IGF2BP1 Harbors Prognostic Significance by Gene Gain and Diverse Expression in Neuroblastoma

VOLUME 33 ⅐ NUMBER 11 ⅐ APRIL 10 2015

JOURNAL OF CLINICAL ONCOLOGY ORIGINAL REPORT

IGF2BP1 Harbors Prognostic Signiﬁcance by Gene Gain and Diverse Expression in Neuroblastoma Jessica L. Bell, Raseswari Turlapati, Tao Liu, Johannes H. Schulte, and Stefan Hüttelmaier

Jessica L. Bell, Raseswari Turlapati, and Stefan Hüttelmaier, Martin Luther Univer- ABSTRACT sity Halle-Wittenberg, Halle; Johannes H. Schulte, West German Cancer Center, Purpose University Hospital Essen, University Chromosomal 17q21-ter gain in neuroblastoma is both a common and prognostically significant Duisburg-Essen, and University Children’s event. The insulin-like growth factor-2 mRNA-binding protein 1 (IGF2BP1) gene is located near the Hospital Essen, Essen; Johannes H. proximal edge of this region. Here, its prognostic value is evaluated in neuroblastoma. Schulte, German Cancer Consortium and German Cancer Research Center, Heidel- Methods berg, Germany; and Tao Liu, Children’s The mRNA expression of IGF2BP family members was first evaluated by microarray data sets. In Cancer Institute Australia for Medical addition, in a separate cohort of 69 tumors, IGF2BP1 gene copy number, mRNA, and protein Research and University of New South abundance were determined and compared with clinical parameters. Wales, Randwick, New South Wales, Australia. Results Published online ahead of print at In two independent microarray data sets, 77% to 100% of tumors had substantial IGF2BP1 mRNA www.jco.org on March 9, 2015. levels measured. High IGF2BP1 transcript abundance was significantly associated with stage 4 Ͻ Ͻ Supported by German Research Foun- tumors (P .001) and decreased patient survival (P .001). IGF2BP1 was also associated with dation Grant No. GRK1591 (S.H.) and MYCN gene amplification and MYCN mRNA abundance. In the 69 neuroblastoma samples, by National Health and Medical IGF2BP1 DNA copy number (increased in 84% of tumors), mRNA, and protein abundance were Research Council project grants (T.L.; significantly higher in stage 4 compared with stage 1 tumors. Importantly, IGF2BP1 protein levels Australian Research Council fellow). were associated with lower overall patient survival (P ϭ .012) and positively correlated with MYCN Authors’ disclosures of potential mRNA, even when excluding MYCN-amplified tumors. Moreover, IGF2BP1 clearly affected MYCN conflicts of interest are found in the expression and neuroblastoma cell survival in vitro. article online at www.jco.org. Author contributions are found at the end of Conclusion this article. In neuroblastoma, IGF2BP1 was expressed in the majority of neuroblastoma specimens analyzed and was Corresponding author: Stefan Hüttel- associated with lower overall patient survival and MYCN abundance. These data demonstrate that IGF2BP1 maier, PhD, Institute of Molecular is a potential oncogene and an independent negative prognostic factor in neuroblastoma. Medicine, Section for Molecular Cell Biology, Martin Luther University Halle- J Clin Oncol 33:1285-1293. © 2015 by American Society of Clinical Oncology Wittenberg, 06120 Halle, Germany; e-mail: stefan.huettelmaier@medizin .uni-halle.de. and invasive potential of cells and promotes their INTRODUCTION 1,3,12 © 2015 by American Society of Clinical proliferation. This was shown to primarily de- Oncology The insulin-like growth factor-2 mRNA-binding pend on IGF2BP1 sustaining elevated MYC expres- 2,13 0732-183X/15/3311w-1285w/$20.00 protein 1 (IGF2BP1) is an oncofetal RNA-binding sion, by preventing MYC mRNA decay. MYC is a DOI: 10.1200/JCO.2014.55.9880 protein that is synthesized de novo in more than 14 paralog of MYCN, and the two proteins share many cancer types, including hematopoietic, liver, and redundancies in function and oncogenic behavior.14 ovarian cancers.1-4 High expression of IGF2BP1 is Neuroblastoma is the most common cause of associated with poor overall prognosis and tumor cancer-related death in infants.15 The deregulation metastasis. Furthermore, its paralog, IGF2BP3, has of neural crest development is considered critical in been touted as a biomarker for a growing list of neuroblastoma initiation. Moreover, MYCN ampli- cancers, including neuroblastoma in one report.1,5,6 fication is an established prognostic marker that is However, because of insufficient paralog specificity used to stratify patients, and countless studies have of antibodies used in many IGF2BP studies, demonstrated that MYCN expression is an impor- paralog-specific expression patterns require fur- tant factor in tumorigenesis and aggressive tumor- ther validation.4,7 During development, IGF2BP cell phenotype. Chromosomal 17q21-ter gain is the family members are essential for the migration of most common chromosomal alteration in neuro- neural crest cells, modulation of neurite develop- blastoma (50% to 95% of patient cases), and this is ment, and regulation of stem-cell properties cen- frequently associated with MYCN amplification at trally within the LIN28/Let-7 networks.8-11 In a 2p24.16-20 The literature provides no consensus cancer context, IGF2BP1 enhances the migratory on the crucial genes located at the 17q locus,

Downloaded from ascopubs.org by University of New South Wales on January 4, 2017 from 129.094.134.108 Copyright © 2017 American Society of Clinical Oncology. All rights reserved. Bell et al although several candidates have been proposed, such as survivin, topo- ily members had prognostic value in neuroblastoma. First, two indepen- isomerase II, ncRAN, and others.21-23 However, it remains unclear dent neuroblastoma microarray data sets, the Versteeg (88 tumors) and whether these candidate genes are collateral DNA, because of their loca- Kocak (476 tumors), were analyzed.23a,23b In these cohorts, all IGF2BP tion, or have additive effects on cancer-cell phenotype. IGF2BP1 is located family members were highly expressed at the mRNA level: IGF2BP1 (77% at 17q21.32, and its role in neuroblastoma has not been investigated thus to 100% of tumors), IGF2BP2 (88% to 100% of tumors), and IGF2BP3 far to our knowledge. (100% of tumors). Using arbitrary median divisions for Kaplan-Meier The chromosomal location (17q) of the IGF2BP1 gene, as well as analyses, we compared the prognostic potency of IGF2BPs with that of the IGF2BP1 protein-directed enhancement of MYC expression, neu- oncogenes reported in neuroblastoma (as discussed by Pugh et al,24 ral crest cell migration, and cell self-renewal strongly, points toward its Cheung and Dyer,24a and Brodeur et al24b) and that of genes located at clinical importance in neuroblastoma. To our knowledge, we show for 17q, whether clinically relevant or irrelevant (Table 1). Strikingly, of the the first time here that IGF2BP1 gene gain and increased IGF2BP1 IGF2BPs, only high IGF2BP1 abundance was significantly associated with mRNA as well as protein expression are clinically significant factors in an increased risk of death in both microarray data sets. IGF2BP2 did not neuroblastoma. In agreement with these findings, we demonstrate reach significance, and high IGF2BP3 expression was associated with that IGF2BP1 promotes neuroblastoma cell viability, MYCN expres- survival only in the Kocak cohort. As expected, a significant association sion, and doxorubicin resistance in vitro. with increased risk of death was also observed for the reported oncogenes ALK and LIN28B in both data sets. However, using median division, no significant association was observed between MYCN expression and pa- METHODS tient survival rates. Notably, the remaining putative oncogenes as well as Study Population two non–clinically relevant control genes (PNMT,MILR1) located at 17q Details on the study population are provided in the Data Supplement. did not achieve significance. Although all IGF2BPs are proposed to modulate IGF2 mRNA translation and/or IGF2 mRNA abundance, high levels Molecular Cell Biology of IGF2 as well as transcripts encoding other factors within the IGF1/2 Protocols are provided in the Data Supplement. pathways (IGF1, IGF1R, IR, IRS1, IRS2) were not associated with decreased patient survival (refer to R2 platform; data not shown).1,25 Be- RESULTS cause IGF2BP1 mRNA abundance compares strikingly well to other oncogenes studied in neuroblastoma, it was our prime candidate for IGF2BP1 mRNA Expression Is Clinically Significant further analysis. in Neuroblastoma Kaplan-Meier analysis of the Versteeg data set, using the P value Because of the oncofetal expression of IGF2BPs, the chromosomal scanning method, confirmed that IGF2BP1 abundance was signifi- location of IGF2BP1 (17q21-ter), and one report suggesting that IGF2BP3 cantly associated with patient risk of death in all tumors as well as is upregulated in neuroblastoma,1,6 we investigated whether IGF2BP fam- non–MYCN-amplified tumors (Figs 1A and 1B). In stage 4 tumors,

Table 1. IGF2BP1 Is Novel Marker of Survival, Performing Well Against Reported Prognostic Markers Versteeg Data Set Kocak Data Set ء(n ϭ 476 tumors) ء(n ϭ 88 tumors)

Gene Alias Locus Survival (%)† P‡ Survival (%)† P‡ IGF2BPs IGF2BP1§ ZBP1, IMP1, CRDBP1 17q21.3 53 .013 68 .002 IGF2BP2 P62, IMP2 3q27.2 62 .564 80 .104 IGF2BP3࿣ KOC-1, IMP3 7p15.3 55 .059 63 Ͻ .001 Reported neuroblastoma oncogenes or markers ALK§ CD246 2p23 53 .022 63 Ͻ .001 ATRX RAD54, XNP Xp21.1 60 .402 73 .798 IGF2 IGF-II, somatomedin-A, preptin 11p15.5 65 .884 75 .868 LIN28B§ CSDD 6q21 52 .012 70 .0042 LMO1 RBTN1, RHOM1 11p15 62 .583 78 .074 MYCN n-Myc, v-myc 2p24.3 53 .055 74 .902 PHOX2B PMX2B, NBPHOX 4p13 63 .758 74 .692 Reported neuroblastoma 17q oncogenes BIRC5࿣ Survivin, API4 17q25.3 43 Ͻ .001 78 .382 PPM1D࿣ WIP1, PP2C- DELTA 17q23.3 53 .021 75 .555 17q examples (not clinically signiﬁcant) PNMT PENT 17q12 60 .279 73 .093 MILR1 MCA-32, allergin-1 17q23.3 59 .307 76 .539

.Accessed and analyzed via R2 visualization platformء †Percentage of patients who survived in high gene expression division grouping. ‡Determined by Kaplan-Meier curve, with median division of mRNA expression levels, and log-rank test. §Statistically signiﬁcant gene in two independent neuroblastoma tumor cohorts; associated with poor patient survival. ࿣Statistically signiﬁcant gene in one cohort.

A All Tumors B Nonamplified Tumors 1.0 Low (n = 77) 1.0 High (n = 11) P < .001 0.8 0.8

0.6 0.6

0.4 0.4 (probability) (probability)

Overall Survival Overall Survival Low (n = 63) 0.2 0.2 High (n = 9) P < .001

0 24 48 72 96 120 144 168 192 216 240 0 24 48 72 96 120 144 168 192 216 240 Duration of Follow-Up (months) Duration of Follow-Up (months) C D 600 10 P = .019 P = .328 (n.s.) 9

8 400 7

5 200 4 IGF2BP1 Expression 3 P < .001 Log2 of IGF2BP1 Expression 0 2 1 2 3 4 4S Nonamplified Amplified (n = 72) (n = 16) Stage MYCN DNA E

14 MYCN nonamplified MYCN amplified 13 All tumors 12 R = 0.466 P < .001 11 MYCN nonamplified 10 R = 0.436 P < .001 9 MYCN amplified n.s. 8 7

Log2 of MYCN Expression 6 5 0 246810 Log2 of IGF2BP1 Expression

Fig 1. High IGF2BP1 expression was significantly associated with lower patient survival and MYCN amplification in microarray data set (Versteeg). (A) High IGF2BP1 mRNA expression was significantly negatively associated with patient survival. Kaplan-Meier graphs demonstrated overall patient survival probability for high and low IGF2BP1 mRNA expression from 88 neuroblastoma tumor samples with microarray mRNA expression data. (B) This was also true when MYCN-amplified samples were excluded from analysis. (C) IGF2BP1 expression according to tumor staging shows IGF2BP1 expression was most elevated in highest-risk grouping of stage 4 tumors. (D) In MYCN DNA–amplified samples (16 samples), IGF2BP1 expression was significantly higher than in non–MYCN-amplified tumors. (E) MYCN and IGF2BP1 were expressed in positive and significant (Pearson’s test) correlation calculated for all tumors; non–MYCN-amplified tumors (alone) also showed correlation. Statistical analyses were performed using R2 microarray analysis and visualization platform software and Versteeg data set. Means and SEs shown. n.s., not statistically significant.

IGF2BP1 was significantly increased, as compared with the less aggres- well as in the non–MYCN-amplified samples (Figs 1D and 1E). Im- sive stage 1 tumors (Fig 1C). Although IGF2BP1 abundance was in- portantly, all these findings and trends were clearly repeated in the creased in MYCN-amplified tumors, the abundance of IGF2BP1 was Kocak microarray data set (Data Supplement). Because high IGF2BP1 positively correlated with the expression of MYCN in all samples as remains significantly associated with reduced overall survival, even www.jco.org © 2015 by American Society of Clinical Oncology 1287

had IGF2BP1 gain (without accounting for statistical confidence). Table 2. Summary of Clinical Characteristics of Tumor Cohorts Importantly, these data confirm the occurrence of IGF2BP1 DNA gain Characteristic No. of Tumors % and confirm that it frequently co-occurs with MYCN amplification. MYCN copy number At the mRNA level, quantitative real-time polymerase chain re- Amplified (Ͼ five copies) 12 17.4 action analysis of the 69 specimens revealed that IGF2BP1 transcript Normal/WT 57 82.6 abundance was the highest in stage 4 tumors and significantly in- INSS tumor stage creased as compared with stage 1 samples (International Neuroblas- 1 14 20.3 2 7 10.1 toma Staging System; Fig 2C). This is consistent with the analyses of 3 9 13.0 the two independent microarray data sets (Fig 1C; Data Supple- 4 27 39.1 ment). Notably, IGF2BP1 transcript abundance was significantly 4S 12 17.4 increased in those tumors with IGF2BP1 gene gains, indicating that Age, months an increase in IGF2BP1 DNA can confer an upregulation of Ͼ 18 34 49.3 IGF2BP1 mRNA abundance (Fig 2C). Moreover, IGF2BP1 mRNA Ͻ 18 35 50.7 Survival levels were increased in MYCN-amplified tumors (Fig 2C). In Event free 48 81.3 agreement with the microarray data analyses (Figs 1D and 1E; Data Progression 21 30.4 Supplement), IGF2BP1 mRNA and MYCN mRNA levels were Died 17 24.6 positively correlated irrespective of MYCN amplification in the 69 No data 1 1.4 neuroblastoma samples analyzed (Fig 2D). Total 69 100

Abbreviations: INSS, International Neuroblastoma Staging System; WT, wild type. IGF2BP1 Is Highly Expressed at Protein Level in Stage 4 Neuroblastoma IGF2BP1 mRNA expression was significantly increased in aggressive neuroblastoma and associated with an overall poor prognosis when excluding MYCN-amplified samples (Data Supplement), we in three independent tumor cohorts. Therefore, we investigated if tested whether high IGF2BP1 expression was an independent prog- IGF2BP1 protein levels were increased, using IGF2BP paralog-specific nostic factor in neuroblastoma. Multivariable analyses following the custom-made monoclonal mouse antibodies and Western blot- Cox regression model were performed in the Kocak cohort (Data ting.1,12,29 In the majority of the 69 specimens analyzed, all IGF2BP Supplement). This confirmed that high IGF2BP1 mRNA abundance family members were expressed (representative Western blot of 30 of was significantly associated with reduced event-free survival (P ϭ 69 samples; Fig 3A). Consistent with the R2 analyses of the Versteeg .0058). As expected, this was also observed in tumors with MYCN gene and Kocak microarray data sets (Table 1), no increased expression in amplification (P ϭ .0134). aggressive neuroblastoma was observed for IGF2BP2 or IGF2BP3. In contrast, IGF2BP1 protein abundance was significantly increased in IGF2BP1 Gene Is Commonly Gained in Neuroblastoma stage 4 tumors (International Neuroblastoma Staging System; Fig 3B) Tissues and Predicts Poor Prognosis and was elevated in both MYCN-amplified and IGF2BP1-gained tu- Because the 17q region is often gained in neuroblastoma, and the mor subsets (Fig 3B). In addition, high IGF2BP1 protein expression IGF2BP1 gene locus is located at 17q21.32, we chose to quantify the was associated with reduced event-free and overall survival. This was DNA copy number of IGF2BP1. For these analyses, we acquired 69 also observed in non–MYCN-amplified tumors (Fig 3C). These find- frozen neuroblastoma specimens of all clinical stages (Table 2). We ings presented a compelling correspondence between increased DNA, used a quantitative polymerase chain reaction strategy, similar to that mRNA, and protein levels in aggressive neuroblastoma. previously used in clinical approaches for MYCN and other genes.26,27 Strikingly, we observed that 58 of the 69 tumors had some level (Ն 20%) of IGF2BP1 DNA copy number increase, whereas no dele- IGF2BP1 Promotes MYCN Expression, Viability, and tions were found. The degree of IGF2BP1 DNA copy gain was signif- Chemotherapy Resistance in Neuroblastoma Cells icantly higher in stage 4 compared with stage 1 tumors (Fig 2A). In carcinoma-derived tumor cell lines, IGF2BP1 reportedly en- Kaplan-Meier analysis, using the cut point of 1.9 copies, revealed high hances cell aggressiveness and survival, partially by preventing MYC IGF2BP1 gene copy number to be a significant factor in determining mRNA degradation and sustaining the elevated expression of MYC, reduced event-free survival (Fig 2B). We were able to confirm that the MYCN paralog.1-3,30 Thus, we investigated if IGF2BP1 protein and IGF2BP1 gain frequently occurred in another neuroblastoma cohort. MYCN mRNA were statistically correlated in the 69-sample tumor Using the TARGET (Therapeutically Applicable Research to Generate cohort. Indeed, we observed this to be the case, suggesting IGF2BP1 Effective Treatments) initiative,28 we found that 23% (78 of 341) of may modulate MYCN levels in neuroblastoma (Fig 4A). Conse- human neuroblastoma samples had IGF2BP1 gene gain or amplifica- quently, we undertook essential in vitro studies to assess the influence tion (Data Supplement), 64% (50 of 78) with IGF2BP1-amplified of IGF2BP1 on MYCN expression and cell survival. First, we tested if neuroblastoma also had MYCN amplification (Data Supplement), IGF2BP1 affected MYCN mRNA expression in a neuroblastoma cell and 56% (50 of 89) with MYCN-amplified neuroblastoma showed line. MYCN mRNA levels were clearly reduced on siRNA knockdown IGF2BP1 amplification. In this database, a stringent threshold of more of IGF2BP1 in MYCN-amplified BE2C cells (Fig 4B). We then dem- than two DNA copies was defined as a gain (we used 1.2-fold for our onstrated that IGF2BP1 knockdown had a strong enough impact to 69 samples); this explains why the incidence in the database is lower severely reduce MYCN at the protein level in three MYCN-amplified than ours. By applying the same cutoff, 43% (30 of 69) of our samples cell lines (Fig 4C). In accord with the loss of MYCN protein, these cell

A B

5 P < .001 1.0

4 0.8

3 0.6

2 0.4 (probability)

1 Event-Free Survival 0.2 Low copy number (n = 34) High copy number (n = 34) P = .0396 IGF2BP1 DNA Copy Number 0 1 2-3 4 4S 0 2,000 4,000 6,000 INSS Tumor Staging Duration of Follow-Up (days) C D

25 P = .028 P = .046 P < .001 1,000 WT 20 MYCN amplified 100 15 All tumors 15 R = 0.526 P < .001 10 (n = 69) Nonamplified 10 1 R = 0.411 P = .007

RNA (log10 axis) (n = 55) 0.1 IGF2BP1 RNA 5 0.01 MYCN (normalized to ACTB + VCL) 0 0.001 1 2-3 4 4S WT Gained WT Amplified 1 0 01 0.1 10 0 0.01 1 0.001 INSS IGF2BP1 MYCN 0.00 IGF2BP1 RNA (log10 axis)

Fig 2. IGF2BP1 DNA copy numbers and mRNA abundance are increased in aggressive neuroblastoma. We extracted and analyzed DNA from 69 frozen neuroblastoma tumors. DNA copies shown are normalized to commercial blood DNA, with 1 indicating normal/wild-type (WT) copy number. (A) IGF2BP1 DNA copy number across International Neuroblastoma Staging System (INSS) staging. IGF2BP1 copy number was signiﬁcantly increased in aggressive stage 4 neuroblastoma. Crosses indicate averages; horizontal lines indicate medians. (B) Kaplan-Meier graph shows increased DNA copy number increases chance of patient death or relapse. Tumor copy number levels were divided by median, with low representing tumors with Ͻ 1.9 DNA copies and high representing those with Ͼ 1.9 copies. (C) mRNA levels of IGF2BP1 as determined by qRT-PCR with 69 tumors divided into different subsets: INSS, MYCN genotype, and IGF2BP1 genotype. WT indicates Ͻ 1.2 copies; gained indicates Ͼ 1.2 copies. One-sided Mann-Whitney test was used as test of difference. mRNA levels were cross normalized to ACTB and VCL template controls. (D) Spearman correlation showed clear rank-based correlation between IGF2BP1 mRNA and MYCN mRNA (for all tumors and also non—MYCN-ampliﬁed tumors alone).

lines also showed a significant decrease in the number of viable cells ing viable cells in the ZBP1-transduced population was double that of when IGF2BP1 was knocked down (Fig 4D). the EV and KH controls (Fig 4F). These findings indicate that Recently, IGF2BP1 has been reported to increase the chemother- IGF2BP1 enhances MYCN expression and chemotherapy resistance apy resistance of cells derived from rhabdomyosarcomas and ovarian of neuroblastoma cells in an RNA binding–dependent manner. cancer.31,32 Therefore, we tested whether the overexpression of ZBP1 (chicken ortholog of IGF2BP1, which has Ͼ 90% protein sequence identity with IGF2BP1) promoted the expression of MYCN and resis- DISCUSSION tance against doxorubicin, a chemotherapeutic used in the treatment of aggressive neuroblastoma. To this end, the non–MYCN-amplified In this study, we evaluated the prognostic significance of the RNA- neuroblastoma cell line, NBLS, was stably transduced with a GFP- binding protein IGF2BP1 in neuroblastoma. To our knowledge, this is tagged vector (either empty vector [EV], ZBP1 [wild type], or RNA the first such study in this malignancy—a surprising fact, given the binding–deficient ZBP1 mutant [KH]). The KH mutant had all role of IGF2BP1 in neural crest development and aggressive cancer HNRNPK-homology domains inactivated by point mutations in the and its location at 17q. Importantly, in three separate and independent GXXG loops, preventing the protein-binding RNA targets.33 The sta- patient cohorts, we were able to show that high IGF2BP1 expression ble overexpression of wild-type ZBP1 resulted in a modest increase in was associated with decreased patient survival and MYCN expression. MYCN proteins levels compared with EV and KH controls (Fig 4E). In neuroblastoma, 17q gain is the most common chromosomal Furthermore, after doxorubicin treatment, the percentage of remain- alteration and arguably a powerful predicting factor for patient survival www.jco.org © 2015 by American Society of Clinical Oncology 1289

A WB: Stage 1 2 3 4 4S Sample ID 1 2 3 4 5 6 8 9 7 11 19 18 22 23 24 25 26 27 28 29 30 12 13 14 15 16 17 20 21 10 IGF2BP1 gain + + + + + + + + + + + + + + + + + + + + + + + + + MYCN amplified + + + + + IGF2BP3 IGF2BP2 IGF2BP1 IGF2BPs

MYCN MYC MYCs

ACTB TUBA4A VCL Loading Controls

B P = .018 P = .029 P = .034 10

IGF2BP1 4 Protein/TUBA4A 2

0 1-3 4 4S WT Gained WT Amplified INSS IGF2BP1 MYCN CDE 1.0 1.0 1.0

0.8 0.8 MYCN 0.8

0.6 0.6 0.6

0.4 0.4 0.4 (probability) (probability)

Low IGF2BP1 protein (n = 28) Overall Survival Low IGF2BP1 protein (n = 28) Low IGF2BP1 protein (n = 26) Event-Free Survival 0.2 0.2 0.2

High IGF2BP1 protein (n = 40) High IGF2BP1 protein (n = 40) Amplified (probability) High IGF2BP1 protein (n = 28) P = .0321 P = .0124 P = .0231 Overall Survival, Non-

0 2,000 4,000 6,000 0 2,000 4,000 6,000 0 2,000 4,000 6,000 Duration of Follow-Up (days) Duration of Follow-Up (days) Duration of Follow-Up (days)

Fig 3. IGF2BP1 protein levels are increased in aggressive neuroblastoma. (A) Representative Western blot (WB) of IGF2BPs, MYC/N oncogenes, and loading control genes in 30 of 69 primary tumors samples analyzed. (B) Protein levels of IGF2BP1 divided into each grouping type: International Neuroblastoma Staging System (INSS), MYCN genotypes, and IGF2BP1 genotypes. Crosses indicate averages; horizontal lines indicate medians. (C, D, E) Kaplan-Meier graph shows increased IGF2BP1 protein is associated with higher chance of patient death and disease progression. WT, wild type. and aggressive disease.18 However, agreed-on strong candidates at this plification. Although the magnitude of the DNA increase in IGF2BP1 was locus are lacking. We have shown that IGF2BP1 is a new candidate of modestinourcohort,ourfindingsarelargelyconsistentwiththeliterature functional importance within 17q. In the majority of the 69 tumor speci- on 17q gain. Several publications have reported 17q as a common event mens analyzed, an increase in IGF2BP1 gene copy number was found. (50% to 95% of neuroblastoma tumors), with 17q-gain magnitudes This was further supported by another larger independent data set, which rarely larger than five-fold normal copy.20,34 Furthermore, in MYCN- in addition showed that IGF2BP1 gain often occurred with MYCN am- overexpressing mice that developed neuroblastoma, the equivalent of

BE2C SiC A B P < .001 P = .0296 IGF2BP1 Protein v MYCN mRNA silGF2BP1 1,000 1.0

100 WT 0.8 MYCN amplified 10 All tumors 0.6 R = 0.355 1 P = .0014 (n = 69) 0.4 RT-qPCR: RNA (log10 axis) 0.1 Nonamplified R = 0.232 P = .0041 0.2 0.01 (n = 55) MYCN Relative mRNA Expression 0.001 0 2 8 IGF2BP1 MYCN 0.5 0.125 0.03125 IGF2BP1 Protein (log2 axis) C D BE2C IMR-32 KELLY P = .0181 P = .0081 P < .001 BE2C IMR-32 KELLY 1.0

0.8 siIGF2BP1 siC siC siIGF2BP1 siC siIGF2BP1 WB: 0.6 1.0 0.41 ± 0.1 1.0 0.39 ± 0.1 1.0 0.31 ± 0.1 0.4 IGF2BP1

1.0 0.47 ± 0.2 1.0 0.31 ± 0.1 1.0 0.2 ± 0.1 0.2

MYCN

Viable Cells Alamar Blue Assay 0 (relative absorbance OD 570/590)

siC siC siC TUBA

silGF2BP1 silGF2BP1 silGF2BP1

E Cell line: NBLS F NBLS Cells Treated With 160 nM Doxorubicin for 72 Hours Vector: EV WT KH 15 WB: 1.0 2.2 1.3 P = .0437

MYCN 590 /A n = 4 570 10

ZBP1-GFP

5 Viable Cells (%) 6A9 Ab 6A9 IGF2BP1 (IGF2BP1) Alamar Blue A

0 VCL EV IGF2BP1-WT KH mutant Overexpression Vector

Loading TUBA4A

Fig 4. IGF2BP1 promotes aggressive neuroblastoma cell phenotype in vitro. Means and SE displayed over panels. (A) Positive correlation of IGF2BP1 protein and MYCN mRNA (by Spearman’s rank correlation method); 69 primary neuroblastoma tumor samples were used for correlation. (B) Real-time quantitative polymerase chain reaction (RT-qPCR) of IGF2BP1 and MYCN expression in BE2C neuroblastoma cells after IGF2BP1 siRNA transient transfection (siIGF2BP1) for 72 hours; siC is control siRNA. GAPDH was used as mRNA reference control. (C) Western blot (WB) of MYCN-amplified neuroblastoma cells lines transiently transfected with siC and siIGF2BP1. TUBA4A was used as loading control. IGF2BP1 and MYCN protein levels are decreased in siIGF2BP1-transfected cells. t testing showed significant differences in MYCN protein levels in MYCN-amplified cell lines: BE2C (P ϭ .034), IMR-32 (P ϭ .013), and KELLY (P ϭ .0013). (D) Alamar blue viability assay was used to determine cell viability; lower absorbance indicates less viable cells. Transfection with IGF2BP1 siRNA decreased cell viability. (E) Non–MYCN-amplified neuroblastoma cells (NBLS) were stably transduced with GFP-tagged lentiviral expression constructs; empty vector (EV), chicken ortholog of IGF2BP1 (ZBP1-GFP/WT), and ZBP1 mRNA–binding mutant (KH mutant). VCL and TUBA4A were used as loading controls. (F) Transduced NBLS cells (10,000 cells/well; 96 well format) were plated and next day treated with doxorubicin (160 nmol/L in concentration for 72 hours). ZBP1-GFP overexpression increased number of viable cells compared with controls after drug treatment. Alamar blue was then used to quantitate cell viability. One-sided t test was used to test significance of differences.

Downloaded from ascopubs.org by University of New South Wales on January 4, 2017 from 129.094.134.108 Copyright © 2017 American Society of Clinical Oncology. All rights reserved. Bell et al human 17q in mice (ie, chromosome 11) was also commonly gained,35 explains why we were unable to repeat the previous findings on IGF2BP3 again indicating 17q gain is a prerequisite in an aggressive neuroblastoma in neuroblastoma with our validated specific antibodies. It is likely that the subset. Most importantly, the DNA copy gain measured in the 69-tumor clinical significance of IGF2BP3 in other work resulted in part from anti- cohort indicated elevated IGF2BP1 mRNA expression and decreased pa- body affinity for IGF2BP1. Our study provides important findings for tient survival. Work by others in breast cancer (by fluorescent in situ the application of IGF2BPs in clinical practice, which we consider hybridization and comparative genomic hybridization) has shown an relevant far beyond neuroblastoma. In addition, with the IGF2BP1- 18% to 30% incidence of IGF2BP1 gain, with one tumor having 20 copies directed antibody used here (which is also suitable for immunohisto- of the gene.36,37 However, these aberrations were not associated with chemistry3), along with genomic DNA primers and expression patient survival, and thus, our findings are the first to our knowledge primers, we are confident we have provided tools for the reliable to reveal the clear clinical significance of IGF2BP1 DNA copy gain. We detection of IGF2BP1 parameters in clinical specimens. expect that IGF2BP1 gain will be clinically important in other cancer It is evident that IGF2BP1 has strong prognostic significance inde- types associated with IGF2BP1, such as testicular, colon, and lym- pendent of MYCN amplification and that the protein may actively en- phatic cancers.1 hance tumor aggressiveness by regulating target mRNAs encoding IGF2BP1 mRNA and protein expression are clearly clinically oncogenic factors. We cannot exclude—and in fact we expect—that significant factors in neuroblastoma. We have analyzed three IGF2BP1 function in neuroblastoma involves additional target mRNAs, independent data sets, all consistently conveying that upregulated such as the reported signaling importance of IGF2BP1 in cancer via tar- IGF2BP1 abundance is associated with aggressive disease. In addition, geting LEF-1 (involved in Wnt signaling and epithelial-to-mesenchymal although IGF2BP1 is correlated with MYCN, it also has a role and transition29) and the central role of IGF2BP1 in the LIN28/Let-7 path- clinical significance independent of MYCN amplification. The Cox way.32,39 These are certainly both attributes of IGF2BP1 that require ur- regression analyses of the Kocak cohort revealed that IGF2BP1 mRNA gent investigation in neuroblastoma, especially because overexpression of expression is an independent prognostic factor, achieving significance LIN28B was demonstrated to induce neuroblastoma in mice.40 In addi- even higher than that of MYCN amplification. tion, IGF2BP1 has been reported to target IGF2 transcripts, influencing The penetrance of IGF2BP1 protein abundance and broad ex- their mRNA translation1,12; this is a factor of known importance in neu- pression of MYCN in neuroblastoma, combined with a high concor- roblastoma.41 However, the activity of many factors within the IGF dance in their aberrations (among four independent cohorts), allow pathway cannot be quantitated effectively or easily with mRNA and us to speculate that these factors actively synergize in neuroblastoma. protein levels. Interestingly, IGF pathway inhibition (by cixutu- In support of this, we report here that IGF2BP1 promotes the expres- mumab) is currently being targeted in phase II clinical trials in neuro- sion of MYCN and enhances the survival of neuroblastoma-derived blastoma,42 and it may also be beneficial to investigate IGF2BP1 cells in vitro. Although the molecular basis of this interaction with expression in relationship to patient response to this drug and possible MYCN regulation awaits further detailed analysis, we have provided use of IGF2BP1 in patient recruitment. Obviously, there are number one presumably powerful mode by which the high expression of of future research avenues for IGF2BP1, a largely ignored IGF pathway IGF2BP1 promotes aggressive neuroblastoma. member, which may prove to be important. As with the IGF2BPs, MYCN protein is expressed in the majority of In conclusion, our studies identify IGF2BP1, located on the 17q neuroblastomas, including many non–MYCN-amplified tumors. This arm, as a gene of striking importance in neuroblastoma. This gene common observation is often understated in the literature and suggests harbors a clear negative prognostic value at the DNA, mRNA, and alternative mechanisms of MYCN deregulation apart from its DNA am- protein levels. Moreover, its expression is positively correlated with plification. These mechanisms are presumably masked in animal and the most prominent oncogene in neuroblastoma—MYCN. Our cell-line models of neuroblastoma, which usually have MYCN amplifica- work, in conjunction with that reported in studies of the cellular tion or high MYC expression. Therefore, such models do not adequately functions of IGF2BP1, creates a novel paradigm, indicating several represent the majority of human tumors and may not reflect more subtle mechanisms to be investigated by which IGF2BP1 could have an complex MYCN regulation. We expect IGF2BP1 is a novel mechanism impact on neuroblastoma phenotype. Although the IGF2BP1 paralog promoting MYCN abundance (and possibly other factors) in neuroblas- IGF2BP3 has received much attention in the literature as a biomarker toma. In non–MYCN-amplified cells, transcriptional output for MYCN of aggressive cancers, we propose that IGF2BP1 is the more promising is comparatively moderate, and thus, mechanisms sustaining expression family member and a potentially drugable target in neuroblastoma. post-transcriptionally would be a substantial selection advantage for neuroblastoma cells. This hypothesis is supported by the IGF2BP1- AUTHORS’ DISCLOSURES OF POTENTIAL CONFLICTS dependent enhancement of MYCN expression and chemotherapy OF INTEREST resistance in the non–MYCN-amplified neuroblastoma cells. IGF2BP2 and IGF2BP3 are highly expressed, without a strong pat- Disclosures provided by the authors are available with this article at tern in neuroblastoma (shown clearly in Fig 3A). At first glance, this www.jco.org. observation seems surprising, because IGF2BP3 is the most prominently reported family member and is apparently unregulated in various cancers, AUTHOR CONTRIBUTIONS including neuroblastoma.6,7,38 However, a vast majority of studies have relied on antibodies with doubtful paralog specificity. A relevant example Conception and design: Jessica L. Bell, Stefan Hüttelmaier is the previous report on IGF2BP3 and its clinical importance in neuro- Provision of study materials or patients: Jessica L. Bell, Johannes H. blastoma, which used the Dako (Carpinteria, CA) IMP-3 (clone 69.1) Schulte, Stefan Hüttelmaier 6 antibody. We and others have published strong evidence that this anti- Collection and assembly of data: Jessica L. Bell, Raseswari Turlapati, body has affinity for all the IGF2BPs, including IGF2BP1.1,4,7,38 This Tao Liu, Johannes H. Schulte

Data analysis and interpretation: Jessica L. Bell, Tao Liu, Johannes H. Manuscript writing: All authors Schulte, Stefan Hüttelmaier Final approval of manuscript: All authors

17. Westermann F, Schwab M: Genetic parame- search to Generate Effective Treatments. http:// REFERENCES ters of neuroblastomas. Cancer Lett 184:127-147, target.cancer.gov/ 2002 29. Zirkel A, Lederer M, Stöhr N, et al: IGF2BP1 1. Bell J, Wächter K, Mühleck B, et al: Insulin- 18. Bown N, Cotterill S, Lastowska M, et al: Gain promotes mesenchymal cell properties and migra- like growth factor 2 mRNA-binding proteins of chromosome arm 17q and adverse outcome in tion of tumor-derived cells by enhancing the expres- (IGF2BPs): Post-transcriptional drivers of cancer pro- patients with neuroblastoma. N Engl J Med 340: sion of LEF1 and SNAI2 (SLUG). Nucleic Acids Res gression? Cell Mol Life Sci 70:2657-2675, 2013 1954-1961, 1999 41:6618-6636, 2013 2. Köbel M, Weidensdorfer D, Reinke C, et al: 19. Chen QR, Bilke S, Wei J, et al: cDNA array- 30. Weidensdorfer D, Stöhr N, Baude A, et al: Expression of the RNA-binding protein IMP1 corre- CGH profiling identifies genomic alterations specific Control of c-myc mRNA stability by IGF2BP1- lates with poor prognosis in ovarian carcinoma. to stage and MYCN-amplification in neuroblastoma. associated cytoplasmic RNPs. RNA 15:104-115, Oncogene 26:7584-7589, 2007 BMC Genomics 5:70, 2004 2009 Gutschner T, Hämmerle M, Pazaitis N, et al: 3. 20. George RE, Attiyeh EF, Li S, et al: Genome- 31. Faye MD, Beug ST, Graber TE, et al: IGF2BP1 Insulin-like growth factor 2 mRNA-binding protein 1 wide analysis of neuroblastomas using high-density controls cell death and drug resistance in rhabdomy- (IGF2BP1) is an important protumorigenic factor in single nucleotide polymorphism arrays. PLoS One osarcomas by regulating translation of cIAP1. Onco- hepatocellular carcinoma. Hepatology 59:1900- 2:e255, 2007 gene [epub ahead of print on April 7, 2014] 1911, 2014 21. Yu M, Ohira M, Li Y, et al: High expression of 32. Boyerinas B, Park SM, Murmann AE, et al: 4. Natkunam Y, Vainer G, Chen J, et al: Expres- ncRAN, a novel non-coding RNA mapped to chro- Let-7 modulates acquired resistance of ovarian can- sion of the RNA-binding protein VICKZ in normal mosome 17q25.1, is associated with poor prognosis cer to taxanes via IMP-1-mediated stabilization of hematopoietic tissues and neoplasms. Haemato- in neuroblastoma. Int J Oncol 34:931-938, 2009 logica 92:176-183, 2007 multidrug resistance 1. Int J Cancer 130:1787-1797, 22. Yoon KJ, Danks MK, Ragsdale ST, et al: 2012 5. Kapoor S: IMP3: A new and important bio- Translocations of 17q21 similar to qter in neuroblas- marker of systemic malignancies. Clin Cancer Res 33. Wachter K, Köhn M, Stöhr N, et al: Subcellular toma cell lines infrequently include the topoisomer- localization and RNP formation of IGF2BPs (IGF2 14:5640, 2008 ase II alpha gene. Cancer Genet Cytogenet 167: mRNA-binding proteins) is modulated by distinct 6. Chen ST, Jeng YM, Chang CC, et al: Insulin- 92-94, 2006 like growth factor II mRNA-binding protein 3 expres- RNA-binding domains. Biol Chem 394:1077-1090, 23. Islam A, Kageyama H, Takada N, et al: High sion predicts unfavorable prognosis in patients with 2013 expression of survivin, mapped to 17q25, is signifi- neuroblastoma. Cancer Sci 102:2191-2198, 2011 34. Bown N: Neuroblastoma tumour genetics: cantly associated with poor prognostic factors and 7. Lederer M, Bley N, Schleifer C, et al: The role Clinical and biological aspects. J Clin Pathol 54:897- promotes cell survival in human neuroblastoma. of the oncofetal IGF2 mRNA-binding protein 3 910, 2001 Oncogene 19:617-623, 2000 (IGF2BP3) in cancer. Semin Cancer Biol 29C:3-12, 35. Hackett CS, Hodgson JG, Law ME, et al: 23a. Molenaar JJ, Koster J, Ebus ME, et al: Copy 2014 Genome-wide array CGH analysis of murine neuro- number defects of G1-cell cycle genes in neuroblas- 8. Zhang HL, Eom T, Oleynikov Y, et al: blastoma reveals distinct genomic aberrations toma are frequent and correlate with high expres- Neurotrophin-induced transport of a beta-actin which parallel those in human tumors. Cancer Res sion of E2F target genes and a poor prognosis. mRNP complex increases beta-actin levels and stim- 63:5266-5273, 2003 Genes Chromosomes Cancer 51:10-19, 2012 ulates growth cone motility. Neuron 31:261-275, 36. Doyle GA, Bourdeau-Heller JM, Coulthard S, 23b. Kocak H, Ackermann S, Hero B, et al: Hox-C9 2001 et al: Amplification in human breast cancer of a gene activates the intrinsic pathway of apoptosis and is 9. Yaniv K, Fainsod A, Kalcheim C, et al: The encoding a c-myc mRNA-binding protein. Cancer associated with spontaneous regression in neuro- RNA-binding protein Vg1 RBP is required for cell Res 60:2756-2759, 2000 blastoma. Cell Death Dis 4:e586, 2013 migration during early neural development. Develop- Ioannidis P, Mahaira L, Papadopoulou A, et al: 24. Pugh TJ, Morozova O, Attiyeh EF, et al: The 37. ment 130:5649-5661, 2003 8q24 copy number gains and expression of the 10. Farina KL, Huttelmaier S, Musunuru K, et al: genetic landscape of high-risk neuroblastoma. Nat Genet 45:279-284, 2013 C-MYC mRNA stabilizing protein CRD-BP in primary Two ZBP1 KH domains facilitate beta-actin mRNA breast carcinomas. Int J Cancer 104:54-59, 2003 localization, granule formation, and cytoskeletal at- 24a. Cheung NK, Dyer MA: Neuroblastoma: De- 38. Findeis-Hosey JJ, Xu H: The use of insulin tachment. J Cell Biol 160:77-87, 2003 velopmental biology, cancer genomics and immuno- like-growth factor II messenger RNA binding 11. Nishino J, Kim S, Zhu Y, et al: A network of therapy. Nat Rev Cancer 13:397-411, 2013 protein-3 in diagnostic pathology. Hum Pathol 42: heterochronic genes including Imp1 regulates tem- 24b. Brodeur GM, Iyer R, Croucher JL, et al: 303-314, 2011 poral changes in stem cell properties. Elife Therapeutic targets for neuroblastoma. Expert Opin 39. Boyerinas B, Park SM, Shomron N, et al: 2:e00924, 2013 Ther Targets 18:277-292, 2014 Identification of let-7-regulated oncofetal genes. 12. Stöhr N, Köhn M, Lederer M, et al: IGF2BP1 25. Tessier CR, Doyle GA, Clark BA, et al: Mam- Cancer Res 68:2587-2591, 2008 promotes cell migration by regulating MK5 and mary tumor induction in transgenic mice expressing Molenaar JJ, Domingo-Fernández R, Ebus PTEN signaling. Genes Dev 26:176-189, 2012 an RNA-binding protein. Cancer Res 64:209-214, 40. 13. Lemm I, Ross J: Regulation of c-myc mRNA 2004 ME, et al: LIN28B induces neuroblastoma and en- decay by translational pausing in a coding region 26. De Preter K, Speleman F, Combaret V, et al: hances MYCN levels via let-7 suppression. Nat instability determinant. Mol Cell Biol 22:3959-3969, Quantification of MYCN, DDX1, and NAG gene copy Genet 44:1199-1206, 2012 2002 number in neuroblastoma using a real-time quanti- 41. El-Badry OM, Romanus JA, Helman LJ, et al: 14. Kohl NE, Legouy E, DePinho RA, et al: Human tative PCR assay. Mod Pathol 15:159-166, 2002 Autonomous growth of a human neuroblastoma cell N-myc is closely related in organization and nucleo- 27. Vandesompele J, Speleman F, Van Roy N, et line is mediated by insulin-like growth factor II. J Clin tide sequence to c-myc. Nature 319:73-77, 1986 al: Multicentre analysis of patterns of DNA gains and Invest 84:829-839, 1989 15. Maris JM: Recent advances in neuroblas- losses in 204 neuroblastoma tumors: How many 42. Weigel B, Malempati S, Reid JM, et al: Phase toma. N Engl J Med 362:2202-2211, 2010 genetic subgroups are there? Med Pediatr Oncol 2 trial of cixutumumab in children, adolescents, and 16. Bown N, Lastowska M, Cotterill S, et al: 17q 36:5-10, 2001 young adults with refractory solid tumors: A report gain in neuroblastoma predicts adverse clinical out- 28. National Cancer Institute: Office of Cancer from the Children’s Oncology Group. Pediatr Blood come. Med Pediatr Oncol 36:14-19, 2001 Genomics: TARGET—Therapeutically Applicable Re- Cancer 61:452-456, 2014

■■■

AUTHORS’ DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST IGF2BP1 Harbors Prognostic Signiﬁcance by Gene Gain and Diverse Expression in Neuroblastoma The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated. Relationships are self-held unless noted. I ϭ Immediate Family Member, Inst ϭ My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO’s conﬂict of interest policy, please refer to www.asco.org/rwc or jco.ascopubs.org/site/ifc. Jessica L. Bell Johannes H. Schulte No relationship to disclose No relationship to disclose Raseswari Turlapati Stefan Hüttelmaier No relationship to disclose No relationship to disclose Tao Liu No relationship to disclose

Acknowledgment We thank Frank Berthold, Barbara Hero, and staff at Cologne Neuroblastoma Tumor Bank (Germany) for organizing tumor specimens for this research and Tobias Irmscher of Martin Luther University for assistance with the chemotherapy-resistance ﬁndings. We much appreciated the use of the R2 visualization platform (Versteeg and Kocak data sets) and TARGET initiative tools for our work.