Oncogene (2008) 27, 1478–1488 & 2008 Nature Publishing Group All rights reserved 0950-9232/08 $30.00 www.nature.com/onc ONCOGENOMICS Deregulated Wnt/b-catenin program in high-risk neuroblastomas without MYCN amplification

X Liu1, P Mazanek1, V Dam1, Q Wang1, H Zhao2, R Guo2, J Jagannathan1, A Cnaan2, JM Maris1,3 and MD Hogarty1,3

1Division of Oncology, The Children’s Hospital of Philadelphia, Philadelphia, PA, USA; 2Department of Biostatistics and Epidemiology, University of Pennsylvania School of Medicine, Philadelphia, PA, USA and 3Department of Pediatrics, University of Pennsylvania School of Medicine, Philadelphia, PA, USA

Neuroblastoma (NB) is a frequently lethal tumor of Introduction childhood. MYCN amplification accounts for the aggres- sive phenotype in a subset while the majority have no Neuroblastoma (NB) is a childhood embryonal malig- consistently identified molecular aberration but frequently nancy arising in the peripheral sympathetic nervous express MYC at high levels. We hypothesized that acti- system. Half of all children with NB present with features vated Wnt/b-catenin (CTNNB1) signaling might account that define their tumorsashigh riskwith poor overall for this as MYC is a b-catenin transcriptional target and survival despite intensive therapy (Matthay et al., 1999). multiple embryonal and neural crest malignancies have A subset of these tumors are characterized by high-level oncogenic alterations in this pathway. NB cell lines without genomic amplification of the MYCN proto-oncogene MYCN amplification express higher levels of MYC and (Matthay et al., 1999) but the remainder have no b-catenin (with aberrant nuclear localization) than MYCN- consistently identified aberration to account for their amplified cell lines. Evidence for aberrant b-catenin–TCF aggressive phenotype. transcriptional activity was demonstrated using expression MycN isa member of the Myc family of genes( MYC, profiles from 73 primary NBs. Findings included increased MYCN, MYCL) that play a central role in diverse WNT ligands (WNT1, WNT6, WNT7A, WNT10B), cellular processes (Grandori et al., 2000) and their DVL1 and TCF7 expression in high-risk NBs without deregulated expression frequently contributes to neo- MYCN amplification, consistent with canonical b-catenin plasia (Cole and McMahon, 1999). NB-derived cell lines signaling. More directly, Patterns of Gene Expression and without MYCN amplification generally express MYC Gene Set Enrichment Analyses demonstrated b-catenin rather than MYCN, often at higher levelsthan normal target genes (for example, MYC, PPARD, NRCAM, tissues (Sadee et al., 1987). MYC amplification (in CD44, TCF7) as coordinately upregulated in high-risk NBs contrast to MYCN) israre in NB (Kohl et al., 1983) yet without MYCN amplification in comparison to high-risk alternative mechanisms of MYC deregulation have not MYCN-amplified or intermediate-risk NBs, supporting been elucidated. pathway activation in this subset. Thus, high-risk NBs A recurring paradigm in embryonal oncogenesis is the without MYCN amplification may deregulate MYC and aberrant appropriation of developmental programs. other oncogenic genes via altered b-catenin signaling Indeed, activation of the Notch, Sonic hedgehog and providing a potential candidate pathway for therapeutic Wnt/b-catenin (CTNNB1) developmental programs inhibition. contribute to malignant transformation (Taipale and Oncogene (2008) 27, 1478–1488; doi:10.1038/sj.onc.1210769; Beachy, 2001; Allenspach et al., 2002; Giles et al., 2003). published online 27 August 2007 Wnt/b-catenin signaling may be of particular relevance to NBs, which arise from migratory neural crest-derived Keywords: neuroblastoma; CTNNB1/b-catenin; MYC; neuroblasts, as this program mediates neural crest cell ONCOGENOMICS MYCN; embryonal cancer fate and neural stem-cell expansion (Chenn and Walsh, 2002; Zechner et al., 2003; Lee et al., 2004). Further, Wnt/b-catenin signaling is aberrantly activated in multi- ple embryonal and neural crest-derived malignancies (Giles et al., 2003). Activation of the canonical Wnt/b-catenin pathway leads to transactivation of target genes, which may play a direct role in tumorigenesis (Goss and Groden, 2000; Correspondence: Dr MD Hogarty, Division of Oncology/The Moon et al., 2002). During the nonactivated state, Children’sHospitalof Philadelphia, 9 North ARC (902C), 3516 b-catenin is sequestered in the cytosol by a complex that Civic Center Boulevard, Philadelphia, PA 19104-4318, USA. E-mail: [email protected] includesAPC, GSK3B, Axin proteinsand others.This Received 25 September 2006; revised 27 June 2007; accepted 1 August structure facilitates phosphorylation of b-catenin target- 2007; published online 27 August 2007 ing it for proteasomal degradation (Goss and Groden, Deregulated b-catenin in neuroblastoma XLiuet al 1479 2000) and b-catenin targetsremain actively repressed. intragenic deletion that hinders b-catenin degradation Canonical pathway activation hinders b-catenin phos- and leadsto canonical signaling(He et al., 1998). phorylation resulting in enhanced protein stability and NB cell lineswithout MYCN amplification express nuclear translocation, where it forms a heterodimeric b-catenin at high levelswith aberrant nuclear localization. transcription factor with proteins of the TCF/LEF Subcellular protein fractions were assessed for expres- family and transactivates target genes that include MYC sion of b-catenin (Figures1c and d). NALM-6 and PB1 (He et al., 1998), CCND1 and others(Miller et al., 1999; controls demonstrate modest b-catenin expression limited Hecht and Kemler, 2000). to the cytosol. HepG2 cells express both wild-type and Malignant cellsmay activate the canonical pathway mutant b-catenin, with the stabilized truncated b-catenin through gain-of-function mutationsin b-catenin, by in higher abundance and enriched in the nuclear fraction inactivation of componentsof the scaffolding complex (positive control for deregulated b-catenin). (Giles et al., 2003), or by autocrine Wnt/receptor signaling NB cellswith MYCN amplification express b-catenin (Bafico et al., 2004). We hypothesized that deregulated comparable to that of the wild-type protein in HepG2. b-catenin signaling may occur in high-risk NBs without The majority iscytosolic, with a minor nuclear contri- MYCN amplification resulting in transactivation of MYC bution (o30% of total). In contrast, MYCN single-copy and other target genes to induce an aggressive phenotype. cell linesexpresshigher levelsof b-catenin overall aswell asa higher amount in the nucleus(similar to that for the mutant allele in HepG2). The absence of significant Results nuclear b-tubulin argues against cytosolic spillover contributing to artifactual localization (see Supplemen- There isa reciprocal relationship between MYC and tary Figure S1). MYCN expression in NB-derived cell lines. Cell lines Immunofluorescence cytology confirmed aberrant with MYCN gene amplification expressed high levels of nuclear localization (Figure 2a). HepG2 cellsdemon- MYCN and undetectable MYC (Figures1a and b). In strated higher b-catenin expression with strong nuclear contrast, NB cell lines without MYCN amplification localization in B25% of cells(arrowhead). The MYCN- expressed MYC in the absence of detectable MYCN amplified cell linesshow b-catenin largely restricted to with the exception of the NBL-S cell line. Thisunique the cytosol (IMR5 shown). In contrast, SK-N-AS and nonamplified cell line hasa prolonged MycN protein LA-N-6 demonstrated higher b-catenin expression with half-life (Cohn et al., 1990). Of note, control HepG2 moderate and marked homogeneousnuclear localiza- cellsexpresshighlevelsof MYC due to a hemizygous tion, respectively. In addition to validating localization,

Figure 1 Reciprocal MYCN and MYC expression (a) in NB at both the mRNA (relative to GAPDH) and protein level. Note marked HepG2 MYC expression secondary to deregulated b-catenin. Immunoblot (IB) validation of Myc protein expression (b) wasdone using fractionated protein from the nuclear (N) and cytosolic (C) compartments. (c)IBofb-catenin and tubulin. Wild-type (wt) and truncated b-catenin (*) are depicted. A single representative experiment is shown. (d) Histogram of expression and localization by densitometry for three replicate experiments (as in (c), with cytosolic (black portion) and nuclear (gray portion) b-catenin detected by densitometry). Wt and mutant alleles for HepG2 are shown separately. Replicates are from independent protein fractionations.

Oncogene Deregulated b-catenin in neuroblastoma XLiuet al 1480 No activating b-catenin mutationsare identified in NB cell linesor primary NBs.No aberrantly migrating b-catenin bands to suggest truncating deletions were present in NB (Figure 1c). Single-strand conformational polymorphism (SSCP) was used to analyse the critical b-catenin regulatory region in NB cell lines( N ¼ 12) and primary NBs( N ¼ 26) and no aberrantly migrating bandswere detected (Table 1). The HepG2 cell line hasa hemizygousmutant b-catenin allele with an exon 3 deletion. Since the SSCP primersamplify within exon 3, HepG2 ishemizygousfor thisPCR product and the banding pattern appearswild type. To exclude similar larger deletions in NB cells, we used primers within exons2 and 4 to amplify a larger genomic fragment encompassing the entire regulatory domain. No internal deletionswere identified in NB while HepG2 demon- strated the internal deletion (data not shown). Despite aberrant nuclear b-catenin, NB-derived cell linesdid not exhibit enhanced b-catenin–TCF transcrip- tional activity. The native b-catenin–TCF luciferase reporter, pTOPflash, was used to assess b-catenin–TCF transcriptional activity in NB and results were compared with those for the control reporter pFOPflash that has mutated b-catenin binding sites (Figure 2b). None of the NB cell lines showed specific b-catenin transactivation. Still, evidence for relative b-catenin target gene upregu- lation (or at least derepression) was demonstrated using expression profiles and Q-PCR (see Supplementary FiguresS2 and S3). We next looked for evidence of b-catenin pathway activation in primary NBs. Expression profiles were obtained from primary NBs(Wang et al., 2006) including 21 from intermediate-risk tumors (IR Group) and 52 from high-risk tumors (32 without MYCN amplification (HR-NA) and 20 with MYCN amplification (HR-A)); see Table 2 and ‘Materialsand methods’section. We defined a b-catenin target gene set using the Wnt/b-catenin database (www.stanford.edu/Brnusse/wntwindow.html). Twenty-six genes were identified and we sought to determine whether these alone could discriminate high- risk tumors with and without MYCN amplification (Figure 3a). Thisunsupervisedmethod segregated HR tumorsinto groupswith 25 of 26 (96%) HR-NA NBs and 19 of 26 (73%) HR-A NBs, and supports the hypothesis that these genes alone are discriminatory, with Figure 2 Fluorescence immunocytology. (a) HepG2 cellsshow coordinate b-catenin pathway activation in HR-NA strong nuclear b-catenin staining (arrowheads) in a subset of nuclei. tumors. Five of seven tumors from children less than 18 SK-N-AS and LA-N-6 (cell lineswithout MYCN amplification) monthsof age segregated with the HR-A tumorswith show higher expression with aberrant nuclear localization in compari- reduced target gene expression. Thus, these findings do son with MYCN-amplified IMR5 cells. Nuclei are visualized with DAPI counterstain. (b) Absence of b-catenin–TCF reporter activity not represent age-related differences whereby expression in NB cell lines transiently co-transfected with either the pTOPflash of thisdevelopmental program in tumorsfrom younger b-catenin reporter construct or the mutant pFOPflash construct, children istemporally but normally upregulated. Only six plotted astheir reporter ratio normalized to luciferaseactivity. tumors met high-risk criteria in the absence of metastases HEK 293 cells transiently transfected to express a mutant-activated b-catenin (S33Y) were used as a positive control. Each experiment (INSS stage 3) and half segregated within the high was performed at least twice in triplicate, and mean and s.d. values b-catenin target gene subset, half in the low expressers were determined. suggesting that canonical b-catenin signaling is not a surrogate metastasis signature. these findings support the immunoblot data and suggest These differences were not the result of the incidental possible differences in b-catenin processing or regulation inclusion of MYCN targetsin the gene listaswe in comparison to HepG2 cells in which deregulation is repeated the analyses 10 times using permuted gene lists caused by the b-catenin internal deletion. consisting of randomly generated 26 gene sets and none

Oncogene Deregulated b-catenin in neuroblastoma XLiuet al 1481

b Table 2 Clinical and genetic data for primary neuroblastomas IR HR-NA HR-A N (%) N (%) N (%)

Total 21 32 20 -Catenin:TCF b reporter activity Age o12 months14 (67) 0 4 (20) 12–18 months5 (24) 2 (6) 1 (5) >18 months2 (9) 30 (94) 15 (75)

INSS* stage 1or2 0 0 0 3 17 (81) 6 (19) 0 4 4 (19) 26 (81) 20 (100) re; SSCP, single-stranded conformational MYCN Amplification 0 0 20 (100)

Shimada grade Favorable 20 (95) 2 (6) 0 Unfavorable 1 (5) 29 (91) 18 (90)

-Catenin exon 3 SSCP mutation analysis Not known 0 1 (3) 2 (10) b 4C RT COG** Risk group Low 0 0 0 Intermediate 21 (100) 0 0 High 0 32 (100) 20 (100)

Abbreviations: COG, Children’s Oncology Group; HR-A, high-risk, MYCN amplified; HR-NA, high-risk NBs, MYCN not amplified; INSS, International Neuroblastoma Staging System; IR, intermediate-risk. -Catenin ++ C/Neg N++ C/+ N ND wt wt wt Neg ND b localization (IC) a segregated the HR-NA and HR-A tumors with similar 31 34 31 precision. Further, the overwhelming majority of targets -catenin and MYC/MYCN analyses (N/N+C)

b regulated by MycN are transactivated. The finding that b-catenin target genes are upregulated in the subset of nonamplified tumors(seePaGE and GSEA analyses below) supports that these changes do not represent a MYCN activation signature. Summary of Similar unsupervised analyses with b-catenin targets pTOPFlash/pFOPFlash luciferase activity normalized for transfection efficiency. b could not readily segregate the IR from HR-NA tumors, though both tumor subsets segregated from the HR-A Cytosolic Nuclear %Nuclear Table 1 subset (Figure 3b). Still, the clade representing lower -Catenin protein expression (IB)

b b-catenin target expression contained 19 of 20 HR-A NBs(95%), 12 of 21 IR tumors(57%) and only 11 of 32

Whole cell lysate HR-NA tumors(34%). Primary NBswithout MYCN amplification have upregulated b-catenin target gene expression. Thirteen of 26 b-catenin targets(50%) demonstrated higher expression in HR-NA tumors and only one (FOSL1) was more highly expressed in HR-A tumors (Figure 4, MycN mRNA/ Protein confidence threshold 0.95). Using a confidence threshold of 0.90 resulted in 17 b-catenin targets(65%) being identified asupregulated in HR-NA tumors.Similarly, comparing the HR-NA tumorswith IR tumorsrevealed upregulated expression for seven b-catenin targets, while no targetswere more highly expressedin the IR tumors By densitometry; w+, weakly positive; wt, wild type.

a (10 are upregulated in HR-NA NBsusinga confidence Protein threshold of 0.80). These findings support increased b-catenin signaling defining a high-risk NB phenotype independent of MYCN amplification. Oligonucleotide array results were previously vali- HepG2(wt protein)(mutant) ++++/++++ Neg/NDNMBNGPIMR5SMS-KAN ++++ Neg/ND Neg/Neg Neg/NegSK-N-SH Neg/NegSK-N-AS ++/+++LA-N-6 ++/+++NBL-S +++/+++ ++++/++ ++/+++ +++ +++/++NALM-6 +++/++ + Neg/Neg Neg/NDPB-1 + Neg/Neg +++ ND +++Abbreviations: Neg/Neg C, ++ cytosolic; +++ IB,polymorphism. ++ +++ immunoblot; IC, ND ++++ immunocytology; N, ++ nuclear; ND, +++ +/+ ++++ not ++ determined; ++++ Neg, negative; w+ ++++ Pos, positive; +++ ++ RT, 22 room ++ w+ temperatu ND ++++/25% of cellswt ++++ 25 ++++ ++++ 29 + ++ ++++ C/Neg N ND ++++ 26 49 +++ +++ C/weak 25 N + C/Neg N + + C/Neg ++ N C/+ 21 N + wt C/Neg N wt + ND + with N staining wt wt wt + wt w+ wt w+ wt 0 wt Pos wt wt wt 0 ND ND Neg Neg Neg ND Neg Neg ND ND ND ND ND ND Cell line Myc mRNA/ dated for 18 genesusingqRT–PCR with each having a

Oncogene Deregulated b-catenin in neuroblastoma XLiuet al 1482

Figure 3 Unsupervised hierarchical clustering of gene expression data. (a) Comparison of HR-NA (unfilled) and HR-A (dark-filled) NBs( N ¼ 52) using only probe sets representing b-catenin targets. The leftward clades contain mostly HR-NA NBs with higher expression of b-catenin targets. Clinical annotation: MYCN, dark box ¼ MYCN amplified; INSS, dark box ¼ INSS stage 4; age, dark box ¼ >18 months. (b) Comparison of IR, HR-NA and HR-A NBs (N ¼ 73). Coding for tumors is shown above in bars: unfilled, HR-NA; dark-filled, HR-A; gray-shaded, IR.

microarray-to-qRT–PCR correlation o0.0001 (Wang enriched in the HR-NA cohort. These included TCF7, et al., 2006). We performed semiquantitative qRT–PCR MET, MYC, GAS, BIRC5, MYCBP, FGF18, FST and for three additional b-catenin target genesfrom random MMP7. No b-catenin target wasenriched in the direction tumors selected from HR-NA and HR-A NBs. qRT– of higher expression in the IR group. These data support PCR data wastightly correlated with oligonucleotide that there are a subset of core b-catenin target genes array expression data for these genes as well (Figure 5). upregulated in high-risk NBs without MYCN amplifica- Distinct cellular processes often coregulate multiple tion (HR-NA) in comparison to IR NBs. genesand modestchangesin a coordinated direction We assessed expression of upstream signaling compo- across such genes may be as biologically meaningful as nents to determine possible mechanisms of pathway single genes with dramatic fold changes, though less activation. WNT1, WNT6, WNT7A and WNT10B were apparent. We used Gene Set Enrichment Analyses all more highly expressed in HR-NA NBs. Soluble (GSEA) to confirm the b-catenin target alterations factorsthat may regulate Wnt/receptor interactionsare between tumor groups(Subramanian et al., 2005). We also abundant in NBs though no differential expression confirmed enrichment of b-catenin target genesin HR- wasapparent. Wnt receptorsinclude the Frizzled ( FZD) NA versus HR-A tumors, with an enrichment score (ES) and LDLreceptor-related protein (LRP) families: FZD2 of 0.45 (P ¼ 0.044) (Figure 6), with 17 genescontributing and LRP5 were both more abundantly expressed in markedly to the ES (making up the ‘leading edge’ subset). HR-A NBs(Table 3). These included CD44, NRCAM, MYC, MET, FST, Wnt signaling activates disheveled homologs (DVL1, TCF7, L1CAM, MYCBP, LEF1, PPARD, JUN, GAS, DVL3) and FRAT2 to initiate negative regulatory effects CCND1, FGF18, BIRC5, PLAUR and MMP7. Statistical on AXIN and GSK3 and increase b-catenin activity. Other significance was confirmed using an empirical phenotype- membersof thisscaffold complex include APC and based permutation test (Subramanian et al., 2005). BTRC. Most of these components (including b-catenin We next sought confirmation that HR-NA NBs had itself) are not differentially expressed (Table 3), which is coordinate up-regulation of b-catenin targetsin compar- not surprising given that pathway components are ison to IR tumors. Despite an ES consistent with primarily regulated post-translationally. However, DVL1 upregulation of the b-catenin targetsin HR-NA NBs isupregulated in HR-NA tumorsin a manner predicted (0.33), the analysis did not meet statistical significance to augment b-catenin activity. (P ¼ 0.37). Still, 9 of the 17 genesthat comprised the Nuclear regulatorsof b-catenin signaling include ‘leading edge subset’ that account for the maximal membersof the TCF/LEF family that act asheterodi- enrichment signal (Subramanian et al., 2005) remained meric transcription partners. TCFsact with corepressors

Oncogene Deregulated b-catenin in neuroblastoma XLiuet al 1483

Figure 4 Box-plots for all differentially expressed b-catenin targets across NB subsets. As labeled across bottom, the left-most box represents expression values from 28 intermediate-risk NBs (IR), the center box 32 HR-NA NBs and the right-most box the 20 HR-A NBs. Each box represents the 25th through 75th percentile with the median marked by a dot; caps span to the 10th and 90th percentile. Statistically significant expression differences (by PaGE analyses) are denoted by the asterisk (*) using a false-discovery rate of 0.1. including the TLE family (Groucho homologs). Additional neural crest stem cells (Chenn and Walsh, 2002; Reya regulatorsinclude CREBBP (CBP), PYGO1 (Pygopos) and Clevers, 2005) and proliferative control of neural and CTNNBIP1 (ICAT). TCF7,themajorTCF/LEF progenitors(Zechner et al., 2003). Similar functionsmay expressed in NBs, was markedly more abundant in HR- contribute to oncogenesis in neural precursors. NA NBs. There was also higher expression of CTNNBIP1 Nearly half of all children with NB present with in these tumors, and lower expression of PYGO1. clinicobiological factorsthat define their tumorsashigh- risk. MYCN amplification occursin B40% yet no consistent pathway alteration accounts for the highly Discussion aggressive phenotype in the remainder. We speculated that deregulated MYC and additional b-catenin targetsmight Embryonal malignanciesfrequently misappropriate be implicated. Interestingly, MYC (rather than MYCN) developmental programsto drive neoplasia(Marisand acts as an essential regulator of earlier neural crest Denny, 2002). We hypothesized that activation of Wnt/ formation in vertebrates, and its expression is dependent b-catenin might contribute to NB progression as path- upon Wnt/b-catenin signaling (Bellmeyer et al., 2003). way mutationsoccur in other pediatric embryonal We demonstrate that b-catenin ishighly expressedand cancersand tumorsof neural crestlineage (reviewed in aberrantly localized in NB cellsabsent MYCN amplifica- Giles et al. (2003)). b-catenin also plays a key role in pre- tion. The nuclear distribution of b-catenin ishomogeneous migratory neural crest development (Lee et al., 2004; in comparison to HepG2 cells and may reflect different Kleber et al., 2005), maintenance and expansion of alterationsin b-catenin regulation in NBsthan those

Oncogene Deregulated b-catenin in neuroblastoma XLiuet al 1484

Figure 5 (a) Duplex qRT–PCR results demonstrating expression of b-catenin targetsrelative to GAPDH for four NBswith (HR-A) or without (HR-NA) MYCN amplification. The NB-derived cell NB69 isshownasa reference. ( b) Correlation between relative expressions obtained using the oligonucleotide arrays and qRT– PCR for the same four primary NBs, with fetal brain as a reference. White and shaded gray bars represent array-based expression scores for the gene indicated normalized to values for the highest expressing tumor, #1204. Black bars represent expression values of Figure 6 Gene Set Enrichment Analysis (GSEA) results: compar- the same targets relative to GAPDH using RT–PCR (raw data isons in gene expression between HR-NA (n ¼ 32) and HR-A shown in panel a), also normalized to #1204. (n ¼ 20) NBs. Differential gene expression is depicted as a gene list, ranked from most differentially upregulated to most downregu- lated in HR-NA tumorsfrom left to right. The location for each of the 26 b-catenin target genes in this rank list is shown in blue resulting from activating mutations in b-catenin. Indeed, (middle). A nonrandom distribution of target genes is apparent, we failed to identify any b-catenin-specific mutations in with upregulation in the HR-NA tumors. This is statistically confirmed with an enrichment score (ES) that reflects the degree to NB. Our cell line findingssuggestcanonical b-catenin which this gene set is overrepresented at the extremes in the analysis signaling activity, yet we were unable to directly (P ¼ 0.04). The ‘leading edge’ subset of genes is circled and demonstrate b-catenin transactivation. b-catenin may definesthe core membersthat participate in defining the phenotype function through interactionswith non-TCFLEF hetero- (b-catenin targetsupregulated in the HR-NA NBsand listedin dimeric partnerssuchasthoseof the SOX or FOXO the text). families(Sinner et al., 2004; Essers et al., 2005). Further, it is possible that relative upregulation of b-catenin target genesmay be achieved through derepression of TCF influences on neural crest stem cells driving sensory complexesat b-catenin target genesin the absence of overt neuron expansion at the expense of autonomic neurons transactivation. (Lee et al., 2004), including the sympathetic ganglia at High-risk NBs without MYCN amplification showed risk of neuroblastic transformation. It is possible that more abundant WNT ligand expression consistent with germ line activation of b-catenin through mutant APC autocrine or paracrine signaling, as has been described for leadsto temporal pathway activation that precludesNB. breast and ovarian carcinomas (Bafico et al., 2004). Hepatoblastomas demonstrate b-catenin pathway muta- WNT5A hasbeen reported to be reduced in NBswith tions in two-thirds of sporadic cases, yet fewer than 1% high-risk features (Blanc et al., 2005), but our datasets of patientswith germ line APC mutations(FAP) are demonstrate low expression in both high (Table 3) and diagnosed with this embryonal tumor (Taipale and lower-risk tumors with no differential expression (data Beachy, 2001). not shown). Still, most tumor-specific pathway deregula- In conclusion, we have defined the b-catenin pathway tion occursthrough activating mutationsdownstream of aspotentially contributing to highly aggressiveNB in Wnt/receptor events. DVL1 isone intermediate demon- the absence of MYCN amplification. The specific strating elevated expression in NBs without MYCN mechanisms whereby the b-catenin pathway isactivated amplification, a direction consistent with activation. in NB remain elusive and it is possible that activation in Extensive studies have not been done to exclude activating additional cellular pathways may contribute, but classi- mutationsin other componentsof the b-catenin scaffold cal tumor-specific mutations in scaffolding and regula- and degradation complex to date, though our data tory proteinshave not been wholly excluded. That the suggest such studies are warranted. Finally, evidence in pathway may contribute to NB tumorigenesisisof support of aberrant b-catenin signaling in high-risk NBs clinical relevance astherapiestargeting thispathway without MYCN amplification comesfrom expression are under development (Moon et al., 2004). The relative profile analyses of b-catenin target genes. importance in NB remainsto be elucidated and We queried colorectal cancer registries for NBs functional evidence supporting a cellular dependence occurring in familial adenomatouspolyposis(FAP) on thispathway hasnot been demonstrated.Still, the kindreds, but did not identify any (F Giardiello, data herein support that preclinical evaluation of personal communication). b-catenin hasinstructive b-catenin-targeted therapiesmay be warranted.

Oncogene Deregulated b-catenin in neuroblastoma XLiuet al 1485 Table 3 Array-based expression data: b-catenin pathway components HR-NA HR-A (n ¼ 32) (n ¼ 20) Gene Probe seta Cytogenetic location Mean e-scoreb Mean e-scoreb Confidence level

Wnt ligands and antagonists WNT1 1853_at 12q13 57.66 41.03 0.969 WNT2 160023_at 7q31 21.23 19.15 WNT2B 33684_at 1p13 1.52 1.91 WNT4 2091_at 1p36-p35 26.07 18.76 WNT5A 31862_at 3p21-p14 21.49 20.89 WNT5A 1669_at 3p21-p14 17.24 15.81 WNT6 2090_i_at 2q35 325.57 255.72 0.992 WNT7A 36763_at 3p25 33.31 20.83 0.998 WNT7A 1887_g_at 3p25 0.10 0.10 WNT7A 1886_at 3p25 0.10 0.10 WNT8B 31769_at 10q24 0.10 0.10 WNT10B 1019_g_at 12q13 41.90 32.65 0.965 WNT10B 1018_at 12q13 30.68 36.51 WNT11 41427_at 11q13.5 1.66 1.13 WNT11 2030_at 11q13.5 90.86 94.59 WNT11 39808_at 11q13.5 17.75 15.80

SFRP1 32521_at 8p12-p11 203.82 200.36 SFRP4 41405_at 7p14.1 139.78 126.34 FRZB 40230_at 2qter 111.15 70.40 DKK1 35977_at 10q11.2 0.42 0.18 DKK4 33594_at 8p11 88.83 85.83

Wnt receptors FZD1 38300_at 7q21 10.11 10.91 FZD2 36799_at 17q21.1 13.28 98.18 0.999 FZD2 628_at 17q21.1 9.50 70.77 0.998 FZD5 34997_r_at 2q33-q34 5.80 4.82 FZD6 34472_at 8q22-q23 37.37 46.98 FZD7 33222_at 2q33 21.25 27.72 FZD9 33578_at 7q11 2.75 2.34

LRP5 41831_at 11q13.4 29.57 36.81 0.962 LRP6 34697_at 12p11-p13 103.35 117.23 b-Catenin scaffold and degradation components DVL1 163_at 1p36 43.11 25.39 0.986 DVL3 40873_at 3q27 85.81 95.02 FRAT2 40173_at 10q24.1 0.27 0.37 FRAT2 40171_at 10q24.1 24.19 23.74 FRAT2 40172_g_at 10q24.1 0.10 0.10

CTNNB1 40777_at 3p21 517.21 551.64

APC 35433_s_at 5q21-q22 6.17 13.46 APC 36182_at 5q21-q22 30.02 27.79 APC 1912_s_at 5q21-q22 68.92 69.93 APCL 34184_at 19p13.3 459.30 468.75 AXIN1 33319_at 16p13.3 226.49 244.56 GSK3A 632_at 19q13.3 144.62 132.29 GSK3B 40645_at 3q13.3 190.62 196.33 GSK3B 1253_at 3q13.3 79.63 84.81 BTRC 39304_g_at 10q24-q25 13.15 9.20 BTRC 39303_at 10q24-q25 0.10 0.15 BTRC 39305_at 10q24-q25 9.50 7.29

Nuclear regulators of b-catenin signaling TCF7 32649_at 5q31.1 51.38 17.79 0.993 TCF7 31440_at 5q31.1 175.62 142.83 0.961 TCF7L2 32025_at 10q25.3 16.03 5.26 LEF1 36021_at 4q23-q25 38.80 21.45

CREBBP 33831_at 16p13.3 141.43 135.06 PYGO1 32015_at 15q21.1 1.65 3.12 0.952

TLE1 41489_at 9q21.32 94.90 68.40 TLE2 40837_at 19p13.3 21.14 25.04

Oncogene Deregulated b-catenin in neuroblastoma XLiuet al 1486 Table 3 (continued ) HR-NA HR-A (n ¼ 32) (n ¼ 20) Gene Probe seta Cytogenetic location Mean e-scoreb Mean e-scoreb Confidence level

TLE3 38234_at 15q22 73.26 70.82 TLE4 40692_at 9q21.32 14.13 5.22

CTNNBIP1 39171_at 1p36.22 98.47 67.74 0.960 CTNNBIP1 33217_at 1p36.22 14.48 7.73

Gray color, upregulated expression. aAffymetrix HG U95Av2 probe set. bProbe profiler-generated expression score.

Materials and methods Immunofluorescence Cells were subconfluently seeded in Nunc Lab-Tek II chamber Cell lines and primary tumors slides (Fisher Scientific, Pittsburgh, PA, USA) before fixation The NB cell lines used are referenced in Thompson et al. (2001) with ice-cold 1:1 methanol:acetone solution for 10 min. Cells and propagated asdescribed.HepG2, NALM-6 (pre-B ALL) were washed in phosphate-buffered saline (PBS) and stored at and PB1 (T-cell lymphoma) cellswere kindly provided by Dr À20 1C after air-drying. Slideswere blocked with 50 mlof Mariusz Ratajczak (University of Pennsylvania). Twenty-six PBS þ 1% bovine serum albumin for 10 min, then incubated tumoral DNA samples were obtained from the former with 40 mlofab-catenin–fluorescein isothyocyanate antibody Children’sCancer Group tumor bank for b-catenin mutation (C19224, Transduction Laboratories) at 10 mgmlÀ1 for 45 min analyses. For oligonucleotide array-based expression profile at room temperature (RT). After washing in PBS (thri- studies, 73 primary NBs were used (Wang et al., 2006) as ce  5 min), cellswere counterstainedwith 4’-6-diamidino-2- detailed below. phenylindole (DAPI; 450 ng mlÀ1) in PBS for 15 min, washed in deionized water thrice for 15 min before mounting with antifade reagent (SlowFade; Molecular Probes, Inc., Eugene, RNA isolation and qRT–PCR analysis OR, USA) and viewed with a Leica DMR Fluorescence RNA wasisolated using RNeasy(Qiagen, Valencia, CA, USA). Microscope. Reverse transcription (RT)–PCR was used for evaluation of MYC, MYCN, CCND1 and NRCAM. RNA (2 mg) was b-Catenin mutation analysis reverse transcribed using the SuperScript II pre-amplification For PCR-based SSCP, DNA was amplified as previously system (Life Technologies, Gaithersburg, MD, USA). One reported (Voeller et al., 1998) with 100 mM dATP including half microliter of product wasusedasa template for PCR Redivue [a-32P] dATP (Amersham Pharmacia, Buckingham- amplification using biotinylated glyceraldehyde-3-phosphate shire, UK). Denatured samples were electrophoresed through dehydrogenase (GAPDH) primersmultiplexed with target gene a 6% non-denaturing mutation detection enhancement gel primers(all primer pairsspannedan intron–exon boundary; (Cambrex) at both 4 1C and (25 1C). Internal deletionswere sequences available upon request). GAPDH primerswere used also screened for with primers within exons 2 (50-ATTGGG at a ratio of biotinylated:non-biotinylated of 1:99 to prevent GCTCTGCTTCGTTG-30) and 4 (50-GTCATTGCTTACC signal saturation (Eggert et al., 2000). PCR wasperformed in TGGTCCTCG-30). 20 ml volumescontaining 400 n M each primer, 0.1 mM each deoxyribonucleotide triphosphate, 1.5 mM MgCl2,1 PCR buffer and one unit of AmpliTaq Gold polymerase (Perkin- b-Catenin/TCF reporter assays Elmer Corp, Branchburg, NJ, USA) for 18 (MYC, MYCN)or Cells were transiently transfected with pTOPflash or pFOP- 30 cycles( CCND1, NRCAM) at an annealing temperature of flash (Morin et al., 1997) provided by Dr Bert Vogelstein (The 55 1C. Detection wasperformed usingthe Southern Light Johns Hopkins University) as described previously (Kolligs system (Tropix Inc., Bedford, MA, USA). et al., 2002). HEK 293 cells transfected to express S33Y mutant b-catenin were used as a positive control. Each experiment wasperformed twice in triplicate, and mean and Protein fractionation and immunoblotting s.d. values were determined. Fractionated nuclear and cytosolic protein lysates were obtained using the NE-PER Extraction Reagents kit (Pierce; Oligonucleotide-array profiling Rockford, IL, USA). Fresh MiniComplete protease inhibitor Expression profiles from diagnostic NBs (Wang et al., 2006) cocktail (Roche) wasadded. Ten microgramsof nuclear using the Affymetrix U95Av2 array (Affymetrix, Santa Clara, protein or 50 mg of cytosolic protein was loaded. Antibodies CA, USA) were mined. Modeling of probe set behavior was included anti-b-catenin (C19220; BD Transduction Labora- conducted using Probe Profiler (Corimbia, Berkeley, CA, tories, San Jose, CA, USA), anti-N-myc (556438, BD USA) software that implements a model-based approach and PharMingen, San Jose, CA, USA), anti-Myc (C33, Santa corrects for nonspecific cross-hybridization effects. A quanti- Cruz Biotechnology, Santa Cruz, CA, USA) and anti-b- tative expression score (e-score) is calculated for each probe set tubulin (D-10, Santa Cruz Biotechnology). Detection wasvia (see the GEO database at http://www.ncbi.nlm.nih.gov/geo/ goat anti-mouse secondary antibody with chemiluminescence (accession number GSE3960)). b-Catenin pathway component detection (Immun-Star; Bio Rad Laboratories, Hercules, CA, geneswere hand curated (seeTable 3); the b-catenin target USA). Ponceau S staining was used to ensure balanced loading gene set was defined a priori using the Wnt/b-catenin database for subcellular fractions but no normalization of target to (www.stanford.edu/Brnusse/wntwindow.html). Probe sets for control signals was used. Semiquantitation was performed twenty-six genes were included: CCND1, MYC, PPARD, using densitometric detection (NIH Image, v1.5 software). NRCAM, TCF7, LEF1, FOSL1, JUN, MMP7, TCF4, VEGF,

Oncogene Deregulated b-catenin in neuroblastoma XLiuet al 1487 FGF18, MYCBP, FZD7, FST, ID2, PLAUR, GAS, CD44, ranked based on the correlation between their expression and BMP4, BIRC5, MET, L1CAM, DKK1, SOX9 and FOXN1 the biological class distinction. The GSEA-P software package (for probe set information, see www.affymetrix.com). was used (Broad Institute, Massachusetts Institute of Technol- For differential gene expression analyses, binary compar- ogy) with the 26 b-catenin target genesasour ‘gene set’,and an isons were performed using e-scores obtained with Probe ES that reflectsthe degree to which a gene setisover- Profiler and standardized for each sample by subtracting the represented at the extremes of the gene rank list in a two-group mean expression value of a probe set and dividing by the s.d.. comparison and corresponds to a weighted statisic like Differential gene expression was measured with the Patterns Kolmogorov–Smironov was calculated. Statistical significance from Gene Expression (PaGE; (Manduchi et al., 2000)) isestimatedby generation of a nominal P-value following algorithm. All runswere done with 1000 permutationson random permutation of the Group labelsto compute a series unlogged data. A confidence level of 0.95 (1-FDR) wasusedto of ESsand define the null distributionfor thismetric define differential expression. (Subramanian et al., 2005).

Unsupervised clustering Acknowledgements Expression values were filtered from the HGU95Av2 data sets for 21 intermediate-risk NBs (IR), 32 high-risk NBs without We thank Garrett Brodeur and Mariucz Ratajczak (University MYCN amplification (HR-NA) and 20 high-risk NBs with of Pennsylvania) for providing cell lines, Kathleen R Cho MYCN amplification (HR-A). Unsupervised agglomerative and Rong Wu (University of Michigan) for assistance with hierarchical clustering was performed on TIGR’s Multi- b-cat:TCF reporter assays, Bert Vogelstein (Johns Hopkins Experiment Viewer. Manhattan distance was used as the University) for b-catenin reporter constructs, the Children’s distance metric and average linkage, which is the average Oncology Group (COG) for primary NB materials, Eric distance between each component of one cluster and each Rappaport (Nucleic Acids/Protein Core at CHOP) for component of the other cluster was used to measure cluster-to- oligonucleotide array assistance, Francis Giardiello (Johns cluster distances. HopkinsHereditary Colorectal Cancer Registry)for informa- tion on neuroblastic tumors in FAP kindreds and Harriet Pais Gene set enrichment analysis for bioinformatics assistance. This work was supported in part GSEA was used to compare profiles across biological classes by P01-CA97323, a Career Development Award from the (Subramanian et al., 2005). Expression values from samples BurroughsWellcome Fund and the Richard and Sheila were stratified by NB risk group in binary comparisons, and Sanford Chair in Pediatric Oncology (MDH).

References

Allenspach EJ, Maillard I, Aster JC, Pear WS. (2002). Notch signaling Hecht A, Kemler R. (2000). Curbing the nuclear activitiesof beta- in cancer. Cancer Biol Ther 1: 466–476. catenin. Control over Wnt target gene expression. EMBO Rep Bafico A, Liu G, Goldin L, HarrisV, AaronsonSA. (2004). An 1: 24–28. autocrine mechanism for constitutive Wnt pathway activation in Kleber M, Lee HY, Wurdak H, Buchstaller J, Riccomagno MM, human cancer cells. Cancer Cell 6: 497–506. Ittner LM et al. (2005). Neural crest stem cell maintenance by Bellmeyer A, Krase J, Lindgren J, LaBonne C. (2003). The proto- combinatorial Wnt and BMP signaling. J Cell Biol 169: 309–320. oncogene c-myc is an essential regulator of neural crest formation Kohl NE, Kanda N, Schreck RR, BrunsG, Latt SA, Gilbert F et al. in xenopus. Dev Cell 4: 827–839. (1983). Transposition and amplification of oncogene-related sequences Blanc E, Roux GL, Benard J, Raguenez G. (2005). Low expression of in human neuroblastomas. Cell 35: 359–367. Wnt-5a gene is associated with high-risk neuroblastoma. Oncogene KolligsFT, Nieman MT, Winer I, Hu G, Van Mater D, Feng Y et al. 24: 1277–1283. (2002). ITF-2, a downstream target of the Wnt/TCF pathway, is Chenn A, Walsh CA. (2002). Regulation of cerebral cortical size by activated in human cancerswith beta-catenin defectsand promotes control of cell cycle exit in neural precursors. Science 297: 365–369. neoplastic transformation. Cancer Cell 1: 145–155. Cohn SL, Salwen H, Quasney MW, Ikegaki N, Cowan JM, Herst CV Lee HY, Kleber M, Hari L, Brault V, Suter U, Taketo MM et al. et al. (1990). Prolonged N-myc protein half-life in a neuroblastoma (2004). Instructive role of Wnt/beta-catenin in sensory fate cell line lacking N-myc amplification. Oncogene 5: 1821–1827. specification in neural crest stem cells. Science 303: 1020–1023. Cole MD, McMahon SB. (1999). The Myc oncoprotein: a critical Manduchi E, Grant GR, McKenzie SE, Overton GC, Surrey S, evaluation of transactivation and target gene regulation. Oncogene Stoeckert Jr CJ. (2000). Generation of patterns from gene expression 18: 2916–2924. data by assigning confidence to differentially expressed genes. Eggert A, Brodeur GM, Ikegaki N. (2000). Relative quantitative RT– Bioinformatics 16: 685–698. PCR protocol for TrkB expression in neuroblastoma using GAPD MarisJM, Denny CT. (2002). Focuson embryonal malignancies. asan internal control. Biotechniques 28: 681–682, 686, 688–691. Cancer Cell 2: 447–450. Essers MA, de Vries-Smits LM, Barker N, Polderman PE, Burgering Matthay KK, Villablanca JG, Seeger RC, Stram DO, HarrisRE, BM, Korswagen HC. (2005). Functional interaction between Ramsay NK et al. (1999). Treatment of high-risk neuroblastoma with beta-catenin and FOXO in oxidative stress signaling. Science 308: intensive chemotherapy, radiotherapy, autologous bone marrow 1181–1184. transplantation, and 13-cis-retinoic acid. Children’s Cancer Group. GilesRH, van EsJH, CleversH. (2003). Caught up in a Wnt storm: N Engl J Med 341: 1165–1173. Wnt signaling in cancer. Biochim Biophys Acta 1653: 1–24. Miller JR, Hocking AM, Brown JD, Moon RT. (1999). Mechanism Goss KH, Groden J. (2000). Biology of the adenomatous polyposis and function of signal transduction by the Wnt/beta-catenin and coli tumor suppressor. J Clin Oncol 18: 1967–1979. Wnt/Ca2+ pathways. Oncogene 18: 7860–7872. Grandori C, Cowley SM, JamesLP, Eisenman RN. (2000). The Myc/ Moon RT, Bowerman B, BoutrosM, Perrimon N. (2002). The promise Max/Mad network and the transcriptional control of cell behavior. and perilsof Wnt signaling through beta-catenin. Science 296: Annu Rev Cell Dev Biol 16: 653–699. 1644–1646. He TC, SparksAB, Rago C, Hermeking H, Zawel L, da CostaLT Moon RT, Kohn AD, De Ferrari GV, KaykasA. (2004). WNT and et al. (1998). Identification of c-MYC asa target of the APC beta-catenin signalling: diseases and therapies. Nat Rev Genet pathway. Science 281: 1509–1512. 5: 691–701.

Oncogene Deregulated b-catenin in neuroblastoma XLiuet al 1488 Morin PJ, SparksAB, Korinek V, Barker N, CleversH, VogelsteinB Taipale J, Beachy PA. (2001). The hedgehog and Wnt signalling et al. (1997). Activation of beta-catenin-Tcf signaling in colon pathwaysin cancer. Nature 411: 349–354. cancer by mutationsin beta-catenin or APC. Science 275: Thompson PM, Maris JM, Hogarty MD, Seeger RC, Reynolds CP, 1787–1790. Brodeur GM et al. (2001). Homozygousdeletion of CDKN2A Reya T, Clevers H. (2005). Wnt signalling in stem cells and cancer. (p16INK4a/p14ARF) but not within 1p36 or at other tumor Nature 434: 843–850. suppressor loci in neuroblastoma. Cancer Res 61: 679–686. Sadee W, Yu VC, RichardsML, PreisPN, Schwab MR, BrodskyFM Voeller HJ, Truica CI, Gelmann EP. (1998). Beta-catenin mutationsin et al. (1987). Expression of neurotransmitter receptors and myc human prostate cancer. Cancer Res 58: 2520–2523. protooncogenes in subclones of a human neuroblastoma cell line. Wang Q, Diskin S, Rappaport E, Attiyeh E, Mosse Y, Shue D et al. Cancer Res 47: 5207–5212. (2006). Integrative genomics identifies distinct molecular classes of Sinner D, Rankin S, Lee M, Zorn AM. (2004). Sox17 and beta-catenin neuroblastoma and demonstrates that multiple genes are targeted cooperate to regulate the transcription of endodermal genes. by regional alterationsin DNA copy number. Cancer Res 66: Development 131: 3069–3080. 6050–6062. Subramanian A, Tamayo P, Mootha VK, Mukherjee S, Ebert BL, Zechner D, Fujita Y, Hulsken J, Muller T, Walther I, Taketo MM Gillette MA et al. (2005). Gene set enrichment analysis: a knowl- et al. (2003). Beta-Catenin signals regulate cell growth and the edge-based approach for interpreting genome-wide expression balance between progenitor cell expansion and differentiation in the profiles. Proc Natl Acad Sci USA 102: 15545–15550. nervous system. Dev Biol 258: 406–418.

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

Oncogene