RNA Interference Screening Identifies a Novel Role for Autocrine Fibroblast Growth Factor Signaling in Neuroblastoma Chemoresist

ORIGINAL ARTICLE RNA interference screening identiﬁes a novel role for autocrine ﬁbroblast growth factor signaling in neuroblastoma chemoresistance

F Salm1, P Cwiek1, A Ghosal2, A Lucia Buccarello2, F Largey2, C Wotzkow1,KHo¨ land1, B Styp-Rekowska3, V Djonov3, I Zlobec4, N Bodmer2, N Gross5, F Westermann6, SC Scha¨fer4 and A Arcaro1

Chemotherapeutic drug resistance is one of the major causes for treatment failure in high-risk neuroblastoma (NB), the most common extra cranial solid tumor in children. Poor prognosis is typically associated with MYCN amplification. Here, we utilized a loss-of-function kinome-wide RNA interference screen to identify genes that cause cisplatin sensitization. We identified fibroblast growth factor receptor 2 (FGFR2) as an important determinant of cisplatin resistance. Pharmacological inhibition of FGFR2 confirmed the importance of this kinase in NB chemoresistance. Silencing of FGFR2 sensitized NB cells to cisplatin-induced apoptosis, which was regulated by the downregulation of the anti-apoptotic proteins BCL2 and BCLXL. Mechanistically, FGFR2 was shown to activate protein kinase C-d to induce BCL2 expression. FGFR2, as well as the ligand fibroblast growth factor-2, were consistently expressed in primary NB and NB cell lines, indicating the presence of an autocrine loop. Expression analysis revealed that FGFR2 correlates with MYCN amplification and with advanced stage disease, demonstrating the clinical relevance of FGFR2 in NB. These findings suggest a novel role for FGFR2 in chemoresistance and provide a rational to combine pharmacological inhibitors against FGFR2 with chemotherapeutic agents for the treatment of NB.

Oncogene (2013) 32, 3944–3953; doi:10.1038/onc.2012.416; published online 1 October 2012 Keywords: neuroblastoma; RNAi screen; FGFR2; BCL2; chemoresistance; cisplatin

INTRODUCTION doxorubicin and vincristine.7,9 Although most high-risk NBs The development of drug resistance is a major cause of treatment initially respond to therapy, 50–60% have a relapse with the failure in many cancer types.1 The multiple mechanisms leading to appearance of more drug-resistant tumors. The identification of chemoresistance are subject of intensive study, but are not yet molecular mechanisms that lead to drug resistance may open a fully understood. Cisplatin is a platinum-based DNA-damaging new window for the design of potential combination therapies to cytotoxic agent2 and is widely used for the treatment of solid treat this patient group. tumors including testicular, ovarian, colorectal, lung, and head and RNA interference (RNAi) libraries allow high-throughput screens, neck cancers.3,4 Cisplatin induces apoptotic cell death, which which are based on systematically silencing gene expression in results from DNA damage-mediated responses. However, order to identify which genes are essential for a specific function 10 several mechanisms to circumvent cisplatin-mediated cell death in the cell. Such screens have already proven to be a highly have been described, which limit its clinical efficacy as an effective research tool for the identification of key determinants of 11–13 anticancer drug. Resistance to cisplatin can occur by different drug sensitivity. Here, we performed a kinome-wide RNAi mechanisms, including overexpression of anti-apoptotic BCL2 screen to identify kinases that, when downregulated, sensitize NB family members.5,6 Alterations in survival pathways are also cells to cisplatin-induced cell death. We report that RNAi targeting associated with drug resistance, for instance overexpression of of FGFR2 caused the strongest impact on cisplatin sensitization. HER-2/neu and over activation of phosphatidylinositol 3 kinase Our data demonstrate a mechanism through which fibroblast (PI3K)/Akt pathway.5,6 growth factor receptor 2 (FGFR2) signaling induces cisplatin Neuroblastoma (NB) is the most common pediatric extracanial resistance through protein kinase C (PKC)-d-dependent induction solid tumor and is responsible for 15% of all cancer-related deaths of BCL2 expression. Thus, our data provide evidence of the in childhood.7 NBs show very diverse clinical behavior, ranging importance of FGFR2 signaling in NB chemoresistance. from a widely disseminated and highly aggressive disease to tumors that differentiate or spontaneously regress with little or no therapy. Amplification of the MYCN oncogene is found RESULTS in approximately 20% of NB and correlates with poor siRNA screen to identify kinases mediating cisplatin resistance disease outcome.8 Current treatments for high-risk NB patients To identify genes conferring cisplatin resistance in NB, we include intense multimodality chemotherapies, including cisplatin, designed a robust RNAi-based loss-of function screen using a

1Department of Clinical Research, University of Bern, Bern, Switzerland; 2Department of Oncology, University Children’s Hospital Zurich, Zurich, Switzerland; 3Institute of Anatomy, University of Bern, Bern, Switzerland; 4Institute of Pathology, University of Bern, Bern, Switzerland; 5Department of Paediatrics, Paediatric Oncology Research, Lausanne University Hospital (CHUV), Lausanne, Switzerland and 6Department Tumor Genetics, German Cancer Research Center, Heidelberg (DKFZ), Germany. Correspondence: Dr A Arcaro, Division of Pediatric Hematology/Oncology, Department of Clinical Research, University of Bern, Tiefenaustrasse 120c, Bern 3004, Switzerland. E-mail: [email protected] Received 15 December 2011; revised 19 July 2012; accepted 31 July 2012; published online 1 October 2012 Role of FGFR2 signaling in neuroblastoma F Salm et al 3945

Figure 2. Validation of cisplatin sensitizing candidate. (a) Validation of the siRNA transfection efficiency. Taqman PCR and western blot showing FGFR2 gene downregulation after 48 h of transfection with two different siRNA sequences. (b) Cisplatin titration in LAN-1 cells transfected with siFGFR2 (white squares) or control siRNA (black circles). EC50 (siFGFR2) ¼ 1.32, EC50 (siControl) ¼ 4.17 in LAN-1. (c) Cell viability assay in LAN-1 cells treated with increasing concentrations of cisplatin (circles), increasing concentrations of cisplatin in presence of SU-5402 at a constant ratio of 1.25 (triangles) and increasing concentrations of SU-5402 (white squares). The combination index according to the Chou–Talalay method is CI ¼ 0.818. Error bars represent the s.d. of three independent experiments. (d) Tumors formed on chorioallantoic membrane (CAM) were treated with dimethylsulphoxide (DMSO) (control) or 20 mM SU-5402, quantification of changes in tumor volume before and after treatment are shown. Lines indicate the mean of each Figure 1. Human kinase siRNA high-throughput screen for cisplatin group, *Po0.05 compared with control treatment. sensitizer kinases. (a, b) Scattered plot of average Z-scores from cisplatin sensitivity screen. (c) Cisplatin sensitivity assay repeated in triplicate for the hit candidate kinases FGFR2 with two siRNA sequences in SH-SY5Y and LAN-1 cells. Survival fractions are shown, because they were not effectively targeted by at least two siRNA *Po0.05 compared with scrambled siRNA control, error bars sequences (Figure 1b and Supplementary Table 1). To validate the represent the s.d. of the mean of two independent experiments. results obtained in the screen, two different individual siRNA sequences targeting FGFR2 and PRKCD were reassayed in both cell lines. As expected, resistance to cisplatin was conferred by PRKCD library of short interference RNA duplexes (siRNA) targeting each and FGFR2 downregulation in SH-SY5Y and LAN-1, respectively. of the 719 known protein and lipid kinases in the human genome. Although not as strong as in LAN-1, the downregulation of FGFR2 The screen was performed in two well-characterized NB cell lines also sensitized SH-SY5Y cells to cisplatin (Figure 1c). The (LAN-1, SH-SY5Y). The cells were systematically transfected with differences in the sensitivity to FGFR2 downregulation might be the siRNAs and subsequently treated with low-dose cisplatin and due to the lower expression levels of FGFR2 in SH-SY5Y compared or vehicle. The dose of cisplatin used in the screen was optimized with LAN-1 (Supplementary Figure 1). Thus, the role of FGFR2 as a for each cell line to achieve a sensitizing effect of 20% decrease in drug resistance candidate gene was further investigated. cell survival (EC20) (data not shown). The results of the screen are represented as Z-scores approximating a normal distribution and are displayed in Figure 1. To rule out off target effects, we only Validation of siRNA screen hit candidates considered a gene as a candidate drug resistance gene if at least To confirm the specificity of the observed effects, the two different two independent siRNAs, caused sensitization to cisplatin scaled siRNA sequences targeting FGFR2 were analyzed for target down- by an average Z-score below—3. Within the candidates in LAN-1 regulation. Both siRNAs efficiently suppressed FGFR2 expression and in SH-SY5Y (Supplementary Table 1) that fulfilled these at mRNA and protein level (Figure 2a), indicating that the high criteria, FGFR2 (Z-score—4) was the top candidate in LAN-1 and throughput screen results were likely to be ‘on-target’. In order to PRKCD (Z-score—3) in SH-SY5Y. Although PRKCD did not have the establish the sensitivity to cisplatin treatment, dose-response lowest Z-score, genes with lower Z-scores were not selected, curves were performed (Figure 2b and Supplementary Figure 2a).

& 2013 Macmillan Publishers Limited Oncogene (2013) 3944 – 3953 Role of FGFR2 signaling in neuroblastoma F Salm et al 3946

Figure 3. FGFR2 silencing sensitizes NB cells to cisplatin-induced apoptosis. (a) Caspase-3 activation was detected by ﬂuorescence-activated cell sorting (FACS) analysis and western blot after transfection of siFGFR2 or scrambled siRNA in absence (black bars) or presence of 1.5 mM cisplatin (grey bars) in SH-SY5Y cells. (b) Cell lysates of siFGFR2 or siCONTROL transfected cells treated with 5 mM cisplatin were analyzed by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE) and western blot for the indicated BCL2 family members. (c) WAC-2 cells were treated with 10 mM cisplatin as single agent (black bars) or combined with 5 mM SU-5402 (grey bars) and analyzed for caspase-3 activation by FACS. (d) BCL2 protein and mRNA levels quantiﬁed by Taqman PCR of LAN-1 cells stimulated for 24 h with increasing concentrations of FGF-2. Error bars represent the s.d. of the mean, *Po0.05 one-way analysis of variance (ANOVA) using Tukey’s post test.

FGFR2 silencing significantly shifted EC50 values to lower cisplatin to cisplatin by triggering apoptosis. Following FGFR2 down- doses (siControl EC50 ¼ 4.17, siFGFR2 EC50 ¼ 1.32 in LAN-1, siControl regulation, the number of apoptotic cells was increased after EC50 ¼ 5.02, siFGFR2 EC50 ¼ 1.59 in SH-SY5Y). Further confirma- cisplatin treatment (Figure 3a and Supplementary Figure 3). tion that FGFR2 signaling modulates resistance to cisplatin was Consistent with these findings, the inhibitor SU-5402 significantly obtained using SU-5402, a pharmacological inhibitor against FGFR. enhanced cisplatin-mediated apoptosis in WAC2 cells (Figure 3c). NB cell lines showed different sensitivities to SU-5402 inhibition, In order to investigate possible implications of FGFR2 in the LAN-1, WAC2 and IMR5 being the most responsive cell lines regulation of apoptosis, we analyzed the expression levels of the (Figure 2c and Supplementary Figure 2b). To quantify drug proapoptotic, as well as the anti-apoptotic members of the BCL2 synergism, the combinational index (CI) was calculated by the family of proteins (Figure 3b). Within the proapoptotic BH3-only Chou–Talalay method (CI ¼ 0.818) indicating synergism between proteins, only Bad was upregulated in response to cisplatin cisplatin and SU-5402 in LAN-1 cells.14 We performed an in vivo treatment, and this was only observed in cells, which had FGFR2 chicken embryo chorioallantoic membrane assay to investigate downregulation. Bim, Bid and Puma levels either decreased or the potential of pharmacological FGFR inhibition on tumor stayed unchanged. Mcl-1 has been previously shown to modulate growth. Application of WAC2 cells on the chorioallantoic cisplatin-induced apoptosis.16 In NB cells, Mcl-1 expression was membrane led to the formation of solid tumors within 3 days. not affected by cisplatin treatment. In contrast, the expression of Treatment of these tumors for 4 consecutive days with SU-5402 the anti-apoptotic proteins BCL2 and BCL-XL increased in response significantly impaired tumor growth when compared with control to cisplatin treatment. In addition, BCL2 expression correlated with treatment (Figure 2d, Po0.05). Together, these results provide the expression of FGFR2, suggesting that BCL2 expression is strong evidence for the role of FGFR2-signaling in NB tumor regulated by FGFR2 signaling. To confirm whether FGFR2 growth and chemoresistance. regulates BCL2 expression, the receptor was stimulated by its main ligand fibroblast growth factor-2 (FGF-2) and probed for BCL2 expression. BCL2 levels dramatically increased after 24 h RNAi targeting of FGFR2 induces apoptosis in combination of growth factor stimulation in a dose-dependent manner with cisplatin (Figure 3d). This effect was evident for protein expression as well Cisplatin induces apoptosis associated with DNA damage.15 We as for mRNA expression, pointing out that the regulation of BCL2 initially investigated whether FGFR2 gene silencing sensitized cells by FGFR2 occurs at least in part at the transcriptional level.

Oncogene (2013) 3944 – 3953 & 2013 Macmillan Publishers Limited Role of FGFR2 signaling in neuroblastoma F Salm et al 3947

Figure 4. Roles of PKC family isoforms in NB cell survival and chemoresistance. (a) Cisplatin sensitivity assay repeated for PKC family members PKC-b1, PKC-d and PKC-z with isoform-speciﬁc siRNA sequences in SH-SY5Y and LAN-1 cells. The effects on cell viability of the siRNA-mediated target downregulation in presence (grey bars) or absence of cisplatin (black bars) was normalized to the non-targeting siControl and expressed as percentages. (b) Cell viability assay of LAN-1 cells treated for 72 h with isoform-selective PKC inhibitors; rottlerin (circles), PKCb1 inhibitor (squares), PKC-z inhibitor (triangles), pan-PKC inhibitor (rhombus). (c) Caspase-3 activation was assessed by ﬂuorescence-activated cell sorting (FACS) in SH-SY5Y cells treated as described in (a) and (b). (d) Protein expression analysis by western blot of LAN-1 cells transfected with siPRKCB1,siPRKCD and siPRKCZ. (e) Cell lysates of LAN-1 cells incubated with rottlerin (PKC-d inhibitor) for 24 h were analyzed for protein expression and phosphorylation of the indicated proteins. Error bars represent the s.d. of the mean of three independent experiments, *Po0.05, **Po0.01, ***Po0.001, two-way analysis of variance (ANOVA) using Bonferroni’s multiple comparison test.

PKC regulates cell survival in NB downstream of FGFR2 targeting of the PKC family member PKC-d, but its isoforms PKC- We next aimed at analyzing the mechanism of the regulation of b1 and PKC-z showed a role in NB survival (Supplementary BCL2 expression by FGFR2. Given that FGFR2 is known to activate Table 2), and were as well validated. The RNAi-mediated down- PKC family members through recruitment and activation of phos- regulation of all three isoforms signiﬁcantly decreased cell viability pholipase C-g,17 we analyzed the possible implications of PKC in in both cell lines, suggesting that PKC signaling may have an FGFR2 signaling in NB. In the initial RNAi screen, we assessed important role in NB survival (Figure 4a). In LAN-1 cells, the changes in cell survival induced by the inhibition of protein and downregulation of PRKCD alone strongly inhibited cell viability, lipid kinases in presence or absence of cisplatin. Not only RNAi thus, its combination with cisplatin had rather modest effects

& 2013 Macmillan Publishers Limited Oncogene (2013) 3944 – 3953 Role of FGFR2 signaling in neuroblastoma F Salm et al 3948

Figure 5. FGFR2 activates PKC-d to modulate cisplatin resistance. (a) LAN-1 cells were incubated with FGF-2 for 15 min and analyzed for phosphorylation of PKC-d by western blot. (b) LAN-1 cells incubated for 24 h with the inhibitors (enzastaurin 10 mM, rottlerin 5 mM, PD-98059 25 mM, LY-29402 10 mM, rapamycin 100 mg/ml, SU-5402 10 mM and dimethylsulphoxide (DMSO) as negative control) and 0.1 mg/ml of FGF-2. BCL2 mRNA levels quantiﬁed by Taqman PCR and protein levels analyzed by western blot are shown. (c) Protein extracts of Myr-PKC-d expressing LAN-1 cells were analyzed by western blot for activated PKC-d (pPKC-d) and p-PKC substrates and the indicated BCL2 family members. (d) Cell viability assay of stably expressing Myr-PKC-d cells incubated for 72 h in presence of FGFR2 inhibitor (SU-5402 5 mM), PKC-d inhibitor (rottlerin 5 mM) or DMSO. (e) Cisplatin titration of stably expressing Myr-PKC-d (EC50 ¼ 12.11) and vector-transfected cells (EC50 ¼ 8.83). Error bars represent the s.d. of the mean of at least two independent experiments. *Po0.05, two-way analysis of variance (ANOVA) using Bonferroni’s multiple comparison test.

(Figure 4a). Among the PKC isoforms, the inhibition of PKC-d had may reveal which of them is important to regulate BCL2, if FGF-2 the strongest impact on cell viability and apoptosis in both cell fails to induce BCL2 expression on pathway inactivation. LAN-1 lines, which was further confirmed with isoform-selective phar- cells were treated with FGF-2 and a panel of inhibitors including macological inhibitors (Figures 4b, c and Supplementary rottlerin and enzastaurin, which target PKC-d and PKC-b1, PD- Figure 3b). 98059 a MEK inhibitor, LY-294002 a pan-PI3K inhibitor, rapamycin Western blot analysis revealed that the downregulation of an mammalian target of rapamycin inhibitor and SU-5402, and PKC-d attenuated important cell survival pathways such as the analyzed their effects on BCL2 expression. The comparative extracellular signal-regulated kinases 1 and 2 (ERK1/2)- and western blot analysis reveals that only rottlerin and enzastaurin mammalian target of rapamycin/S6-signaling pathway. Rottlerin strongly inhibited FGF-2-mediated upregulation of BCL2 decreased the levels of BCL2, BCLXL and MCL1 and the phospho- (Figure 5b). In contrast, the inhibition of the PI3K pathway or rylation of ERK1/2 and ribosomal protein S6 kinase, which was MEK/ERK pathway had no effect on BCL2 expression. SU-5402 confirmed by the effect of PRKCD siRNA (Figures 4d, e). Moreover, reduced the levels of BCL2, but only modestly compared with the the downregulation of PRKCD, but not PRKCB1 or PRKCZ, induced observed effects by rottlerin or enzastaurin. apoptosis and downregulation of BCL2 (Figures 4c, d), suggesting its involvement in the control of apoptosis regulated by BCL2 expression and raising the hypothesis that PKC-d may act Activation of PKC-d increases drug resistance in NB cells downstream of FGFR2 to regulate BCL2 expression. Our results had shown that FGFR2 signaling uses PKC-d to activate a prosurvival mechanism involving BCL2 and BCLXL regulation. To further support this model, we stably transfected LAN-1 cells with FGF-2-induced BCL2 expression is mediated by PKC signaling but a myristoylated catalytically active PRKCD mutant. The expression not mitogen-activated extracellular signal-regulated kinase (MEK)/ of the constitutively active PKC-d mutant was confirmed as the ERK or PI3K/AKT/mammalian target of rapamycin/S6K signaling cells showed higher levels of PKC substrate phosphorylation In view of our previous results, we analyzed PKC-d activation in compared with cells transfected with the empty expression vector response to FGF-2 stimulation. PKC-d was activated by phospho- (Figure 5c). Western blot analysis of the anti-apoptotic BCL2 family rylation in a dose-dependent manner in response to FGF-2 members, showed that activation of PKC-d correlated with (Figure 5a). We next investigated the possibility that FGF-2- increased Bcl-XL and BCL2 levels, although the increases in BCL2 mediated activation of PKC-d could account for BCL2 regulation. expression were rather modest (Figure 5c). We investigated Inhibition of different signaling pathways downstream of FGFR2 whether activation of PKC-d could reverse the sensitivity of cells

Oncogene (2013) 3944 – 3953 & 2013 Macmillan Publishers Limited Role of FGFR2 signaling in neuroblastoma F Salm et al 3949 to FGFR2 inhibition. Therefore, we treated the stably transfected cells with the inhibitor SU-5402 and assessed cell viability. The PRKCD-Myr expressing cells were less sensitive to FGFR2 inhibition than cells transfected with the empty vector (Figure 5d). Inhibition of PKC-d by rottlerin lead to a signiﬁcant decrease of cell viability in both transfected cell lines, but the difference in the sensitivity to PKC-d inhibition was less marked than in the case of SU-5402. Finally, cells transfected with activated PKC-d displayed enhanced survival under normal culture conditions and were more resistant against cisplatin treatment (EC50 shift from 8.83 to 12.11, Figure 5). Together, these results show the ability of activated PKC-d to uncouple from FGFR2 signaling and to provide NB cells a more chemoresistant phenotype.

Importance of FGFR2 signaling in NB chemoresistance To further confirm the potential role of FGFR2 in NB chemoresistance, we used a cell line model derived from LAN-1 cells that were selected for resistance against chemotherapeutic agents including cisplatin.18 We analyzed the expression of FGFR2 and FGF-2 in LAN-1R cells (Figure 6a). FGFR2 was significantly overexpressed compared with its expression in LAN-1 cells. Moreover, LAN-1R cells expressed FGF-2 in protein and mRNA level, whereas in the other NB cell lines FGF-2 was only detected at mRNA level. In LAN-1R, the elevated levels of FGF-2 and FGFR2 Figure 6. Protein expression of LAN-1R cells. (a) Protein extracts of were correlated BCL2 expression, which is in agreement with LAN-1R cells were analyzed by western blot for FGFR2, FGF-2 and the previous observations, that BCL2 was overexpressed in BCL2. (b) Semi-quantitative PCR of FGF-2 in NB cell lines. comparison with the parental cell line.18 These results show the importance of FGFR2 signaling in NB acquired chemoresistance. Further evidence of the clinical relevance of FGFR2 in NB was Targeting BCL2 and BCLXL inhibits cell survival and enhances the obtained from a DNA microarray study comprising 251 NB patient effects of cisplatin in NB cells samples. This analysis revealed that FGFR2 expression is upregu- Given that NB cells induced BCL2 as a pro-survival mechanism to lated in stage 4 NB (Po0.001) and MYCN-amplified tumors overcome cell death, targeting BCL2 with pharmacological (Po0.001) (Figure 8g). inhibitors provides a rational to efficiently induce cell death in To further explore the link between FGFR2 and MYCN, we used NB. Here, we tested the effects of ABT-737 in the NB cell lines two different inducible NB cell line models where MYCN can be 19 SH-SY5Y, LAN-1. In both cell lines, increasing concentrations of either up or downregulated. This analysis revealed that FGFR2 ABT-737 impaired cell viability and had a strong synergistic effect expression was increased on induction of MYCN in SH-SY5Y with cisplatin (CI ¼ 0.512 for SH-SY5Y and CI ¼ 0.411 for LAN-1, (Figure 9a). In addition, shRNA-mediated downregulation of MYCN Figures 7a, b). In addition, we investigated if NB cell lines that in IMR5 strongly decreased FGFR2 expression (Figure 9b). On the display a chemoresistant phenotype were also sensitive to BCL2 contrary, tetracycline had no effect in the parental cell lines inhibition. We therefore investigated the sensitivity to ABT-737 in SH-SY5Y and IMR5 (Supplementary Figure 4), indicating that LAN-1R cells and the above described LAN-1 cells, which stably FGFR2 levels are regulated by induction of MYCN. express activated PKC-d. Also in both cell lines, ABT-737 inhibited A recent study using a murine NB model induced by MYCN has 20 cell viability and had synergy with cisplatin, although compared implicated miR-125b, a known regulator of FGFR2. We therefore, with the parental cell line LAN-1, the combinational effect with investigated a potential relationship between miR-125b and cisplatin in LAN-1R was less significant (LAN-1-PRKCD-myr FGFR2 in NB cell lines. Indeed, an inverse correlation between CI ¼ 0.473, LAN-1R CI ¼ 0.633, Figures 7c, d). mi-R125b and FGFR2 expression could be documented (Pearson test, r ¼À0.8, P ¼ 0.015). Expression of FGF-2 and FGFR2 in primary NB samples In order to investigate the clinical relevance of FGFR2 and FGF-2 in NB tumor samples, a total of 91 samples of NBs available as a DISCUSSION tissue microarray were probed for protein expression by The poor outcome of high-risk NB patients with current multi- immunohistochemical staining (Figures 8a–d). FGFR2 was modal therapies justifies the development of novel targeted detected in all NB primary tumors, whereas 80% was scored as therapies for this common childhood cancer. The development positive (53 of 66), 3% as strong positive (2 of 66) and 17% as of chemoresistance has been demonstrated to be a major cause weak positive (11 of 66). Among the metastatic tumors, 81% were for treatment failure in NB, as well as in many other human positive (21 of 26) and 19% were weak positive (5 of 26). cancers.1,7 Cisplatin is commonly used in current multimodal Expression of FGF-2 was quantified positive for 30% of primary therapeutic regimens in high-risk NB and multiple resistance tumors (20 of 67), strong positive for 3% (2 of 67), weak positive mechanisms have been described against this drug in human for 61% (41 of 67) and negative for only 6% (4 of 67). Interestingly, cancer.2–4 It has been demonstrated that the deregulation of the expression of FGFR2 correlated with FGF-2 expression intracellular signaling pathways controlling apoptosis has a major (P ¼ 0.045, Figure 8, left table). The tissue microarray was then role in cisplatin resistance in human cancer.2 By using high- probed for BCL2 (Figures 8e, f) and analyzed for possible throughput RNAi screens in established NB cell lines, we have correlations with FGFR2 and FGF-2. Whereas FGFR2 showed no identified novel potential drug targets involved in NB survival and correlation with BCL2, FGF-2 levels were shown to correlate with resistance to cisplatin. A previous report using RNAi screening BCL2 expression (P ¼ 0.035, Figure 8, right table). identified CHK1 as a novel target.21 In our screen, the

& 2013 Macmillan Publishers Limited Oncogene (2013) 3944 – 3953 Role of FGFR2 signaling in neuroblastoma F Salm et al 3950

Figure 7. Targeting ABT-737 in NB cell lines. (a–d) Cell viability assay of SH-SY5Y (a), LAN-1 (b), PRKCD-myr LAN-1 transfectant (c) and LAN-1R (d) treated for 72 h with ABT-737 (circles), cisplatin (rhombus) and the combination of ABT-737 and cisplatin (triangles) at a constant ratio (1:2.5). CIs were calculated (SHSY5Y CI ¼ 0.512; LAN-1 CI ¼ 0.411; (PRKCD-myr LAN-1 transfectant CI ¼ 0.473; LAN-1R CI ¼ 0.633). Error bars represent the s.d. of the mean of three independent experiments.

downregulation of CHK1 did not show significant effects on cell that FGFR2 is activated by an autocrine loop involving FGF2 in NB survival. The different results might be explained due to the cells. This model is based on the following observations: (i) FGF-2 biological variability within the cell lines used. However, in is consistently expressed in primary NB and NB cell lines; (ii) FGF-2 agreement with this screen, we also identified in our cell lines activates FGFR2 signaling in NB cell lines at nanomolar BMPR1A and RPS6KB1 as hit candidate kinases. Moreover, our concentrations; (iii) pharmacological inhibition of FGFR2 tyrosine screen identified AURKB as an important regulator of cell survival kinase activity blocks the effect of FGF-2 on BCL2 expression; (iv) in NB cells harboring MYCN amplification, consistent with a the expression of FGF-2 and FGFR2 is upregulated in NB cell lines previous study in which the related isoform AURKA was identified with acquired chemoresistance. The observation that the from a shRNA screen as a gene that is required for the growth of expression of FGF-2 and FGFR2 is upregulated in NB cell lines MYCN-amplified NB cells.22 with acquired chemoresistance points to a more general role of The comparative analysis of the kinases identified as chemo- this signaling pathway in the resistance of NB cells to other sensitizer kinases in the NB cell lines used in our study revealed chemotherapeutic agents, such as doxorubicin. The fact that that FGFR2 was the most promising candidate. In support of our FGFR2 is overexpressed in higher stage NB underscores the findings, the FGFR2 has been recently implicated in modulating relevance of FGFR2 as a target in these tumors. However, the value resistance to cisplatin kinase in ovarian cancer.23 LAN-1 cells have of FGFR2 as a prognostic factor remains to be elucidated. MYCN amplification, whereas SH-SY5Y is MYCN single copy. This Mechanistically, FGFR2 signaling altered the expression profile may in part explain that there were significant differences in the of anti-apoptotic BCL2 and BCLXL in NB cells, resulting in a identity of the chemosensitizer kinases identified in the two NB decrease in the sensitivity of the cells to cisplatin-mediated cell lines. Intriguingly, MYCN expression also correlated with FGFR2 apoptosis. PKC isoforms were shown to have a major role in FGFR2 expression in primary NB, which may account for the stronger signaling to BCL2 and BCLXL, in contrast to other signaling impact of FGFR2 silencing on cisplatin sensitivity in LAN-1, as pathways, such as PI3K/mammalian target of rapamycin and Erk. compared with SH-SY5Y. Accordingly, MYCN was shown to BCL2 family proteins have been extensively studied for their role regulate FGFR2 expression in NB cell lines. In addition to MYCN, in modulating cancer chemoresistance. Previous studies have also the miR-125b may also control FGFR2 expression in NB. documented the potential of targeting BCL2 family proteins with Alterations in FGFR2 have been widely identified in human ABT-737 in NB.30,31 Previous reports have documented that FGF-2 cancers. Gene amplification or mutations of FGFR2 were reported signaling protects small cell lung cancer cells from chemotherapy- in different cancers.24–27 However, a recent study failed to detect induced apoptosis by modulating the expression levels of BCL2 any FGFR2 mutation in NB.28 Genetic alterations may induce family proteins.32,33 Intriguingly, PKC isoforms were reported to be aberrant FGFR2 signaling activation because of the release of involved in transducing FGF-2/FGFR-1 signals to BCL2 family FGFR2 from auto inhibition, or through creation of FGF signaling proteins in small cell lung cancer cells, but the mechanism autocrine loops.17 FGF-2 was previously identified in advanced- involved post-transcriptional events.32 In contrast, FGF-2-mediated stage tumors (stages 3 and 4) compared with low-stage tumors upregulation of BCL2 occurred at the transcriptional level in NB (stages 1, 2 and 4S).29 Consistent with this study, our data indicate cells, which may be accounted for by differences in FGFR isoform

Oncogene (2013) 3944 – 3953 & 2013 Macmillan Publishers Limited Role of FGFR2 signaling in neuroblastoma F Salm et al 3951

Figure 8. Analysis of the expression of FGFR2 and FGF-2 in NB patients. (a–e) Representative NB samples of tissue microarray (TMA) analysis showing the different immunoreactivities for FGFR2 weakly positive samples (a), positive samples (b), for FGF-2-negative samples (c) and positive samples (d) and for BCL2-negative samples (e) and positive samples (f). Tables indicate the association of FGF2 (neg þ weak pos vs pos) and FGFR2 (neg þ weak pos vs pos) on the left and the association of FGF2 (neg þ weak pos vs pos) and BCL2 (neg vs pos) on the right (n ¼ 64, P-values from the w2-test are indicated). (g) DNA microarray analysis of 251 NB patient samples showing the expression of FGFR2 in correlation to patient age (41.5 vs o1.5 years, left panel), to tumor stage (stages 1, 2, 3, 4S vs stage 4, middle panel) or to MYCN amplification status (right panel). P-values from the Wilcoxon test are indicated. involvement. Several transcription factors may be involved in be relevant for the future development of novel prognostic transducing the signal from FGFR2/PKC to BCL2, including nuclear biomarkers in NB patients. factor-kB.34 PKC is known to regulate the inhibitor of kappa B Together, our findings indicate that FGFR2 is a determinant for kinase and induce nuclear factor-kB activation B and T cells, a cisplatin response in NB and suggest a therapeutic option to process that is important for the adaptive immune response.35,36 improve the efficacy of chemotherapeutic agents in the treatment Although the main role is attributed to the isoform PKC-b1, it has of high-risk NB patients. been as well reported that nuclear factor-kB is activated by phospholipase C-g and PKC-d.37 Importantly, our data have a direct relevance to translational cancer research in NB, as there are MATERIALS AND METHODS several pharmacological inhibitors of FGFR in clinical trials, Cell culture and reagents including ZD-4547, BGJ398 and FP-1039, which are entering All NB cell lines were cultured as previously described in refs18,19,41. The phase I or II for the treatment breast cancer or gastric cancers with compounds SU-5402, rottlerin, PKC-b1 inhibitor, PKC-z inhibitor, LY-29402, FGFR2 amplification, endometrial cancers with FGFR2 mutations or PD-98059, FGF-2 were purchased from Calbiochem (Darmstadt, Germany), advanced solid cancers with FGFR1, 2, 3 amplifications.17 In enzastaurin was purchased from Selleck Chemicals (Houston, TX, USA) and addition, enzastaurin is also in clinical trials for other human ABT-737 from Chemietek (Indianapolis, IN, USA). cancers,36,38 such as glioblastoma, multiple myeloma, non- Hodgkin’s lymphoma, breast cancer and non-small cell lung High throughput screen method cancer. Among them, enzastaurin has already proven antitumor NB cell lines (104 cells per well for LAN-1 and 7.5 Â 103 cells for SH-SY5Y) 39,40 effects in myeloma and non-small cell lung cancer patients. were plated in 96-well plates and transfected 24 h later with siRNA The ability to detect FGF-2 in the urine of cancer patients may also (Silencer Kinase siRNA Library Ambion, Applied Biosystems, Foster City, CA,

& 2013 Macmillan Publishers Limited Oncogene (2013) 3944 – 3953 Role of FGFR2 signaling in neuroblastoma F Salm et al 3952 Cell viability and apoptosis Cell viability was assessed by MTS assay using the CellTiter 961 Aqueous One Solution Cell proliferation Assay (Promega, Madison, WI, USA) according to the manufacturer’s instructions. For the quantiﬁcation of apoptosis, cells were stained with PE-Active caspase-3 Apoptosis Kit (BD Pharmingen, San Diego, CA, USA) and analyzed by ﬂow cytometry according to the manufacturer’s instructions.

Chick chorioallantoic membrane assay Fertilized chicken eggs (gallus gallus) purchased at a local hatchery were incubated in a humidiﬁed incubator at 37 1C. On embryonic day 3, a 2 cm window was opened in the shell and covered with tape to continue incubation. On embryonic day 7, a silicon ring with 1 cm inner diameter was placed on the chorioallantoic membrane, and 4 million WAC2 cells suspended in 20 ml phosphate-buffered saline were applied after gentle laceration of the membrane. Subsequently tumors were treated with either FGFR2-speciﬁc inhibitor SU5402 (20 mM) or dimethylsulphoxide (solvent control) for 4 consecutive days.

Plasmid transfection Stable transfections were performed with Lipofectamine Plus (Invitrogen) following the manufacturer’s instructions. Cells were transfected with PRKCD constructs cloned into pWZL Neo Myr Flag retroviral plasmid Figure 9. Regulation of FGFR2 expression by MYCN and miR-125b. (plasmid 220603, Addgene, Cambridge, MA, USA) and empty vector as SH-SY5Y cells stably expressing a tetracycline-regulated MYCN control. At 48 h after transfection, cells were diluted 1:10 in culture medium transgene (a) and IMR5 cells stably expressing a tetracycline- containing G418 (0.8 mg/ml, Calbiochem). A mixed population of resistant regulated shRNA MYCN transgene (b), were treated with 100 nM cells was expanded and analyzed after selection. tetracycline for the indicated time points and analyzed for FGFR2 protein expression. (c) mRNA expression of FGFR2 was correlated to miR-125b expression in NB cell lines (from left to right LAN-1, WAC- Immunohistochemistry 2, SH-SY5Y, SHEP007, IMR5, LAN-5, SKNAS and LAN-1R). The Pearson Paraffin-embedded tissue arrays mounted with 91 biopsies of NB patients correlation coefficient is indicated. were obtained from the University Medical Center Hamburg-Eppendorf, Germany. For immunohistochemistry, the sections were prepared and stained as described in ref. Nikolova et al.45, except the epitope retrieval USA) at a final concentration of 30 nMm, using Lipofectamine 2000 was done by Proteinase K digestion for 5 min at 37 1C. The antibodies following the manufacturer’s instructions (Invitrogen, Carlsbad, CA, USA). anti-FGF-2 rabbit polyclonal antibody (ab16828) and anti-FGFR2 rabbit Additional controls were added to each plate of the library. Negative polyclonal antibody (ab10648) were purchased from Abcam (Cambridge, controls: siRNA for GAPDH (Ambion), siCONTROL non-targeting siRNA Pool MA, USA). (Dharmacon, Waltham, MA, USA), ONTarget Plus non targeting siRNA (Dharmacon) and positive controls: siRNA for PLK1, EG5 and siCONTROL DNA microarray analysis TOX (Dharmacon). At 24 h following transfection, cells were treated with cisplatin (2.5 mM for LAN-1 and 0.75 mM for SH-SY5Y, Bristol-Myers Squibb, Generation of tumor gene expression profiles was previously described in ref.Oberthuer et al.46 and the clinical characteristics of the NB patients New York, NY, USA) or vehicle. Cell viability was assessed after 48 h. The 47 screen was performed in triplicate. Median centered Z-scores of the (n ¼ 251) in ref. Sagulenko et al. Raw and normalized microarray data are surviving fractions were calculated as described in Zhang et al.42 available at ArrayExpress (accession: E-TABM-38).

Validation of high throughput screen Statistical analysis Two distinct siRNA sequences targeting each candidate hit kinase were All statistical tests were performed with the GraphPad Prism software used to revalidate results from the screen (against FGFR2: siRNA IDs 118292 (La Jolla, CA, USA), except the CI values that were calculated with the and 1215; against PRKCD siRNA IDs 773 and 775; against PRKCB1 siRNA IDs Calcusyn software (Cambridge, UK). The statistical signiﬁcance of 103309 and 261007 against PRKCZ siRNAs ID 103575 and 103679, Ambion). differences between groups was assessed with one-way or two-way Validation of RNAi gene silencing was evaluated 48 h after transfection by analysis of variance, using either Tukey’s or Bonferroni’s multiple western blotting for protein expression and by quantitative reverse comparison test; Po0.05 was indicated with one asterisk, Po0.01 with a transcriptase–PCR (Taqman) for mRNA expression. double asterisk, Po0.001 with a triple asterisk.

Antibodies and western blotting Protein expression was analyzed by immunoblotting as described in Arcaro CONFLICT OF INTEREST et al.43 The following antibodies were used: anti-Bek-1, anti-PKC-d, anti- The authors declare no conﬂict of interest. PKC-b1, anti-PKC-z, anti-caspase-3 (Santa Cruz Biotechnology, Santa Cruz, CA, USA), anti-phosphoThr-PKC-d, anti-Puma, anti-Bcl-XL, anti-Bad (Cell Signaling Technology, Danvers, MA, USA), anti-Mcl-1, anti-BCL2, anti-Bim, ACKNOWLEDGEMENTS anti-Bid, anti-Bak, anti-Bax (Epitomics, Burlingame, CA, USA) and anti-b- This work was supported by grants from the Gottfried und Julia Bangerter-Rhyner- actin (Sigma, Buchs, Switzerland). Stiftung and Jubila¨umsstiftung Swiss Life. The research leading to these results has received funding from the European Union’s Seventh Framework Programme Quantitative reverse transcriptase–PCR (FP7/2007–2013) under grant agreement no. 259348. Real-time quantitative PCR was performed using Assays-on-Demand Gene Expression products (Applied Biosystems) FGFR2 (Hs01552926_m1), PRKCD (Hs00178914), and PRKCZ (Hs00177051_m1), PRKCB1 (Hs01034075), BCL2 REFERENCES (Hs00153350_m1), GAPDH (Hs99999905_m1), hsa-miR-125b (TM000449) 1 Wilson TR, Longley DB, Johnston PG. Chemoresistance in solid tumours. Ann and U6-snRNA (TM001973) according to manufacturer’s instructions. Oncol 2006; 17(Suppl 10): x315–x324. Relative mRNA expression levels were calculated using the comparative 2 Siddik ZH. Cisplatin: mode of cytotoxic action and molecular basis of resistance. threshold cycle (CT) method.44 Oncogene 2003; 22: 7265–7279.

Oncogene (2013) 3944 – 3953 & 2013 Macmillan Publishers Limited Role of FGFR2 signaling in neuroblastoma F Salm et al 3953 3 Kelland L. The resurgence of platinum-based cancer chemotherapy. Nat Rev 25 Fanale D, Amodeo V, Corsini LR, Rizzo S, Bazan V, Russo A. Breast cancer genome- Cancer 2007; 7: 573–584. wide association studies: there is strength in numbers. Oncogene 2011; 31: 4 Galanski M. Recent developments in the field of anticancer platinum complexes. 2121–2128. Recent Pat Anticancer Drug Discov 2006; 1: 285–295. 26 Byron SA, Pollock PM. FGFR2 as a molecular target in endometrial cancer. Future 5 Erovic BM, Pelzmann M, Grasl M, Pammer J, Kornek G, Brannath W et al. Mcl-1, Oncol 2009; 5: 27–32. vascular endothelial growth factor-R2, and 14-3-3sigma expression might predict 27 Katoh M. Cancer genomics and genetics of FGFR2 (review). Int J Oncol 2008; 33: primary response against radiotherapy and chemotherapy in patients with locally 233–237. advanced squamous cell carcinomas of the head and neck. Clin Cancer Res 2005; 28 Shukla N, Ameur N, Yilmaz I, Nafa K, Lau CY, Marchetti A et al. Oncogene mutation 11(24 Pt 1): 8632–8636. profiling of pediatric solid tumors reveals significant subsets of embryonal 6 Michaud WA, Nichols AC, Mroz EA, Faquin WC, Clark JR, Begum S et al. rhabdomyosarcoma and neuroblastoma with mutated genes in growth signaling Bcl-2 blocks cisplatin-induced apoptosis and predicts poor outcome following pathways. Clin Cancer Res 2012; 18: 748–757. chemoradiation treatment in advanced oropharyngeal squamous cell carcinoma. 29 Eggert A, Ikegaki N, Kwiatkowski J, Zhao H, Brodeur GM, Himelstein BP. High-level Clin Cancer Res 2009; 15: 1645–1654. expression of angiogenic factors is associated with advanced tumor stage in 7 Maris JM. Recent advances in neuroblastoma. N Engl J Med 2010; 362: 2202–2211. human neuroblastomas. Clin Cancer Res 2000; 6: 1900–1908. 8 Brodeur GM, Seeger RC, Schwab M, Varmus HE, Bishop JM. Amplification of N-myc 30 Fang H, Harned T, Kalous O, Maldonado V, Declerck YA, Reynolds CP. Synergistic in untreated human neuroblastomas correlates with advanced disease stage. activity of fenretinide and the Bcl-2 family protein inhibitor ABT-737 against Science 1984; 224: 1121–1124. human neuroblastoma. Clin Cancer Res 2011; 17: 7093–7104. 9 Goldsmith KC, Hogarty MD. Targeting programmed cell death pathways with 31 Goldsmith KC, Lestini BJ, Gross M, Ip L, Bhumbla A, Zhang X et al. BH3 response experimental therapeutics: opportunities in high-risk neuroblastoma. Cancer Lett profiles from neuroblastoma mitochondria predict activity of small molecule Bcl-2 2005; 228: 133–141. family antagonists. Cell Death Differ 2010; 17: 872–882. 10 MacKeigan JP, Murphy LO, Blenis J. Sensitized RNAi screen of human kinases and 32 Pardo OE, Wellbrock C, Khanzada UK, Aubert M, Arozarena I, Davidson S et al. phosphatases identifies new regulators of apoptosis and chemoresistance. Nat FGF-2 protects small cell lung cancer cells from apoptosis through a complex Cell Biol 2005; 7: 591–600. involving PKCepsilon, B-Raf and S6K2. EMBO J 2006; 25: 3078–3088. 11 Iorns E, Turner NC, Elliott R, Syed N, Garrone O, Gasco M et al. Identification of 33 Pardo OE, Arcaro A, Salerno G, Raguz S, Downward J, Seckl MJ. Fibroblast growth CDK10 as an important determinant of resistance to endocrine therapy for breast factor-2 induces translational regulation of Bcl-XL and Bcl-2 via a MEK-dependent cancer. Cancer Cell 2008; 13: 91–104. pathway: correlation with resistance to etoposide-induced apoptosis. J Biol Chem 12 Holzel M, Huang S, Koster J, Ora I, Lakeman A, Caron H et al. NF1 is a tumor 2002; 277: 12040–12046. suppressor in neuroblastoma that determines retinoic acid response and disease 34 Mattson MP, Meffert MK. Roles for NF-kappaB in nerve cell survival, plasticity, and outcome. Cell 2010; 142: 218–229. disease. Cell Death Differ 2006; 13: 852–860. 13 Diep CH, Munoz RM, Choudhary A, Von Hoff DD, Han H. Synergistic effect 35 Lucas PC, McAllister-Lucas LM, Nunez G. NF-kappaB signaling in lymphocytes: a between erlotinib and MEK inhibitors in KRAS wild-type human pancreatic cancer new cast of characters. J Cell Sci 2004; 117(Pt 1): 31–39. cells. Clin Cancer Res 2011; 17: 2744–2756. 36 Guo B, Su TT, Rawlings DJ. Protein kinase C family functions in B-cell activation. 14 Chou TC. Drug combination studies and their synergy quantification using the Curr Opin Immunol 2004; 16: 367–373. Chou-Talalay method. Cancer Res 2010; 70: 440–446. 37 Lee CH, Shieh DC, Tzeng CY, Chen CP, Wang SP, Chiu YC et al. Bradykinin-induced 15 Sorenson CM, Barry MA, Eastman A. Analysis of events associated with cell cycle IL-6 expression through bradykinin B2 receptor, phospholipase C, protein kinase arrest at G2 phase and cell death induced by cisplatin. J Natl Cancer Inst 1990; 82: Cdelta and NF-kappaB pathway in human synovial fibroblasts. Mol Immunol 2008; 749–755. 45: 3693–3702. 16 Sheridan C, Brumatti G, Elgendy M, Brunet M, Martin SJ. An ERK dependent 38 Ma S, Rosen ST. Enzastaurin. Curr Opin Oncol 2007; 19: 590–595. pathway to Noxa expression regulates apoptosis by platinum-based che- 39 Herbst RS, Oh Y, Wagle A, Lahn M. Enzastaurin, a protein kinase Cbeta-selective motherapeutic drugs. Oncogene 2010; 29: 6428–6441. inhibitor, and its potential application as an anticancer agent in lung cancer. Clin 17 Turner N, Grose R. Fibroblast growth factor signalling: from development to Cancer Res 2007; 13(15 Pt 2): s4641–s4646. cancer. Nat Rev Cancer 2010; 10: 116–129. 40 Ghobrial IM, Munshi NC, Harris BN, Shi P, Porter NM, Schlossman RL et al. A phase I 18 Flahaut M, Muhlethaler-Mottet A, Martinet D, Fattet S, Bourloud KB, Auderset Ket al. safety study of enzastaurin plus bortezomib in the treatment of relapsed or Molecular cytogenetic characterization of doxorubicin-resistant neuroblastoma cell refractory multiple myeloma. Am J Hematol 2011; 86: 573–578. lines: evidence that acquired multidrug resistance results from a unique large 41 Boller D, Schramm A, Doepfner KT, Shalaby T, von Bueren AO, Eggert A et al. amplification of the 7q21 region. Genes Chromosomes Cancer 2006; 45: 495–508. Targeting the phosphoinositide 3-kinase isoform p110delta impairs growth and 19 Muth D, Ghazaryan S, Eckerle I, Beckett E, Pohler C, Batzler J et al. Transcriptional survival in neuroblastoma cells. Clin Cancer Res 2008; 14: 1172–1181. repression of SKP2 is impaired in MYCN-amplified neuroblastoma. Cancer Res 70: 42 Zhang JH, Chung TD, Oldenburg KRA. Simple statistical parameter for use in 3791–3802. evaluation and validation of high throughput screening assays. J Biomol Screen 20 Terrile M, Bryan K, Vaughan L, Hallsworth A, Webber H, Chesler L et al. miRNA 1999; 4: 67–73. expression profiling of the murine TH-MYCN neuroblastoma model reveals 43 Arcaro A, Khanzada UK, Vanhaesebroeck B, Tetley TD, Waterfield MD, Seckl MJ. similarities with human tumors and identifies novel candidate miRNAs. PLoS One Two distinct phosphoinositide 3-kinases mediate polypeptide growth factor- 2011; 6: e28356. stimulated PKB activation. EMBO J 2002; 21: 5097–5108. 21 Cole KA, Huggins J, Laquaglia M, Hulderman CE, Russell MR, Bosse K et al. RNAi 44 Giulietti A, Overbergh L, Valckx D, Decallonne B, Bouillon R, Mathieu C. An screen of the protein kinome identifies checkpoint kinase 1 (CHK1) as a overview of real-time quantitative PCR: applications to quantify cytokine gene therapeutic target in neuroblastoma. Proc Natl Acad Sci USA 2011; 108: expression. Methods 2001; 25: 386–401. 3336–3341. 45 Nikolova Z, Djonov V, Zuercher G, Andres AC, Ziemiecki A. Cell-type specific and 22 Otto T, Horn S, Brockmann M, Eilers U, Schuttrumpf L, Popov N et al. Stabilization estrogen dependent expression of the receptor tyrosine kinase EphB4 and its of N-Myc is a critical function of Aurora A in human neuroblastoma. Cancer Cell ligand ephrin-B2 during mammary gland morphogenesis. J Cell Sci 1998; 2009; 15: 67–78. 111(Pt 18): 2741–2751. 23 Cole C, Lau S, Backen A, Clamp A, Rushton G, Dive C et al. Inhibition of FGFR2 and 46 Oberthuer A, Berthold F, Warnat P, Hero B, Kahlert Y, Spitz R et al. Customized FGFR1 increases cisplatin sensitivity in ovarian cancer. Cancer Biol Ther 2010; 10: oligonucleotide microarray gene expression-based classification of neuro- 495–504. blastoma patients outperforms current clinical risk stratification. J Clin Oncol 2006; 24 Turner N, Lambros MB, Horlings HM, Pearson A, Sharpe R, Natrajan R et al. 24: 5070–5078. Integrative molecular profiling of triple negative breast cancers identifies 47 Sagulenko V, Muth D, Sagulenko E, Paffhausen T, Schwab M, Westermann F. amplicon drivers and potential therapeutic targets. Oncogene 2010; 29: Cathepsin D protects human neuroblastoma cells from doxorubicin-induced cell 2013–2023. death. Carcinogenesis 2008; 29: 1869–1877.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc)

& 2013 Macmillan Publishers Limited Oncogene (2013) 3944 – 3953