1 N-Myc Regulates Expression of the Detoxifying Enzyme Glutathione

Author Manuscript Published OnlineFirst on December 27, 2011; DOI: 10.1158/0008-5472.CAN-11-1885 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.

N-Myc regulates expression of the detoxifying enzyme glutathione transferase GSTP1, a marker of poor outcome in neuroblastoma

Jamie I. Fletcher1, Samuele Gherardi2, Jayne Murray1, Catherine A. Burkhart1,3, Amanda Russell1, Emanuele Valli2, Janice Smith1, André Oberthuer4, Lesley J. Ashton1, Wendy B. London5, Glenn M. Marshall1, Murray D. Norris1, Giovanni Perini2, Michelle Haber1.

1Children’s Cancer Institute Australia for Medical Research, Lowy Cancer Research Centre Randwick, NSW, Australia. 2Department of Biology, University of Bologna, Bologna, Italy. 3Current address: Cleveland BioLabs, Inc, Buffalo, New York, USA. 4Children’s Hospital, Department of Pediatric Oncology and Hematology, University of Cologne and Center for Molecular Medicine Cologne, Cologne, Germany. 5Children’s Oncology Group Statistics and Data Center, Dana-Farber Harvard Cancer Care and Children’s Hospital Boston, MA, USA.

Running title: Expression and regulation of GSTP1 in neuroblastoma

Key words: Multidrug resistance, tumor, chemotherapy, glutathione S-transferase, MYCN.

Abbreviations: Glutathione S-transferase P1 (GSTP1), ATP-binding cassette (ABC), real-time polymerase chain reaction (RT-PCR), chromatin immunoprecipitation (ChIP), peroxisome proliferator-activated receptor gamma (PPAR-γ), β2-microglobulin (B2M); tetracycline (tet).

Footnotes Financial support: National Health and Medical Research Council, Australia (JF, MDN, GMM, MH), Cancer Institute New South Wales (JF, MDN, GMM, MH), Cancer Council New South Wales (JF, MDN, GMM, MH), the Italian Association for Research on Cancer (GP), the Italian Ministry of University and Research (GP), the University of Bologna (GP) and the Children’s Oncology Group (USA). Corresponding authors: Michelle Haber, Children’s Cancer Institute Australia for Medical Research, Lowy Cancer Research Centre, PO Box 81, Randwick, New South Wales 2031, Australia. E-mail: [email protected], or Giovanni Perini, University of Bologna, Department of Biology, via F. Selmi 3, 40126 Bologna, Italy. E-mail: [email protected]. Conflict of interest: The authors declare no competing financial interests. 1

Notes: MDN, GP and MH contributed equally to this work. Children’s Cancer Institute Australia for Medical Research is affiliated with the University of New South Wales and Sydney Children’s Hospital, Randwick, Sydney, Australia. Word count = 4762; 4 figures; 1 table.

Abstract

Amplification of the transcription factor MYCN is associated with poor outcome and a multidrug resistant phenotype in neuroblastoma. N-Myc regulates the expression of several ATP-binding cassette (ABC) transporter genes, thus affecting global drug efflux. Because these transporters do not confer resistance to several important cytotoxic agents used to treat neuroblastoma, we explored the prognostic significance and transcriptional regulation of the phase II detoxifying enzyme, glutathione S-transferase P1 (GSTP1). Using quantitative real-time PCR, GSTP1 gene expression was assessed in a retrospective cohort of 51 patients, and subsequently in a cohort of 207 prospectively accrued primary neuroblastomas. These data along with GSTP1 expression data from an independent microarray study of 251 neuroblastoma samples were correlated with established prognostic indicators and disease outcome. High levels of GSTP1 were associated with decreased event-free and overall survival in all three cohorts. Multivariable analyses, including age at diagnosis, tumor stage and MYCN amplification status, were conducted on the two larger cohorts, independently demonstrating the prognostic significance of GSTP1 expression levels in this setting.

Mechanistic investigations revealed that GSTP1 is a direct transcriptional target of N-Myc in neuroblastoma cells. Together, our findings reveal that N-Myc regulates GSTP1 along with ABC transporters that act to control drug metabolism and efflux. Further, they imply that strategies to jointly alter these key multidrug resistance mechanisms may have therapeutic implications to manage neuroblastomas and other malignancies driven by amplified Myc family genes.

Introduction

Neuroblastoma is a solid tumor of embryonal neural crest origin and is the most common extracranial solid tumor of early childhood, accounting for 15% of all cancer related deaths in children (1). Patients diagnosed over one year of age often have disseminated, metastatic disease, often with MYCN amplification. While neuroblastomas typically respond to initial therapy regardless of risk group (2), the majority of patients with high-risk disease relapse with tumors that are refractory to treatment. There are no salvage regimens known to be curative for these patients

(3). We have previously demonstrated that the ATP binding cassette transporter C1 (ABCC1), a key component of phase III detoxification, is a powerful independent prognostic marker in neuroblastoma (4, 5). Consistent with this observation, ABCC1 is able to efflux a wide range of chemotherapeutics (6), including many agents of importance in neuroblastoma therapy. However, several important chemotherapeutic agents to which resistance is observed, including cisplatin and cyclophosphamide, are not substrates of ABCC1. Amongst the well-characterized contributors to multidrug resistance in other types of tumors are cellular detoxification pathways. In particular, the

Phase II conjugating enzyme glutathione S-transferase P1 (GSTP1) is up-regulated in a wide range of tumors and high GSTP1 expression is often associated with multidrug resistance (7-9). GSTP1 catalyses the conjugation of reduced glutathione to otherwise harmful electrophilic compounds, including cytotoxic drugs, and can sequester a range of non-substrate ligands (10). In addition to these roles, GSTP1 has been proposed to inhibit the mitogen-activated protein kinase (MAPK) pathway through direct interaction with c-Jun N-terminal kinase 1 (JNK1), decreasing sensitivity to drug-induced apoptosis (11). We have therefore extended our studies to investigate the prognostic value of this gene in neuroblastoma.

Here we demonstrate that GSTP1 expression is an independent prognostic indicator of clinical outcome in primary neuroblastoma and, like ABCC1, is a direct transcriptional target of the

N-Myc oncoprotein. The regulation of multiple components of drug metabolism pathways has implications for understanding drug insensitivity in neuroblastoma patients with MYCN

Materials and Methods

Patient tumor samples

All patient material used for gene expression analyses has been previously described (5, 12).

In brief, a small discovery cohort of 51 primary tumors from previously untreated patients was obtained from either the Sydney Children’s Hospital, Sydney, Australia, or the Neuroblastoma

Tumor Bank of the Pediatric Oncology Group, USA (“SCH/POG” cohort) (12) and a large validation cohort of 207 patients enrolled onto COG Neuroblastoma Biology Study 9047 was obtained from the Tumor Bank of the Children’s Oncology Group, USA (“COG” cohort) (5). The study was approved by the individual institutional review boards, and informed consent was obtained for patients registered on the study. Publicly available data for GSTP1 expression from a customized oligonucleotide microarray dataset of primary tumors from 251 neuroblastoma patients from the German Neuroblastoma Trials NB90-NB2004 were also analysed as a second independent validation cohort (“Oberthuer cohort”; GSTP1 probe = A_23_P202658). The patient cohort and microarray study are previously published (13). A summary of patient data for each cohort is provided as Supplementary data Tables 1, 2 and 3.

Treatment

Treatments subsequently administered to patients were specific to their disease stage, age and tumor biology and have been previously described for each of the SCH/POG (4), COG (5) and

Oberthuer cohorts (14).

TH-MYCN mouse tumor samples

The TH-MYCN transgenic mouse model of neuroblastoma, which has the expression of the human MYCN oncogene targeted to mouse neuroectodermal cells via the rat tyrosine hydroxylase promoter, has been previously described (15-17). When medium palpable tumors (approximately 1 cm2) were detected, they were resected, measured, and snap-frozen in liquid nitrogen for subsequent analysis. All experimental procedures were approved by the University of New South Wales

Animal Care and Ethics Committee and were conducted under the Animal Research Act, 1985

(New South Wales, Australia) and the Australian Code of Practice for the Care and Use Animals for

Scientific Purposes (1997).

RT-PCR methods

For tumor samples, total cytoplasmic RNA was isolated from human and mouse tumor material using a guanidinium thiocyanate-phenol-chloroform extraction method (4) and cDNAs were synthesized from 1 μg of RNA using random hexanucleotide primers and Moloney murine leukemia virus reverse transcriptase. Gene expression was determined by real-time polymerase chain reaction (RT-PCR) using either a Prism 7700 or 7900 Sequence Detection System (Applied

Biosystems, Foster City, CA) with the following oligonucleotide primers and probes: human

GSTP1, beta2-microglobulin (B2M) and MYCN, and mouse Gstp1, Gstp2 (the two mouse genes corresponding to human GSTP1) and B2m. All sequences are shown in Supplementary data Table 4.

For cell lines, RNA samples were prepared using Tri-Reagent (Sigma-Aldrich, St Louis,

MO) and treated with DNase (DNAfree; Ambion, Austin, TX). Reverse transcription and PCR were performed using a SuperScript reverse transcription-PCR kit (Invitrogen, Carlsbad, CA). RT-PCR was performed using iQ SYBR Green Supermix and the iQCycler thermocycler (Bio-Rad,

Hercules, CA) with primers and probes for human MYCN, GSTP1, NUC and GUSB. All sequences are shown in Supplementary data Table 4.

Gene expression levels were determined using the ΔΔCt method, normalized to the B2M control, and expressed relative to a calibrator (18).

Western blot analysis

Frozen tissue (approximately 50 mg) was pulverized on dry ice with a mortar and pestle in

500 μL lysis buffer (2 x PBS, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS) containing a protease inhibitor cocktail (Sigma-Aldrich, St Louis, MO). Samples were centrifuged at 13,000

RPM in a benchtop centrifuge and the supernatants retained and quantitated by BCA assay (Pierce,

Rockford, IL). Samples (20 μg) were electrophoresed on a 4-20% SDS-PAGE gel and transferred to nitrocellulose membrane (GE Healthcare, Rydalmere, NSW, Australia). The membrane was blocked with 5% skim milk powder in Tris buffered saline with 0.05% Tween 20, then incubated with an anti-GSTP1 antibody (mouse monoclonal 3F2C2, Abcam, Cambridge, UK) followed by a horseradish peroxidase conjugated secondary antibody. Membranes were developed using

Supersignal reagent (Pierce, Rockford, IL), visualized on film and quantitated relative to α-tubulin by densitometry using QuantityOne software (BioRad, Hercules, CA).

Human neuroblastoma cell lines and culture

SH-EP Tet-21N cells (19) were cultured in Dulbecco’s modified Eagle’s medium containing

10% heat-inactivated tetracycline-free fetal bovine serum, while BE(2)-C and LA-N-5 cells were cultured in RPMI medium 1640 containing 20% fetal bovine serum and 50 mg/mL gentamycin. For

MYCN repression, SH-EP Tet-21N cells were treated with 2μg/mL tetracycline for the indicated time. Cell lines were obtained from: Manfred Schwab (SH-EP Tet-21N), German Collection of

Microorganisms and Cell Cultures (BE(2)-C) and Deborah Tweddle (LA-N-5) between 1997 and

2005, and were systematically validated for MYCN copy number and expression of specific markers such as MYCN, upon receipt and during usage by dr. Giuliano Della Valle who runs the cytogenetic facility, Department of Biology, University of Bologna, Italy.

Chromatin immunoprecipitation (ChIP) assays

The promoter region of GSTP1 was analysed using bioinformatics techniques to identify putative N-Myc binding sites, including the canonical E-box sequence CACGTG and other non- canonical sequences such as CATGTG and CACGCG. Promoter regions from -2000 to +2000 base pairs with respect to the transcription start site were considered for analysis. To test binding of N-

Myc, quantitative ChIP was performed as previously described (20, 21) using either SH-EP Tet-

21N cells cultured in the absence of tetracycline (presence of N-Myc), BE(2)-C or LA-N-5 cells.

Antibodies employed in this study were: IgG (Santa Cruz Biotechnology, Santa Cruz, CA, sc-

2027); N-Myc monoclonal antibody B8.4.B (22) and Max C-17 (Santa Cruz Biotechnology, sc-

197). Specific pairs of primers used for quantitative ChIP are listed in Supplementary data Table 5.

Statistical analysis

The relationship between levels of target gene expression and patient survival was analyzed using the Kaplan-Meier method with curves analyzed by log-rank test. Standard errors were calculated according to the method of Peto and Peto (23). Comparisons of outcome between subgroups in multivariable analyses were performed using Cox’s proportional hazards regression model. The relationship between levels of target gene expression was analysed using Spearman’s rank correlation. Differences in target gene expression between samples were assessed using a two- sided Student’s t test.Statistical analyses were conducted using either Stata (StataCorp, College

Station, TX) or SPSS (IBM, Chicago, IL). Event Free Survival (EFS) time was calculated from the time of enrolment onto the relevant study, or from diagnosis (SCH samples), to the time of the first occurrence of an event (relapse, progressive disease, secondary malignancy, or death) or to the date of last contact if no event occurred. Death was the only event considered for the calculation of overall survival time. All samples were de-identified, and researchers who performed the gene expression analysis were blinded to all clinical characteristics and outcome of the subjects.

Tumors were categorized as having high expression of GSTP1 based on a range of cut points, including the median, upper quartile, and upper decile PCR values. The statistical

Results

Prognostic significance of GSTP1 in primary human neuroblastoma

In order to assess the prognostic significance of GSTP1 expression in neuroblastoma, we initially used RT-PCR to analyse its expression in a small cohort of 51 primary untreated tumors.

When patients were dichotomized around the median real-time PCR value, high expression of

GSTP1 was significantly associated with poorer event free survival (P = 0.002, Figure 1A) with 5- year survival rates of 92% ± 5.4% and 51% ± 11% for tumors with low and high GSTP1 expression, respectively, and also significantly associated with poorer overall survival (P = 0.006, Figure 1B), with 5-year survival rates of 96% ± 3.9% and 63% ± 10% for tumors with low and high GSTP1 expression, respectively. Furthermore, GSTP1 expression was significantly higher in tumors with

MYCN amplification than in non-amplified tumors (Figure 1C, P = 0.024). To verify that GSTP1 gene expression accurately reflected expression at the protein level, we conducted Western blotting on six representative tumor samples, three with high GSTP1 gene expression and three with low gene expression (Figure 1D). Protein levels were found to correlate strongly with gene expression in these samples (Figure 1E; r = 0.96, P = 0.002). Full-length Western blots are presented in

Supplementary Figure S1.

Next, we extended these observations to two large patient cohorts. In the first, we analysed

GSTP1 expression by RT-PCR in 207 primary tumors (COG cohort). In the second, we analysed publicly available microarray data (13) for GSTP1 expression in 251 primary tumors (Oberthuer cohort).

In the COG cohort, high GSTP1 expression was associated with worse outcome when patients were dichotomized around either the median or upper quartile real-time PCR values, although this failed to reach statistical significance at these cut-points (not shown). However at the

Figure 2A). The 5-year EFS rate for patients with high GSTP1 expression was 52% ± 11%, compared with 75% ± 3.3% for low GSTP1 expression. The prognostic significance of GSTP1 was also examined for patients with tumors lacking MYCN amplification. While these patients would be expected to have a relatively favourable outcome compared to patients with MYCN amplification, high levels of GSTP1 expression remained predictive of poorer outcome with a 5-year EFS rate of

60% ± 13% compared to 80% ± 3% for patients with low GSTP1 expression in their tumors (P =

0.025; Figure 2B). Similarly, within poor prognosis patients (INSS stage 3 and 4), high levels of

GSTP1 expression identified a subset with particularly poor outcome and a 5-year EFS rate of only

31% ± 14% compared to 52% ± 6% for patients compared to patients with low GSTP1 expression

(P = 0.022; Figure 2C). In each case, similar effects were observed for overall survival (not shown).

Within the Oberthuer cohort, GSTP1 was strongly associated with EFS at the upper decile

(P < 0.001; Figure 2D), and also at the upper quartile and median (not shown). At the upper decile, the 5-year EFS rates for patients with high GSTP1 expression were 24% ± 13% compared with 74%

± 3% for low GSTP1 expression. In patients with tumors lacking MYCN amplification, 5-year EFS rates were 26% ± 20% and 78% ± 3% for those with high and low GSTP1 expression respectively

(P < 0.001, Figure 2E). For those with poor prognosis (stage 3,4), high GSTP1 expression was associated with significantly worse outcome with 5-year EFS rates of 15% ± 12% and 55% ± 6% for high and low GSTP1 expression respectively (P = 0.007; Figure 2F). Within the stage 4 patients only, EFS was not significantly different for those with low and high GSTP1 expression in their tumors (P = 0.227), however overall survival rates retained significance (P = 0.049) with 5 year survival rates of 51% ± 8.5% and 37% ± 14% for high and low GSTP1 expression respectively

(Supplementary Figure S2).

Multivariable analysis was performed to determine whether GSTP1 expression had prognostic significance independent of established prognostic indicators for neuroblastoma. Tumor stage (1, 2, or 4S v 3 or 4), age at diagnosis (<12 mo v 12 mo), MYCN status (amplified v single

(Table 1). In the COG cohort, GSTP1 expression retained independent prognostic significance for

EFS (hazard ratio = 2.6; 95% CI 1.2 to 5.5; P = 0.011) and overall survival (hazard ratio = 2.21;

95% CI 1.0 to 4.8; P = 0.047). Likewise in the Oberthuer cohort, GSTP1 expression retained prognostic significance for both EFS (relative risk = 2.0; 95% CI 1.1 to 3.8; P = 0.029) and overall survival (relative risk = 2.3; 95% CI 1.1 to 4.9; P = 0.024). Finally, in a multivariable analysis for stage 4 patients only, including GSTP1 expression and age, GSTP1 expression approached significance for overall survival (relative risk = 2.1; 95% CI 0.99 to 4.6; P = 0.054), but was not significant for event-free survival (Supplementary data Table 6).

Expression of GSTP1 correlates with MYCN in neuroblastoma tumors

As amplification of the transcription factor MYCN is a common feature of poor outcome in neuroblastoma, and GSTP1 expression was higher in MYCN-amplified tumors than in those without

MYCN amplification (Figure 1C), we sought to determine whether GSTP1 expression was regulated by MYCN. First, we assessed the correlation between GSTP1 expression and MYCN expression in the COG cohort after conducting RT-PCR analysis to determine relative MYCN expression levels.

Using Spearman’s rank correlation, GSTP1 levels were found to correlate with MYCN expression in both tumors with MYCN-amplification (r = 0.61, P = 0.003; Figure 3A) and those without (r =

0.73, P < 0.001; Figure 3B). We also extended this analysis to the TH-MYCN transgenic mouse model of neuroblastoma (15), where the human MYCN transgene is expressed in cells of neuroectodermal origin by use of a tyrosine hydroxylase promoter. Using RT-PCR analysis, MYCN levels were found to correlate particularly closely with the expression of mouse Gstp1 (r = 0.91, P <

0.001; Figure 3C). Similar results were observed for Gstp2 (not shown). Finally, we assessed the expression of GSTP1 using RT-PCR at various time points after tetracycline-induced suppression of exogenous MYCN expression in the neuroblastoma cell line SH-EP Tet-21N (19). GSTP1 expression was decreased after tetracycline exposure, along with MYCN (not shown).

Collectively, these data support regulation of GSTP1 expression by N-Myc in cultured neuroblastoma cell lines, human neuroblastoma tumors, and tumors from a biologically relevant mouse model of neuroblastoma.

N-Myc directly regulates GSTP1 expression

In light of these results, we assessed whether N-Myc directly regulates GSTP1 expression.

The promoter region of GSTP1 was examined bioinformatically for putative N-Myc-binding sites within 2000bp on either side of the transcriptional start site. Two canonical (CACGTG) and four non-canonical (CACGCG, CATGTG) E-box sites were identified over five distinct regions of the promoter (Figure 4A). Next, we used ChIP assays to examine the physical association of N-Myc with each of the identified regions using the three independent human neuroblastoma cell lines SH-

EP Tet-21N, LA-N-5 and BE(2)-C. In each cell line, the GSTP1 promoter was directly bound by N-

Myc, as well as its dimerization partner Max, at a non-canonical E-box proximal to the putative transcriptional start site (Fig. 4A). The fold enrichment for GSTP1 was comparable to that of

APEX1, a well-known N-Myc target gene (25) indicating that the N-Myc/Max complex is strongly bound to the promoter.

Finally, to determine whether N-Myc directly activates GSTP1 transcription, we cloned the

GSTP1 promoter upstream of a luciferase reporter gene and tested activity in SH-EP Tet-21N cells as a function of N-Myc expression. Luciferase activity was found to be dependent on N-Myc expression for the GSTP1 reporter construct, as for ABCC1, a positive control (26) (Figure 4B).

Together, these data demonstrate that GSTP1 is a direct transcriptional target of N-Myc in neuroblastoma cells.

Discussion

Using three independent patient cohorts, we have shown that high expression of GSTP1 is a marker of poor prognosis in primary neuroblastoma. Furthermore, when included in multivariable

Downloaded from cancerres.aacrjournals.org on October 1, 2021. © 2011 American Association for Cancer Research. Author Manuscript Published OnlineFirst on December 27, 2011; DOI: 10.1158/0008-5472.CAN-11-1885 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited. analyses with the established prognostic indicators of age at diagnosis, MYCN amplification status and tumor stage as variables, GSTP1 expression retained independent prognostic significance. This result raises the possibility that GSTP1 contributes to the clinical behaviour of neuroblastoma, consistent with its known roles in drug resistance and tumor biology. While neuroblastomas typically respond to initial therapy regardless of risk group (2), the majority of patients with high- risk disease relapse with tumors that are refractory to treatment. It is as yet unclear whether this reflects the acquisition of multidrug resistance following exposure to chemotherapy, as a result of induction of one or more of the mechanisms which have previously been implicated as mediating multidrug resistance in neuroblastoma, such as MRP1 over-expression (5) or p53 mutation (27), or the selection and expansion of pre-existent multidrug resistant tumor clones, as we have previously described for acute lymphoblastic leukaemia (28). Future studies involving matched pairs of pre- and post-treatment tumor samples would be valuable to determine whether levels of GSTP1 or other drug-resistance associated genes are elevated at relapse compared to levels at the time of diagnosis.

We also provide evidence that GSTP1 is a direct transcriptional target of N-Myc. Firstly, expression of GSTP1 correlates with that of MYCN in both human tumors and the tumors derived from the TH-MYCN mouse neuroblastoma model. Secondly, the N-Myc protein, in complex with its dimerization partner Max, directly interacts with a non-canonical E-box proximal to the putative transcriptional start site of GSTP1. Finally, repression of MYCN expression reduces activity in a luminescent reporter assay using the GSTP1 promoter. Notably, both the multidrug transporter

ABCC1, a Phase III detoxification component that can efflux a number of glutathione conjugated chemotherapeutic agents (29) and ABCC4, which can efflux both irinotecan and topotecan (12), are also independently prognostic of neuroblastoma outcome (4, 5, 12, 30) as well as directly regulated by N-Myc (26). MYCN amplification can therefore contribute to multidrug resistance in neuroblastoma through the upregulation of a suite of drug resistance genes covering a range of chemotherapeutic drugs. While previous in vitro studies suggest co-operativity between ABCC1 and GSTs (31, 32), GSTP1 has a relatively weak affinity for many anti-cancer drugs (7), implying

It is notable that of the two cohorts analyzed by RT-PCR, GSTP1 expression was significantly associated with poor outcome when dichotomized around the median RT-PCR value in the initial SCH/POG 51 sample cohort, while in the larger COG cohort, the association between high GSTP1 expression and poor outcome was evident but not significant at this cut-point and in fact did not achieve prognostic significance until the data was dichotomized around the upper decile

RT-PCR value. One contributing factor for this difference is most likely due to the differences between the patient characteristics of the two cohorts. Specifically, in the retrospectively accrued

SCH/POG cohort of 51 tumors, 23.5% of tumors were MYCN-amplified and 55% were INSS stage

3 or 4 (12), whilst in the prospectively accrued COG cohort (5), only 10.1% of tumors were MYCN- amplified and 42% were INSS stage 3 or 4. The under-representation of MYCN amplified, advanced stage tumors in the COG cohort can be related to the frequent diagnosis of stage 4 disease based on bone marrow sampling rather than tumor biopsy in that cohort, as we have previously described (5).

As GSTP1 is transcriptionally regulated by MYCN, and given the strong correlation between MYCN amplification and GSTP1 overexpression, it is therefore not surprising that GSTP1 expression is more predictive of outcome in a cohort with a higher proportion of MYCN-amplified tumors and less predictive in a cohort where MYCN-amplified tumors are under-represented. Indeed, those

MYCN amplified patients lost to the COG cohort would most likely only further strengthen the association between high GSTP1 expression and poor patient outcome evident in that cohort. The distributions of RT-PCR values for each cohort are shown in Supplementary Figure S3.

The identification of GSTP1 as an N-Myc target gene also has importance for malignancies with high expression or amplification of c-Myc. N-Myc and c-Myc share a common set of target genes and are functionally interchangeable in the mouse (33). Furthermore, GSTP1 was one of a large number of genes identified as potential c-Myc transcriptional targets in a high-throughput

ChIP study (25) and GSTP1 is transcriptionally regulated by c-Myc in osteosarcoma cell lines (GP,

While the most familiar role for GSTP1 is the conjugation of reduced glutathione to cytotoxic drugs, GSTP1 may possess other functions that contribute to chemoresistance, including inhibition of JNK1 (34), and activation of extracellular signal regulated kinase (ERK) (35). In addition to potential roles in mediating drug resistance in neuroblastoma, GSTP1 could potentially impact other tumor properties through its ability to conjugate the cyclopentenone prostaglandins

12,14 15d-Δ -PGJ2 and PGA2 (36), leading to their export through ABCC1 and ABCC3 (37), and its

12,14 ability to sequester 15d-Δ -PGJ2 in the cytosol independently of conjugation (38). Both mechanisms may prevent cyclopentenone prostaglandins from binding to the nuclear localized gamma isoform of the peroxisome proliferator-activated receptor (PPAR-γ), activation of which is anti-proliferative, pro-differentiation and pro-apoptotic (39) in cancer cells. In neuroblastoma,

12,14 activation of PPAR-γ by 15d-Δ -PGJ2 has been shown to induce apoptosis (40) and differentiation (41). While we cannot exclude these additional functions in our experiments, analysis of microarray data available for the Oberthuer cohort did not reveal any evidence of an association between GSTP1 expression and that of JNK1 (MAPK8), ERK1 (MAPK1) or PPARγ

(PPARG; Supplementary Figure S4), or with pathways associated with these genes (data not shown).

GSTP1 is up-regulated in numerous human tumors and high GSTP1 expression is associated with chemoresistance (7-9). To our knowledge, this is the first study to reveal the prognostic significance of GSTP1 expression in neuroblastoma and the first to demonstrate that GSTP1 is a direct transcriptional target of the MYCN oncogene, which is frequently amplified in poor-outcome neuroblastoma. These observations have significance both for MYCN-amplified neuroblastoma and for other diseases with amplification of Myc family oncogenes.

References

1. Maris JM, Hogarty MD, Bagatell R, Cohn SL. Neuroblastoma. Lancet. 2007;369:2106-20. 2. Brodeur GM, Maris JM. Neuroblastoma. In: Pizzo PA, Poplack DG, editors. Principles and Practices of Pediatric Oncology. 5th ed. Philadelphia: JB Lippincott Company; 2006. p. 933-70. 3. Maris JM. Recent advances in neuroblastoma. N Engl J Med. 2010;362:2202-11. 4. Norris MD, Bordow SB, Marshall GM, Haber PS, Cohn SL, Haber M. Expression of the gene for multidrug-resistance-associated protein and outcome in patients with neuroblastoma. N Engl J Med. 1996;334:231-8. 5. Haber M, Smith J, Bordow SB, Flemming C, Cohn SL, London WB, et al. Association of high-level MRP1 expression with poor clinical outcome in a large prospective study of primary neuroblastoma. J Clin Oncol. 2006;24:1546-53. 6. Deeley RG, Cole SP. Substrate recognition and transport by multidrug resistance protein 1 (ABCC1). FEBS Lett. 2006;580:1103-11. 7. Laborde E. Glutathione transferases as mediators of signaling pathways involved in cell proliferation and cell death. Cell Death Differ. 2010;17:1373-80. 8. Ekhart C, Rodenhuis S, Smits PH, Beijnen JH, Huitema AD. An overview of the relations between polymorphisms in drug metabolising enzymes and drug transporters and survival after cancer drug treatment. Cancer Treat Rev. 2009;35:18-31. 9. Lo HW, Ali-Osman F. Genetic polymorphism and function of glutathione S-transferases in tumor drug resistance. Curr Opin Pharmacol. 2007;7:367-74. 10. Hayes JD, Flanagan JU, Jowsey IR. Glutathione transferases. Annu Rev Pharmacol Toxicol. 2005;45:51-88. 11. Townsend DM, Tew KD. The role of glutathione-S-transferase in anti-cancer drug resistance. Oncogene. 2003;22:7369-75. 12. Norris MD, Smith J, Tanabe K, Tobin P, Flemming C, Scheffer GL, et al. Expression of multidrug transporter MRP4/ABCC4 is a marker of poor prognosis in neuroblastoma and confers resistance to irinotecan in vitro. Mol Cancer Ther. 2005;4:547-53. 13. Oberthuer A, Berthold F, Warnat P, Hero B, Kahlert Y, Spitz R, et al. Customized oligonucleotide microarray gene expression-based classification of neuroblastoma patients outperforms current clinical risk stratification. J Clin Oncol. 2006;24:5070-8. 14. Simon T, Berthold F, Borkhardt A, Kremens B, De Carolis B, Hero B. Treatment and outcomes of patients with relapsed, high-risk neuroblastoma: results of German trials. Pediatr Blood Cancer. 2011;56:578-83. 15. Weiss W, Aldape K, Mohapatra G, Feuerstein B, Bishop J. Targeted expression of MYCN causes neuroblastoma in transgenic mice. EMBO Journal. 1997;16:2985-95. 16. Norris MD, Burkhart CA, Marshall GM, Weiss WA, Haber M. Expression of N-myc and MRP genes and their relationship to N-myc gene dosage and tumor formation in a murine neuroblastoma model. Med Pediatr Oncol. 2000;35:585-9. 17. Burkhart CA, Cheng AJ, Madafiglio J, Kavallaris M, Mili M, Marshall GM, et al. Effects of MYCN antisense oligonucleotide administration on tumorigenesis in a murine model of neuroblastoma. J Natl Cancer Inst. 2003;95:1394-403. 18. Winer J, Jung CKS, Shackel I, Williams PM. Development and Validation of Real-Time Quantitative Reverse Transcriptase-Polymerase Chain Reaction for Monitoring Gene Expression in Cardiac Myocytes in Vitro. Analyt Biochem. 1999;270:41-9. 19. Lutz W, Stohr M, Schurmann J, Wenzel A, Lohr A, Schwab M. Conditional expression of N-myc in human neuroblastoma cells increases expression of alpha-prothymosin and ornithine decarboxylase and accelerates progression into S-phase early after mitogenic stimulation of quiescent cells. Oncogene. 1996;13:803-12. 20. Porro A, Perini G. In vivo dual cross-linking chromatin immunoprecipitation: detecting chromatin proteins not directly bound to DNA. Epigenome Network of Excellence: Section Protocols. 2007;http://www.epigenome-noe.net. 16

21. Weinmann AS, Farnham PJ. Identification of unknown target genes of human transcription factors using chromatin immunoprecipitation. Methods. 2002;26:37-47. 22. Wenzel A, Cziepluch C, Hamann U, Schurmann J, Schwab M. The N-Myc oncoprotein is associated in vivo with the phosphoprotein Max(p20/22) in human neuroblastoma cells. EMBO J. 1991;10:3703-12. 23. Peto R, Peto J. Asymptotically efficient rank invariant test procedures. J R Stat Soc A. 1972;135:185-98. 24. London WB, Castleberry RP, Matthay KK, Look AT, Seeger RC, Shimada H, et al. Evidence for an age cutoff greater than 365 days for neuroblastoma risk group stratification in the Children's Oncology Group. J Clin Oncol. 2005;23:6459-65. 25. Fernandez PC, Frank SR, Wang L, Schroeder M, Liu S, Greene J, et al. Genomic targets of the human c-Myc protein. Genes Dev. 2003;17:1115-29. 26. Porro A, Haber M, Diolaiti D, Iraci N, Henderson M, Gherardi S, et al. Direct and coordinate regulation of ATP-binding cassette transporter genes by Myc factors generates specific transcription signatures that significantly affect the chemoresistance phenotype of cancer cells. J Biol Chem. 2010;285:19532-43. 27. Xue C, Haber M, Flemming C, Marshall GM, Lock RB, MacKenzie KL, et al. p53 determines multidrug sensitivity of childhood neuroblastoma. Cancer Res. 2007;67:10351-60. 28. Choi S, Henderson MJ, Kwan E, Beesley AH, Sutton R, Bahar AY, et al. Relapse in children with acute lymphoblastic leukemia involving selection of a preexisting drug-resistant subclone. Blood. 2007;110:632-9. 29. Cole SP, Deeley RG. Transport of glutathione and glutathione conjugates by MRP1. Trends Pharmacol Sci. 2006;27:438-46. 30. Henderson MJ, Haber M, Porro A, Munoz MA, Iraci N, Xue C, et al. ABCC multidrug transporters in childhood neuroblastoma: clinical and biological effects independent of cytotoxic drug efflux. J Natl Cancer Inst. 2011;103:1236-51. 31. Morrow CS, Smitherman PK, Townsend AJ. Combined expression of multidrug resistance protein (MRP) and glutathione S-transferase P1-1 (GSTP1-1) in MCF7 cells and high level resistance to the cytotoxicities of ethacrynic acid but not oxazaphosphorines or cisplatin. Biochem Pharmacol. 1998;56:1013-21. 32. Meijerman I, Beijnen JH, Schellens JH. Combined action and regulation of phase II enzymes and multidrug resistance proteins in multidrug resistance in cancer. Cancer Treat Rev. 2008;34:505-20. 33. Malynn BA, de Alboran IM, O'Hagan RC, Bronson R, Davidson L, DePinho RA, et al. N- myc can functionally replace c-myc in murine development, cellular growth, and differentiation. Genes Dev. 2000;14:1390-9. 34. Adler V, Yin Z, Fuchs SY, Benezra M, Rosario L, Tew KD, et al. Regulation of JNK signaling by GSTp. EMBO J. 1999;18:1321-34. 35. Yin Z, Ivanov VN, Habelhah H, Tew K, Ronai Z. Glutathione S-transferase p elicits protection against H2O2-induced cell death via coordinated regulation of stress kinases. Cancer Res. 2000;60:4053-7. 36. Bogaards JJ, Venekamp JC, van Bladeren PJ. Stereoselective conjugation of prostaglandin A2 and prostaglandin J2 with glutathione, catalyzed by the human glutathione S-transferases A1-1, A2-2, M1a-1a, and P1-1. Chem Res Toxicol. 1997;10:310-7. 37. Paumi CM, Wright M, Townsend AJ, Morrow CS. Multidrug resistance protein (MRP) 1 and MRP3 attenuate cytotoxic and transactivating effects of the cyclopentenone prostaglandin, 15- deoxy-Delta(12,14)prostaglandin J2 in MCF7 breast cancer cells. Biochemistry. 2003;42:5429-37. 38. Paumi CM, Smitherman PK, Townsend AJ, Morrow CS. Glutathione S-transferases (GSTs) inhibit transcriptional activation by the peroxisomal proliferator-activated receptor gamma (PPAR gamma) ligand, 15-deoxy-delta 12,14prostaglandin J2 (15-d-PGJ2). Biochemistry. 2004;43:2345- 52.

39. Michalik L, Desvergne B, Wahli W. Peroxisome-proliferator-activated receptors and cancers: complex stories. Nat Rev Cancer. 2004;4:61-70. 40. Kim EJ, Park KS, Chung SY, Sheen YY, Moon DC, Song YS, et al. Peroxisome proliferator-activated receptor-gamma activator 15-deoxy-Delta12,14-prostaglandin J2 inhibits neuroblastoma cell growth through induction of apoptosis: association with extracellular signal- regulated kinase signal pathway. J Pharmacol Exp Ther. 2003;307:505-17. 41. Han SW, Greene ME, Pitts J, Wada RK, Sidell N. Novel expression and function of peroxisome proliferator-activated receptor gamma (PPARgamma) in human neuroblastoma cells. Clin Cancer Res. 2001;7:98-104.

Table 1

Event-free survival Overall survival Hazard Hazard Factor ratio 95% CI P ratio 95% CI P COG-209 Tumor stage 4.6 2.1 to 10.0 <0.001 9.8 3.4 to 28.7 <0.001 Age 2.4 1.1 to 5.1 0.025 2.5 1.1 to 6.1 0.037 MYCN status 1.8 0.9 to 3.6 0.086 2.1 1.0 to 4.1 0.04 GSTP1 expression 2.6 1.2 to 5.5 0.011 2.2 1.0 to 4.8 0.047 Oberthuer Tumor stage 2.7 1.6 to 4.8 <0.001 7.7 2.6 to 23.1 <0.001 Age 0.8 0.4 to 1.4 0.385 1.5 0.6 to 4.1 0.411 MYCN status 2.3 1.3 to 4.1 0.005 4.3 2.1 to 8.9 <0.001 GSTP1 expression 2.0 1.1 to 3.8 0.029 2.3 1.1 to 4.9 0.024

Variables adjusted for in the multivariable analysis included tumor stage (1, 2, 4S vs 3, 4), age at diagnosis (<12 months vs 12 months), MYCN status (amplified vs non-amplified), and GSTP1 expression (dichotomized at the upper decile low vs high).

Figure and table legends

Table 1: Multivariable Cox regression analysis of prognostic factors for neuroblastoma outcome.

Figure 1: GSTP1 is marker of poor-prognosis in neuroblastoma.

Kaplan-Meier curves for event free survival (A) and overall survival (B) in 51 NB patients from the

SCH/COG cohort according to the expression of GSTP1. The level of GSTP1 expression was determined by RT-PCR and tumors dichotomized around the median value. Survival curves were analysed by log-rank test. (C) GSTP1 is expressed at a higher level in MYCN-amplified tumors by comparison with those without MYCN amplification. Graph shows relative expression as determined by ΔΔCT method ± SD (see Materials and Methods) with P value determined by

Student’s t-test. (D) GSTP1 protein levels in six representative neuroblastoma tumors. Relative expression at mRNA level is shown above the blots, while protein expression relative to α-tubulin, as determined using densitometry, is shown below. The drug resistant neuroblastoma cell line

BE(2)-C is used as a positive control. (E) GSTP1 mRNA levels are closely correlated with GSTP1 protein levels in primary neuroblastoma tumors (r = 0.96, P = 0.002). The relationship was analysed using linear regression.

Figure 2: GSTP1 is prognostic for outcome in two large, independent neuroblastoma cohorts.

Kaplan-Meier curves for neuroblastoma patients from the COG and Oberthuer cohorts showing

EFS for the overall cohorts (A, D), patients with tumors lacking MYCN amplification (B, E), and those from patients of INSS stage 3 or 4 (C, F), according to the expression of GSTP1. In each case, tumors were dichotomized into “low” and “high” expression around the upper decile value.

Survival curves were analysed by log-rank test. Five year EFS rates are indicated.

Figure 3: GSTP1 gene expression correlates with that of MYCN.

Expression of GSTP1 correlates with that of MYCN in tumors from the COG cohort either with

MYCN-amplification (A), or without (B), while Gstp1 expression correlates with that of the MYCN transgene in tumors from the TH-MYCN mouse neuroblastoma model (C). The relationships were analysed using Spearman’s rank correlation.

Figure 4: GSTP1 is a direct transcriptional target of N-Myc.

(A) N-Myc associates with the GSTP1 promoter. Quantitative ChIP was applied to the neuroblastoma cell lines BE(2)-C, LA-N-5, and SH-EP Tet21/N cells expressing MYCN. Relative enrichment is compared to the pre-immune serum. The results represent the mean ±S.E. of five independent ChIP experiments in which each region was amplified by RT-PCR in triplicate. Bent arrow, transcription start site; gray arrow, canonical E-box; black arrow, noncanonical E-box; black boxes, amplicons. The chromosome and coordinates (bp) are also given. Quantitative ChIP for

APEX1 in BE(2)-C cells is also shown. (B) N-Myc regulates GSTP1 in a luciferase reporter assay.

Firefly luciferase activity was determined following transient transfection of either a GSTP1 or

ABCC1 reporter construct into SH-EP Tet21/N cells. Cells were either not treated (NT; MYCN

ON), or treated with tetracycline for 24 or 48 h (TET 24h, or TET 48h; MYCN OFF). Cloned regions are shown at left, with bent arrows denoting the transcription start site, the cloned DNA region (bp) indicated below the promoter map and fragment coordinates expressed relative to the transcription start point. The results are the means of the relative fluorescence units (RFU) ±S.E. of three independent transfections.

A B 100 Low (n=25) Low (n=25) 92% 100 96% 80 80 60 High (n=26) High (n=26) 51% 60 63% 40 EFS (%)

OS (%) 40 20 P = 0.002 20 0 P = 0. 006 0 0 1000 2000 3000 4000 0 1000 2000 3000 4000 Analysis time (d) Analysis time (d)

C DE primary tumors E(2)-C

B 3 5 6 46 57 90 mRNA

GSTP1

-tubulin

1 .03 .3 .16 .54 .681.4 protein

COG cohort Oberthuer cohort A D All patients All patients 100 100 Low (n=186) Low (n=227) 75 75% 75 74%

50 52% 50

FS (%) High (n=21) FS (%) High (n = 24) E 25 E 25 24% P P = 0.002 < 0.001 0 0 0 2000 4000 6000 0 2000 4000 6000 Analysis time (d) Analysis time (d)

B E No MYCN amplification No MYCN amplification

100 Low (n=169) 100 80% Low (n=202) 78% 75 75 60% High (n=15) 50 50 FS (%)

FS (%) High (n=15) E E 25 25 26% P = 0.025 P < 0.001 0 0 0 500 1000 1500 2000 2500 0 2000 4000 6000 Analysis time (d) Analysis time (d)

C F

Stage 3 and 4 patients Stage 3 and 4 patients 100 100

75 75 Low (n=74) Low (n=87) 55% 50 52% 50 EFS (%) 31% EFS (%) 25 25 Hig h (n= 18) High (n=12) 15% P = 0.022 P = 0.007 0 0 0 500 1000 1500 2000 2500 0 2000 4000 6000 Analysis time (d) Analysis time (d)

A Tumors with MYCN amplification r 150 = 0.61 P = 0.003 n = 21 100 ve expression) i i

50 (relat

0 GSTP1 0 50000 100000 150000 MYCN (ΔΔCT) B Tumors without MYCN amplification 200 r = 0.73 P < 0.001 150 n = 184 expression) 100

(relative 50

GSTP1 0 5000 10000 15000 MYCN (ΔΔCT)

C TH-MYCN mouse tumors

15 expression) 10 r = 0.91

(relative 5 P < 0.001 n = 20 0 MYCN 0 5 10 15 20 Gstp1 (relative expression)

Figure 4: GSTP1 is a direct transcriptional target of N-Myc

N-Myc regulates expression of the detoxifying enzyme glutathione transferase GSTP1, a marker of poor outcome in neuroblastoma

Jamie I Fletcher, Samuele Gherardi, Jayne Murray, et al.

Cancer Res Published OnlineFirst December 27, 2011.

Updated version Access the most recent version of this article at: doi:10.1158/0008-5472.CAN-11-1885

Supplementary Access the most recent supplemental material at: Material http://cancerres.aacrjournals.org/content/suppl/2011/12/27/0008-5472.CAN-11-1885.DC1

Author Author manuscripts have been peer reviewed and accepted for publication but have not yet been Manuscript edited.

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

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

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/early/2011/12/23/0008-5472.CAN-11-1885. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.