sign in sign up
<<
Home , Ephrin, Ephrin B2, Ephrin receptor, PDGFRA

PDGFRβ reverses EphB4 signaling in alveolar rhabdomyosarcoma

M. Imran Aslama,b,c,d, Jinu Abrahamc, Atiya Mansoore, Brian J. Drukera,b,f,1, Jeffrey W. Tynera,g,1,2, and Charles Kellerc,1,2

aKnight Institute, Oregon Health and Science University, Portland, OR 97239; bDivision of Hematology and Medical Oncology, Department of Medicine, Oregon Health and Science University, Portland, OR 97239; cPediatric Cancer Biology Program, Papé Family Pediatric Research Institute, Department of Pediatrics, Oregon Health and Science University, Portland, OR 97239; dHoward Hughes Medical Institute Medical Research Fellows Program, Chevy Chase, MD 20815; eDepartment of Pathology, Oregon Health and Science University, Portland, OR 97239; fHoward Hughes Medical Institute, Portland, OR 97239; and gDepartment of Cell and Developmental Biology, Oregon Health and Science University, Portland, OR 97239

Contributed by Brian J. Druker, March 21, 2014 (sent for review January 28, 2013) Alveolar rhabdomyosarcoma (aRMS) is an aggressive myogenic progression in the malignancy. Particularly in , childhood malignancy, not infrequently presenting as incurable loss of EphB4 expression was shown to be critical at the ade- metastatic disease. To identify therapeutic targets, we performed noma-carcinoma transition (11). an unbiased kinome RNA interference screen in primary cell The platelet derived receptors (PDGFRs) α and cultures from a genetically engineered, conditional mouse model of β have been shown to be expressed both on tumor cells and aRMS. We identified B4 (EphB4) as a target that is stromal cells of mesenchymal origin in the tumor microenviron- widely expressed in human aRMS and that portends a poor clinical ment (12, 13). In turn, the PDGFRs are involved in multiple outcome in an expression level-dependent manner. We also uncov- malignant processes through activating or driving on- ered cross-talk of this with another receptor tyro- cogenic signaling pathways in both epithelial and hematologic sine kinase, PDGFRβ, which facilitates PDGF -dependent, malignancies (14–16).InaRMS,PDGFRα is a transcriptional ephrin ligand-independent activation of EphB4 converging on the target of a translocation-mediated fusion of PAX3 and Akt and Erk1/2 pathways. Conversely, EphB4 activation by its cog- FOXO1 (PAX3:FOXO1) and a therapeutic target in preclinical nate ligand, EphrinB2, did not stimulate PDGFRβ; instead, apopto- studies using the small molecule inhibitor or antibody- sis was paradoxically induced. Finally, we showed that small- mediated receptor blockade (17). Although cross-talk between molecule inhibition of both PDGFRβ and EphB4 by resulted the PDGFRs and the Eph receptors has not been directly in- in a significant decrease in tumor cell viability in vitro, as well as vestigated in tumorigenesis, evidence of an existing functional decreased tumor growth rate and significantly prolonged survival interface between the receptors has been implicated in select non- in vivo. To our knowledge, these results are the first to identify malignant biological systems (18). Furthermore, targeting RTKs EphB4 and its cross-talk with PDGFRβ as unexpected vital determi- nants of tumor cell survival in aRMS, with EphB4 at the crux of Significance a bivalent signaling node that is either mitogenic or proapoptotic.

Effective targeted therapies to complement already intensive MEDICAL SCIENCES | pediatric | muscle chemotherapy are much needed for the childhood muscle cancer rhabdomyosarcoma, yet few targeted agents have been he childhood cancer alveolar rhabdomyosarcoma (aRMS) is identified that improve long-term survival. In particular, the Ta malignancy for which molecularly targeted therapies are in alveolar subtype of rhabdomyosarcoma accounts for dispro- great need. More than half of aRMS patients present with unre- portionate mortality. Herein, the receptor EphB4 sectable or metastatic disease, yet despite maximally intensive is identified as a potential two-way switch for alveolar rhabdo- chemotherapy, survival of these patients has remained unchanged myosarcoma. Whereas the typical EphB4 ligand, EphrinB2, drives for four decades (1). For adults with aRMS, survival probability is tumor cells toward apoptosis, the interaction between EphB4 even more dismal (2). Thus, viable targeted therapies to incor- and another , PDGFRβ, drives tumors to porate into clinical trials for aRMS are greatly anticipated. We proliferate in the presence of the PDGFRβ ligand, PDGF-BB. The have previously identified receptor tyrosine kinases (RTKs) to be Food and Drug Administration-approved dual EphB4-PDGFRβ high-value target candidates in aRMS based on clinical survival inhibitor, dasatinib, is found to have significant preclinical ac- data (3), which has led to the present studies performing an un- tivity, which is clinically relevant because EphB4 and PDGFRβ are biased examination of the tyrosine kinome. independent poor prognostic factors in this childhood disease. Ephrin receptors (Eph ) are the largest family of RTKs and bind to Eph receptor-interacting (ephrin) ligands, which are Author contributions: M.I.A., J.A., J.W.T., and C.K. designed research; M.I.A., J.A., and C.K. glycosylphosphatidylinositol-linked or transmembrane ligands at performed research; M.I.A., J.W.T., and C.K. contributed new reagents/analytic tools; M.I.A., sites of cell–cell contact. This receptor-ligand axis is atypical for J.A., A.M., B.J.D., J.W.T., and C.K. analyzed data; and M.I.A., B.J.D., J.W.T., and C.K. wrote RTKs in that it generates a bidirectional signaling cascade in both the paper. the cell expressing the receptor and the cell presenting the ligand. Conflict of interest statement: B.J.D. is principal investigator or coinvestigator on Novartis and Bristol-Myers Squibb (BMS) clinical trials. His institution has contracts with these Although both Eph receptors and are demonstrated to be companies to pay for patient costs, nurse and data manager salaries, and institutional widely expressed in a variety of malignancies, their contribution overhead. B.J.D. does not derive salary, nor does his laboratory receive funds from these to tumorigenesis has been controversial. This disparity has been contracts. Oregon Health and Science University (OHSU) and B.J.D. have a financial in- because of the ability of Eph receptors, in particular, to either terest in MolecularMD. OHSU has licensed technology used in some of these clinical trials to MolecularMD. This potential individual and institutional conflict of interest has been contribute to tumorigenesis or to perform a role in tumor sup- reviewed and managed by OHSU. C.K. previously employed a technician who is a family pression, depending upon the cellular context (4–8). In cancer, member of a BMS employee. these signaling proteins may support progression, not necessarily Freely available online through the PNAS open access option. through association with their respective ligands, but through 1To whom correspondence may be addressed. E-mail: [email protected], tynerj@ohsu. cross-talk using other oncogenic signaling pathways, such as Akt, edu, or [email protected]. the Src family kinases, or EGFR (4, 9, 10). Another interesting 2J.W.T. and C.K. contributed equally to this work. ’ and important aspect of the Eph receptors function in cancer is This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. the regulation of their expression with respect to the stage of 1073/pnas.1403608111/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1403608111 PNAS | April 29, 2014 | vol. 111 | no. 17 | 6383–6388 Downloaded by guest on September 26, 2021 with small-molecule inhibitors is a major clinical trial interest area of the Children’s Oncology Group (19). Thus far, the importance of EphB4 in tumorigenesis has been interrogated by methods involving inhibition of extracellular do- main interaction with its ligand, as well as in vivo silencing of the receptor in murine tumor xenograft models (4, 8); however, the potential benefit of clinical small-molecule inhibitors targeting EphB4 kinase activity in cancer is yet to be explored. In this re- port, we have identified EphB4 as a druggable target in aRMS, a result obtained after investigating a surprising dichotomy for the signaling role of EphB4 through EphrinB2 versus through cross- talk with PDGFRβ and PDGF ligand. Results RNA-Interference–Assisted Identification Screen. With the aim of identifying novel tyrosine kinase gene targets in aRMS, we first established primary tumor cell cultures from our genetically engineered, conditional mouse model of aRMS, as previously de- scribed (20, 21). Primary cell culture U23674 was derived from a primary limb tumor of a mouse comparable to Intergroup Rhab- domyosarcoma Study Group (IRSG) surgical stage T2b/N1/M0, and culture U48484 was derived from a pulmonary metastasis of a Fig. 1. RNA-interference screen identifying EphB4 and PDGFRβ are required different tumor-bearing animal comparable to IRSG surgical stage for tumor cell survival. (A) Cell viability result for the RNAi-assisted protein T2b/N0/M1. We performed an RNA-interference assisted protein identification screen of primary (U23674) and metastatic (U48484) mouse target identification (RAPID) screen of validated siRNAs to in- aRMS tumor cultures. Dashed lines are ±2 SD with respect to median viability dividually silence each member of the mouse tyrosine kinome of cells. (B) Expression of EphB4, PDGFRβ, and EphrinB2 in U23674 and (22, 23). Cell viability was assessed by MTS assay to determine the U48484 in addition to C2C12, an immortalized murine myoblast cell line. (C) siRNAs that have deleterious effects on cell viability. Representative immunoblot of U23674 cells showing expression and knock- β The statistical threshold used to classify a significant decrease in down efficiency of EphB4 and PDGFR in conditions identical to RNAi screen in tumor cell viability after siRNA gene silencing was as previously A. White bar separates lane cropped from a lane on same gel. (D) Cell viability assessed by MTS assay on day 4 of murine aRMS cells and C2C12 transfected described (22, 23). This method identified Ntrk4, Pdgfrβ,andEphB4 β with nonspecific targeting siRNA, in addition to siRNA targeting EphB4 or in U23674, and Pdgfr , EphB4,andEphB6 in U48484. Interestingly, Pdgfrß alone or in combination, EphrinB2 and Plk1 (+ control). *P < 0.05 Pdgfrβ and EphB4 were both identified as targets in both cell lines compared with nonspecific siRNA by Student t test. For all experiments we (Fig. 1A). Expression of these RTKs at the protein level was vali- included an siRNA targeted to Plk1 to serve as a positive control. dated by immunoblotting (Fig. 1B), and knockdown confirmed at C Pdgfrβ the protein level (Fig. 1 ). The average knockdown of and − EphB4 in U23674 cells was ∼98% and 78%, respectively, and ∼95% with PAX3:FOXO1 aRMS (r = 0.82) or human skeletal muscle and ∼98% in U48484, respectively (Fig. S1 A and B). We further (r = 0.80) (Fig. 2A). Immunoblot of PDGFRβ and EphB4 in four validated the RAPID screen results using pooled EphB4 and Pdgfrβ previously described human RMS cell lines, Rh5, Rh30, Rh3 (a siRNA in aRMS compared with a control mouse myoblast cell line subclone of Rh28), and Rh18 was also performed (Fig. 2B). Rh5, (C2C12) (Fig. 1D). To determine if decrease in tumor cell viability Rh30, and Rh3 are PAX3:FOXO1 fusion-positive aRMS, whereas from EphB4 silencing was a ligand-dependent or -independent Rh18 is a fusion-negative rhabdomyosarcoma. Both RTKs were phenotype, we silenced EphrinB2 and saw no appreciable de- present at the protein level in Rh5, Rh30, and Rh18, although the crease in tumor cell viability (Fig. 1D), suggesting EphB4’son- cell line Rh3 did not express EphB4. cogenic effect is independent of its cognate ligand, EphrinB2. Immunohistochemistry on a human aRMS tissue microarray There was no cooperative decrease in tumor cell viability when consisting of 19 unique cases with 31 sections confirmed ex- both receptors were silenced concurrently (Fig. 1D). Results of pression of EphB4 (scores of 0–3) (Fig. S2B) and PDGFRβ all MTS assays were confirmed by flow cytometry (Fig. S1C). (scores of 0–3) (Fig. S3A) at the protein level. The majority of To control for potential off-target effects causing the observed tumor sections demonstrated strong staining of both EphB4 and decrease in cell viability, we used the individual duplex siRNAs PDGFRβ, although there were sections with weak expression of that comprise the Pdgfrβ and EphB4 siRNA pools used in the either receptor. Scoring by staining intensity is summarized in aforementioned studies to transfect U23674 and U48484 cells. SI Methods. Fig. 2C includes representative images of human The decrease in relative tumor cell viability we observed corre- aRMS tissue stained for EphB4 and PDGFRβ, as well as their lated with the individual duplex siRNA that most efficiently si- ligands EphrinB2 and PDGF-BB, respectively. In addition, wild- lenced Pdgfrβ or EphB4 (Fig. S1 D–G). type human skeletal muscle was stained for EphB4 and PDGFRβ (Fig. S3 B and C). Additional images stained for the receptors and EphB4 and EphrinB2 Is Expressed in Human and Mouse aRMS and their corresponding ligands can be found in Figs. S4 and S5.To EphB4 Is a Poor Prognostic Indicator in Fusion-Positive Human control for antibody specificity, antibodies were preincubated with aRMS. No previous studies have examined expression or func- blocking peptides before sections were stained (Figs. S4 and S5). tional significance of EphB4 or its ligand in aRMS. We examined PDGFRβ expression has been shown to be a strongly negative expression of EphB4 and EphrinB2 in human and mouse aRMS. prognostic factor in rhabdomyosarcoma, reducing 5-y survival By quantitative RT-PCR (RT-qPCR), EPHB4 and EPHRINB2 probability by >35% when adjusting for known clinical covariates expression were elevated in human aRMS compared with nor- (3). A similar analysis of this IRSG-IV microarray dataset (3) mal skeletal muscle, whereas in mouse aRMS only EphB4 levels demonstrated that EPHB4 expression showed a >70% decrease were elevated (Fig. S2A). In both human skeletal muscle and in 5-y survival between high and low EPHB4 levels in fusion aRMS, we saw a direct correlation between EphB4 and PDGFRβ positive aRMS (Fig. 2D). + expression level, with PAX3:FOXO1 aRMS demonstrating Taken together, these results are consistent with a role of a higher Pearson’s correlation coefficient (r = 0.91) compared potential clinical significance for EphB4 in aRMS. Furthermore,

6384 | www.pnas.org/cgi/doi/10.1073/pnas.1403608111 Aslam et al. Downloaded by guest on September 26, 2021 unidentified target of PAX3:FOXO1. An indirect relationship is likely because previous ChIP-Seq performed on global/genome- wide PAX3:FOXO1 binding sites in human aRMS, as well as various PAX3 isoform binding sites, did not identify EphB4 or PDGFRβ as direct targets of these transcription factors (25, 26). In our murine aRMS primary cell culture U23674, we observed that decreased Pax3:Foxo1 levels further decreases cell viability when silencing PDGFRβ or EphB4 individually (Fig. S6E). To confirm that EphB4 is a target of Pax3:Foxo1, we in- troduced a pCDNA3 vector expressing the Pax3:Foxo1 fusion transcript, or an empty vector, into murine myoblast (C2C12) cells. Accordingly, we observed increased expression of EphB4 at both the mRNA and protein level in C2C12 cells transfected with the fusion transcript compared with control C2C12 cells (Fig. S7). Pax3:Foxo1 expression was confirmed at the protein level in these cells (Fig. S7B). These data are consistent with EphB4 being either a direct or indirect target of Pax3:Foxo1a.

Activation of PDGFRβ and EphB4 in Murine aRMS Can Be Blocked by Tyrosine Kinase Inhibitors. We next determined the activation Fig. 2. EphB4 and EphrinB2 levels in mouse aRMS are similar to human aRMS status of EphB4 and Pdgfrβ in U23674 and U48484 cells by and tumor microarray of human aRMS shows strong expression of EphB4 and PDGFRβ by immunohistochemistry. (A) Correlation of PDGFRβ and EphB4 be- immunoprecipitating both RTKs from cell lysates and probing tween human normal skeletal muscle (SKM) and aRMS tumors. r indicates cal- with antiphospho-tyrosine (p-Tyr) antibody, revealing that both culated Pearson’s correlation coefficient. (B) Immunoblot of previously estab- receptors are phosphorylated at baseline (Fig. 4A). The primary lished human aRMS cell lines positive for PAX3:FOXO1: Rh5, Rh30, Rh3, and fusion cell cultures (U23674 and U48484) were treated with imatinib negative Rh18 show expression of PDGFRβ and EphB4. (C) Immunohistochemistry (targets PDGFRβ) and dasatinib (targets PDGFRβ and EphB4) of human aRMS tumors stained for EphB4, EphrinB2, PDGFRβ and PDGF-BB. for 72 h and cell viability assessed. These inhibitors decreased μ (Scale bar, 50 m.) (D) EPHB4 expression portends a poor overall survival for tumor cell viability and of their targeted patients in PAX3:FOXO1 fusion positive aRMS (P = 0.032). Histogram signifies frequency of varying EPHB4 expression level in fusion positive aRMS patients. receptors (Fig. 4 A and B). Dasatinib proved more potent; IC50s were achieved in the nanomolar range, whereas the imatinib IC50 was in the micromolar range. The IC50 for dasatinib in Rh5 significant correlating expression of PDGFRβ and EPHB4 sug- and Rh30 (human aRMS which express PDGFRβ and EphB4) gested potential interactions for further investigation. (Fig. 2B) were 35 and 200 nM, respectively (Fig. S8A).

Knockdown of PAX3:FOXO1 Regulates EphB4 Levels in Murine and siRNA-Mediated Knockdown of Pdgfrβ or EphB4 Induces Apoptosis and Human aRMS. Previously, other RTKs such as PDGFRα and Abrogates Anchorage-Independent Colony Formation. To determine MEDICAL SCIENCES c-Met have been established as downstream transcriptional tar- whether the decrease in cell viability observed after knockdown gets of PAX3:FOXO1 (17, 24). To investigate whether EphB4 and PDGFRβ are directly or indirectly regulated by the PAX3: FOXO1 , we silenced the in both U23674 and U48484 using siRNA targeted to eYFP. This ap- proach can be taken in our conditional mouse model because eYFP is expressed by means of an internal ribosomal entry site on a bicistronic transcript with Pax3:Foxo1 (Pax3:Foxo1-IRES- eYFP). We are thus able to knock down Pax3:Foxo1 in tumor cells by targeting eYFP. Silencing of Pax3:Foxo1 led to decreased levels of EphB4 protein (Fig. 3A), but Pdgfrβ expression sur- prisingly increased. This result for Pdgfrβ is the opposite of what we have observed for Pdgfrα (17). The same phenomenon was observed in Rh5 (human aRMS) when PAX3:FOXO1 was si- lenced (Fig. 3A). Furthermore, silencing of PAX3:FOXO1 in murine and human aRMS resulted in a decrease and increase in EphB4 and PDGFRβ levels, respectively (Fig. S6 A and B). We hypothesized that the increase observed in PDGFRβ protein may represent a compensatory response to loss of EphB4 after silencing of PAX3:FOXO1 or a tumor cell response to stress. To test the former hypothesis, we studied the effect on PDGFRβ protein level after direct silencing of EphB4 in murine and hu- man aRMS cells. Indeed, we observed that EphB4 silencing resulted in a significant increase in PDGFRβ protein (Fig. 3B), but had no effect on PDGFRβ mRNA levels in murine and hu- man aRMS (Fig. S6 C and D); this suggests a posttranslational β mechanism may be responsible for increased PDGFRβ protein. Fig. 3. Effect of PAX3:FOXO1 on EphB4 and PDGFR expression and sensi- β tivity to siRNA mediated silencing. (A) Effect of Pax3:Foxo1 silencing by Silencing of PDGFR in both murine and human aRMS resulted eYFP-targeted siRNA on expression of EphB4 and PDGFRβ by immunoblot in in a significant decrease of EphB4 at the protein (Fig. 3B) and U23674 and U48484 murine aRMS cells, and PAX3:FOXO1 silencing in human mRNA level (Fig. S6 C and D). Decreased EphB4 expression aRMS cell line Rh5 demonstrates an identical phenomenon. (B) Effect on from PAX3:FOXO1 knockdown suggests EphB4 is a previously PDGFRβ expression after EphB4 silencing in U23674, U48484, and Rh5 cells.

Aslam et al. PNAS | April 29, 2014 | vol. 111 | no. 17 | 6385 Downloaded by guest on September 26, 2021 of each receptor individually, we observed a reduction in phosphorylation of Erk 1/2 and Akt. FAK phosphorylation was reduced after silencing of EphB4, but not Pdgfrβ in both U23674 and U48484 cells (Fig. 4E). To further interrogate signaling from Pdgfrβ and EphB4 receptor stimulation, cells were serum- starved then stimulated with their respective ligands. PDGF-BB stimulated phosphorylation of Pdgfrβ, Akt, FAK, and Erk 1/2 (Fig. 5A). EphrinB2 stimulation of either cell line did not yield an increase in FAK, Akt, or Erk 1/2 phosphorylation (Fig. 5A). These results were also observed in the human aRMS cell lines Rh5 and Rh30 (Fig. 5B). Because EphrinB2 stimulation did not result in activation of Akt, Erk, or FAK, we immunoprecipitated EphB4 and probed with an anti–p-Tyr antibody to confirm EphB4 phosphorylation upon EphrinB2 stimulation (Fig. 6A). Taken together, these data suggest that with PDGF-BB ligand stimulation, PDGFRβ activates the common downstream effec- tors Akt and Erk1/2, whereas EphB4 signaling through Akt and Erk 1/2 appears to be independent of its cognate ligand, EphrinB2, in human and murine aRMS. Therefore, to test if Pdgfrβ could activate EphB4 (and thus, hypothetically, downstream mediators Akt and Erk 1/2), we looked for evidence of the ability of Pdgfrβ to phosphorylate EphB4. We immunoprecipitated EphB4 with and without stimulation of Pdgfrβ with PDGF-BB and showed that phosphorylation of EphB4 increases after PDGF-BB treat- ment (Fig. 6B). Imatinib was able to inhibit this effect despite EphB4 not being a direct target of imatinib. This result suggested

Fig. 4. PDGFRβ and EphB4 inhibition abrogates growth of mouse tumor primary cell cultures. (A) Small molecule inhibition of Pdgfrβ phosphoryla- tion by imatinib or dasatinib and EphB4 phosphorylation by dasatinib. DMSO was used as a vehicle control. (B) A 72 h MTS assay on cells treated with increasing concentrations of imatinib and dasatinib. IC50 for each drug and cell line are shown. (C) Knockdown of Pdgfrβ and EphB4 results in tumor cells undergoing apoptosis as measured by Annexin V. Early apoptotic cells + + are Annexin V only, late apoptotic cells are Annexin V and 7-AAD .(D) Anchorage-independent colony formation is decreased significantly in mu- rine aRMS U23674 and U48484 cells when Pdgfrβ and EphB4 are silenced by siRNA. Relative number of colonies normalized to nonspecific siRNA. (E) Immunoblot showing signaling downstream EphB4 and PDGFRβ shown after siRNA mediated silencing of each kinase in U23674 and U48484 cells. For PDGFRβ and EphB4 levels after corresponding siRNAs, see Fig. 3B.*P < 0.05 compared with nonspecific siRNA by Student t test.

of Pdgfrβ and EphB4 was from either cell cycle redistribution and/or apoptosis, we performed cell cycle analysis of U23674 and U48484 cells by FACS after transfection with nonspecific Pdgfrβ- or EphB4-targeted siRNA, revealing cell cycle distribution was not significantly altered (Fig. S8B). Silencing of EphB4 or Pdgfrβ in- dividually or simultaneously resulted in significant apoptosis as determined by FACS (Fig. 4C). Silencing of Pax3:Foxo1 was used as a comparator of tumor cell apoptosis (Fig. 4C). The ability of murine aRMS cells to form colonies in an anchorage-independent Fig. 5. PDGFRβ but not EphB4 stimulation by their respective ligands acti- colony formation assay was significantly inhibited upon Pdgfrβ and vates downstream cell survival and proliferative pathways in murine and EphB4 silencing (Fig. 4D and Fig. S8C). Taken together, these human aRMS. (A) Immunoblots showing U23674 and U48484 cells stimulated data indicate a vital role of Pdgfrβ and EphB4 in tumor cell sur- (after overnight serum starvation) with PDGF-BB (10 ng/mL, 15 min) alone or μ vival and transformation. in the presence of imatinib (100 nM, 1 h) or with EphrinB2 (2 g/mL, 15 min) alone or in the presence of dasatinib (100 nM, 1 h). Fc only (Fc portion lacking clustered ligand, 2 μg/mL for 15 min) lane represents control for Akt, Erk 1/2 Are Downstream of PDGFRβ and EphB4, Whereas FAK Is EphrinB2. (B) Immunoblots showing Rh5 and Rh30 cells stimulated (after Downstream of EphB4 and EphB4 Is Phosphorylated upon PDGF-BB overnight serum starvation) with PDGF-BB (10 ng/mL, 15 min) alone or in the Stimulation. We next assessed the downstream signaling pathways presence of imatinib (100 nM, 1 h) or with EphrinB2 (2 μg/mL, 15 min) alone affected by silencing of Pdgfrβ and EphB4. Upon knockdown or in the presence of dasatinib (100 nM, 1 h).

6386 | www.pnas.org/cgi/doi/10.1073/pnas.1403608111 Aslam et al. Downloaded by guest on September 26, 2021 that EphrinB2 was similarly stimulating a distinct signaling cascade. Because Crkl is a monogamous substrate for phos- phorylation by Abl (6), we examined Crkl phosphorylation upon EphrinB2 treatment. p-Crkl was stimulated by EphrinB2 treat- ment in both murine and human aRMS cells, U48484 and Rh5, respectively (Fig. 6E). The murine aRMS primary cell cultures U23674 and U48484 along with human aRMS cell lines Rh5 and Rh30 in the presence of Fc only control or EphrinB2 for a 3 d period showed that all cell lines underwent apoptosis upon EphrinB2 treatment (Fig. 6 F and G, respectively). These results indicate that although EphrinB2 does stimulate EphB4, the signaling cascade initiated by EphrinB2 is distinct from that stimulated through cross-talk with PDGFRβ.

Dasatinib-Mediated Inhibition of both PDGFRβ and EphB4 Improves Survival More than Imatinib-Mediated Inhibition of PDGFRβ Alone in an Orthotopic Allograft Model of aRMS. To determine whether dasatinib’s increased efficacy relative to imatinib was conserved in vivo, we orthotopically engrafted (SCID/hairless/outbred) mice with U23674 cells and treated these mice with dasatinib (15 or 50 mg/kg) or imatinib (100 mg/kg). Once tumors were 0.2 cc3, we commenced daily treatment with drug or vehicle control. Survival was significantly extended with dasatinib at a dose of 50 mg/kg compared with all other treatment groups (P < 0.05) (Fig. 7A). To confirm drug administration in vivo inhibited Pdgfrβ and EphB4 phosphorylation, immunoblotting of tumor lysates were performed. Immunoblot of tumors treated with daily vehicle, Fig. 6. PDGFRβ and EphB4 cross-talk; EphrinB2 signals through EphB4 to imatinib, and both high- or low-dose dasatinib showed that induce apoptosis in murine and human aRMS. (A) Representative immuno- β blot indicating EphrinB2 treatment (2 μg/mL, 15 min) results in phosphory- Pdgfr and EphB4 phosphorylation was decreased in dasatinib- lation of EphB4 and (B) PDGF-BB treatment (10 ng/mL, 15 min) results in treated mice, whereas only Pdgfrβ phosphorylation was de- EphB4 phosphorylation in murine aRMS (U48484) and human aRMS (Rh30). creased in mice treated with imatinib (Fig. 7 B and C). Taken PDGF-BB mediated phosphorylation is inhibited by imatinib. (C) A graded together, these data suggest that blockade of only PDGFRβ increase in PDGF-BB concentrations of 0, 100, 200, 500, and 1,000 ng/mL activity (by imatinib) can impart some inhibition of tumor growth, serves as a mitogen in U23674 and U48484 cells. (D) A graded increase in whereas direct blockade of both Pdgfrβ and EphB4 activity PDGF-BB concentrations of 0, 100, 200, 500, and 1,000 ng/mL serves as (by dasatinib) imparts the greatest effect on inhibition of tu- MEDICAL SCIENCES a mitogen in Rh5 and Rh30 cells. (E) In murine aRMS U48484 and human aRMS Rh5, EphrinB2 (2 μg/mL, 15 min) treatment stimulates p-Crkl, which mor growth. dasatinib inhibits. (F) EphrinB2 (20 μg/mL) treatment versus Fc only for 3 d of mouse aRMS and (G) human aRMS cells results in tumor cells undergoing Discussion + apoptosis as measured by Annexin V. Early apoptotic cells are Annexin V Through an unbiased RNAi screen targeted to the tyrosine only, late apoptotic cells are Annexin V and 7-AAD+.*P < 0.01, **P < 0.05 by kinome, we identified EphB4 as a novel target critical to tumor Student t test compared with respective control. cell viability. We also found EphB4 to be a poor prognostic indicator in PAX3:FOXO1 fusion-positive aRMS. EphB4 has that the phosphorylation of EphB4 is mediated by an imatinib target, such as PDGFRβ (one of only a few imatinib targets). Fi- nally, we observed that PDGF-BB ligand increased by two- to threefold in a dose-dependent manner, affirming a mito- genic phenotype in murine and human aRMS (Fig. 6 C and D, respectively). In addition, the demonstrated ability of kinase inhibitors imatinib or dasatinib to ablate this mitogenic pheno- type in murine primary cell culture of aRMS (U23674) (Fig. S8D) suggests that PDGF-BB ligand-mediated growth is a kinase- dependent process.

EphrinB2 Treatment Induces Apoptosis in Murine and Human aRMS. Although PDGF-BB activated PDGFRβ, EphB4, Akt, and Erk 1/2, resulting in stimulating mitogenesis of tumor cells, EphrinB2 Fig. 7. Dasatinib significantly slows aggressive murine aRMS tumor growth versus imatinib in vivo. (A) Kaplan–Meier survival curve of SCID/hairless/ was not capable of stimulating these downstream signaling path- outbred mice treated with vehicle daily, dasatinib 15 mg·kg·d, dasatinib ways, despite inducing phosphorylation of EphB4, suggesting that 50 mg·kg·d or imatinib 100 mg·kg·d, with endpoint measurement being days canonical EphB4 signaling may serve a different role. In prior for tumor volume to reach 2.0 cm3. Mantel–Cox analysis of survival curves reports, the Eph RTK family has been shown to function as tumor revealed P < 0.05 between comparison of vehicle and dasatinib 15 mg/kg suppressors (5, 11) or as (4, 27) in a wide variety of and imatinib 100 mg/kg and P < 0.05 between 50 mg/kg dasatinib and all β malignancies. An EphrinB2 induced EphB4 signaling cascade other treatment groups. (B) Immunoblot confirming inhibition of PDGFR phosphorylation by imatinib and dasatinib in vivo and (C) inhibition of mediated by the Abl-Crkl pathway resulting in apoptosis has EphB4 phosphorylation by dasatinib only (imatinib unable to inhibit EphB4 been shown in mammary carcinoma (6). Based on our previous activation) from randomly selected tumor-bearing mice. White bar separates observations in these murine and human aRMS, we hypothesized lanes cropped from the same gel.

Aslam et al. PNAS | April 29, 2014 | vol. 111 | no. 17 | 6387 Downloaded by guest on September 26, 2021 previously been identified to contribute toward tumor suppres- in Eph kinase-ephrin interactions (9). We show dasatinib is ef- sion, apoptosis, or an increased malignant phenotype in a variety ficacious at physiologically achievable serum concentrations (31) of other (4–6, 9, 11, 27, 28). In our studies, we further but acknowledge that while dasatinib inhibits both PDGFRβ and identified a previously undescribed mechanism of cross-talk be- EphB4, it also inhibits many other tyrosine (and /threo- tween EphB4 with another RTK, PDGFRβ, which is known to nine) kinases. Therefore, its antagonistic effects on aRMS cells be activated and expressed in aRMS (29, 30). in culture and in the xenograft studies could be due to inhibition Our studies demonstrated that both cell survival (Akt) and cell of not only PDGFRβ and EphB4, but could also be a result of its proliferation (Erk) signaling pathways are downstream of both inhibition of other kinases as well. It might also stand to reason RTKs, suggesting that both RTKs converge on similar pathways. that clinical trials targeting EphB4 in aRMS would benefit well We show EphB4 stimulation by its cognate ligand, EphrinB2, from a biomarker-driven strategy to identify which patients are was unable to stimulate Akt or Erk; however, PDGFRβ stimu- lation with PDGF-BB ligand was able to phosphorylate EphB4 best suited for EphB4-targeted therapy. as well as activate these downstream pathways in both murine Methods and human aRMS. Signaling interrogation revealed p-FAK was Human Samples. Human samples (normal skeletal muscle and tumor RNA, abrogated with EphB4 knockdown, but not by Pdgfrβ knockdown. β and human tissue microarrays) were provided by the Cooperative Human However, PDGFR stimulation via PDGF-BB was able to stim- Tissue Network or Children’s Oncology Group Biorepository. Studies were ulate FAK phosphorylation, presumably through cross-talk with performed under an Institutional Review Board-approved protocol. EphB4. These results allude to EphB4 having an ability to activate β oncogenic signaling pathways independent of PDGFR or its Other Methods. Detailed descriptions of cell culture, RT-qPCR, immunohis- cognate ligand, EphrinB2, as we also describe. Based on our tochemistry, immunoblotting and immunoprecipitation, RNA-interference results (summarized in Fig. S9), as well as previous literature studies, in vitro growth inhibition and proliferation assays, cell viability and corroborating the phenomenon of ephrin kinase signaling initiated apoptosis assays, in vivo studies, and statistical considerations are given in by noncanonical pathways, we deduce that EphB4 in aRMS may SI Methods. serve as a scaffold to activate other signaling pathways or associate with other proteins that are able to activate the receptor, similar to ACKNOWLEDGMENTS. We thank Stephanie Willis, Elaine Huang, and Chris Eide its cross-talk with PDGFRβ. Such a ligand-independent function for technical assistance, as well as other B.J.D. laboratory members for their invaluable support. M.I.A. is a Howard Hughes Medical Institute Medical Research of Eph receptors has been described in other model systems, such Fellow. B.J.D. is an investigator of the Howard Hughes Medical Institute. This work as prostate cancer and cells (9). was supported by Grant 5R01CA133229 and in part by a grant from the Joanna To translate our results to the clinic, we chose to test the Food McAfee Childhood Cancer Foundation (to C.K.); a grant from the William and Drug Administration-approved, dual PDGFRβ and EphB4 Lawrence and Blanche Hughes Foundation (to J.W.T.); a grant from the and Society (to J.W.T.); a grant from the V Foundation for Cancer inhibitor, dasatinib. A promising secondary attribute of dasatinib Research (to J.W.T.); and National Cancer Institute Grant 4R00CA151457 (to is its ability to inhibit the Src family kinases, which is relevant J.W.T.). The Cooperative Human Tissue Network and the Children’sOncology because the Src family kinases are responsible for reverse signaling Group Biorepository are funded by the National Cancer Institute.

1. Arndt CA, Crist WM (1999) Common musculoskeletal tumors of childhood and ado- 17. Taniguchi E, et al. (2008) PDGFR-A is a therapeutic target in alveolar rhabdomyo- lescence. N Engl J Med 341(5):342–352. sarcoma. 27(51):6550–6560. 2. Sultan I, Qaddoumi I, Yaser S, Rodriguez-Galindo C, Ferrari A (2009) Comparing adult 18. He S, et al. (2010) Soluble EphB4 inhibition of PDGF-induced RPE migration in vitro. and pediatric rhabdomyosarcoma in the surveillance, epidemiology and end results Invest Ophthalmol Vis Sci 51(1):543–552. program, 1973 to 2005: An analysis of 2,600 patients. J Clin Oncol 27(20):3391–3397. 19. Hawkins DS, Spunt SL, Skapek SX; COG Soft Tissue Sarcoma Committee (2013) 3. Blandford MC, et al. (2006) Rhabdomyosarcomas utilize developmental, myogenic Children’s Oncology Group’s 2013 blueprint for research: Soft tissue . Pediatr growth factors for disease advantage: A report from the Children’s Oncology Group. Blood Cancer 60(6):1001–1008. Pediatr Blood Cancer 46(3):329–338. 20. Nishijo K, et al. (2009) Credentialing a preclinical mouse model of alveolar rhabdo- 4. Xia G, et al. (2006) EphB4 receptor tyrosine kinase is expressed in bladder cancer and myosarcoma. Cancer Res 69(7):2902–2911. provides signals for cell survival. Oncogene 25(5):769–780. 21. Keller C, et al. (2004) Alveolar rhabdomyosarcomas in conditional Pax3:Fkhr mice: 5. Oricchio E, et al. (2011) The Eph-receptor A7 is a soluble tumor suppressor for fol- of Ink4a/ARF and Trp53 loss of function. Dev 18(21):2614– licular lymphoma. Cell 147(3):554–564. 2626. 6. Noren NK, Foos G, Hauser CA, Pasquale EB (2006) The EphB4 receptor suppresses breast 22. Tyner JW, et al. (2008) RNAi screening of the tyrosine kinome identifies therapeutic cancer cell tumorigenicity through an Abl-Crk pathway. Nat Cell Biol 8(8):815–825. targets in . Blood 111(4):2238–2245. 7. Kumar SR, et al. (2009) Preferential induction of EphB4 over EphB2 and its implication 23. Tyner JW, et al. (2009) RNAi screen for rapid therapeutic target identification in in colorectal cancer progression. Cancer Res 69(9):3736–3745. leukemia patients. Proc Natl Acad Sci USA 106(21):8695–8700. 8. Kertesz N, et al. (2006) The soluble extracellular domain of EphB4 (sEphB4) antagonizes 24. Ginsberg JP, Davis RJ, Bennicelli JL, Nauta LE, Barr FG (1998) Up-regulation of MET but EphB4-EphrinB2 interaction, modulates , and inhibits tumor growth. Blood not neural cell adhesion molecule expression by the PAX3-FKHR fusion protein in 107(6):2330–2338. alveolar rhabdomyosarcoma. Cancer Res 58(16):3542–3546. 9. Miao H, et al. (2009) EphA2 mediates ligand-dependent inhibition and ligand- 25. Wang Q, Kumar S, Mitsios N, Slevin M, Kumar P (2007) Investigation of downstream independent promotion of cell migration and invasion via a reciprocal regula- target genes of PAX3c, PAX3e and PAX3g isoforms in melanocytes by microarray tory loop with Akt. Cancer Cell 16(1):9–20. analysis. Int J Cancer 120(6):1223–1231. 10. Leroy C, et al. (2009) Quantitative phosphoproteomics reveals a cluster of tyrosine 26. Cao L, et al. (2010) Genome-wide identification of PAX3-FKHR binding sites in kinases that mediates SRC invasive activity in advanced colon carcinoma cells. Cancer rhabdomyosarcoma reveals candidate target genes important for development and Res 69(6):2279–2286. cancer. Cancer Res 70(16):6497–6508. 11. Batlle E, et al. (2005) EphB receptor activity suppresses colorectal cancer progression. 27. Masood R, et al. (2006) EphB4 provides survival advantage to squamous cell carci- Nature 435(7045):1126–1130. noma of the head and neck. Int J Cancer 119(6):1236–1248. 12. Andrae J, Gallini R, Betsholtz C (2008) Role of platelet-derived growth factors in 28. Noren NK, Lu M, Freeman AL, Koolpe M, Pasquale EB (2004) Interplay between EphB4 physiology and medicine. Genes Dev 22(10):1276–1312. on tumor cells and vascular ephrin-B2 regulates tumor growth. Proc Natl Acad Sci USA 13. Ostman A, Heldin CH (2007) PDGF receptors as targets in tumor treatment. Adv 101(15):5583–5588. Cancer Res 97:247–274. 29. Cen L, et al. (2007) Phosphorylation profiles of protein kinases in alveolar and em- 14. Steller EJ, et al. (2013) PDGFRB promotes liver metastasis formation of mesenchymal- bryonal rhabdomyosarcoma. Mod Pathol 20(9):936–946. like colorectal tumor cells. Neoplasia 15(2):204–217. 30. Armistead PM, et al. (2007) Expression of receptor tyrosine kinases and apoptotic 15. Noel P, Mesa RA (2013) Eosinophilic myeloid . Curr Opin Hematol 20(2): molecules in rhabdomyosarcoma: Correlation with overall survival in 105 patients. 157–162. Cancer 110(10):2293–2303. 16. Heinrich MC, et al. (2003) PDGFRA activating mutations in gastrointestinal stromal 31. Johnson FM, et al. (2010) Phase 1 pharmacokinetic and drug-interaction study of tumors. Science 299(5607):708–710. dasatinib in patients with advanced solid tumors. Cancer 116(6):1582–1591.

6388 | www.pnas.org/cgi/doi/10.1073/pnas.1403608111 Aslam et al. Downloaded by guest on September 26, 2021