Supplemental Figure S8). The combination of LY2874455 + TZ1 resulted in 53% apoptosis while the combination of LY2874455 + JQ1 + TZ1 induced 71% apoptosis.

Transcriptome profiling reveals increased SOX18 activity following treatment with TZ-1.

The previous results showed that inhibition of IGF-1R by TZ-1 resulted in increased expression of multiple RTKs and that inhibition of BRD4 by JQ1 treatment was able to block this TZ-1-mediated increased RTK expression. We examined whether TZ-1 treatment resulted in increased transcription factor activity that resulted in increased RTK expression. To identify the transcription factors mediating these changes, we used a transcription factor (TF) activation profiling array to quantify DNA-binding activities of 96 transcription factors. Rh41 parental cells were treated with TZ-1 (10μg/mL), JQ1 (500nM) alone or in combination for 24 Hr and nuclear extracts were incubated with the TF activation-profiling arrays. The only transcription factor that showed increased activity following TZ-1 treatment was SOX18 (increased 18-fold; Supplemental Figure S9). SOX18 expression levels were assessed in Rh41 cells by qRTPCR and a 1.7 fold increase was seen with TZ-1 treatment that was blocked by combination with JQ1 (Figure 4A). In the Rh41 parental line, increased SOX18 mRNA and protein expression was detected following TZ-1 treatment, and SOX18 levels were high in TZ1-resistant cells in the presence or absence of antibody (Figure 4B, 4C). Exogenous overexpression of SOX18 resulted in an 8-fold increase in IRβ and a 2-fold increase in PDGFRα expression, but did not result in a substantial increase in the other RTKs tested (Figure 4D). Knockdown of SOX18 (45% knockdown efficiency) in Rh41 parental cells, however, decreased transcript levels for nearly all 12 RTKs tested (Figure 4E). Further, knockdown of SOX18 together with inhibition of IGF-1R resulted in marked cell death as determined by caspase 3/7 activity (Figure 4F).

Figure 4. Knockdown of SOX18 results in reduced cell proliferation and increased cell death following IGF-1R inhibition.

A) Relative mRNA levels of SOX18 was determined by RT-qPCR in Rh41 parental cells treated with TZ-1 (10μg/mL) or JQ1 (500nM) for 48 Hr and fold expression is shown. B). Western blot showing levels of SOX18 protein following TZ-1 treatment in Rh41 parental or TZ-1-resistant cells for 48 Hr. Results from the quantification of the western blot is graphed. C) Relative mRNA levels of SOX18 was determined by RT-qPCR in Rh41 parental or resistant cells treated with TZ-1 (10μg/mL) for 48 Hr and fold expression is shown. D) Relative mRNA levels of multiple RTKs were determined by RT-qPCR in Rh41 Control and SOX18-overexpressing cells. E) Relative mRNA expression of multiple RTKs were determined by RT-qPCR in siControl or siSOX18-transfected Rh41 parental cells, fold change is shown. F) Rh41 shControl and shSOX18 stable cells were plated at 1.5 × 10⁴ cells/well in a 96 well plate. The next day they were treated with the IGF-1R targeting monoclonal antibody TZ1 (10μg/mL) in the presence of 5μM Caspase 3–7 reagent (Essen BioSciences) and imaged with the Incucyte every 2 Hr. The kinetic response to TZ1 treatment was plotted for shControl and shSOX18 stable cells and representative images at 24 Hr post-treatment for each is shown.

Knockdown of SOX18 or BRD4 suppressed TZ-1-induced RTK expression in intrinsically resistant cells but did not sensitize cells to TZ-1.

We investigated the effect of knocking down SOX18 in the cell lines that were intrinsically resistant to TZ-1: ES2, ES4, Rh30 and RD (Figure 5A). Treatment with TZ-1 resulted in modest increase in RTK mRNA levels in these cells that was detected by RT-qPCR. Knocking down SOX18 reduced expression of a limited number RTKs that was sustained when treated with TZ-1 (Figure 5B). Notably, FGFR1 was reduced in three cell lines, ALK was reduced in the EWS cells, while PDGFRα and PDGFRβ were reduced in the RMS cells. While a change in gene expression was detected following knockdown of SOX18, it did not result in a decrease in cell viability even in the presence of TZ-1 (Figure 5C).

Figure 5. Intrinsically resistant cells have a slight increase in RTK expression following TZ1 treatment that can be blocked by knockdown of SOX18.

A) Relative mRNA levels of SOX18 following TZ1 treatment (10μg/mL) in ES2, ES4, Rh30 and RD cells were determined by RT-qPCR and fold change is shown. B) Relative mRNA levels of multiple RTKs were determined by RT-qPCR in ES2, ES4, Rh30 and RD cells treated with IgG or TZ-1 (10μg/mL) for 48 Hr. Fold change in expression is shown. C) Cell proliferation was evaluated over 72 Hr in cells transfected with siControl or siSOX18 and treatment with IgG (10μg/mL) or TZ1 (10μg/mL). (***p < 0.001, ****p < 0.0001, n = 3 by 2 way ANOVA. Bars are means +/− SD).

We undertook similar knockdown experiments using siBRD4 (Figure 6A). Knockdown of BRD4 had greatest effect in blocking TZ1-induced RTK expression in Rh30 cells. In other cell lines, the effects were more variable although knockdown of BRD4 prevented TZ-1 induction of ALK and ERBB3 in ES2 and ES4 cells (Figure 6B). However, the effect of BRD4 suppression on TZ1-mediated inhibition of proliferation was modest (Figure 6C).

Figure 6. Intrinsically resistant cells have a slight increase in RTK expression following TZ-1 treatment that can be blocked by knockdown of BRD4.

A) Relative mRNA levels of BRD4 following TZ-1 treatment (10μg/mL) in ES2, ES4, Rh30 and RD cells were determined by RT-qPCR and fold change is shown. B) Relative mRNA levels of multiple RTKs were determined by RT-qPCR in ES2, ES4, Rh30 and RD cells treated with IgG or TZ-1 (10μg/mL) for 48 Hr. Fold change in expression is shown. C) Cell proliferation was evaluated over 72 Hr in cells transfected with siControl or siBRD4 and treatment with IgG (10μg/mL) or TZ-1 (10μg/mL). (***p < 0.001, ****p < 0.0001, n = 3 by 2 way ANOVA. Bars are means +/− SD).

Discussion

An extensive literature supports the IGF-1R-dependence of sarcoma cells for proliferation. Further, the role of IGF-1R in maintaining cell viability under stress conditions, including cancer chemotherapeutic agents (37–40) (with some exceptions (41)) has been reported. These observations have led to clinical testing of IGF-1R-targeted antibodies in pediatric sarcoma patients. These antibodies show specificity for the IGF-IR although they may also inhibit chimeric receptors formed through heterodimerization with the insulin receptor. To date, there have been very few serious side effects resulting from this treatment. Hyperglycemia, when present, has been mild and has only been seen with some of the antibodies tested (42, 43). However, the antitumor activity of antibodies that block receptor-ligand binding has been disappointing. Although IGF-1R-targeted antibodies have caused tumor regressions in preclinical sarcoma models, such responses have been quite infrequent (11, 14, 15, 44, 45). Further, studies in mice may overestimate the activity of IGF-1R-targeted antibodies because levels of circulating IGF-2 in mice are either very low or undetectable (46). In humans, circulating IGF-2 could signal through the IN-R to circumvent IGF-1R blockade. However, the overarching problem is intrinsic resistance in both preclinical animal models and clinical trials (47) (17–19, 25, 42, 43). Several mechanisms including cancer cell intrinsic, and cancer cell autonomous, have been proposed to explain resistance (20, 22, 24, 48–50). The aim of our study was to identify mechanisms of resistance to IGF-1R targeted therapies in panels of RMS and EWS cell lines. Of 20 cell lines, only one, Rh41, was sensitive to IGF-1R blockade in vitro. Thus, 95% of cell lines were intrinsically resistant. These intrinsically resistant lines share a complex profile of de novo phosphorylated (activated) RTKs. Further, blockade of IGF-1R did not lead to significant changes in other phospho-RTKs, suggesting that there was no requirement for compensation when IGF-1R signaling is blocked. While other explanations to account for intrinsic resistance can be entertained (e.g. RAS mutation in RD and Rh36 cells), the observation common to all of the sarcoma lines insensitive to IGF-1R blockade, was complex patterns of RTK phosphorylation. The conclusion from these studies is that these other RTKs drive redundant signaling that makes IGF-1R inhibition inconsequential, at least under these experimental conditions.

In contrast to the cell lines with intrinsic resistance, TZ-1 treatment resulted in enhanced levels of multiple phosphorylated RTKs in Rh41 cells, suggesting compensation for IGF-1R blockade. The Rh41 cell line has been used previously to select for resistance to both IGF-1R antibody (MAB391) and small molecule IGF-1R/IN-R dual inhibitors (23). Huang et al., reported increased AXL in the line resistant to MAB391, but did not report rapid induction of multiple RTKs in parental Rh41 cells in response to antibody exposure. Rh41 cells resistant to either SCH717454 (AB1R) or TZ-1 (AB2R) in the presence of TZ-1 showed enhanced expression of several RTKs (FGFR4, AXL, EGFR), whereas PDGFRα/β and VEGFR1 seemed elevated to a greater level in cells further selected for resistance to TZ-1. This suggests that these receptors are activated in response to inhibition of IGF-1R and this adaptive kinome reprogramming provides a method of escape from targeted therapy (34, 51, 52). The results from the RPPA analysis validated the increase in AXL signaling in the Rh41 parental cell line and the ErbB3 increase in the Rh41-AB1R cell line. Interestingly, BRD4 expression was increased in the Rh41 parent cell line following treatment (Supplemental Figure S6), suggesting that BRD4 may contribute to the development of resistance over time. The RPPA signature that correlated with increased resistance to TZ-1 had increased ErbB receptor-family activity but less PI3K/AKT signaling through a reduction in AKT S473/T308 and p70S6.

In vitro, Rh41 cells were the most sensitive compared to other sarcoma cell lines to SCH717454 and TZ-1, however the sensitivity to TZ-1 in vivo had not been evaluated. Of note, TZ-1 treatment of Rh41 xenografts was far more effective than reported for MAB391 (49), or two other IGF-1R-targeted antibodies tested (SCH717454 (11) and IMC-A12 (12, 53)). MAB391, SCH717454 and IMC-A12 slowed Rh41 tumor growth by >50%, although because tumors increased in volume by >25% during treatment this still represents progressive disease. In contrast, TZ-1 induced complete regression of parental Rh41 tumor and partial regression (>50% tumor volume reduction) of the TZ-1 resistant line. Similarly, EFS was reduced in Rh41TZ-1R xenografts compared to the parental line (EFS 63.87 ± 11.2 days vs 28.20 ± 20.2 days). Thus, in vivo, Rh41 is highly sensitive to TZ-1 and the line selected for acquired resistance in vitro was less sensitive to treatment than the parental line, thus supporting the relevance of the in vitro model, and constitutive activation of RTKs as a mechanism for both acquired and intrinsic resistance to IGF1R-targeted antibodies.

Because of the rapid increase in multiple RTKs, we considered that the action of epigenetic super enhancers such as BET might be involved. The antiproliferative effects of JQ1 in Ewing sarcoma cell lines has been attributed to downregulation of IGF-1 and decreased autocrine signaling (54), however, it was not demonstrated that exogenous IGF-1 could restore proliferation in the presence of JQ1, and BET inhibitors may have pleiotropic effects (55). Of note, rhabdomyosarcomas are largely driven by IGF-2 (2, 56) rather than IGF-1 as in Ewing sarcoma, thus distinguishing these studies. Synergy between JQ1 and BMS-754807, has been reported for Ewing sarcoma cells (57), however, whereas antibodies are specific for IGF-1R, small molecule drugs equally inhibit IGF-1R and IN-R, hence may have different effects on signaling. Further, Rh41 xenografts are not sensitive to BMS-754807 (58), hence the effects of TZ-1 are clearly different from small molecule inhibitors, that inhibit IGF-1R and IN-R equally.

BET sensitivity has been described for Rh41 cells previously (59, 60). In the Gryder et al study (60) it was proposed that JQ1 reduced enhancer recruitment to the Pax3-Foxo1 fusion oncogene, leading to inhibition of proliferation. However, another alveolar rhabdomyosarcoma cell line driven by the same Pax3-Foxo1 fusion, Rh30, is not sensitive to JQ1 in vitro (59), thus questioning the generality of the effect of JQ1 on enhancer recruitment. At least part of the antitumor activity of JQ1 in xenograft models may be due to JQ1 mediated inhibition of angiogenesis (59). In Rh41 cells treatment with JQ1 repressed TZ-1-mediated overexpression of IGF-1R, IRB, AXL, PDGFRα/β, VEGFR-1 but not FGFR1–4 and ALK. Further, there was an additive inhibitory effect on viability when JQ1 was combined with TZ-1. Because of the rapid increase in RTK expression, the synergism is unlikely mediated through the suppression of compensatory activation of MAPK signaling, which is common in adult tumors treated with inhibitors targeting the PI3K/AKT axis (29). However, although a comprehensive analysis of kinome reprogramming was not performed in our study, our data do not exclude the possibility that BET inhibition suppresses the activation of other kinases that may compensate the blockade of IGF1R signaling in Ewing sarcoma cells.

Because JQ1 abrogated RTK mRNA induction by TZ-1, we screened Rh41 cells for transcription factors that were suppressed by JQ1 in the presence of TZ-1 antibody. Only one, SOX18, was identified. Interestingly, SOX18 levels were elevated by treatment with TZ-1 in parental Rh41 cells, and were elevated in Rh41-AB2R, TZ-1 resistant cells in the presence or absence of TZ-1. Further, knockdown of SOX18 decreased expression of multiple RTKs, whereas overexpression of SOX18 induced only IRβ in Rh41 cells. While knockdown of SOX18 induced a low level of apoptosis, this was exacerbated by co-treatment with TZ-1.

While in cells intrinsically resistant to TZ-1, the antibody did not change phospho-RTK profiles, there was a small effect on RTK mRNA levels. TZ-1 increased SOX18 mRNA and siSOX18 suppressed mRNA levels for some RTKs although there was no common pattern of suppression in the cell lines studied. Further, knockdown of SOX18 or BRD4 did not sensitize cells to TZ-1 in these cell lines.

In summary, of the 20 childhood sarcoma cell lines tested, Rh41 cells appear to be unique due to their sensitivity to antibodies that block ligand binding to the IGF-1R. Notable differences between Rh41 cells and intrinsically resistant lines is the complexity of activated RTKs in resistant lines, and the failure of IGF-1R blockade to induce compensation. In contrast, IGF-1R blockade leads to rapid kinome reprogramming in Rh41 cells and many of the upregulated RTKs are associated with acquired resistance. Our conclusion from this study is that intrinsic resistance to IGF-1R blockade is through signaling redundancy, such that suppressing IGF-1R signaling is not essential for proliferation and survival.