Insulin-Mediated Signaling Facilitates Resistance to PDGFR Inhibition in Proneural Hpdgfb-Driven Gliomas Damian A

Published OnlineFirst January 30, 2017; DOI: 10.1158/1535-7163.MCT-16-0616

Cancer Biology and Signal Transduction Molecular Cancer Therapeutics Insulin-Mediated Signaling Facilitates Resistance to PDGFR Inhibition in Proneural hPDGFB-Driven Gliomas Damian A. Almiron Bonnin1,2, Cong Ran1,2, Matthew C. Havrda1,2, Huan Liu1,2, Yasuyuki Hitoshi1,2,3, Zhonghua Zhang1,2, Chao Cheng1,2,4, Matthew Ung1,4, and Mark A. Israel1,2,5

Abstract

Despite abundant evidence implicating receptor tyrosine the expression and activation of the insulin receptor (IR)/insulin kinases (RTK), including the platelet-derived growth factor recep- growth-like factor receptor (IGF1R) for tumor cell proliferation tor (PDGFR), in the pathogenesis of glioblastoma (GBM), the and survival. Cotargeting IR/IGF1R and PDGFR decreased the clinical use of RTK inhibitors in this disease has been greatly emergence of resistant clones in vitro. Our ﬁndings characterize a compromised by the rapid emergence of therapeutic resistance. To novel model of glioma recurrence that implicates the IR/IGF1R study the resistance of proneural gliomas that are driven by a signaling axis in mediating the development of resistance to PDGFR-regulated pathway to targeted tyrosine kinase inhibitors, PDGFR inhibition and provide evidence that IR/IGF1R signaling we utilized a mouse model of proneural glioma in which mice is important in the recurrence of the proneural subtype of glioma develop tumors that become resistant to PDGFR inhibition. in which PDGF/PDGFR is most commonly expressed at a high We found that tumors resistant to PDGFR inhibition required level. Mol Cancer Ther; 16(4); 705–16. 2017 AACR.

Introduction recur. Therefore, significant efforts are currently under way to find more effective approaches to treat GBM including the use of Receptor tyrosine kinases (RTK) recognize extracellular signals tyrosine kinase inhibitors (TKI). and activate intracellular adaptor proteins, which in turn mediate TKIs have been used in the clinic to treat tumors in which there downstream effectors such as phosphatidylinositol-4,5-bisphos- was evidence for the pathologic activation of the RTK that they phate 3-kinase (PI3K)/protein kinase B (AKT) or mitogen-acti- target, and in some cases this approach has dramatically improved vated protein kinases (MAPK)/extracellular signal–regulated patient survival (1, 2). Examples of this strategy currently being kinases (ERK1/2) to modulate important pathologic properties used for treatment include epidermal growth factor receptor of tumors including proliferation, resistance to apoptosis, and cell (EGFR) inhibition for lung cancer, human epidermal growth motility (1). Activation of RTKs has been linked to the initiation, factor 2 (HER2) inhibition for breast cancer, and breakpoint maintenance, and progression of many different tumor types cluster region (BCR)-Abelson (ABL) inhibition for chronic mye- including glioblastoma (GBM; refs. 2, 3). GBM is one of the most loid leukemia (CML; ref. 5). Recent transcriptomic analysis of common malignant intracranial tumors in adults (4). Despite high-grade gliomas resulted in the characterization of four clin- optimal medical management, the median survival of patients ically relevant subtypes: proneural, neural, classical, and mesen- with GBM is only 12 to 15 months (4). Surgery, chemotherapy, chymal glioblastoma (6). Each of these subtypes is defined by a and radiation can contribute to an initial response, but most specific molecular signature (6). Genetic alterations resulting in GBMs progress despite therapy or become resistant to therapy and increased platelet-derived growth factor (PDGF)/platelet-derived growth factor receptor (PDGFR) signaling are a characteristic feature of the proneural subgroup of gliomas (6–8). The impor- 1Department of Molecular and Systems Biology, Geisel School of Medicine at tance of PDGF signaling in the proneural subgroup of high-grade Dartmouth, Hanover, New Hampshire; 2Norris Cotton Cancer Center, Geisel gliomas, which is the least responsive to therapy, provided a School of Medicine at Dartmouth, Lebanon, New Hampshire; 3Department of strong rationale for clinical trials to evaluate PDGFR inhibitors 4 Neurosurgery, Rosai Hospital, Kumamoto, Japan. Department of Biomedical in the treatment of this type of glioma (9). However, the clinical Data Science, Geisel School of Medicine at Dartmouth, Hanover, New Hampshire; efficacy of RTK inhibitors has been shown to be compromised by 5Departments of Medicine and Pediatrics, Geisel School of Medicine at Dartmouth, Hanover, New Hampshire. the rapid emergence of resistance in both glioma (4, 10) and other tumor types (11). Note: Supplementary data for this article are available at Molecular Cancer Therapeutics Online (http://mct.aacrjournals.org/). Multiple mechanisms of resistance to RTK-targeted therapy, including mutation of the active site, amplification of the targeted D.A.A. Bonnin and C. Ran contributed equally to this article. receptor, and activation of alternative RTKs, have been identified Corresponding Author: Mark A. Israel, Geisel School of Medicine at Dartmouth, in a variety of tumors (12). Given this diversity of mechanisms One Medical Center Drive, Lebanon, NH 03756. Phone: 603-653-3611; Fax: 603- known to contribute to the development of RTK resistance, we 653-9003; E-mail: [email protected] sought to characterize therapeutic resistance to PDGF/PDGFR doi: 10.1158/1535-7163.MCT-16-0616 inhibition so that novel clinical strategies to inhibit these key 2017 American Association for Cancer Research. drivers of tumor progression could be developed. This is

www.aacrjournals.org 705

Almiron Bonnin et al.

especially important in the context of proneural gliomas where Antibodies detecting cleaved and uncleaved PARP, total the current standard of therapy has shown the least benefit relative PDGFRa/b, phospho-PDGFRa/b, total AKT, phospho-AKT, to the other subtypes of glioma (6). total ERK1/2, phospho-ERK1/2, phospho-ErbB2, total ErbB2, phospho-IGF1R/IR, total IR, and total IGF1R were purchased Materials and Methods from Cell Signaling Technology. Each antibody was diluted at 1:1,000 in TBST containing 5% BSA for overnight immuno- Cell cultures blotting. An HRP-conjugated actin antibody (Sigma) was used Primary mouse tumor sphere cells (TSCs) from CNS glioma or at 1:50,000 to detect b-actin as a loading control. The RTK from flank allografts were isolated as previously described (13). Antibody Array was purchased from R&D Systems, Inc. Fol- The first two passages of TSCs were maintained in EGF-containing lowing treatment of TSC-S1 and TSC-R1 with doxycycline (20 ng/mL) DMEM/F12 with 1X B27 supplement with insulin. (1 mg/mL) for 48 hours, whole-cell lysates were used to probe TSCs were then propagated in DMEM/F12 with 1X B27 supple- RTK antibody arrays, upon which a series of antibodies to RTKs ment with or without insulin, in a humidified 5% CO2 atmo- were spotted in duplicate. sphere at 37 C.

Mouse allograft Immunohistochemistry To establish doxycycline-resistant allografts from doxycy- Tumor tissues were perfusion fixed by immersion in 4% PFA cline-sensitive TSCs, we injected 100,000 doxycycline-sensitive overnight, and then embedded in paraffin. Immunohistochem- TSCs into Nu/Nu or C57BL/6J syngeneic mice. Following for- ical analysis was performed by the Norris Cotton Cancer Center mation of a palpable tumor, animals were treated with oral Research Pathology Shared Resource (Lebanon, NH) using stan- doxycycline and were observed for regression in allograft size. dard techniques. Antibodies detecting PDGFRa/b and IR/IGF1R Oral doxycycline was always administered at 1 g/L in the were purchased from Cell Signaling Technology. Each was diluted animal's drinking water. All tumors initially regressed. After at 1:500. three months of treatment, recurrent flank allografts were excised and cultured. All animal care was in accordance with Real-time PCR approved IACUC-approved protocols. RNA was isolated using the RNeasy Kit (Qiagen) and reverse transcribed using the iScript cDNA synthesis kit (Bio-Rad). To Basso mouse scale assessment examine mRNA levels, we utilized the iQ SYBR Green Supermix To assess neurological deficits in tumor-bearing mice, func- (Bio-Rad) and MyIQ Real-Time PCR system (Bio-Rad). Expres- tional hind limb strength is evaluated using the Basso mouse scale sion data were analyzed relative to untreated or DMSO control as previously described (14). and normalized to 18S Ct values. Primer sequences are provided in Supplementary Table S2. Drugs AG1295 (Sigma-Aldrich; ref. 15), OSI-906 (Selleckchem; ref. 16), Imatinib (Sigma-Aldrich), LY294002 (Sigma-Aldrich; Generation and application of RTK signature ref. 17) and UO126 (Qiagen; ref. 18) stock solutions were The RTK signature was derived from previously published data fi prepared in DMSO and stored in 20 C. Doxycycline (Santa from our lab (20). The microarray pro le of untreated GFAP/tTA: Cruz Biotechnology) is diluted in PBS as a stock and stored in TRE/hPDGFB cells was compared with doxycycline-treated cells P 20 C. All drugs were diluted into fresh media immediately by taking the log2-fold change of all genes in the array. A value – before experimentation. was computed for each gene and log10 transformed. These values were used to generate an upregulated and downregulated profile; fi SiRNA transfections in the upregulated pro le, all downregulated genes were assigned fi TSCs were seeded at 50,000 cells/mL and transfected with a zero value and vice versa for the downregulated pro le. The Neofx (Ambion) per the manufacturer's manual. Pooled siRNA BASE algorithm was used to calculate an RTK score for each patient fi targeting IR (Thermo Scientific), IGF1R (Thermo Scientific), and by taking the tumor gene expression pro le and the upregulated the control siRNA siGlo (Thermo Scientific) were transfected at 25 and downregulated signatures as input (21). nmol/L final concentration. Data queries and statistical analysis Cell growth and cell death All data are representative of three or more independent experi- TSCs were seeded into 6 or 96 well plates at a concentration of ments, presented as the mean SEM. We used the two-tailed 50,000 cells/mL and grown for 7 days. As a surrogate measurement student t test to evaluate differences between two groups, and one- of cell number, total ATP was examined in lysed cells based on the way ANOVA to evaluate differences in multiple groups, followed luminescence produced by the reaction of ATP with luciferase and by the Tukey correction. Dose–response curves were compared D-luciferin (ATPlite, PerkinElmer). For some experiments, cell using two-way ANOVA, and each time point was compared with counts were determined on a Nexolom cell counter. Cell viability the one-way ANOVA and corrected using the Tukey correction. fl was monitored by uorescence-activated cell-sorting (FACS) anal- The frequency of recurrence in doxycycline and OSI-906 treated ysis of propidium iodide (PI) exclusion assay (19). TSCs was analyzed using a contingency table. Gene expression data of human GBM or human proneural GBM were obtained Western blotting and RTK antibody array through The Cancer Genome Atlas (TCGA). Heat map and hier- Cells lysed in RIPA buffer were size fractionated in 8% SDS- archical clustering were performed by GENE-E software. All PAGE gels and subsequently transferred to PVDF (Millipore). statistical analyses were performed by Graphpad Prism 5 software.

706 Mol Cancer Ther; 16(4) April 2017 Molecular Cancer Therapeutics

GBM Resistance to PDGFR Inhibition

Results TSCs derived from recurrent proneural hPDGFB-driven tumors are resistant to PDGF/PDGFR inhibition Seeking an in vivo model to study the mechanism of resistance While hPDGFB suppression mediated by doxycycline expo- to RTK inhibition in high-grade gliomas, we took advantage of the sure (Supplementary Fig. S2A) dramatically inhibited the PDGF/PDGFR-driven mouse model of proneural glioma devel- growth of TSC-S lines, it had no significant effect on the growth oped previously in our laboratory (22). This model uses the glial of TSC-R cultures, regardless of whether they were derived from fibrillary acidic protein (GFAP) promoter to drive the expression de novo hPDGFB-driven CNS glioma or hPDGFB-driven flank of the tetracycline transactivator (tTA) which regulates expression allografts (Fig. 1A and Supplementary S2A). Further validating of a transgene encoding human platelet-derived growth factor B theresistanceofTSC-RstoPDGF/PDGFR signaling inhibition, by binding to a tetracycline responsive element that is located just TSC-Rs exhibited a significantly decreased sensitivity to treat- upstream of the transgene (TRE; ref. 11). This GFAP/tTA:TRE/ ment with Imatinib, a pharmacologic inhibitor of PDGFR that hPDGFB mouse model produces tumors that were characterized readily suppressed the growth of TSC-Ss (Fig. 1B; ref. 25). To in our laboratory to faithfully model the proneural subtype of examine the effect of PDGF/PDGFR signaling on the viability of glioma (20) as defined by Verhak and colleagues (6). Expression these TSCs, we used a dye-exclusion assay. After exposing TSCs of the hPDGFB ligand in the GFAP/tTA:TRE/hPDGFB mouse to doxycycline or AG1295, a chemical inhibitor of PDGFR, we drives tumorigenesis. PDGFB is recognized by both PDGFRA and incubated these cultures with propidium-iodide (PI) and eval- PDGFRB homodimers or heterodimers (23). Exposure to either uated them. Inhibition of PDGFR using either doxycycline or tetracycline or doxycycline inhibits transcription of the hPDGFB AG1295 induced cell death in TSC-S cultures but not in TSC-R transgene to decrease PDGFR signaling, thereby mimicking cultures (Fig. 1C and 1D). To confirm the induction of apo- important aspects of the therapeutic activity of RTK inhibitors ptotic cell death, we examined PARP cleavage following inhi- (19, 20, 22). bition of PDGF/PDGFR in the doxycycline-sensitive and resis- To study the effectiveness of targeted RTK inhibition on tant TSCs (Fig. 1E). Cleaved PARP, which provides evidence of established PDGF/PDGFR-driven high-grade proneural glio- apoptosis, was detected only in TSC-Ss and not in TSC-Rs mas, we treated GFAP/tTA:TRE/hPDGFB mice showing evi- following the inhibition of PDGF/PDGFR signaling. This pro- dence of glioma. We withdrew oral doxycycline administration vides additional evidence that cell survival of TSC-Rs was not from adult GFAP/tTA:TRE/hPDGFB mice to allow hPDGFB affected by inhibition of PDGF/PDGFR signaling (Fig. 1C–E). production. Then, when these mice exhibited symptoms of a These in vitro studies indicate that TSCs derived from our in vivo CNS tumor, we reexposed them to oral doxycycline. In this models retain the PDGF sensitivity of the tumors from which mouse model, progressive paralysis is an effective surrogate for they were derived. tumor development (22), and it can be measured using the Basso Mouse Scale for Locomotion (14, 22, 24). We evaluated Proneural TSCs resistant to PDGF/PDGFR inhibition require mice during the course of oral doxycycline administration and PI3K/AKT and MAPK/ERK signaling for cellular proliferation found that even though all mice (n ¼ 30) treated with doxy- Because the function of the PI3K/AKT and MAPK/ERK1/2 cycline exhibited evidence of tumor regression, some animals signaling pathways in driving cellular proliferation is well relapsed after a period of remission (n ¼ 8) developing dis- characterized (26, 27), we examined whether PDGFR activation abilities in locomotion we could recognize as evidence of regulated these second messengers in TSCs. As expected, treat- tumor progression (Supplementary Fig. S1A and S1B). Tumors ment of all TSC lines with doxycycline reduced hPDGFB mRNA arising in these animals that became resistant to treatment were to undetectable levels (Supplementary Fig. S2A) and conse- used to prepare TSCs, designated TSC-R1, TSC-R2, and TSC-R3 quently decreased the levels of PDGFR phosphorylation as (Supplementary Table S1). These cultures were evaluated in shown in several representative cultures (Fig. 2A). Also, treat- subsequent experiments. ment with AG1295, a PDGFR inhibitor, resulted in a marked We also developed an allograft model to study resistance to reduction of PDGFR phosphorylation in all treated cell cultures inhibition of PDGFR in glioma driven to grow by PDGF/PDGFR. as shown in several representative cultures (Fig. 2B; ref. 15). In We prepared TSCs from individual tumors that arose in GFAP/ sensitive lines, doxycycline, as well as pharmacologic inhibition tTA:TRE/hPDGFB animals that had not been exposed to doxycy- of PDGFR with AG1295 or imatinib, suppressed AKT and cline. The growth of these TSC cultures was sensitive to doxycy- ERK1/2 phosphorylation as shown for several representative cline, and they were designated TSC-S1 and TSC-S2 (Fig. 1A; cultures (Fig. 2A and B), reduced cellular growth, and increased Supplementary Fig. S1A and Supplementary Table S1). Subse- cell death (Fig. 1A–D). Consistent with these data, expression quently, we injected these TSCs into the flank of syngeneic of a constitutively activated AKT (myristoylated-AKT) rescued a animals. Once flank tumors were approximately 300 mm3, ani- representative sensitive TSC line, TSC-S2, from doxycycline- mals bearing these tumor allografts were treated with oral doxy- mediated growth suppression (Supplementary Fig. S3A) illus- cycline (Supplementary Fig. S1C and S1D). As we observed in the trating the importance of AKT activation in PDGF/PDGFR- spontaneous model described above, we found that all mice mediated growth of TSC-Ss. In TSC-R cultures, however, PDGFR treated with doxycycline exhibited tumor regression (n ¼ 17), inhibition had less of an effect on the canonical PDGFR but some animals (n ¼ 13) relapsed after a period of remission of downstream targets, AKT and ERK1/2 (Fig. 2A and B). Whether 1 to 3 months (Supplementary Fig. S1D). These recurrent tumors the PDGFR inhibition was mediated by doxycycline treatment grew despite continued oral doxycycline treatment. TSCs from or pharmacological inhibition, it had less or no detectable these relapsed tumors provided a second set of cellular reagents effect on the phosphorylation of AKT and ERK1/2 as well as from resistant tumors, designated TSC-R4, TSC-R5, and TSC-R6 no significant effect on cellular growth in TSC-Rs (Figs. 1A (Supplementary Table S1) with which to pursue these studies of and 2A). These findings indicate that TSC-Rs are able to main- therapeutic resistance. tain AKT and ERK1/2 activation in the absence of PDGFR

www.aacrjournals.org Mol Cancer Ther; 16(4) April 2017 707

Almiron Bonnin et al.

AB150 150 S1 S1 S2 S2 125 125

R1 R1 100 100 R2 R2 R3 R3 75 R4 75 R4 R6 R6 50 50 Relative cell growth Relative cell growth

(normalized to untreated, %) 25 25 (normalized to untreated, %)

0 0 -8 -7 -6 -5 -4 -3 -2 -1 0 1 0 0 0 0 0.125 0.25 0.5 1 2 4 8 1 10 1 10 10 1 10 1 10 10 Imatinib (µmol/L) Doxycycline (µg/mL)

CD 50 80 70 40 60 30 50 40 20 30 % PI (+) cells 10 % PI (+) cells 20 10 0 0 Doxycycline –––––––+ ++++++ AG1295 –––––––+++++++ (1 µg/mL) S1 S2 R1 R2 R3 R4 R6 (10 µmol/L) S1 S2 R1 R2 R3 R4 R6 TSC TSC

E TSC S1 S2 R1 R2

Doxycycline: +– +–+–+– MW (kDa) 116 PARP - uncleaved

89 PARP - cleaved

42 Actin

Figure 1. TSCs derived from recurrent GFAP/tTA:TRE/hPDGFB tumors are resistant to hPDGFB suppression. A, Relative cell growth measured by ATP content of TSC-Ss and TSC-Rs cultures treated with doxycycline for 7 days. Data points represent ﬁve independent experiments conducted in triplicate presented as the mean SEM. When comparing all TSC-Ss lines to all TSC-Rs lines, P < 0.0001. B, Relative cell growth measured by ATP content of TSC-Ss and TSC-Rs cultures treated with imatinib for 7 days. Data points represent ﬁve independent experiments conducted in triplicate presented as the mean SEM. When comparing all TSC-Ss lines to all TSC-Rs lines, P < 0.0001. C, Cell death measured by FACS analysis of PI positivity in TSCs grown in the presence (þ) or absence () of doxycycline (1 mg/mL) for 72 hours. Data represent three independent experiments conducted in triplicate presented as the mean SEM; , P < 0.01; , P < 0.001. D, Cell death measured by FACS analysis of PI positivity in TSCs grown in the presence (þ) or absence () of AG1295 (10 mmol/L) for 72 hours. Data represent three independent experiments conducted in triplicate presented as the mean SEM; , P < 0.001. E, Western blot analysis of PARP cleavage in representative sensitive and resistant TSCs in the presence (þ) or absence () of doxycycline (1 mg/mL) for 48 hours.

signaling. This loss of PI3K/AKT and MAPK/ERK regulation by To evaluate whether the persistent activation of PI3K/AKT and PDGF/PDGFR provides evidence that PI3K/AKT and MAPK/ MAPK/ERK1/2 mediates the resistance of these glioma-derived ERK may still be key mediators of proliferation and survival in TSCs to PDGFR inhibition, we assessed the growth of these TSC-Rs. Moreover, it suggests TSC-Rs have developed an alter- cells after treatment with AG1295, LY294002, or UO126, phar- native mechanism for maintaining the activation of these path- macologic agents that speciﬁcally target PDGFR, PI3K, or ERK1/2, ways, thereby conferring resistance to PDGFR inhibition. respectively (Fig. 2C; refs. 1, 26, 27). In TSC-S cultures, we

708 Mol Cancer Ther; 16(4) April 2017 Molecular Cancer Therapeutics

GBM Resistance to PDGFR Inhibition

TSC A B TSC S1 S2 R4 R6 S1 R4 S2 R6 Doxycycline: – + – + – + – + AG1295: – + – + – + – + MW (kDa) MW (kDa) 185 Phospho-PDGFRα/β 185 Phospho-PDGFR α/β

60 Phospho-AKT 60 Phospho-AKT

Total AKT 60 60 Total AKT

Phospho-ERK1/2 44/42 44/42 Phospho-ERK1/2

44/42 Total ERK1/2 44/42 Total ERK1/2

TSC-R4 D LY294002: – + – UO126: – – + C MW (kDa) * * ** 125 ** * * * * 60 Phospho-AKT ** * * ** * ** * * 100 60 Total AKT 75 DMSO AG1295 50 LY294002 44/42 Phospho-ERK1/2 UO126 25 Relative cell growth (normalized to DMSO, %) 0 44/42 Total ERK1/2 S1 S2 R1 R2 R3 R4 R5 R6 TSC Actin 42

E S1 S1. ErBB2 1 3 5 S2. PDGFR 2 S3. ErBB4 4 S4. MuSK S5. IR/IGF1R

R1 R1. ErBB2 1 3 2 5 R2. PDGFR 4 R3. ErBB4 R4. MuSK R5. IR/IGF1R

Figure 2. The proliferation of TSCs resistant to hPDGFB suppression is repressed by inhibition of the PI3K or MAPK pathway. A, Western blot analysis of phosphorylated PDGFR and phosphorylated and total AKT and ERK1/2 in TSCs grown in the presence (þ) or absence () of doxycycline (1 mg/mL) for 48 hours. B, Western blot analysis of phosphorylated PDGFR and phosphorylated and total AKT and ERK1/2 in TSCs grown in the presence (þ) or absence () of AG1295 (10 mmol/L) for 48 hours. C, Relative cell growth measured by ATP content of TSCs treated with DMSO, AG1295, LY294002, or UO126 at 10 mmol/L for 120 hours. DMSO was used as a control. Data represent three or more independent experiments conducted in triplicate presented as the mean þ/ SEM. , P < 0.05; , P < 0.01. D, Western blot analysis of phosphorylated and total AKT and ERK1/2 in TSC-R4 grown in the presence (þ) or absence () of LY294002 (10 mmol/L) or UO126 (10 mmol/L) for 48 hours. E, RTK Antibody Array (R&D Systems, Inc.). Following treatment of TSC-S1 and TSC-R1 with doxycycline (1 mg/mL) for 48 hours, whole-cell lysates were used to probe RTK antibody arrays, upon which a series of antibodies to RTKs were spotted in duplicate. Expressed RTKs are framed and numbered. The corresponding RTK is listed on the right.

www.aacrjournals.org Mol Cancer Ther; 16(4) April 2017 709

Almiron Bonnin et al.

observed growth inhibitory effects with PDGFR, PI3K, or ERK1/2 decreased in average by more than 50% in TSC-Rs (Fig. 3B inhibitors, as expected. In TSC-S cultures, the most significant and 3C). Consistent with these data, insulin withdrawal resulted effect of these various inhibitors on growth was observed follow- in the inhibition of the canonical RTK downstream target, AKT, in ing AG1295 inhibition of PDGFR (Fig. 2C), most likely because it TSC-Rs, and a representative culture demonstrates this in Figure leads to the inhibition of both the PI3K/AKT and MAPK/ERK1/2 3A. The proliferation of TSC-R3, which did not require the axis (Fig. 2B). In TSC-R cultures, however, we observed a different activation of PI3K or MAPK (Fig. 2C), was unaffected by insulin growth response following such drug treatments (Fig. 2C). None treatment (Fig. 3B). Importantly, doxycycline-sensitive TSC-Ss, of the resistant lines were sensitive to PDGFR inhibition, though which were isolated and cultured in the same manner as TSC-Rs, all but one of them (TSC-R3) were sensitive to the growth were insensitive to insulin withdrawal (Fig. 3B and 3C). We inhibitory effect of the PI3K inhibitor, LY294002, and the observed no altered cellular growth or changes in cell death ERK1/2 inhibitor, UO126 (Fig. 2C). The activity of these drugs following insulin withdrawal from the TSC-S cultures (Fig. 3B in a representative TSC-R cell line is demonstrated in Figure 2D. and 3C), and in the absence of insulin, the downstream growth The growth inhibitory effect of LY294002 and UO126 was sta- regulatory mediators, AKT and ERK1/2, were not activated in tistically significant in all TSC-R lines except for TSC-R3 (Fig. 2C). TSC-Ss, as demonstrated in a representative TSC-S culture These findings support a model of resistance to RTK inhibition in in Figure 3A. These findings suggest a functional role for IR and which the molecular alterations maintaining proliferation are IGF1R in the proliferation and survival of TSC-Rs. upstream of the PI3K/AKT and MAPK/ERK1/2 axis. To examine further the effect of IR/IGF1R inhibition on TSC- Rs, we evaluated the response of TSCs to OSI-906, which IR/IGF1R is required for proliferation of proneural hPDGFB- inhibits IR and IGF1R activation (ref. 16; Fig. 3D). Comparing driven TSCs after they become resistant to PDGF/PDGFR the relative cell growth and cell death of TSC-Rs and TSC-Ss in inhibition the presence of OSI-906, we found that TSC-Rs are much more To determine if an alternative RTK upstream of AKT and ERK1/2 sensitive to OSI-906 than the TSC-Ss from which they may be responsible for the maintenance of AKT and ERK1/2 were derived (Fig. 3E and F, and Supplementary Table S1). activation and the proliferation of TSC-R cultures, we examined We confirmed this finding utilizing pooled siRNAs to modu- the phosphorylation of a panel of RTKs in representative TSC late IR expression. We found that inhibition of IR expression cultures (Fig. 2E). We found increased phosphorylation of three led to a dramatic growth inhibitory effect in TSC-Rs when RTKs in the resistant line: proto-oncogene Neu (ErBB2), muscle compared with the effect seen in a representative TSC-S culture specific kinase (MuSK), and insulin receptor (IR)/insulin growth- (Supplementary Figs. S5C and S5D). The same siRNAs were like factor receptor (IGF1R; Fig. 2E). However, the increase in used in Supplemental Figs. 5C and 5D. Moreover, inhibition of MuSK and ErBB2 could not be validated by Western blot in these phosphorylation of IR/IGF1R using OSI-906 also resulted in cell lines or others. inhibition of AKT and ERK1/2 activation in TSC-R1 but not in Based on these data and previously reported findings support- TSC-S2(Fig.3D).Thesefindings are consistent with the obser- ing a role for IR/IGF1R in maintaining the activation of AKT and vation that upon PDGF/PDGFR inhibition sensitive TSCs stop ERK in GBM (3), we conducted a histologic analysis to charac- growing and die, while the resistant TSCs do not (Fig. 1A–E). terize the expression of IR/IGF1R throughout the process of Resistant cells are clearly no longer dependent on PDGF/ remission and recurrence in response to doxycycline-mediated PDGFR signaling, because they have acquired an alternative hPDGFB suppression in our allograft model. Recurrent tumors RTK, IR/IGF1R, to drive proliferation and survival through AKT that grew despite hPDGFB suppression invariably expressed IR/ and ERK1/2 activation. IGF1R uniformly throughout the tissue analyzed and at a gener- Proliferation and survival modulation by IR/IGF1R activa- ally higher level than seen in tissue from primary tumors (Sup- tion is not evident in the doxycycline-sensitive cultures, TSC-Ss plementary Fig. S4A–S4C). Consistent with this observation, the (Fig. 3B and 3C), despite the clearpresenceofIR/IGF1Rinthese expression of IR and IGF1R was elevated in TSC-R cultures sensitive cells (Fig. 3A; Supplementary Figs. S5C and S5D). compared with the levels in the parental TSC-S cultures from Therefore, we hypothesized that the functional potential of IR/ which they were derived, as demonstrated in pairs of such cell IGF1R signaling may be masked during robust PDGF/PDGFR lines when examined for the steady-state level of encoding mRNA growth stimulatory pathway signaling in the sensitive TSCs. To (Supplementary Fig. S5A and S5B) or in a representative pair of test this hypothesis, we incubated sensitive cells, TSC-Ss, in a such TSCs when examined for expression of pro-IR, total IR, or dose range (25–100 pg/mL) of doxycycline that we determined IGF1R protein (Fig. 3A). We also observed that total protein levels in preliminary experiments was adequate to inhibit growth but of IR and IGF1R decreased when insulin was present (Fig. 3A), did not cause cell death. We then evaluated the effect of insulin consistent with the increased protein turnover of RTKs following on the proliferation of these cells. Although IR/IGF1R signaling ligand binding (1). These data illustrate that IR/IGF1R is expressed did not enhance the proliferationofsensitiveTSCwhenthe in some cells within sensitive tumors and in TSC-Ss, but it PDGF/PDGFR pathway was active (Fig. 3B), when PDGF/ becomes much more abundant in resistant tumors and TSC-Rs. PDGFR activation was diminished the effect of insulin on This enhanced expression of IR/IGF1R in vitro and in vivo suggests proliferation of TSC-S became obvious (Fig. 4A and 4B). Con- that activation of a preexisting IGF1R pathway might contribute to sistent with these observations, we found that when hPDGFB the development of resistance to PDGFR inhibition in these expression was suppressed in the sensitive TSCs, insulin proneural glioma cells. contributed to maintaining the phosphorylation levels of We used insulin as a tool to determine if there was a functional downstream mediators shared by the PDGFR and IR/IGF1R role in TSC-Rs for the increased expression of IR and IGF1R, which pathways,includingSHP2,AKT,andERK1/2(Fig.4C).We respond to both insulin and IGF (28–32). We found that in the observed that AKT was particularly sensitive to IR/IGF1R inhi- absence of insulin, cell death increased and relative cell growth bition during doxycycline-mediated hPDGFB suppression, as

710 Mol Cancer Ther; 16(4) April 2017 Molecular Cancer Therapeutics

GBM Resistance to PDGFR Inhibition

A TSC B S2 R6 + insulin Insulin: +– +– - insulin MW (kDa) Phospho-IR/IGF1R 95 150

200 Pro-IR * * * *** * 100 95 Total IR 50 95 Total IGF1R Relative cell growth 60 0 Phospho-AKT S1 S2 R1 R2 R3 R4 R5 R6 (normalized to + insulin group, %) TSC 60 Total AKT

44/42 Phospho-ERK1/2

44/42 Total ERK1/2

TSC C D R1 S2

OSI-906 (µmol/L) : 0 1 5 0 1 5 50 ** *** MW (kDa) *** 40 95 Phospho-IR/IGF1R

30 95 Total IGF1R

Cell death 60 Phospho-AKT 10 (PI-positive cells, %) 60 Total AKT 0 Insulin: –+ –+ –+ –+ –+ Phospho-ERK1/2 R1 S1 R4 S2 R6 44/42

TSC 44/42 Total ERK1/2

E F 100 120 110 100 TSCs: 80 ** 90 80 S1 * 70 60 60 S2 50 R4 40 40 R6 Cell death 30 Relative cell growth 20 20 10 (PI-positive cells, %) 0 (normalized to DMSO group, %) 0.0 0.5 1.0 1.5 2.0 2.5 3.0 0 OSI-906: O .5 5 O .5 5 OSI-906 Concentration (µmol/L) S 0 S 0 M M D D S2 R6 TSC

Figure 3. Sensitivity to IGF1R/IR inhibition of TSCs derived from recurrent tumors. A, Western blot analysis of pro-IR and phosphorylated and total IGF1R/IR, AKT, and ERK1/2 in TSC-S2 and TSC-R6 in the presence (þ) or absence () of Insulin (200 ng/mL) for 48 hours in serum-free medium. B, Relative cell growth measured by ATP content of TSCs cultured in the presence (þ) or absence () of insulin (200 ng/mL) for 96 hours in serum-free medium. Data represent three independent experiments conducted in triplicate presented as the mean SEM. , P < 0.05; , P < 0.001. C, Cell death measured by FACS analysis of propidium iodide (PI) positivity of TSCs in the presence (þ) or absence () of insulin (200 ng/mL) for 72 hours in serum-free medium. Data represent three independent experiments conducted in triplicate presented as the mean SEM. , P < 0.01; , P < 0.001. D, Western blot analysis of phosphorylated and total IR/IGF1R, AKT, and ERK1/2 in TSC-R1 and TSC-S2 treated with DMSO or OSI-906 (5 mmol/L) for 48 hrs in serum-free medium. E, Relative cell growth measured by ATP content of TSCs treated with an IR/IGF1R inhibitor, OSI-906, at the indicated concentrations for 96 hours in serum-free medium. Data represent three independent experiments conducted in triplicate presented as the mean SEM. When comparing all TSC-S lines to all TSC-R lines, P < 0.0001. F, Cell death measured by FACS analysis of PI positivity of TSC-S2 and TSC-R6 treated with DMSO or OSI-906 (5 mmol/L) for 72 hrs in serum-free medium. Data represent three independent experiments conducted in triplicate presented as the mean SEM. , P < 0.05; , P < 0.01.

www.aacrjournals.org Mol Cancer Ther; 16(4) April 2017 711

Almiron Bonnin et al.

A Insulin B – + * 120 ** Doxycycline: 110 100 90 ** 0 pg/mL 80 70 60 50 40 30 Relative cell growth 20

(normalized to DMSO, %) 10 25 pg/mL 0 Doxycycline: ––++ Insulin: – + – + D

100 pg/mL 100

50 Relative cell growth

(normalized to DMSO, %) 0 C TSC-S2 LY294002: ––+ + Insulin: – + – + Insulin: ++ + ––– Doxycycline (pg/mL): 0 50 100 0 50 100 MW (kDa) 60 Phospho-AKT E

60 Total AKT 100 44/42 Phospho-ERK1/2

50 44/42 Total ERK1/2 Number of spheres

42 Actin (normalized to DMSO, %) 0 OSI-906 (µmol/L): 0 0.25 0.5 0 0.25 0.5 Doxycycline : –––+++ (0.5 ng/mL)

Figure 4. The emergence of TSCs resistant to hPDGFB suppression is decreased by IGF1R/IR inhibition during in vitro drug treatment. A, Phase-contrast microscopy (40) of TSC-S2 cultured at the indicated concentrations of doxycycline in the presence (þ) or absence () of insulin (200 ng/mL) for 48 hours. B, Relative cell growth measured by counting TSC-S2 cells cultured in the presence (þ) or absence () of doxycycline (100 pg/mL) and insulin (200 ng/mL) for 7 days. Data represent three independent experiments conducted in triplicate presented as the mean SEM. , P < 0.05; , P < 0.01. C, Western blot analysis of phosphorylated SHP2 as well as total and phosphorylated AKT and ERK1/2 in TSC-S2 grown in the presence (þ) or absence () of insulin (200 ng/mL) at the indicated concentrations of doxycycline for 48 hours. D, Relative cell growth measured by counting TSC-S2 cells cultured in the presence (þ) or absence () of LY294002 (10 mmol/L) and insulin (200 ng/mL) for 7 days. Data represent three independent experiments conducted in triplicate presented as the mean SEM. , P < 0.001. E, TSC-S2 cells were plated into 24-well plates (5,000 cells per well) and maintained in the presence (þ) or absence () of doxycycline (1 mg/mL) and OSI-906 at the indicated concentrations. The presence of spheres was examined at 6 weeks. Data represent three independent experiments conducted in triplicate presented as the mean SEM. , P < 0.01.

indicated by the sharp decrease in its phosphorylation seen is already expressed and is either active or capable of being following insulin-withdrawal from TSC-Ss in the presence of activated in TSCs from sensitive tumors. During suppression of doxycycline (Fig. 4C). Unlike our observation that insulin PDGFR, the driver RTK, IR/IGF1R is able to mediate prolifer- partially rescues the inhibition of cell growth mediated by ation of these sensitive cells through the activation of down- doxycycline (Fig. 4B), the inhibition of cell growth mediated stream mediators shared with PDGFR. by the PI3K/AKT inhibitor LY294002 was unaffected by insulin To further characterize the role of IR/IGF1R in the develop- in TSC-Ss and is demonstrated here in a representative line ment of resistance to PDGFR inhibition in hPDGFB-driven in Figure 4D. These ﬁndings suggest that the IR/IGF1R pathway proneural TSCs, we examined the effect of IR/IGF1R inhibition

712 Mol Cancer Ther; 16(4) April 2017 Molecular Cancer Therapeutics

GBM Resistance to PDGFR Inhibition

Figure 5. IGF1R is associated with survival and RTK activity in a subgroup of proneural GBMs with lower PDGFRA levels. A, Heat map and dendogram show hierarchical clustering of 95 proneural samples from the TCGA database based on RTK expression. B, Kaplan–Meier survival curves comparing IGF1R high (z > 0.5) and IGF1R low (z 0.5) proneural GBM patients within cluster B shown in A. Log- rank test comparing the survival curves has a P value of 0.0408.

on the emergence of TSC populations resistant to PDGFR however, resulted in a very significant reduction in the frequen- inhibition (Fig. 4E). We plated 5,000 freshly dispersed cy with which tumorspheres appeared (Fig. 4E). The effect of hPDGFB-driven proneural cells in 24-well plates and incubated adding OSI-906 to the doxycycline-mediated hPDGFB suppres- theminthepresenceofinsulinfor 6 weeks in increasing doses sion was dose dependent and this addition of OSI-906 achieved of OSI-906 alone or in combination with doxycycline using approximately a 90% reduction of tumorsphere formation at a DMSO as the vehicle. The concentrations of doxycycline and dose of 0.5 mmol/L when compared with the DMSO control. OSI-906 used were previously shown to significantly inhibit the OSI-906 alone, however, had no effect on tumorsphere forma- proliferation of TSC-Ss and TSC-Rs, respectively (Fig. 1A tion. We interpreted these findings to indicate that OSI-906– and 3D). After this prolonged incubation, we observed the mediated IR/IGF1R inhibition suppressed the expansion of the emergence of tumor spheres in most of the cultures exposed to subset of cells in the TSC-S cultures that were able to develop doxycycline alone. The addition of OSI-906 to doxycycline, resistance to doxycycline.

www.aacrjournals.org Mol Cancer Ther; 16(4) April 2017 713

Almiron Bonnin et al.

IGF1R is associated with survival and RTK activity in a subgroup breast, and head and neck cancers to RTK therapy in the clinic of proneural GBMs with lower PDGFRA and PDGFRB levels (41–48). However, these recent reports suggesting a role for IR/ In an initial effort to explore the clinical relevance of our IGF1R in the sensitization of tumors to various TKIs including current findings, we queried TCGA database seeking to iden- PDGFR (39–48) have not addressed the role of IR/IGF1R in tify evidence within the proneural subgroup of human GBM, acquired resistance and recurrence. suggesting a pathologic role for IGF1R. Utilizing Verhak's Our studies demonstrate that activation of IR/IGF1R is a gene expression-based molecular classification of GBM, we novel bypass RTK associated with the emergence of resistance to identified 95 proneural GBMs with complete mRNA data in PDGFRinhibitioninPDGFR-drivenproneuralGBMs.These TCGA provisional dataset (6). Euclidian hierarchical cluster- studies provide a strong rationale for deciphering the mecha- ing, based on the expression of the 58 known human RTKs, nism underlying IR/IGF1R activation in this setting. Unlike the identified two distinct patient groups: cluster "A" and cluster mechanism proposed by Akhavan, in which transcription of the "B" (Fig. 5A). Interestingly, cluster "B" has a 2.4-fold lower bypass RTK is suppressed by activation of the driver RTK in average expression of PDGFRA (P < 0.0001) and a 1.5-fold EGFRvIII mutant GBMs (49), we found that IR/IGF1R was lower average expression of PDGFRB (P < 0.0001) than was expressed regardless of PDGFR activity in our model of pro- observed in cluster "A." neural glioma (Supplementary Figs. S4 and S6). IR/IGF1R was To develop a measure of RTK activity for each GBM sample, we expressed and was either active or capable of becoming acti- derived a "RTK signature" from the expression profile of TSC-Ss vated in TSCs from primary, drug sensitive tumors (Fig. 4). As treated with and without doxycycline. We assumed that RTK resistance to PDGF/PDGFR inhibition emerges and IR/IGF1R drivers share common downstream pathways, and used this becomes the new driver of proliferation, tumor cells which can signature to derive a RTK activity score for each patient. This respond to insulin become a dominant feature of the recurrent activity score was positively correlated with the expression of tumor (Fig. 3; Supplementary Fig. S4). In resistant tumor cells, PDGFRA (P < 0.0001) and PDGFRB (P < 0.0001) in the proneural the original driver RTK, PDGFR, continues to be expressed and GBMs analyzed, which asserts its validity as a measure of RTK is phosphorylated in the presence of PDGF (Fig. 2; Supple- activity. mentary Fig. S6A and S6B), but it no longer mediates viability To test RTKs that could potentially be driving tumor progres- and proliferation through AKT and ERK1/2 activation (Figs. 1 sion in these clusters, we correlated the mRNA expression level of and 2). Rather, there is an apparent loss of dependence of AKT each RTK with RTK activity and survival. While many RTKs are and ERK1/2 activation from PDGFR signaling as a new asso- differentially expressed between clusters "A" and "B", we found ciation between IR/IG1R and AKT and ERK1/2 becomes more that in cluster "B," which has lower PDGFRA and PDGFRB levels prominent. than were observed in cluster "A," IGF1R was the only RTK These observations might be explained by the selection of a amongst all 58 known human RTKs examined, whose expression subpopulation of cells, from within the primary sensitive PDGFR- was positively correlated with both RTK activity (Spearman r ¼ driven tumors that has acquired a growth dependence on IR/ 0.2305, P ¼ 0.0384) and decreased survival (Spearman r ¼ IGF1R and thereby mediates resistance to PDGF/PDGFR inhibi- 0.3326, P ¼ 0.0068). Furthermore, in this cluster, IGF1R expres- tion upon treatment. Going forward, the development of TKI sion was associated with shorter patient survival (Fig. 5B). We combination therapies, targeting not only the driver RTKs, but interpret these data as being consistent with the possibility that also the bypass RTKs, may significantly enhance treatment out- IGF1R-mediated RTK signaling might serve as an alternative comes in the proneural subgroup of GBM. pathway driving proneural GBM progression in the absence of high PDGF/PDGFR signaling. Disclosure of Potential Conflicts of Interest No potential conflicts of interest were disclosed. Discussion IR and IGF1R are homologous proteins (3, 30, 32, 33) which Authors' Contributions share common ligands and common downstream targets (28– Conception and design: D.A. Almiron Bonnin, C. Ran, M.C. Havrda, H. Liu, 32). These RTKs, which are widely coexpressed, are known to M.A. Israel heterodimerize activating various downstream targets impor- Development of methodology: D.A. Almiron Bonnin, C. Ran, M.C. Havrda, – H. Liu, M.A. Israel tant for growth and survival (28, 34 38). Previous reports on Acquisition of data (provided animals, acquired and managed patients, IR/IGF1R in the context of therapeutic resistance to targeted provided facilities, etc.): D.A. Almiron Bonnin, C. Ran, Z. Zhang RTK therapy have recognized the sensitization of various Analysis and interpretation of data (e.g., statistical analysis, biostatistics, tumors to RTK therapy that results when an inhibitor of IR/ computational analysis): D.A. Almiron Bonnin, C. Ran, M.C. Havrda, IGF1R is used in combination with a second TKI, including C. Cheng, M. Ung, M.A. Israel PDGFR inhibitors (39–48). For example, preclinical studies Writing, review, and/or revision of the manuscript: D.A. Almiron Bonnin, a b C. Ran, M.C. Havrda, M.A. Israel have shown that dual targeting of IGF1R and PDGFR / act Administrative, technical, or material support (i.e., reporting or organizing synergistically to inhibit the growth of high-grade gliomas (39– data, constructing databases): D.A. Almiron Bonnin, C. Ran, Y. Hitoshi, 41). Similarly, it has been reported that that coinhibition of IR/ M.A. Israel IGF1R and EGFR improves the treatment of subcutaneous GBM Study supervision: D.A. Almiron Bonnin, C. Ran, M.C. Havrda, Z. Zhang, xenografts in a synergistic fashion (46). Consistent with these M.A. Israel reports, insulin and IGF1 have been shown to antagonize EGFR inhibition in EGFR-dependent GBM patient-derived mouse Acknowledgments xenografts (46). In the clinic, coinhibition of IR/IGF1R and The authors wish to thank Drs. D. Compton, L. Witters, and A. Eastman for EGFR has been suggested to improve the response of lung, suggestions and assistance and Tabatha Richardson for her administrative

714 Mol Cancer Ther; 16(4) April 2017 Molecular Cancer Therapeutics

GBM Resistance to PDGFR Inhibition

support. The authors also wish to thank Dr. T. Tosteson for his consultation on The costs of publication of this article were defrayed in part by the payment statistical analysis. of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. Grant Support Support was generously provided by the Theodora B. Betz Foundation (M.A. Israel), the Jordan and Kyra Memorial Foundation (M.A. Israel), and the Andrea Received September 16, 2016; revised December 6, 2016; accepted December Clark Nelson Medical Research Endowment (M.A. Israel). 22, 2016; published OnlineFirst January 30, 2017.

References 1. Lemmon MA, Schlessinger J. Cell signaling by receptor tyrosine kinases. 21. Cheng C, Yan X, Sun F, Li LM. Inferring activity changes of transcription Cell 2010;141:1117–34. factors by binding association with sorted expression profiles. BMC Bio- 2. Casaletto JB, McClatchey AI. Spatial regulation of receptor tyrosine kinases informatics 2007;8:452–. in development and cancer. Nat Rev Cancer 2012;12:387–400. 22. Hitoshi Y, Harris BT, Liu H, Popko B, Israel MA. Spinal glioma: platelet- 3. Carrasco-García E, Saceda M, Martínez-Lacaci I. Role of receptor tyrosine derived growth factor B–mediated oncogenesis in the spinal cord. Cancer kinases and their ligands in glioblastoma. Cells 2014;3:199–235. Res 2008;68:8507–15. 4. Mrugala MM, Kesari S, Ramakrishna N, Wen PY. Therapy for recurrent 23. Cao Y.Multifarious functions of PDGFs and PDGFRs in tumor growth and malignant glioma in adults. Expert Rev Anticancer Ther 2004;4: metastasis. Trends Mol Med 19:460–73. 759–82. 24. Basso DM, Beattie MS, Bresnahan JC. A sensitive and reliable locomotor 5. Giamas G, Man YL, Hirner H, Bischof J, Kramer K, Khan K, et al. rating scale for open field testing in rats. J Neurotrauma 1995;12:1–21. Kinases as targets in the treatment of solid tumors. Cell Signal 2010; 25. Szerlip NJ, Pedraza A, Chakravarty D, Azim M, McGuire J, Fang Y, et al. 22:984–1002. Intratumoral heterogeneity of receptor tyrosine kinases EGFR and PDGFRA 6. Verhaak RGW, Hoadley KA, Purdom E, Wang V, Qi Y, Wilkerson MD, et al. amplification in glioblastoma defines subpopulations with distinct growth An integrated genomic analysis identifies clinically relevant subtypes of factor response. Proc Natl Acad Sci USA 2012;109:3041–6. glioblastoma characterized by abnormalities in PDGFRA, IDH1, EGFR and 26. H€agerstrand D, Lindh MB, Pena~ C, Garcia-Echeverria C, Nister M, Hof- NF1. Cancer Cell 2010;17:98. mann F, et al. PI3K/PTEN/Akt pathway status affects the sensitivity of high- 7. Lei L, Sonabend AM, Guarnieri P, Soderquist C, Ludwig T, Rosenfeld grade glioma cell cultures to the insulin-like growth factor-1 receptor S, et al. Glioblastoma models reveal the connection between adult inhibitor NVP-AEW541. Neuro-Oncol 2010;12:967–75. glial progenitors and the proneural phenotype. PLoS ONE 2011;6: 27. Cully M, You H, Levine AJ, Mak TW. Beyond PTEN mutations: the PI3K e20041. pathway as an integrator of multiple inputs during tumorigenesis. Nat Rev 8. Ozawa T, Riester M, Cheng Y-K, Huse JT, Squatrito M, Helmy K, et al. Cancer 2006;6:184–92. Most human non-GCIMP glioblastoma subtypes evolve from a 28. Pollak M. The insulin receptor/insulin-like growth factor receptor family as common proneural-like precursor glioma. Cancer Cell 2014;26: a therapeutic target in oncology. Clin Cancer Res 2012;18:40. 288–300. 29. Pollak M.Insulin and insulin-like growth factor signalling in neoplasia. Nat 9. Study of Platelet Derived Growth Factor Receptor (PDGFR) in Recurrent Rev Cancer 2008;8:915–28. Malignant Gliomas. https://clinicaltrials.gov/show/NCT01140568. 30. Fernandez AM, Torres-Aleman I. The many faces of insulin-like peptide 10. Drappatz J, Norden AD, Wen PY. Therapeutic strategies for inhibiting signalling in the brain. Nat Rev Neurosci 2012;13:225–39. invasion in glioblastoma. Expert Rev Neurother 2009;9:519–34. 31. Buck E, Gokhale PC, Koujak S, Brown E, Eyzaguirre A, Tao N, et al. 11. Alexander PB, Wang X-F. Resistance to receptor tyrosine kinase inhibition Compensatory insulin receptor (IR) activation on inhibition of insulin- in cancer: molecular mechanisms and therapeutic strategies. Front Med like growth factor-1 receptor (IGF-1R): rationale for cotargeting IGF-1R and 2015;9:134–8. IR in cancer. Mol Cancer Ther 2010;9:2652. 12. Logue JS, Morrison DK. Complexity in the signaling network: insights 32. Siddle K.The insulin receptor and type I IGF receptor: Comparison of from the use of targeted inhibitors in cancer therapy. Gen Devel 2012; structure and function. Prog Growth Factor Res 1992;4:301–20. 26:641–50. 33. De Meyts P, Whittaker J. Structural biology of insulin and IGF1 receptors: 13. Karajannis MA, Legault G, Fisher MJ, Milla SS, Cohen KJ, Wisoff JH, et al. implications for drug design. Nat Rev Drug Discov 2002;1:769–83. Phase II study of sorafenib in children with recurrent or progressive low- 34. Bailyes EM, Nave BT, Soos MA, Orr SR, Hayward AC, Siddle K. Insulin grade astrocytomas. Neuro-Oncol 2014;16:1408–16. receptor/IGF-I receptor hybrids are widely distributed in mammalian 14. Basso DM, Fisher LC, Anderson AJ, Jakeman LB, McTigue DM, Popovich tissues: quantification of individual receptor species by selective immu- PG. Basso mouse scale for locomotion detects differences in recovery after noprecipitation and immunoblotting. Biochem J 1997;327:209–15. spinal cord injury in five common mouse strains. J Neurotrauma 2006; 35. Pollak M.Insulin and insulin-like growth factor signalling in neoplasia. Nat 23:635–59. Rev Cancer 2008;8:915–28. 15. Kovalenko M, Gazit A, Bohmer€ A, Rorsman C, Ronnstrand€ L, Heldin C-H, 36. Gong Y, Ma Y, Sinyuk M, Loganathan S, Thompson RC, Sarkaria JN, et al. et al. Selective platelet-derived growth factor receptor kinase blockers Insulin-mediated signaling promotes proliferation and survival of glio- reverse sis-transformation. Cancer Res 1994;54:6106–14. blastoma through Akt activation. Neuro-Oncol 2015. doi: 10.1093/ 16. Mulvihill MJ, Cooke A, Rosenfeld-Franklin M, Buck E, Foreman K, Landfair neuonc/nov096 D, et al. Discovery of OSI-906: a selective and orally efficacious dual 37. Benyoucef S, Surinya Katharina H, Hadaschik D, Siddle K. Characterization inhibitor of the IGF-1 receptor and insulin receptor. Fut Med Chem of insulin/IGF hybrid receptors: contributions of the insulin receptor L2 2009;1:1153–71. and Fn1 domains and the alternatively spliced exon 11 sequence to ligand 17. Vlahos CJ, Matter WF, Hui KY, Brown RF. A specific inhibitor of phospha- binding and receptor activation. Biochem J 2007;403:603. tidylinositol 3-kinase, 2-(4-morpholinyl)-8-phenyl-4H-1-benzopyran-4- 38. Belfiore A, Frasca F, Pandini G, Sciacca L, Vigneri R. Insulin receptor one (LY294002). J Biol Chem 1994;269:5241–8. isoforms and insulin receptor/insulin-like growth factor receptor hybrids 18. Favata MF, Horiuchi KY, Manos EJ, Daulerio AJ, Stradley DA, Feeser WS, in physiology and disease. Endocr Rev 2008;30:586–623. et al. Identification of a Novel Inhibitor of Mitogen-activated Protein 39. Bielen A, Perryman L, Box GM, Valenti M, de Haven Brandon A, Martins V, Kinase Kinase. J Biol Chem 1998;273:18623–32. et al. Enhanced efficacy of IGF1R inhibition in pediatric glioblastoma by 19. Ran C, Liu H, Hitoshi Y, Israel MA. Proliferation-independent control of combinatorial targeting of PDGFRa/b;. Am Assoc Cancer Res 2011; tumor glycolysis by PDGFR-mediated AKT activation. Cancer Res 10:1407–18. 2013;73:1831–43. 40. Carapancea M, Cosaceanu D, Budiu R, Kwiecinska A, Tataranu L, Ciubotaru 20. Rahme GJ, Zhang Z, Young AL, Cheng C, Bivona EJ, Fiering SN, et al. PDGF V, et al. Dual targeting of IGF-1R and PDGFR inhibits proliferation in high- engages an E2F-USP1 signaling pathway to support ID2-mediated survival grade gliomas cells and induces radiosensitivity in JNK-1 expressing cells. of proneural glioma cells. Cancer Res 2016;76:2964–76. J Neuro-Oncol 2007;85:245–54.

www.aacrjournals.org Mol Cancer Ther; 16(4) April 2017 715

Almiron Bonnin et al.

41. Chen HX, Sharon E. IGF-1R as an anti-cancer target—trials and tribula- way are related with trastuzumab resistance in HER2 breast carcinomas. Br J tions. Chinese J Cancer 2013;32:242–52. Cancer 2012;106:1367–73. 42. Villanueva J, Vultur A, Lee JT, Somasundaram R, Fukunaga-Kalabis M, 46. Ma Y, Tang N, Thompson RC, Mobley BC, Clark SW, Sarkaria JN, et al. InsR/ Cipolla AK, et al. Acquired resistance to BRAF inhibitors mediated by a RAF IGF1R pathway mediates resistance to EGFR inhibitors in glioblastoma. kinase switch in melanoma can be overcome by cotargeting MEK and IGF- Am Assoc Cancer Res 2016;22:1767–76. 1R/PI3K. Cancer Cell 2010; 18:683–95. 47. Niederst MJ, Engelman JA. Bypass mechanisms of resistance to receptor 43. Soroceanu L, Kharbanda S, Chen R, Soriano RH, Aldape K, Misra A, et al. tyrosine kinase inhibition in lung cancer. Sci Signal 2013;>6:re6–re. Identiﬁcation of IGF2 signaling through phosphoinositide-3-kinase regu- 48. Study of Erlotinib (Tarceva) in Combination With OSI-906 in Patients latory subunit 3 as a growth-promoting axis in glioblastoma. Proc Nat Acad With Advanced Non-small Cell Lung Cancer (NSCLC) With Activating Sci 2007;104:3466–71. Mutations of the Epidermal Growth Factor Receptor (EGFR) Gene. https:// 44. Furnari FB, Cloughesy TF, Cavenee WK, Mischel PS. Heterogeneity of clinicaltrials.gov/show/NCT01221077. epidermal growth factor receptor signalling networks in glioblastoma. Nat 49. Akhavan D, Pourzia AL, Nourian AA, Williams KJ, Nathanson D, Babic I, Rev Cancer 2015;15:302–10. et al. De-Repression of PDGFRb transcription promotes acquired resistance 45. Gallardo A, Lerma E, Escuin D, Tibau A, Munoz J, Ojeda B, et al. Increased to EGFR tyrosine kinase inhibitors in glioblastoma patients. Cancer Discov signalling of EGFR and IGF1R, and deregulation of PTEN/PI3K/Akt path- 2013;3:534–47.

716 Mol Cancer Ther; 16(4) April 2017 Molecular Cancer Therapeutics

Insulin-Mediated Signaling Facilitates Resistance to PDGFR Inhibition in Proneural hPDGFB-Driven Gliomas

Damian A. Almiron Bonnin, Cong Ran, Matthew C. Havrda, et al.

Mol Cancer Ther 2017;16:705-716. Published OnlineFirst January 30, 2017.

Updated version Access the most recent version of this article at: doi:10.1158/1535-7163.MCT-16-0616

Supplementary Access the most recent supplemental material at: Material http://mct.aacrjournals.org/content/suppl/2017/01/28/1535-7163.MCT-16-0616.DC1

Cited articles This article cites 45 articles, 14 of which you can access for free at: http://mct.aacrjournals.org/content/16/4/705.full#ref-list-1

Citing articles This article has been cited by 2 HighWire-hosted articles. Access the articles at: http://mct.aacrjournals.org/content/16/4/705.full#related-urls

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

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

Permissions To request permission to re-use all or part of this article, use this link http://mct.aacrjournals.org/content/16/4/705. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.