PDGF Engages an E2F-USP1 Signaling Pathway to Support ID2-Mediated Survival of Proneural Glioma Cells Gilbert J

Published OnlineFirst March 7, 2016; DOI: 10.1158/0008-5472.CAN-15-2157

Cancer Molecular and Cellular Pathobiology Research

PDGF Engages an E2F-USP1 Signaling Pathway to Support ID2-Mediated Survival of Proneural Glioma Cells Gilbert J. Rahme1,2,3, Zhonghua Zhang2,3,4, Alison L. Young2,3,4, Chao Cheng1,2,3, Eric J. Bivona5, Steven N. Fiering1,2,3, Yasuyuki Hitoshi2,3,4, and Mark A. Israel1,2,3,4

Abstract

Glioblastoma is the most aggressive primary brain tumor and activated Usp1. Furthermore, PDGF-mediated expression of responds poorly to currently available therapies. Transcriptomic USP1 led to the stabilization of Inhibitor of DNA-binding 2 characterization of glioblastoma has identified distinct molec- (ID2), which we found to be required for glioma cell survival. ular subtypes of glioblastoma. Gain-of-function alterations Genetic ablation of Id2 delayed tumor-induced mortality, and leading to enhanced platelet-derived growth factor (PDGF) pharmacologic inhibition of USP1, resulting in decreased ID2 signaling are commonly observed in the proneural subtype of levels, also delayed tumorigenesis in mice. Notably, decreased glioblastoma and can drive gliomagenesis. However, little is USP1 expression was associated with prolonged survival in known about the downstream effectors of PDGF signaling in patients with proneural glioblastoma, but not with other sub- glioblastoma. Using a mouse model of proneural glioma and types of glioblastoma. Collectively, our findings describe a comparative transcriptomics, we determined that PDGF signal- signaling cascade downstream of PDGF that sustains proneural ing upregulated ubiquitin-specificpeptidase1(Usp1)topro- glioblastoma cells and suggest that inhibition of the PDGF– mote the survival of murine proneural glioma cells. Mechanis- E2F–USP1–ID2 axis could serve as a therapeutic strategy for tically, we found that PDGF signaling regulated the expression of proneural glioblastoma featuring increased PDGF signaling. the E2F transcription factors, which directly bound to and Cancer Res; 76(10); 2964–76. 2016 AACR.

Introduction characterized by frequent amplifications in the genomic region encoding the receptor tyrosine kinase (RTK), platelet-derived Glioblastoma is the most lethal and aggressive primary brain growth factor receptor a (PDGFRA; ref. 5). Multiple studies tumor. Despite expansive diagnostic efforts and aggressive com- indicate that increased PDGF signaling is sufficient to drive bination therapy, including surgery, ionizing radiation, and che- gliomagenesis in murine models, emphasizing the importance motherapy, the median survival of patients with glioblastoma of PDGF signaling in glioblastoma (7–12). Transcriptomic char- remains a modest 12–15 months (1, 2). Although described as a acterization of PDGF-induced high-grade glioma in mice revealed single histopathologic entity, transcriptomic profiling of glioblas- a proneural glioblastoma molecular signature, indicating that toma has led to the identification of multiple subtypes (3, 4) with increased PDGF signaling is a driver of proneural gliomagenesis the latest report identifying four subtypes: proneural, neural, (11, 12). Little is known, however, about the precise molecular classical, and mesenchymal glioblastoma (5). Each of these sub- pathways downstream of PDGF signaling causing malignancy in types has an associated profile of genetic alterations (5, 6). proneural glioma. Proneural glioblastoma, which is resistant to current therapy, is Among the cellular processes that are critical for glioblastoma pathology, protein degradation is emerging as an important

1 regulator of tumorigenesis and a target for cancer treatment Department of Genetics, Dartmouth College, Hanover, New Hamp- – shire. 2Norris Cotton Cancer Center, Dartmouth College, Hanover, New (13 15). Ubiquitination and its reverse, deubiquitination, are Hampshire. 3Geisel School of Medicine, Dartmouth College, Hanover, major regulatory pathways that control protein degradation and New Hampshire. 4Department of Pediatrics, Dartmouth College, Han- contribute to cell-cycle regulation, stem cell maintenance, and over, New Hampshire. 5Digital Library Technologies Group, Dartmouth College, Hanover, New Hampshire. cellular survival, as well as other biologic activities of importance for tumorigenesis (16, 17). Although it is known that the cellular Note: Supplementary data for this article are available at Cancer Research localization, posttranslational modification, and altered structure Online (http://cancerres.aacrjournals.org/). of proteins can regulate their degradation (18), the repertoire of Current address for Y. Hitoshi: Department of Neurosurgery, Amakusa Medical molecules that regulates the destruction of specific targets is Center, 854-1 Jikiba Kameba, Amakusa, Kumamoto, Japan. incompletely known. Deubiquitinases (DUB), which include Corresponding Author: Mark A. Israel, Geisel School of Medicine at Dartmouth, over 100 genes classified into five subgroups, catalyze the removal One Medical Center Drive, Lebanon, NH 03756. Phone: 603-653-3611; Fax: 603- of ubiquitin molecules that are covalently bound to a substrate, 653-9003; E-mail: [email protected] thus diminishing its proteasomal degradation (17). Recent work doi: 10.1158/0008-5472.CAN-15-2157 has identified a role for USP1, a member of the ubiquitin-specific 2016 American Association for Cancer Research. peptidase (USP) group of DUBs, in the regulation of tumor

2964 Cancer Res; 76(10) May 15, 2016

A PDGF-E2F-USP1-ID2 Axis Is Required for Proneural Glioma

pathogenesis (19–21) including glioblastoma (22). Inhibition of Transfections USP1 by multiple strategies has been suggested as a potential PDGF-GSCs were transfected using LipoD293 (Signagen) as strategy to enhance the therapeutic efficacy of chemotherapy per product guidelines. and radiation in treating cancer (22–24). USP1, upon forming a molecular complex with UAF1 (WDR48), deubiquitinates sub- Annexin V/PI flow cytometry analysis strates including PCNA, FANCI, FANCD2, and the short-lived Annexin V/PI staining was performed using the Annexin V/PI Inhibitor of DNA Binding (ID) proteins ID1, ID2, and ID3 that Apoptosis Detection Kit II (BD Pharmingen) following the man- each have a half-life of about 10 minutes (19, 20, 25–28). ufacturer's guidelines. The FL1 channel was used to detect Members of the Id gene family, including Id2, have been Annexin V-FITC and the FL3 channel was used to detect PI. proposed as mediators of glioblastoma pathology and are known to be highly expressed in a number of different tumor Western blot analysis types (28–30). Western blot analysis was done as described previously (33). While studying the molecular events underlying the critical The antibodies used and the working dilutions are shown in role of PDGF signaling in proneural glioma, we confirmed the Supplementary Table S1. Antibodies that recognize ID proteins regulation of multiple gene groups by PDGF signaling and were validated in a previous report from our laboratory (33). discovered a previously unknown role for PDGF signaling in the regulation of protein degradation. We found that PDGF RNA extraction and qRT-PCR regulates the transcription of molecules key for protein degra- For microarray analysis, RNA was purified using the TRIzol dation, including E3 ubiquitin ligases, ubiquitin, SUMO (small method. For qRT-PCR, RNA was extracted using the RNeasy kit as ubiquitin-like modifier)-related genes, and DUBs, among per product guidelines (Qiagen). Data were quantified relative to which we found USP1 to be critical for survival of PDGF-driven untreated control in each GSC culture, and normalized either only glioma spheroid cells (referred to as PDGF-GSCs in this study). to 18S Ct values, or 18S Ct values followed by another normal- DDCt In addition, we found that PDGF signaling drives the expres- ization to Id2 Ct values (Supplementary Fig. S6), using the 2 sion of USP1 by upregulating the expression of E2f1, E2f2,and method. Primers are provided in Supplementary Table S2. E2f3 (E2f1-3), transcription factors that interact with the reti- noblastoma protein (pRB), and thereby increases their inter- Chromatin immunoprecipitation analysis action with the Usp1 promoter. Furthermore, we determined Chromatin immunoprecipitation (ChIP) was performed as that ID2 is a critical USP1 substrate that is required for survival DDC described previously (33). Data were analyzed using the 2 t in PDGF-GSCs. Targeting the PDGF-E2F-USP1-ID2 axis by method with normalization to the IgG isotype control ChIP of deletion of the Id2 gene or pharmacologic inhibition of USP1 cells that were treated in a manner identical to the experimental delayed tumorigenesis of PDGF-driven glioma. Importantly, cells. Primer sequences are provided in Supplementary Table S2. data from The Cancer Genome Atlas (TCGA) suggest the presence of a PDGF-USP1 axis in proneural glioblastoma, and that USP1 is a prognostic factor in proneural glioblastoma, but Lentiviral infection not in other subtypes of glioblastoma. These results suggest that Lentiviruses were prepared using 293FT cells with an optimized targeting the PDGF-E2F-USP1-ID2 axis can be a beneficial Signagen protocol. A three-vector second-generation system using treatment strategy for proneural glioma. These findings may pCMV-Delta8.9, pVSVG, and the pLKO.1 transfer vector was used. be applicable to other gliomas in which alterations in PDGF 293FT cells were transfected with the following ratio of plasmid signaling are common including pediatric high-grade glioma DNA: 6.5:4.5:10 (pCMV-Delta8.9, VSV-G, pLKO.1 transfer vector, and diffuse intrinsic pontine glioma (DIPG) as well as to respectively). After overnight incubation, the transfected cells tumors that share the pathobiology of being driven by PDGF were washed gently with 1 PBS without calcium or magnesium signaling though occurring at other sites (31, 32). and incubated in GSC media for 48 hours to prepare a lentiviral supernatant. Conditioned media was collected, spun down at 210 g for 5 minutes, filtered through a 0.45-mm filter (Milli- Materials and Methods pore), aliquoted, and frozen at 80C. Mission pLKO.1-puro Cell culture vectors encoding shRNA sequences were purchased from Sigma. PDGF-GSCs were cultured in petri dishes in GSC media shRNA sequences are available in Supplementary Table S3. Mis- sion SHC002 was used as a mammalian nontarget shRNA control (DMEM/F12 without L-glutamine (Corning) supplemented with 1% penicillin/streptomycin and 1X B27 (Gibco) without vitamin (ShControl; Sigma). A). PDGF-GSCs were dissociated following incubation in Accu- tase (Millipore) for 5 minutes at 37C. Doxycycline hyclate (100 Statistical analysis ng/mL; Sigma), pimozide (5 mmol/L; Sigma), palbociclib isethio- GraphPad Prism 6 was used for statistical analysis. Two-tailed nate (5 mmol/L, PD0332991, PD; Selleckchem), and AG1295 (10 Student t test was used to evaluate differences between two mmol/L; Santa Cruz Biotechnology) were added to cultures for groups. For experiments involving multiple comparisons, we times indicated in the figure legends. MG132 (1 mmol/L; Sigma) performed one-way ANOVA with the Tukey post hoc test to was added to cultures for 1 hour or 30 minutes when treating in evaluate differences. We evaluated bivariate correlations using combination with doxycycline hyclate or AG1295, respectively. the Pearson correlation test. Two-way ANOVA test with Sidak Primary astrocytes and primary mouse embryo fibroblast cells multiple comparison was performed to evaluate significant dif- were grown on tissue culture dishes in growth media [DMEM with ferences in the mouse pimozide treatment experiment (Fig. 5F). L-glutamine (Corning) supplemented with 1% penicillin/strep- Log-rank (Mantel–Cox) was used to perform significance analysis tomycin (Corning) and 10% FBS (Corning)]. on survival curves with two groups. We examined survival curves

www.aacrjournals.org Cancer Res; 76(10) May 15, 2016 2965

Rahme et al.

with more than two groups using the log-rank (Mantel–Cox) and the gene in the USP family most significantly regulated by PDGF the log-rank test for trend. (data not shown) and had approximately a 3-fold decrease Complete methods can be found in the Supplementary Mate- following the inhibition of PDGF signaling in PDGF-GSCs. Fur- rials and Methods. thermore, USP1 was highly expressed in human glioblastoma relative to normal brain tissue (Supplementary Fig. S3A), and was Results variably expressed in the different glioblastoma subtypes (Sup- plementary Fig. S3B). Importantly, USP1 can be therapeutically PDGFB-driven gliomas in mice belong to the proneural subtype targeted by pimozide (23, 27), an antipsychotic drug of the and are addicted to PDGF signaling diphenylbutylpiperidine class that has limited toxicities, has been To study the molecular determinants of malignancy down- approved for use in humans, and can cross the blood–brain stream of PDGF signaling in proneural glioma, we used a previ- barrier (34). These characteristics make USP1 a possible thera- ously characterized model from our laboratory (Gfap-tTa/Tre- peutic target in glioblastoma. Hence, we focused our study on PDGFB). In this model, human PDGFb (PDGFB) was expressed USP1. under the control of a Tet-off system driven by the glial fibrillary We first validated the results of our microarray. Inhibition of acidic protein (Gfap) promoter. High-grade glioma (glioma grade PDGF signaling by doxycycline hyclate significantly decreased III–IV) developed in these mice upon doxycycline hyclate removal Usp1 mRNA (Fig. 2C) and protein (Fig. 2D) expression in PDGF- (Supplementary Fig. S1; ref. 8). In characterizing these mice, we GSCs. Importantly, exposure of neural stem cells (NSC) to performed transcriptomic analysis of glioma tissues and PDGF- recombinant PDGF-B increases USP1 mRNA (Supplementary Fig. GSCs derived from this model. Our analysis revealed that these S4A) and protein (Supplementary Fig. S4B) expression. Further- mouse glioma and PDGF-GSCs belong to the proneural subtype. more, Usp1 promoter activity, which was high in PDGF-GSCs, was A heatmap of Spearman r correlation scores demonstrates this significantly decreased when PDGF signaling was blocked by correlation (Fig. 1A). Further analysis using hierarchical clustering AG1295 (Fig. 2E). These findings show that PDGF signaling can of Spearman correlation scores (Supplementary Fig. S2A), as well upregulate Usp1 expression. Inhibition of USP1 has been previ- as a heatmap of expression values of the glioblastoma subtyping ously shown to decrease proliferation of leukemia cells and signature gene set (Supplementary Fig. S2B; ref.11), and clustering spheroid formation of human glioblastoma cells (22, 27). To of these expression values (Supplementary Fig. S2C) confirmed examine whether Usp1 affects the survival of PDGF-GSCs, we this interpretation. We evaluated PDGF-GSCs derived from infected these cells with lentiviruses encoding Usp1-targeting tumors arising in three different mice and referred to them as shRNAs (short hairpin RNAs) or a nontarget control (ShControl) GSC1, GSC2, and GSC3 in this study. We first tested whether the and performed Annexin V/PI staining. We found that decreased expression of PDGFB in PDGF-GSCs was still responsive to USP1 expression led to cell death of PDGF-GSCs (Fig. 2F and G). doxycycline hyclate. Doxycycline hyclate addition to the media of PDGF-GSCs suppressed PDGFB mRNA (Fig. 1B). Consistent PDGF signaling drives USP1 expression by regulating the with this finding, the level of phosphorylated PDGFR (pPDGFR) expression of E2F1-3 and their binding to the USP1 promoter was reduced when PDGF-GSCs were treated with doxycycline To explore the mechanism by which PDGF signaling drives the hyclate (Fig. 1C). We found decreased cellular proliferation (Fig. expression of Usp1,wefirst analyzed the 1KB DNA sequence 50 to 1D and E) and decreased sphere size (Fig. 1F) following treatment the first exon of Usp1 using the decipherment of DNA elements of PDGF-GSCs with doxycycline hyclate or AG1295, a chemical (DECODE) algorithm (http://www.sabiosciences.com/chipqpcr- inhibitor of pPDGFR. To determine whether the decrease in cell search.php). We found multiple transcription factor binding number in these cultures was associated with cellular death, we motifs in this promoter region including those corresponding to performed Annexin V/PI staining, and found that both doxycy- the PAX-6, NKX5-1, LUN-1, NF-Y, and the E2F factors. Among cline hyclate and AG1295 treatment induced cellular death these transcription factors, only the E2F transcription factors (Fig. 1G and H). E2F1-3 were regulated by PDGF signaling in our microarray analysis (Figs. 2A and B and 3A). The E2F transcription factor PDGF signaling regulates Usp1 expression to promote survival binding site motifs we identified in the mouse Usp1 promoter are of PDGF-GSCs conserved in the human USP1 promoter (data not shown). We To discover the molecular determinants of malignancy down- validated our microarray results by determining that E2F1-3 stream of PDGF signaling in proneural glioma, we performed mRNA (Fig. 3B) and protein (Fig. 3C) levels were lower in comparative transcriptomic analysis on untreated and doxycy- PDGF-GSCs following treatment with doxycycline hyclate. Fur- cline hyclate–treated PDGF-GSCs. We detected 1,706 genes that thermore, we found that E2F1-3 bind the Usp1 promoter only were significantly regulated by PDGF signaling (Fig. 2A). Of these, during PDGF exposure (Fig. 3D). When we inhibited the activity 885 genes were upregulated and 821 were downregulated. As of E2F1-3 by inducing active pRB using the cyclin-dependent expected, gene ontology annotation of these genes revealed that kinase 4/6 (CDK4/6) inhibitor palbociclib (PD0332991, PD) in cell cycle and proliferation-related genes constituted the major PDGF-GSCs, USP1 mRNA (Fig. 3E) and protein (Fig. 3F) levels downregulated gene group in cells following PDGF signaling decreased even in the presence of PDGF signaling. To determine inhibition (Fig. 2B). Interestingly, a small percentage of genes the E2F-binding motifs located in the Usp1 promoter at which regulating ubiquitination and protein degradation were down- E2F1-3 binding is specifically regulated by PDGF signaling, we regulated in cells in which PDGF signaling had been inhibited. We sequentially deleted the 1KB Usp1 promoter used in our expres- sought to explore this previously unknown finding that suggests a sion construct (Fig. 2E), 200 bp at a time from the 50 end. We then role for PDGF signaling in the regulation of protein degradation. tested the activity of these constructs in the presence or absence of Of the genes, we found in the ubiquitination and protein degra- PDGF signaling. We found that the activity and responsiveness of dation group, Usp1, a negative regulator of ubiquitination, was the promoter reporter construct to PDGF inhibition by AG1295

2966 Cancer Res; 76(10) May 15, 2016 Cancer Research

A PDGF-E2F-USP1-ID2 Axis Is Required for Proneural Glioma

Figure 1. PDGFB-driven gliomas in mice belong to the proneural subtype and are addicted to PDGF signaling. A, heatmap of correlation (Spearman) r values of glioblastoma (GBM) subtyping signature genes in mouse and human samples (TCGA) with human glioblastoma subtype–specific centroid values. Full details of analysis are available in Supplementary Materials and Methods (dataset in Supplementary Table S4). B and C, PDGFB mRNA levels (B) and Western blot analysis (C) of PDGFRA and pPDGFR in lysates of PDGF-GSCs treated with doxycycline hyclate (Dox; 48 hours). D and E, growth curves of GSCs treated with doxycycline hyclate (D) or with AG1295 (E) compared with mock-treated cells. F, phase contrast images of PDGF-GSCs treated with DMSO, doxycycline hyclate, or AG1295 for four days. G and H, pseudo-colored flow cytometry evaluation of Annexin V/PI-stained GSC1 treated with DMSO, doxycycline hyclate, or AG1295 as indicated (G) and quantification of live cells (Annexin V /PI ) in these cultures (H). Mean (B, D, E) 1 SD or mean (H) 95% confidence interval of three independent experiments are shown. , P < 0.0005.

was unchanged despite our trimming the promoter region to 0.8 of which was located in the 0.2 kb promoter segment as well. A kb, 0.6 kb, and 0.4 kb. Unexpectedly, these promoter constructs summary of the promoter segments we used and the location of had enhanced activity when compared with the 1 kb promoter the E2F-binding motifs in them can be found in Supplementary construct. Of particular note, however, was our ﬁnding that the Fig. S5. 0.2 kb promoter construct was still responsive to pPDGFR inhi- Deletion of the E2F-binding motif in the 0.4 kb proximal bition, although it had lower activity than the 0.4 kb promoter to the ﬁrstexonoftheUsp1 gene [Del-E2F(1)], but not dele- (Fig. 3G). We concluded that the critical transcription factor tion of the distal E2F biding motif [Del-E2F(2)], rendered binding motifs that mediate PDGF regulation of USP1 expression this promoter construct inactive and it was not responsive to driven by PDGF signaling must be present in the 0.4 kb promoter AG1295 (Fig. 3H) or PD (Fig. 3I). Collectively, these results segment. Within this fragment were two E2F-binding motifs, one indicate that regulation of Usp1 expression by PDGF signaling

www.aacrjournals.org Cancer Res; 76(10) May 15, 2016 2967

Rahme et al.

Figure 2. PDGF signaling regulates USP1 expression to promote survival of PDGF-GSCs. A, heatmap representation of unsupervised hierarchical clustering of z-score values of all significantly regulated genes, as defined in Supplementary Materials and Methods, in GSC1 following doxycycline hyclate (Dox) treatment for 48 hours (n ¼ 3 independent replicates in each treatment arm; dataset in Supplementary Table S5). B, pie charts describing the general overview of Gene Ontology classes upregulated or downregulated in doxycycline hyclate–treated GSC1 cells. The top four downregulated genes in the ubiquitination and protein degradation class are shown below the displaced red fraction. USP1 mRNA (C) and protein (D) levels determined by Western blot analysis of lysates of PDGF-GSCs treated with doxycycline hyclate (48 hours). E, evaluation of Usp1 promoter activity in PDGF-GSCs treated with AG1295 as indicated in Supplementary Materials and Methods. F and G, pseudo- colored flow cytometry evaluation of Annexin V/PI-stained PDGF-GSCs transduced with lentiviruses encoding ShControl or Usp1 targeting ShRNAs (ShUSP1A and B; F) and quantification of live cells (Annexin V /PI ) in these analyses (G). Mean (C) 1 SD or mean (E and G) 95% confidence interval of three independent experiments are shown. , P < 0.0005.

is driven by enhanced expression of E2f1-3 and the binding of tional targets of USP1 (26, 27), hence we examined whether they these factors to an E2F-binding motif within 200 bp of the ﬁrst were regulated by PDGF signaling in a manner that parallels USP1 Usp1 exon. regulation. Inhibition of PDGF signaling by doxycycline hyclate led to a decrease in ID2 protein, and an inconsistent decrease in PDGF signaling stabilizes ID2 by regulating USP1 to promote ID1 protein levels (Fig. 4A). ID3 and ID4 proteins were not survival of PDGF-GSCs expressed at detectable levels in these cells. The mRNA expression We sought to identify molecules responsible for USP1-medi- of Id genes was not downregulated by doxycycline hyclate (Fig. 4B, ated survival in PDGF-GSCs. ID proteins are known posttransla- Supplementary Fig. S6). As ID2 was consistently regulated in

2968 Cancer Res; 76(10) May 15, 2016 Cancer Research

A PDGF-E2F-USP1-ID2 Axis Is Required for Proneural Glioma

Figure 3. PDGF signaling drives USP1 expression by regulating the expression of E2F1-3 and their binding to the USP1 promoter. A, heatmap representation of z-score values for E2F1-3 genes in GSC1 following doxycycline hyclate (Dox) treatment for 48 hours. Data are from the microarray shown in Fig. 2A. B and C, E2F1-3 mRNA (B) and protein (C) levels in PDGF-GSCs treated with doxycycline hyclate (48 hours). D, ChIP analysis using antibodies specific for E2F1, E2F2, and E2F3 to determine binding of E2F1-3 to the USP1 promoter in PDGF-GSCs following doxycycline hyclate treatment for 48 hours. E and F, effect of palbociclib (þPD, 5 mmol/L, 24 hours) treatment on USP1 mRNA (E) and protein levels of USP1, pPDGFR, pRB, and p-pRB (F). G, evaluation of the activity of truncated Usp1 promoter constructs in PDGF-GSCs treated with AG1295 (10 mmol/L) as indicated in Supplementary Materials and Methods. H and I, evaluation of the activity of the 0.4 KB Usp1 promoter construct (WT; H) or this same construct in which either the first or second E2F binding site was deleted in PDGF-GSCs treated with AG1295 (10 mmol/L) or palbociclib (PD; 5 mmol/L; I) as indicated in Supplementary Materials and Methods. Mean (B, D, E, G, H, I) 1SD of three independent experiments is shown. N.S., nonspecific; , P < 0.0005.

PDGF-GSCs and was expressed at a higher level than Id1 at the lentiviruses encoding Usp1-speciﬁc ShRNAs or a ShControl and mRNA level (Supplementary Fig. S6), we focused on ID2. We compared the levels of ID2 following infection. We found that treated cells with the combination of a proteasome inhibitor, decreased USP1 expression led to a decrease in ID2 protein MG132, and either doxycycline hyclate or AG1295 to evaluate levels (Fig. 4F). USP1 has been shown to bind ID2 and cleave ID2 degradation in the absence of PDGF signaling. We found that ubiquitin molecules covalently bound to ID2 (26). We evalu- the decrease in ID2 observed following inhibition of PDGF ated whether USP1 interacts with ID2 in PDGF-GSCs and signaling was primarily a result of proteasomal degradation (Fig. whether this interaction was regulated by PDGF signaling. We 4C and D). Consistent with these data, we observed that ID2 was foundthatUSP1bindsID2whenPDGF-GSCsarestimulated ubiquitinated in PDGF-GSCs when PDGF signaling was inhibited byPDGF(Fig.4G).Wethenchecked whether ID2, like USP1 (Fig. 4E). To determine whether ID2 was a target of USP1 (Fig.2FandG),isimportantforthesurvivalforPDGF-GSCs. stabilization in PDGF-GSCs, we infected these cells with We infected PDGF-GSCs with lentiviruses encoding Id2-speciﬁc

www.aacrjournals.org Cancer Res; 76(10) May 15, 2016 2969

Rahme et al.

Figure 4. PDGF signaling stabilizes ID2 by regulating USP1 to promote survival of PDGF-GSCs. A and B, Western blot analysis of ID family members (A) and evaluation of Id2 mRNA levels in lysates of PDGF-GSCs treated with doxycycline hyclate (Dox; 48 hours; B). C and D, Western blot analysis of pPDGFR and ID2 (C) and ID2 in lysates (D) of PDGF-GSCs treated with indicated drugs. Doxycycline hyclate and AG1295 were added for 24 hours. E, Western blot analysis of ubiquitin in anti-ID2 or control IgG immunoprecipitates from lysates of GSC1 that were treated with indicated drugs. Doxycycline hyclate was added for 24 hours. Duration of MG132 treatment is indicated in Materials and Methods. F, Western blot analysis of USP1 and ID2 in GSC1 infected with ShControl and ShUSP1A and B lentiviruses. G, Western blot analysis of USP1 in anti-ID2 or control IgG immunoprecipitates of lysates from GSC1 that were treated with doxycycline hyclate for 48 hours. H, Western blot analysis of ID2 in lysates of GSC1 cells infected with lentiviruses encoding ShControl or Id2 targeting ShRNAs(ShId2A,ShId2B).IandJ,pseudo-coloredflow cytometry evaluation of Annexin V/PI–stained GSC1 transduced with ShControl and ShId2A/B lentiviruses (I) and quantification of live cells (Annexin V/PI) in these cultures (J). Mean (B) 1SD or mean (J) 95% confidence interval of three independent experiments are shown. , P < 0.0005.

fl fl ShRNAs and determined that decreased ID2 levels led to cell an Id2 / mouse and crossed it to a previously reported transgenic death of PDGF-GSCs (Fig. 4H–J). Our results provide evidence mouse in which the Gfap promoter, the same promoter element that ID2 is posttranslationally stabilized by PDGF-mediated driving gliomagenesis, regulated Cre recombinase expression fl fl regulation of ID2 ubiquitination through USP1 to support (Gfap-Cre; ref. 35). From this cross, we obtained Gfap-Cre/Id2 / tumor cell survival. mice that conditionally delete Id2 in Gfap-expressing cells such as postnatal NSCs (36) and astrocytes (37), and we verified that such Deletion of Id2 delays tumorigenesis in a mouse model of a deletion occurs (Supplementary Fig. S7). We then crossed the fl fl fl fl PDGF-driven glioma Id2 / mouse or the Gfap-Cre/Id2 / into the Gfap-tTa/Tre-PDGFB We sought to determine whether loss of Id2 has an impact on mouse model. From these crosses, we obtained mice that carried mortality induced by PDGF-driven glioma in vivo. We developed the doxycycline hyclate–regulatable PDGFB allele and Id2 that

2970 Cancer Res; 76(10) May 15, 2016 Cancer Research

A PDGF-E2F-USP1-ID2 Axis Is Required for Proneural Glioma

Figure 5. Deletion of Id2 delays tumorigenesis in a mouse model of PDGF-driven glioma. A, Kaplan–Meier survival curve of the mice with indicated genotypes treated from the time of conception throughout their life with doxycycline hyclate (Doxþ/þ) or treated with doxycycline hyclate from the time of conception until weaning, postnatal day 20 (Doxþ/). P values reflect comparison of mice on different doxycycline hyclate treatment regimens to Gfap-tTa/Tre- PDGFB/Id2fl/fl mice (Doxþ/). Median survival (B) and tumor incidence (C) at the end of experiment of mice shown in A.

was either expressed normally (designated Gfap-tTa/Tre-PDGFB/ observe a decrease in the weight of mice receiving pimozide fl fl Id2 / ), or deleted in Gfap-expressing cells (designated Gfap-tTa/ during the course of treatment (Fig. 6J). To determine whether fl fl Tre-PDGFB/Id2 / /Gfap-Cre; ref. 8). We maintained these mice on pimozide administration decreased ID2 in PDGF-driven glio- doxycycline hyclate throughout development in utero and post- mas in vivo, we performed a Western blot analysis of ID2 in natally until day 20 (Doxþ/). We found that deletion of Id2 lysates from eight independent tumors derived from animals significantly enhanced median survival by 100 days (Fig. 5A treated with seven treatments of pimozide or drug vehicle. Our and B). Mice bearing the genotypes described above did not results show that pimozide administration decreased ID2 develop tumors when doxycycline hyclate was continuously protein levels in PDGF-driven glioma (Fig. 6K). These data provided in drinking water (Doxþ/þ) (Fig. 5A and C). Tumor indicate that the PDGF-E2F-USP1-ID2 pathway is critical for incidence was very similar in the mice studied (Fig. 5C). These PDGF-driven glioma and raises the possibility that pimozide findings indicate that loss of Id2 can significantly delay tumor could contribute to the clinical management of PDGF-driven induced mortality in mice engineered to develop PDGF-driven glioma. proneural glioma. USP1 mRNA expression correlates with PDGFRA mRNA Pimozide inhibits USP1-mediated ID2 stabilization and delays expression and is associated with survival of patients with tumorigenesis of PDGF-driven glioma proneural glioblastoma Our results provide strong evidence that a PDGF-E2F-USP1- To examine whether the expression levels of PDGFRA and ID2 axis is important for PDGF-driven glioma. We next exam- USP1 are correlated in glioblastoma tissues and whether USP1 ined whether pharmacologic inhibition of this axis, achieved expression is associated with the survival of patients with by inhibition of USP1 by pimozide, can affect PDGF-driven glioblastoma, we analyzed gene expression data from TCGA. glioma. We first treated PDGF-GSCs with pimozide and exam- In a cohort of 206 patients (39), we found that mRNA expres- ined ID2 protein levels in these cells. Inhibition of USP1 by sion of PDGFRA and USP1 correlated directly in all glioblas- pimozide led to decreased ID2 protein levels (Fig. 6A) and toma patients analyzed (Fig. 7A). When we analyzed this induced cell death in PDGF-GSCs (Fig. 6B and C). We com- correlation in the four glioblastoma subtypes, we found that pared the sensitivity to pimozide of PDGF-GSCs, GSCs USP1 expression correlated with PDGFRA expression only in derived from a v-erb-B/Trp53 / glioma model (38), and the proneural subtype (Fig. 7A). Correlation analysis of EGFR astrocytes and mouse embryo fibroblasts from wild-type and other RTKs (Supplementary Table S6) with USP1 showed mice. We found that PDGF-GSCs are significantly more sen- no direct correlation. We then sought to determine whether sitive to pimozide than the other cell types, including EGF- USP1 expression was associated with the overall survival of driven glioma cells, which represent other subtypes of glio- human glioblastoma patients. We found that patients in the blastoma (Fig. 6D and E). We then determined whether proneural subtype whose tumors had low USP1 expression had pimozide could inhibit PDGF-driven glioma in vivo.We asignificantly increased overall survival when compared with implanted syngeneic grafts of PDGF-GSCs subcutaneously in the remaining proneural glioblastoma patients (Fig. 7B). This C57BL/6J mice and, upon palpable tumor detection, gavaged association was not observed for other glioblastoma subtypes the mice with pimozide (60 mg/kg) or a vehicle control three (Fig. 7B). Taken together, these data demonstrate the correla- times a week. Treatment with pimozide significantly delayed tion of PDGFRA and USP1 mRNA levels in glioblastoma the growth of PDGF-driven gliomas in these mice (Fig. 6F–I). patients, particularly in proneural glioblastoma, and reveal that We measured the total body weight of mice that were low USP1 mRNA expression is associated with better overall treated with pimozide and the vehicle control and did not survival of glioblastoma patients.

www.aacrjournals.org Cancer Res; 76(10) May 15, 2016 2971

Rahme et al.

Figure 6. Pimozide inhibits USP1-mediated ID2 stabilization and delays tumorigenesis of PDGF-driven glioma. A, Western blot analysis of ID2 in lysates of PDGF-GSCs treated with DMSO or pimozide (5 mmol/L) for 24 hours. B and C, pseudo- colored flow cytometry evaluation of Annexin V/PI–stained PDGF-GSCs treated with DMSO or pimozide (48 h; B) and quantification of live cells (Annexin V /PI ) in these cultures (C). D, survival response of indicated cell types to pimozide. E, IC50 derived from data in D. , significance observed when comparing any PDGF-GSC to any other cell type included in the graph. F, tumor volume of implanted PDGF- GSCs (n ¼ 14 in vehicle group; n ¼ 9in pimozide group). Appearance (G), volume (H), and weight (I) of tumors extracted from mice described in F after the last pimozide treatment. J, total body weight of mice with syngeneic implants of glioma receiving pimozide or vehicle treatment regimens. K, Western blot analysis of ID2 and USP1 in lysates from extracted tumors. All data points, mean 1 SEM are shown (H and I). Mean values 95% confidence interval (C) or 1SEM (D, E, F, J) of three independent experiments are shown. , P < 0.005; , P < 0.0001.

Discussion ally required by individual tumors will provide more effective and less toxic cancer treatments. We found a therapeutically The promise of precision medicine for oncology is that targetable signaling axis critical for survival of PDGF-GSCs in a therapeutic interventions tailored to target pathways function- mouse model of PDGF-driven proneural glioma and evidence

2972 Cancer Res; 76(10) May 15, 2016 Cancer Research

A PDGF-E2F-USP1-ID2 Axis Is Required for Proneural Glioma

of this axis in human proneural glioblastoma. The Gfap-tTa/Tre- anaphase-promoting complex, which is known to ubiquitinate PDGFB mouse we studied, like other mouse models of PDGF- ID2 (46), was downregulated when PDGF signaling was inhib- driven glioma, displays a molecular signature similar to human ited, suggesting that other ubiquitin ligases may ubiquitinate proneural glioblastoma (Fig. 1 and Supplementary Fig. S2; ID2 in PDGF-driven glioma. refs.11, 12). Importantly, GSCs derived from our model With the importance of ID2 for the survival of PDGF-GSCs required PDGF signaling for survival, providing a unique in vitro being clear (Fig. 4), we sought to examine the in vivo role reagent for use in dissecting the critical role of PDGF in these of Id2 in PDGF-driven glioma. Deletion of Id2 in tumor cells gliomas (Fig. 1). In evaluating gene networks regulated by led to a significant increase in the duration of survival of mice PDGF signaling, we found several gene groups involved in who develop glioma without expression of Id2 (Fig. 5), but did metabolism, proliferation and cell cycle, signal transduction, not fully eliminate tumor formation in our model. We have not apoptosis, and protein degradation that were differentially yet pursued this finding, but several explanations are worthy regulated by PDGF. While the importance of growth factors of consideration. Some, such as incomplete Cre recombination for metabolism and proliferation has been widely recognized in vivo, seem less important than others. Others may be more (40), few reports link growth factor signaling to regulation of important, such as detection of redundancy in ID activity. In cells protein degradation (41, 42). The novelty of this finding without Id2, the presence of dysregulated PDGF during tumori- stimulated our interest in studying these genes. We found and genesis may facilitate the selection of cells expressing other Id characterized a previously unknown role for PDGF signaling in genes, such as Id1. Such a finding would be consistent with the the regulation of protein degradation and in supporting tumor observations of others that in a transgenic PDGF-driven glioma cell survival, which had not been previously described (Fig. 2). mouse model loss of Id1 led to increased overall survival (47). We found that USP1, a therapeutically targetable DUB, is Furthermore, the development of tumors in all animals suggests regulated by PDGF signaling (Fig. 2). Importantly, this pathway that ID2 may be able to contribute to different stages of tumor is also active in NSCs, which are known to be glioma-initiating development (48). We interpret the finding of delayed death in cells, and suggests a possible mechanism by which PDGF animals deleting Id2 in Gfap-expressing cells as indicating an signaling can drive gliomagenesis (Supplementary Fig. S4). important role for Id2 in tumor progression. PDGF stimulation led to an increase in Usp1 expression to Our findings indicating that a PDGF-E2F-USP1-ID2 axis is promote tumor cell survival. These findings are consistent with required for the survival of PDGF-GSCs raises the possibility of recent work of others who reported a role for USP1 in the a therapeutic benefit from inhibition of USP1 in patients with maintenance of human glioblastoma-derived GSCs (22), and PDGF-driven glioma. We used pimozide, a drug that is known extend those findings by defining a mechanism by which to inhibit USP1 (23), is active in the CNS (34), and approved survival is mediated and a recognizable glioma subgroup in for human use as an anti-psychotic drug (34) to examine which USP1 is likely to be an efficacious therapeutic target. pharmacologic targeting of the PDGF-E2F-USP1-ID2 axis in The E2F family of transcription factors encodes activators and established glioma. In vitro,PDGF-GSCsdisplayedamuch repressors. E2F1-3 are transcriptional activators that control cell- higher sensitivity to pimozide than normal cells or GSCs cycle progression (43). We found that PDGF signaling drives the derived from a mouse brain tumor model not driven by PDGF. expression of E2f1-3, leading to enhanced binding to the USP1 In vivo, inhibition of USP1, by pimozide, led to decreased ID2 promoter and causing increased Usp1 expression. These results levels and delayed the growth of PDGF-driven glioma in these link transcription driven by E2F activator proteins to regulation of tumors (Fig. 6). Using a subcutaneous syngeneic model to protein ubiquitination and to tumor cell survival in the context of monitor critically the effect of USP1 inhibition in vivo,we PDGF-driven proneural glioma. found that tumor growth decreased, suggesting the potential Our findings indicating that USP1 is required for tumor cell of using pimozide in treating brain tumor patients, especially as survival suggested that substrates of USP1 could be important for pimozide is known to cross the blood–brain barrier (34). maintaining the survival of PDGF-GSCs. The ID family of tran- Future experimentation using orthotopic models of PDGF- scriptional regulators has been implicated in several aspects of driven glioma will provide additional opportunities to examine tumorigenesis and in the pathology of glioma (28, 29). Given that such an effect in furtherdetail.Ourdataareconsistentwiththe ID1, ID2, and ID3 are known to be deubiquitinated and stabilized significant correlation between PDGFRA and USP1 expression by USP1 (26), we hypothesized that these USP1 substrates could in human glioblastoma specimens from the TCGA database, be important for maintaining PDGF-GSC survival. While pre- and with our finding that USP1 expression may be of prog- vious studies have described regulation of Id gene mRNA nostic value for patients with proneural glioblastoma (Fig. 7). expression by PDGF signaling in human and mouse fibroblasts In patients with proneural glioblastoma, we posit that low (44, 45), we found that PDGF signaling did not upregulate ID expression of USP1 couldleadtolowlevelsofID2protein, gene mRNA expression in PDGF-GSCs (Supplementary Fig. decreased tumor cell survival, and as shown by others, S6). PDGF signaling, however, did increase protein levels of increased tumor cell sensitivity to radiotherapy (22). Parallel ID2 by regulating Usp1 to decrease ID2 ubiquitination and in vitro studies using PDGF-responsive cultures derived from subsequent proteasomal degradation (Fig. 4). Consistent with human proneural PDGF-driven glioblastoma will allow further the role of USP1 in mediating PDGF-GSC survival (Fig. 2), exploration of this signaling cascade. Importantly, the molec- inhibition of Id2 expression by shRNA indicated that ID2 also ular pathway described here also might be implicated in other was required for survival of PDGF-GSCs (Fig. 4). This obser- tumors in which enhanced PDGF signaling has been charac- vation suggests strongly that other molecules in the same terized. Such tumor types include pediatric high-grade glioma pathway as ID2 and USP1 could be important for PDGF-GSC (32), DIPG (32), myeloid tumors (49), GIST (50), and others survival, and may have potential as therapeutic targets. Inter- (31). Pimozide administration might have therapeutic activity estingly, CDC20, an E3 ubiquitin ligase, a component of the in patients with these tumor types. Our data provide strong

www.aacrjournals.org Cancer Res; 76(10) May 15, 2016 2973

Rahme et al.

Figure 7. USP1 mRNA expression correlates with PDGFRA mRNA expression and is associated with survival of patients with proneural glioblastoma (GBM). A, dot plot of USP1 and PDGFRA mRNA levels in human glioblastoma tissues. Molecular subtype, Pearson r value, and P value are indicated above each graph. All data points with trend line are shown in graph. B, Kaplan–Meier survival curve showing overall survival of glioblastoma patients grouped by USP1 expression level. Molecular subtype is indicated above the graph and P values are indicated in the graphs.

2974 Cancer Res; 76(10) May 15, 2016 Cancer Research

A PDGF-E2F-USP1-ID2 Axis Is Required for Proneural Glioma

preclinical evidence that pimozide administration may have Acknowledgments therapeutic activity in patients with proneural glioma through The authors are grateful for stimulating discussion with all Israel laboratory inhibition of the PDGF-E2F-USP1-ID2 axis. members. The authors thank Ms. Tabatha Richardson at Dartmouth for her editorial advice. Resources of the Norris Cotton Cancer Center (NCCC) Geno- Disclosure of Potential Conﬂicts of Interest mics Shared Resource, NCCC Immunology Shared Resource, Transgenics and Genetic Constructs Shared Resource (Ms. Jennifer L. Fields), and the Center for ﬂ No potential con icts of interest were disclosed. Comparative Medicine and Research (CCMR) at Dartmouth were used for this study. Authors' Contributions Conception and design: G.J. Rahme, S.N. Fiering, M.A. Israel Grant Support Development of methodology: G.J. Rahme, A.L. Young, E.J. Bivona, Y. Hitoshi, The study was supported by grants from the Hitchcock Foundation (G.J. M.A. Israel Rahme), the Jordan and Kyra Memorial Foundation (M.A. Israel), and the Acquisition of data (provided animals, acquired and managed patients, Theodora B. Betz Foundation (M.A. Israel). G.J. Rahme is supported by a provided facilities, etc.): G.J. Rahme, A.L. Young, S.N. Fiering fellowship from the Albert J. Ryan foundation. Analysis and interpretation of data (e.g., statistical analysis, biostatistics, The costs of publication of this article were defrayed in part by the payment of computational analysis): G.J. Rahme, Z. Zhang, C. Cheng, M.A. Israel page charges. This article must therefore be hereby marked advertisement in Writing, review, and/or revision of the manuscript: G.J. Rahme, C. Cheng, accordance with 18 U.S.C. Section 1734 solely to indicate this fact. S.N. Fiering, M.A. Israel Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): Z. Zhang, Y. Hitoshi Received August 7, 2015; revised February 4, 2016; accepted February 11, Study supervision: M.A. Israel 2016; published OnlineFirst March 7, 2016.

References 1. Omuro A, DeAngelis LM. Glioblastoma and other malignant gliomas: a to temozolomide in a schedule-dependent manner. Invest New Drugs clinical review. JAMA 2013;310:1842–50. 2013;31:1169–81. 2. Louis DN, Ohgaki H, Wiestler OD, Cavenee WK, Burger PC, Jouvet A, et al. 15. Bota DA, Alexandru D, Keir ST, Bigner D, Vredenburgh J, Friedman HS. The 2007 WHO classification of tumours of the central nervous system. Proteasome inhibition with bortezomib induces cell death in GBM stem- Acta Neuropathol 2007;114:97–109. like cells and temozolomide-resistant glioma cell lines, but stimulates 3. Phillips HS, Kharbanda S, Chen R, Forrest WF, Soriano RH, Wu TD, et al. GBM stem-like cells' VEGF production and angiogenesis: Laboratory Molecular subclasses of high-grade glioma predict prognosis, delineate a investigation. J Neurosurg 2013;119:1415–23. pattern of disease progression, and resemble stages in neurogenesis. Cancer 16. Turcu FER, Ventii KH, Wilkinson KD. Regulation and cellular roles of ubiqui- Cell 2006;9:157–73. tin-specific deubiquitinating enzymes. Annu Rev Biochem 2009;78:363. 4. Liang Y, Diehn M, Watson N, Bollen AW, Aldape KD, Nicholas MK, et al. 17. Lill JR, Wertz IE. Toward understanding ubiquitin-modifying enzymes: Gene expression profiling reveals molecularly and clinically distinct sub- from pharmacological targeting to proteomics. Trends Pharmacol Sci types of glioblastoma multiforme. Proc Natl Acad Sci U S A 2005;102: 2014;35:187–207. 5814–9. 18. Hunter T. The age of crosstalk: phosphorylation, ubiquitination, and 5. Verhaak RG, Hoadley KA, Purdom E, Wang V, Qi Y, Wilkerson MD, et al. beyond. Mol Cell 2007;28:730–8. Integrated genomic analysis identifies clinically relevant subtypes of glio- 19. Huang TT, Nijman SM, Mirchandani KD, Galardy PJ, Cohn MA, Haas W, blastoma characterized by abnormalities in PDGFRA, IDH1, EGFR, and et al. Regulation of monoubiquitinated PCNA by DUB autocleavage. Nat NF1. Cancer Cell 2010;17:98–110. Cell Biol 2006;8:339–47. 6. Brennan CW, Verhaak RG, McKenna A, Campos B, Noushmehr H, Salama 20. Nijman SM, Huang TT, Dirac AM, Brummelkamp TR, Kerkhoven RM, SR, et al. The somatic genomic landscape of glioblastoma. Cell 2013;155: D'Andrea AD, et al. The deubiquitinating enzyme USP1 regulates the 462–77. Fanconi anemia pathway. Mol Cell 2005;17:331–9. 7. Dai C, Celestino JC, Okada Y, Louis DN, Fuller GN, Holland EC. PDGF 21. Cotto-Rios XM, Jones MJ, Busino L, Pagano M, Huang TT. APC/CCdh1- autocrine stimulation dedifferentiates cultured astrocytes and induces dependent proteolysis of USP1 regulates the response to UV-mediated oligodendrogliomas and oligoastrocytomas from neural progenitors and DNA damage. J Cell Biol 2011;194:177–86. astrocytes in vivo. Genes Dev 2001;15:1913–25. 22. Lee JK, Chang N, Yoon Y, Yang H, Cho H, Kim E, et al. USP1 targeting 8. Hitoshi Y, Harris BT, Liu H, Popko B, Israel MA. Spinal glioma: platelet- impedes GBM growth by inhibiting stem cell maintenance and radio- derived growth factor B-mediated oncogenesis in the spinal cord. Cancer resistance. Neuro Oncol 2015:nov091. Res 2008;68:8507–15. 23. Chen J, Dexheimer TS, Ai Y, Liang Q, Villamil MA, Inglese J, et al. Selective 9. Assanah M, Lochhead R, Ogden A, Bruce J, Goldman J, Canoll P. Glial and cell-active inhibitors of the USP1/UAF1 deubiquitinase complex progenitors in adult white matter are driven to form malignant gliomas by reverse cisplatin resistance in non-small cell lung cancer cells. Chem Biol platelet-derived growth factor-expressing retroviruses. J Neurosci 2006;26: 2011;18:1390–400. 6781–90. 24. Liang Q, Dexheimer TS, Zhang P, Rosenthal AS, Villamil MA, You C, et al. A 10. Jiang YW, Boije M, Westermark B, Uhrbom L. PDGF-B can sustain self-renewal selective USP1-UAF1 inhibitor links deubiquitination to DNA damage and tumorigenicity of experimental glioma-derived cancer-initiating cells by responses. Nat Chem Biol 2014;10:298–304. preventing oligodendrocyte differentiation. Neoplasia 2011;13:492–503. 25. Cohn MA, Kowal P, Yang K, Haas W, Huang TT, Gygi SP, et al. A UAF1- 11. Lei L, Sonabend AM, Guarnieri P, Soderquist C, Ludwig T, Rosenfeld S, et al. containing multisubunit protein complex regulates the Fanconi anemia Glioblastoma models reveal the connection between adult glial progeni- pathway. Mol Cell 2007;28:786–97. tors and the proneural phenotype. PLoS One 2011;6:e20041–e41. 26. Williams SA, Maecker HL, French DM, Liu J, Gregg A, Silverstein LB, et al. 12. Ozawa T, Riester M, Cheng Y-K, Huse JT, Squatrito M, Helmy K, et al. Most USP1 deubiquitinates ID proteins to preserve a mesenchymal stem cell human non-GCIMP glioblastoma subtypes evolve from a common pro- program in osteosarcoma. Cell 2011;146:918–30. neural-like precursor glioma. Cancer cell 2014;26:288–300. 27. Mistry H, Hsieh G, Buhrlage SJ, Huang M, Park E, Cuny GD, et al. Small- 13. Adams J. The proteasome: a suitable antineoplastic target. Nat Rev Cancer molecule inhibitors of USP1 target ID1 degradation in leukemic cells. Mol 2004;4:349–60. Cancer Ther 2013;12:2651–62. 14. Vlachostergios PJ, Hatzidaki E, Befani CD, Liakos P, Papandreou CN. 28. Lasorella A, Benezra R, Iavarone A. The ID proteins: master regulators of Bortezomib overcomes MGMT-related resistance of glioblastoma cell lines cancer stem cells and tumour aggressiveness. Nat Rev Cancer 2014;14:77–91.

www.aacrjournals.org Cancer Res; 76(10) May 15, 2016 2975

Rahme et al.

29. Niola F, Zhao X, Singh D, Sullivan R, Castano A, Verrico A, et al. Mesen- 40. Galluzzi L, Kepp O, Vander Heiden MG, Kroemer G. Metabolic targets for chymal high-grade glioma is maintained by the ID-RAP1 axis. J Clin Invest cancer therapy. Nat Rev Drug Discov 2013;12:829–46. 2013;123:405. 41. Li YP, Chen Y, John J, Moylan J, Jin B, Mann DL, et al. TNF-alpha acts via p38 30. Israel MA, Hernandez M-C, Florio M, Andres-Barquin PJ, Mantani A, Carter MAPK to stimulate expression of the ubiquitin ligase atrogin1/MAFbx in JH, et al. Id gene expression as a key mediator of tumor cell biology. Cancer skeletal muscle. FASEB J 2005;19:362–70. research 1999;59(7 Suppl):S1726–S30. 42. Sacheck JM, Ohtsuka A, McLary SC, Goldberg AL. IGF-I stimulates muscle 31. Heldin C-H. Targeting the PDGF signaling pathway in tumor treatment. growth by suppressing protein breakdown and expression of atrophy- Cell Commun Signal 2013;11:97. related ubiquitin ligases, atrogin-1 and MuRF1. Am J Physiol Endocrinol 32. Wu G, Diaz AK, Paugh BS, Rankin SL, Ju B, Li Y, et al. The genomic Metab 2004;287:E591–601. landscape of diffuse intrinsic pontine glioma and pediatric non-brainstem 43. Polager S, Ginsberg D. E2F–at the crossroads of life and death. Trends Cell high-grade glioma. Nat Genet 2014;46:444–50. Biol 2008;18:528–35. 33. Rahme GJ, Israel MA. Id4 suppresses MMP2-mediated invasion of glio- 44. Barone MV, Pepperkok R, Peverali FA, Philipson L. Id proteins control blastoma-derived cells by direct inactivation of Twist1 function. Oncogene growth induction in mammalian cells. Proc Natl Acad Sci U S A 1994;91: 2015;34:53–62. 4985–8. 34. Finder R, Brogden R, Sawyer PR, Speight T, Spencer R, Avery G. Pimozide: a 45. Christy BA, Sanders LK, Lau LF, Copeland NG, Jenkins NA, Nathans D. An review of its pharmacological properties and therapeutic uses in psychiatry. Id-related helix-loop-helix protein encoded by a growth factor-inducible Drugs 1976;12:1–40. gene. Proc Natl Acad Sci U S A 1991;88:1815–9. 35. Garcia AD, Doan NB, Imura T, Bush TG, Sofroniew MV. GFAP-expressing 46. Lasorella A, Stegmuller€ J, Guardavaccaro D, Liu G, Carro MS, Rothschild G, progenitors are the principal source of constitutive neurogenesis in adult et al. Degradation of Id2 by the anaphase-promoting complex couples cell mouse forebrain. Nat Neurosci 2004;7:1233–41. cycle exit and axonal growth. Nature 2006;442:471–74. 36. Imura T, Kornblum HI, Sofroniew MV. The predominant neural stem cell 47. Barrett LE, Granot Z, Coker C, Iavarone A, Hambardzumyan D, Holland isolated from postnatal and adult forebrain but not early embryonic EC, et al. Self-renewal does not predict tumor growth potential in mouse forebrain expresses GFAP. J Neurosci 2003;23:2824–32. models of high-grade glioma. Cancer Cell 2012;21:11–24. 37. Jacque CM, Vinner C, Kujas M, Raoul M, Racadot J, Baumann NA. 48. Havrda MC, Paolella BR, Ran C, Jering KS, Wray CM, Sullivan JM, et al. Determination of glial ﬁbrillary acidic protein (GFAP) in human brain Id2 mediates oligodendrocyte precursor cell maturation arrest and is tumors. J Neurol Sci 1978;35:147–55. tumorigenic in a PDGF-rich microenvironment. Cancer Res 2014;74: 38. Weiss WA, Burns MJ, Hackett C, Aldape K, Hill JR, Kuriyama H, et al. 1822–32. Genetic determinants of malignancy in a mouse model for oligodendro- 49. Golub TR, Barker GF, Lovett M, Gilliland DG. Fusion of PDGF receptor b to glioma. Cancer Res 2003;63:1589–95. a novel ets-like gene, tel, in chronic myelomonocytic leukemia with t (5; 39. Chin L, Meyerson M, Aldape K, Bigner D, Mikkelsen T, VandenBerg 12) chromosomal translocation. Cell 1994;77:307–16. S, et al. Comprehensive genomic characterization deﬁnes 50. Heinrich MC, Corless CL, Duensing A, McGreevey L, Chen C-J, Joseph N, human glioblastoma genes and core pathways. Nature 2008;455: et al. PDGFRA activating mutations in gastrointestinal stromal tumors. 1061–68. Science 2003;299:708–10.

2976 Cancer Res; 76(10) May 15, 2016 Cancer Research

PDGF Engages an E2F-USP1 Signaling Pathway to Support ID2-Mediated Survival of Proneural Glioma Cells

Gilbert J. Rahme, Zhonghua Zhang, Alison L. Young, et al.

Cancer Res 2016;76:2964-2976. Published OnlineFirst March 7, 2016.

Updated version Access the most recent version of this article at: doi:10.1158/0008-5472.CAN-15-2157

Supplementary Access the most recent supplemental material at: Material http://cancerres.aacrjournals.org/content/suppl/2016/05/18/0008-5472.CAN-15-2157.DC1

Cited articles This article cites 49 articles, 12 of which you can access for free at: http://cancerres.aacrjournals.org/content/76/10/2964.full#ref-list-1

Citing articles This article has been cited by 2 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/76/10/2964.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://cancerres.aacrjournals.org/content/76/10/2964. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.