(2009) 28, 1605–1615 & 2009 Macmillan Publishers Limited All rights reserved 0950-9232/09 $32.00 www.nature.com/onc ORIGINAL ARTICLE Patched1 deletion increases N- stability as a mechanism of initiation and progression

WD Thomas1,4, J Chen1,4, YR Gao1, B Cheung1, J Koach1, E Sekyere1, MD Norris1, M Haber1, T Ellis2, B Wainwright2 and GM Marshall1,3

1Children’s Cancer Institute Australia for Medical Research, Randwick, New South Wales, Australia; 2Institute for Molecular Bioscience, University of Queensland, Brisbane, Queensland, Australia and 3Centre for Children’s Cancer and Blood Disorders, Sydney Children’s Hospital, Randwick, New South Wales, Australia

Medulloblastoma tumorigenesis caused by inactivating Introduction mutations in the PATCHED1 (PTCH1) is initiated by persistently activated (Shh) signaling More than half of all child cancer types arise in in granule neuron precursors (GNPs) during the late embryonal cells that have persisted beyond birth, to stages of cerebellar development. Both normal cerebellar later undergo changes that result in clinical tumors development and Shh-driven medulloblastoma tumorigen- (Pizzo and Poplack, 2002). However, the mechanisms of esis require N-Myc expression. However, the mechanisms embryonal cancer initiation, promotion and progression by which N-Myc affects the stages of medulloblastoma are poorly understood (Knudson, 1971; Felsher, 2008). initiation and progression are unknown. Here we used a During normal embryonal neurodevelopment there is mouse model of Ptch1 heterozygosity and medulloblasto- initially a proliferation and migration of neural crest ma to show that increased N-Myc expression character- cells, although, only 15–40% of post-migratory cells ized the earliest selection of focal GNP hyperplasia survive to form mature neural tissues (Ferrer et al., destined for later tumor progression. Step-wise loss of 1992; Lossi and Merighi, 2003). Persistent embryonal Ptch1 expression, from tumor initiation to progression, cells with tumorigenic capacity must possess both the led to incremental increases in N-Myc protein, rather than capacity to resist cell death imposed by trophic factor mRNA, expression. Increased N-Myc resulted in en- withdrawal during the final stages of neurodevelopment hanced proliferation and death resistance of perinatal and undergo secondary changes that characterize later GNPs at tumor initiation. Sequential N-Myc protein tumor promotion and progression. phosphorylation at serine-62 and serine-62/threonine-58 Medulloblastoma is postulated to originate from characterized the early and late stages of medulloblasto- persistent granule neuron precursors (GNPs) in the ma tumorigenesis, respectively. Shh pathway activation external granular layer (EGL) during the final stages of led to increased Myc protein stability and reduced cerebellar development (Kadin et al., 1970; Zakhary expression of key regulatory factors. Taken together our et al., 2001). GNPs rapidly proliferate under the data identify N-Myc protein stability as the result of loss paracrine influence of a Sonic Hedgehog (Shh) signal of Ptch1, which distinguishes normal cerebellar develop- originating from the Purkinje cell layer, before migrat- ment from medulloblastoma tumorigenesis. ing and undergoing terminal differentiation at the Oncogene (2009) 28, 1605–1615; doi:10.1038/onc.2009.3; internal granular layer (IGL; Corrales et al., 2006). published online 23 February 2009 Cellular responses in the hedgehog pathway are controlled by the Shh acting through the Keywords: N-Myc; PP2A-B56; Sonic Hedgehog; medullo- Patched1 (Ptch1) and (Smo) membrane blastoma initiation; Myc . In the absence of Shh, Ptch1 maintains Smo in an inactive state, thus silencing intracellular signaling. With the binding of Shh, Ptch1 inhibition of Smo is released and the signal is transduced. As activity of the pathway is dependent on the Ptch1/Smo interaction, removal of Ptch1 from a cell activates the pathway in a cell-autonomous manner. Once activated, this pathway leads to increased expression of downstream transcrip- Correspondence: Professor GM Marshall, Molecular Carcinogenesis tion factors, including the Gli family and N-Myc Program, Children’s Cancer Institute Australia for Medical Research (Murone et al., 1999; Pomeroy et al., 2002; Kenney and Centre for Children’s Cancer and Blood Disorders, PO Box 81, et al., 2003; Oliver et al., 2003; Wechsler-Reya, 2003). Randwick, Sydney, NSW 2031, Australia. Failure to switch off growth-promoting signals to GNPs E-mail: [email protected] 4These authors contributed equally to this work. during cerebellar development, such as the Shh Received 20 August 2008; revised 10 December 2008; accepted 29 pathway, results in medulloblastoma tumorigenesis December 2008; published online 23 February 2009 (Dahmane and Ruiz i Altaba, 1999; Wechsler-Reya Medulloblastoma and N-Myc WD Thomas et al 1606 and Scott, 1999; Ruiz i Altaba et al., 2004). Overall, the protein, but not the transcript, level during tumorigen- Shh pathway is proposed to be prevalent in B25% of esis, and, that an activated Shh signal can directly human cancers (Lum and Beachy, 2004; Fogarty et al., increase Myc protein stability. Despite high levels 2005), with up to 25% of human of S62- and S62/T58-N-Myc phosphorylation in having an activated Shh signal (Pomeroy et al., 2002). tumor tissue, N-Myc protein expression was markedly Targeted deletions of a single Ptch1 allele in mice increased, and, this correlated with low expression produced cerebellar tumors at a median age of 1 year in levels of key factors involved in Myc protein 14–20% of mice (Goodrich et al., 1997; Wetmore et al., degradation. 2000). The low frequency and long latency of tumor- igenesis in these murine models suggest the existence of collaborating oncogenic events. Studies of N-Myc mRNA expression in Ptch1 þ /À Results tumors are unclear, with some reporting raised expres- sion (Oliver et al., 2003), whereas others showing little Selection of N-Myc-expressing focal hyperplasias in the difference (Lelievre et al., 2008). Elevated N-Myc perinatal cerebellum expression is present in a significant proportion of Mice with conditional and tissue-specific hemizygous human medulloblastoma (Pomeroy et al., 2002; deletion of the Ptch1 gene (Ptch1 þ /À) under the control Eberhart et al., 2004), and is required for Shh-driven of a Nestin promoter developed cerebellar tumors at a medulloblastoma tumorigenesis (Hatton et al., 2006). median age of 25.6 (±2.79) weeks in >70% of mice by Although most studies of N-Myc oncogenesis in human 52 weeks of age (Supplementary Figure S1), and were cancer tissues have documented genomic amplification histologically similar to human medulloblastoma or enhanced transcription, stabilization of the N-Myc (Goodrich et al., 1997; Hahn et al., 2000). To better degradation pathway has only recently been proposed as define the process of tumor initiation and progression, another mechanism of deregulation of the N-Myc we undertook a histologic examination of the cerebel- growth signal in embryonal cancer (Knoepfler and lum of wild-type (n ¼ 75) and Ptch1 þ /À (n ¼ 83) mice Kenney, 2006), as has been reported for Myc-driven (Figure 1). We observed the physiologic EGL regression tumorigenesis in Burkitt’s lymphoma (Bahram et al., at postnatal day 14 (P14), with the regression complete 2000). A better understanding of post-transcriptional by (Figures 1a–d). However, in cerebella from both Myc (c-Myc) regulation in cancer cells is crucial as the wild-type and Ptch1 þ /À mice we observed the elevated expression of the Myc protein is observed in continued presence of GNP remnants at the EGL. We B70% of human tumors, whereas only B20% show have designated these persisting GNPs as focal hyper- amplification or translocation in regions of high plasia. A focal hyperplasia consisted of >30 densely transcriptional activity (Nesbit et al., 1999; Arnold and packed cells having a high nuclear-to-cytoplasmic ratio Sears, 2008). by hematoxylin and eosin (H&E) staining (Figure 1d; Physiologic studies in normal GNPs show N-Myc Hansford et al., 2004). Surprisingly, wild-type litter- expression is important for cell replication, but N-Myc mates also demonstrated focal hyperplasia on the protein degradation is crucial for cell-cycle exit (Kenney surface of the cerebellum at P21, although at a much et al., 2004; Sjostrom et al., 2005; Benassi et al., 2006). lower frequency (Figure 1h). Focal hyperplasias were N-Myc transcription is increased during Shh signaling, most often observed as multiple, noncontiguous lesions with the Cdk1 complex reported to regulate the on the cerebellum of Ptch1 þ /À mice (n ¼ 6.3±1.2 at P21; phosphorylation of serine-62-N-Myc (S62P-N-Myc, Figures 1d and h). Similar histologic findings have been formerly referred to as S54-N-Myc), priming N-Myc noted in other studies of Ptch1 þ /À mouse models of for protein degradation (Oliver et al., 2003; Kenney medulloblastoma (Corcoran and Scott, 2001; Kim et al., et al., 2004; Sjostrom et al., 2005) before phosphoryla- 2003; Oliver et al., 2005). In our studies, focal tion of threonine-58 (T58P, formerly referred to as T50- hyperplasias diminished in frequency from P21, before N-Myc) by GSK-3b (Sears, 2004; Sjostrom et al., 2005). the later emergence of tumors. We designated collec- On both sites being phosphorylated the prolyl isomer- tions of >5000 cells as diffuse hyperplasia. These larger ase, Pin1, interacts with Myc to promote dephosphor- collections of malignant, or premalignant, cells were ylation of S62P by the protein phosphatase 2A (PP2A) present at a very low frequency from 3 weeks, and complex (Sears, 2004; Yeh et al., 2004). The Fbxw7 persisted after the smaller, but more numerous, focal is then able to target T58P and stimulate hyperplasias were no longer evident (Figure 1f). Thus, proteasomal degradation, a process that may be many persistent GNPs in focal hyperplasias from inhibited if USP28 binds to Fbxw7 (Popov et al., 2007). Ptch1 þ /À mice at week 3 spontaneously regressed or Here we used a Ptch1 þ /À murine model of medullo- matured by week 10, implying a process of selection of blastoma, to show that elevated N-Myc expression in GNPs that had become competent to progress to diffuse GNPs soon after birth enhanced cell replication, and hyperplasia and later tumor formation. resistance to trophic factor withdrawal at tumor To better understand the role of N-Myc in tumor initiation. Loss of Ptch1 expression from the remaining progression we examined N-Myc and proliferating cell allele and further increases in N-Myc protein expression nuclear antigen (PCNA) expression, a proliferative characterized later tumor progression. We demonstrate marker, in tissues representing the different stages of that increased N-Myc expression occurred at the tumor initiation and progression (Figure 2a). Notably,

Oncogene Medulloblastoma and N-Myc WD Thomas et al 1607

Figure 1 Medulloblastoma tumor initiation is characterized by focal granule cell precursor (GNP) hyperplasia. Photomicrographs of hematoxylin and eosin staining of the cerebellum. Arrows indicate areas of hyperplasia or tumor. (a and b) Developing cerebellar with external granule cell layer (EGL) partially regressed. (c) Developing cerebellum with no evidence of hyperplasia or EGL in wild type mice. (d) Developed cerebellum in Ptch1 þ /À mice with regressed EGL and multiple focal hyperplasias (arrows indicate hyperplasias). (e) Adult wild-type mice cerebellum. (f) Adult Ptch1 þ /À mice cerebellum with advanced diffuse hyperplasia. (g) Cerebellum tumor in a Ptch1 þ /À mice (week 24). (i–vi) Enlarged areas of photomicrographs as indicated by boxes (scale bar, 0.05 mm). (h) Analysis of focal hyperplasia from four sagittal sections of the cerebellum in wild-type and Ptch1 þ /À mice. Focal hyperplastic lesions were defined as >30 cells. Diffuse hyperplasias were defined as >5000 cells. only a small proportion (13%) of focal hyperplasias (Figure 3c), indicating that the physiologic process of expressed high levels of N-Myc (Figure 2b). In contrast, migration is not inhibited for the majority of the GNPs all diffuse hyperplasia lesions and tumors were observed from the Ptch1 þ /À EGL. The proliferating nature of the to express high N-Myc diffusely (Figure 2b). Similarly, GNPs was confirmed when we showed that the number the pattern of PCNA expression correlated with N-Myc of PCNA-positive GNPs was markedly increased in P7 in focal and diffuse hyperplasias, and tumor tissue Ptch1 þ /À mice (Figure 3d). Viable GNP cell counts (Figures 2b and c). Together, these observations suggest taken after dissociation of the cerebellum at P7 reflected that one factor allowing selection of persistent GNPs for our in vivo observation, that the EGL from Ptch1 þ /À tumor progression was an acquired increase in N-Myc mice demonstrated increased cellularity before regres- expression level, which promoted cell proliferation. sion (Figure 3bii). An in situ terminal transferase dUTP nick-end labeling (TUNEL) assay of the EGL of both P7 wild-type and Ptch1 þ /À mice did not show any EGL regression during development is preceded significant difference for the proportion of apoptotic by increased GNP cell number in the cerebellum of cells (data not shown). These results show Ptch1 þ /À mice / Ptch1 þ À mice have significantly increased numbers of replicating Granule neuron precursors normally undergo a rapid, GNPs in the EGL, which have retained the ability to postnatal expansion in the EGL, which is complete by migrate subsequently to the IGL. P21 in mice. We compared histologic features of the EGL in Ptch1 þ /À mice at P7, when the EGL achieves its maximal dimension before regression. We observed a N-Myc expression is required for the viability of postnatal significant increase in the width of the EGL in Ptch1 þ /À Ptch1 þ /À GNPs mice, in comparison to wild-type mice, for both the As GNPs are produced in excess of requirements, whole and outer layer of the EGL (Figures 3a and b), extensive cell death in the developing central nervous but not the inner layer, suggesting that Ptch1 þ /À mice system is necessary to shape the structure and systems have increased numbers of mitotically active GNPs that permit physiologic functioning. We hypothesized (Wechsler-Reya and Scott, 1999). An increased number that medulloblastoma tumorigenesis required GNPs of cells were also observed migrating through the from Ptch1 þ /À mice to be dependent on the continued molecular layer of the cerebellum of Ptch1 þ /À mice presence of N-Myc for viability on trophic factor

Oncogene Medulloblastoma and N-Myc WD Thomas et al 1608

Figure 2 Progression from focal to diffuse hyperplasia associates with N-Myc expression. (a) Immunohistochemical staining in cerebellum with hyperplastic lesions and tumor from Ptch1 þ /À mice. Focal hyperplasia lesions (week 3) were either negative or positive for N-Myc and proliferating cell nuclear antigen (PCNA; scale bar, 70 mm). Diffuse hyperplasia (week 10) and tumor (week 25) were positive for N-Myc and PCNA expression (scale bar, 0.25 mm). Isotype IgG controls were negative (data not shown). (b) The bar graph indicates percentage of focal hyperplasia (3–6 weeks), diffuse hyperplasia (4–10 weeks) and tumors demonstrating expression of N-Myc and PCNA (nX8). (c) Merged photomicrographs of focal hyperplasia from a 4-week-old cerebella (scale bar, 0.25 mm) with DAPI (4,6-diamidino-2-phenylindole, blue), PCNA (green) and N-Myc (red) expression (scale bar, 50 mm).

withdrawal. First, we established primary cultures of to decreased N-Myc expression, we observed a GNPs from P7 wild-type and Ptch1 þ /À mice, in media significantly decreased viability of Ptch1 þ /À GNPs, both containing Shh and insulin, an activator of the PI-3 with and without Shh (Figure 4d). To exclude toxicity of kinase signal that is required for GNP survival (Kenney the shN-Myc, comparable experiments with pooled et al., 2004). Shh was withdrawn for 24 h and viability N-Myc siRNA were performed with similar results was assessed. We observed a significant decrease in the (data not shown). Other trophic factor withdrawal viability of primary wild-type GNPs, following Shh assays, such as PI-3 kinase inhibition or N2 serum withdrawal, in comparison to its control. In contrast, withdrawal, resulted in a decreased survival of Ptch1 þ /À Ptch1 þ /À mice GNPs showed no significant difference in GNPs in comparison to wild-type GNPs (Supplemen- survival after Shh withdrawal (Figure 4a). Increased tary Figure S2). Importantly, the above results indicated apoptosis measured by Annexin V binding took place in that at the time of tumor initiation, N-Myc expression wild-type GNPs when Shh was withdrawn compared to mediated death resistance to trophic factor withdrawal Ptch1 þ /À GNPs (Figure 4b). To identify whether the in GNPs from Ptch1 þ /À mice. increased N-Myc expression observed in Ptch1 þ /À GNPs was responsible for increased survival after Shh with- drawal, we transiently transfected scrambled shRNA, or N-Myc expression is progressively upregulated by shRNA targeted against N-Myc (shN-Myc) into Ptch1 þ /À post-translational mechanisms during tumorigenesis GNPs. We obtained a 70% decrease in N-Myc in Shh-pathway-activated GNPs expression for shN-Myc-transfected Ptch1 þ /À GNPs, in We next studied the effect of Ptch1 þ /À on N-Myc comparison to scrambled GNPs (Figure 4c). In response transcription and translation. We found no significant

Oncogene Medulloblastoma and N-Myc WD Thomas et al 1609

Figure 3 Medulloblastoma initiation is characterized by an increase in the replicating cells in a thickened outer external granular layer (EGL). (a) hematoxylin and eosin (H&E) staining of the cerebellum and EGL of P7 mice. Boxes represent enlarged areas with markers indicating width of the inner and outer layer of the EGL (scale bar, 35 mm). (b) (i) Analysis of EGL thickness, the statistical significance was determined by comparison to wild type (nX8). (ii) The number of viable GNPs determined by Trypan blue exclusion from dissociated EGL from P7 cerebellum. (c) (i) Photomicrographs of H&E staining of P14 mice cerebellum (scale bars, 35 mm). Arrows indicate the molecular layer containing migrating cells. (ii) Analysis of the number of cells within the molecular layer of P14 mice cerebellum (nX8). (d) (i) Photomicrographs of the EGL in P7 mice with H&E and proliferating cell nuclear antigen (PCNA) staining (scale bars, 35 mm). Arrows indicate cells stained with primary antibody PCNA (brown). (ii) Analysis of PCNA expression in the EGL of P7 mice cerebellum (nX8). The data are presented as mean±s.e.m.; *Po0.05 and **Po0.01 were considered significant. difference in N-Myc mRNA expression between P7 Different forms of phosphorylated N-Myc are expressed wild-type GNPs, P7 Ptch1 þ /À GNPs, and Ptch1 þ /À tumor during tumor initiation and progression tissue by either real-time PCR (Figure 5a) or reverse To determine the mechanism for enhanced N-Myc transcription (RT)–PCR (data not shown). N-Myc mRNA protein stability, we examined the expression of is expressed at high levels in primary GNPs as expected with phosphorylated S62- and T58-N-Myc, which regulates in vivo Shh stimulation. In contrast, N-Myc protein the physiologic N-Myc protein degradation process. expression was significantly elevated in Ptch1 þ /À GNPs We observed increased expression of S62P-N-Myc in (>1.4-fold) and further increased in Ptch1 þ /À tumor tissue both Ptch1 þ /À mouse P7 GNPs (>1.9-fold) and tumor (>3-fold), compared with wild-type GNPs (Figure 5b). As (>5.0-fold), in comparison to wild-type P7 GNPs expected, P7 Ptch1 þ /À GNPs expressed lower Ptch1 mRNA (Figure 6a). A similar increase in total N-Myc and protein levels than P7 wild-type cells, whereas tumor protein expression was also observed. Unexpectedly, tissue demonstrated complete loss of Ptch1 expression increased expression of the S62P/T58P-N-Myc (>2.4- (Figure5candd).WenextexaminedtherelativemRNA fold) protein form was only seen in tumor, along with expression levels of involved in the Shh pathway by decreased expression of T58P-N-Myc (>3.0-fold) RT–PCR (Figure 5d). We observed a significant increase (Figure 6a). When GNPs and tumor were treated with in expression of Gli1, and inhibition of Gli3 mRNA in cycloheximide and actinomycin-D for over 1 h, we Ptch1 þ /À tumor tissue, in contrast to P7 GNPs from determined that N-Myc protein had increased stability wild-type and Ptch1 þ /À mice. in Ptch1 þ /À GNPs in comparison to wild-type and that

Oncogene Medulloblastoma and N-Myc WD Thomas et al 1610 125 Shh Shh P7 GNPs P7 GNPs Shh withdrawn 70 Shh withdrawn 1.5 ** 100 60 ** 50 1.0 75 1.0 40 ** 50 30 * 0.5

of N-Myc 0.5 20 25 (N-Myc/Actin)

Alamar blue reduction (%) 10 % cells Annexin V positive N-Myc protein expression 0.0 0 0 Relative mRNA expression 0.0 +/- +/-

wild-type Ptch1 wild-type Ptch1 +/- +/-

tumor tumor

c Shh withdrawn wild-type Ptch1 Ptch1

i wild-type 125 Shh N-Myc shRNA N-Myc scrambled shRNA 100 N-Myc actin 75 * β ** 2M 50 wild-type GNPs ii +/- +/- Ptch1 GNPs N-Myc 25 6 tumor Alamar blue reduction (%) wild-type Ptch1 tumor actin 0 ** Ptch1 +/- scrambled N-Myc 4 Ptch1 P7 GNPs shRNA transfected actin Ptch1+/- GNPs ** * (RT-PCR ratio) 2

Figure 4 N-Myc expression is required for the effect of Ptch1 Relative expression P7 GCPs deletion on granule neuron precursor (GNP) viability and ** persistence at tumor initiation. (a) Histogram quantifying survival of primary P7 GNPs after Shh withdrawal. Cell viability was 0 ** examined using the alamarBlue assay, measured as optical density Ptch1 Gli1 Gli2 Gli3 (OD) units of absorbance, and expressed as the absorbance of Figure 5 Hemizygous Ptch1 deletion leads to increased N-Myc treated over control samples (n ¼ 5). (b) Early apoptosis was expression by post-transcriptional mechanisms. (a) Expression of measured by Annexin V binding after Shh withdrawal for 6 h in N-Myc mRNA by real-time RT-PCR. Data represent mean±s.e.m. primary P7 GNPs (nX3). (c) Expression of N-Myc with transiently for experiments performed in quadruplicate. (b)Immunoblotand transfected shRNA N-Myc or scrambled. (i) N-Myc mRNA analysis for N-Myc protein in P7 granule neuron precursors (GNPs) expression with b2M to verify sample loading. (ii) Immunoblot of and tumor. The identical blot was subjected to anti-actin antibody to N-Myc with anti-actin antibody to verify sample loading. verify sample loading. (c) Expression of Ptch1 protein in P7 GNPs (d) Histogram quantifying survival after transiently transfection and tumor by immunoblot subject to anti-actin antibody to verify with shRNA targeting N-Myc or scrambled using the alamarBlue sample loading. (d) Expression of Shh signaling pathway genes in P7 assay, measured as OD units of absorbance, and expressed as the GNPs and tumor. Data is presented as a ratio of the densitometric absorbance of treated over control samples. Data represent volume of the PCR product of Shh target and the control gene, b2M. mean±s.e.m. for experiments performed (n ¼ 6). *Po0.05 and All experiments were performed on at least three independent **Po0.01 were considered significant. occasions. Statistical significance was determined by comparing values for tumor and P7 Ptch1 þ /À mice GNPs with those of P7 wild- type GNPs. The data are presented as mean±s.e.m. (n>4); *Po0.05 and **Po0.01 were considered significant. this stability was significantly enhanced in the tumor (Figure 6b). Recent evidence suggests that the Cdk1 complex, An important barrier to analyses of Shh signaling in which contains cyclin B1, is required for S62-N-Myc medulloblastoma tumor cell lines or GNPs in vitro is the phosphorylation (Sjostrom et al., 2005). Furthermore, loss of Shh responsiveness when these cells are placed in Ptch1 can bind directly to phosphorylated cyclin B1 culture (Sasai et al., 2006). To better understand the preventing its accumulation in the nucleus (Barnes et al., mechanism by which the Shh signal regulates N-Myc 2005). We determined that there was an >60-fold protein expression, we next used an Shh-inducible increase in cyclin B1 expression in Ptch1 þ /À GNPs, and, in vitro model of the Shh signal in NIH3T3 murine a >160-fold increase in tumor tissue, compared to wild- fibroblasts (Shh-Light II cells). We demonstrated that type GNPs (Figure 6d). We were unable to observe any with the addition of Shh to the culture medium these difference between Ptch1 þ /À GNPs and tumor, in cells readily upregulate Myc and phosphorylated Myc, comparison to wild-type GNPs, for either the active or but not N-Myc, expression (Figure 6c) by post- inactive form of GSK-3b (Supplementary Figure S4), an transcriptional mechanisms with mRNA expression effector of T58-N-Myc phosphorylation in some settings unchanged (Supplementary Figure S3). When the cells (Kenney et al., 2004; Knoepfler and Kenney, 2006). were treated with cycloheximide over 1 h, in the presence Collectively, these data suggest that those consecutive and absence of Shh, we showed that Myc protein N-Myc phosphorylation events, which mark N-Myc for stability was increased, following Shh signal activation degradation, proceed despite increasing N-Myc protein (Figure 6d). levels during tumor progression in Ptch1 þ /À mice.

Oncogene Medulloblastoma and N-Myc WD Thomas et al 1611 P7 GNPs 2.0 S62P-N-Myc S62P/T58P-N-Myc S62P-N-Myc T58P-N-Myc 100 1.5 ** S62P/T58P- N-Myc 1.0 wild-type GNPs T58P-N-Myc ** Ptch1+/- GNPs % remaining N-Myc 0.5 * tumor actin * 10 Protein expression (N-Myc/actin) +/- 0.0 0 15304560 wild-type Ptch1+/- tumor Minutes tumor Ptch1

wild-type P7 GNPs

- Shh + Shh 0.85 ** cytoplasm ** N-Myc 0.45 nuclear 100 total ** Myc 0.05

S62P-Myc * 0.02 ** S62/PT58P- - Shh Myc % remaining Myc 0.01 + Shh % cyclin B1 positive actin 10 0.00 0 15304560 wild-type Ptch1+/- tumor Shh-light Minutes II cells P7 GNPs Figure 6 Activation of the Shh pathway phosphorylates both Myc and N-Myc. Tumor initiation is associated with S62P-N-Myc whereas phosphorylation of both S62P- and T58P-N-Myc is associated with tumor progression. (a) Immunoblot and analysis of the different forms of N-Myc in P7 granule neuron precursors (GNPs) and tumor. The data are presented as mean±s.e.m. (nX3) with significance determined by comparison to wild-type GNPs. (b) Pulse-chase analysis of endogenous N-Myc stability in P7 GNPs and tumor after treatment with actinomycin-D and cycloheximide for the indicated times. (c) Immunoblot showing expression of N-Myc, Myc and phosphorylated Myc after Shh-Light cells are stimulated with Shh for 24 h (actin, loading controls). (d) Pulse-chase analysis of endogenous Myc stability in Shh-Light II cells stimulated with Shh and control after treatment with cycloheximide for the indicated times. (e) Quantitation of cyclin B1 expression. The data are presented as mean±s.e.m. (nX5). Statistical significance was determined by comparing values for with those of wild-type GNPs. The data are presented as mean±s.e.m. (nX3). *Po0.05 and **Po0.01 were considered significant.

The Shh pathway regulates the N-Myc degradation N-Myc proteins are DNA-binding proteins that can pathway by multiple mechanisms activate or repress transcription of effector target genes As we observed increased S62P/T58P-N-Myc expression (Blackwood and Eisenman, 1991). Thus, we examined only on progression to tumor we assessed the mRNA transcript expression of well-known N-Myc target genes expression levels of key proteins associated with the Myc in wild-type and Ptch1 þ /À P7 GNPs, and tumor tissue degradation pathway in wild-type and Ptch1 þ /À P7 (Figure 7b). None of the nine N-Myc target gene GNPs and tumor tissue. Evidence (Yeh et al., 2004) transcript levels was altered by Shh signal activation in suggests that the Myc proteolysis effector proteins, Pin1 the Ptch1 þ /À GNPs. However, we observed significant and the PP2A–B56 complex, direct S62P dephosphor- repression of Activin-A, Id2 and Pax-3, and in contrast, ylation leading to an increase in Myc protein degrada- upregulation of and Tert, in medulloblastoma tion. We observed decreased expression of Pin1, PP2A- tumor tissue from Ptch1 þ /À mice, compared to GNPs, B56 isoforms a, b, g and d (data not shown) in tumor suggesting these proteins may be involved in the later (Figure 7a). Analysis further downstream on Fbxw7 and stages of tumor progression. USP28, both known to influence the stability of T58P- Myc, identified the two Fbxw7 isoforms a-and b-mRNA to be similarly reduced in tumor tissue only (Figure 7a). The additional control mechanism for Myc Discussion degradation, USP28, which associates with Fbxw7, was not altered. Protein analyses of Fbxw7a and -b did not The final stages of cerebellar development require show reduced expression in tumor in comparison to transient, N-Myc-driven GNP proliferation to provide GNPs (Supplementary Figure S5). the necessary population of mature granule neurons for

Oncogene Medulloblastoma and N-Myc WD Thomas et al 1612 Shh- N-Myc Gli pathway

P7 GCPs Cdk1 complex Pin1 wild-type P7 GNPs 1.5 Ptch1+/- P7 GNPs PP2A-B56α S62P- Ptch1+/- tumor N-Myc CIP2A 1.0 Fbxw7α GSK-3β ** ** Fbxw7β ** ** 0.5 ** USP28 S62P-T58P- β N-Myc 2M 0.0

+/- α α β Relative Expression (RT-PCR Ratio) Pin1

tumor CIP2A Pin1 Fbxw7 Fbxw7 USP28 Ptch1 wild-type PP2A-B56

3.5 PP2A-B56α CIP2A wild-type GNPs complex 3.0 Ptch1+/- GNPs tumor 2.5 T58P- N-Myc 2.0

1.5 ** Fbxw7 USP28

(RT-PCR ratio) 1.0 **

Relative expression * * 0.5 T58P- * Ubiquination 0.0 ** N-Myc

Id2 Tert Odc Pax-3 Mdm2 Mcm7 Abcc1 Ptma Activin-A

Proteosome degradation Figure 7 The Shh pathway regulates the Myc degradation pathway correlating with changes in N-Myc transcriptional target genes. (a) Reverse transcription (RT)–PCR of genes associated with Myc degradation pathway in P7 granule cell precursors (GNPs) and tumor, and analysis of the relative mRNA expression levels. Data are presented as a ratio of the densitometric volume of the PCR product test genes and control gene b2M.(b) mRNA expression levels of N-Myc target genes. Data are presented as a ratio of the densitometric volume of the PCR product on the control gene, mouse b-actin. All experiments were performed on at least three independent occasions. The data are presented as mean±s.e.m; *Po0.05 and **Po0.01 were considered significant. (c) Hypothetical model of the regulatory associations with the N-Myc degradation pathway.

a functional IGL, and, yet continued N-Myc expression data raise important issues regarding the regulation of in GNPs is a key factor contributing to Shh-driven normal cerebellar development and factors initiating and medulloblastoma tumorigenesis (Hatton et al., 2006). maintaining medulloblastoma tumorigenesis. Our data suggest that in the first week of life GNPs with Post-translational modification of Myc proteins, such hemizygous Ptch1 deletion acquire a higher N-Myc as phosphorylation, constitutes another level of Myc protein expression level as an early event, increasing regulation that may have a significant function in many their number and resistance to the physiologic influences cellular processes including , growth, apoptosis that shape terminal cerebellar development. Indeed, only and tumorigenesis. Our data identify an increased S62P- a small proportion of GNP hyperplasias in Ptch1 þ /À N-Myc expression at tumor initiation, and later during cerebella persist beyond the first weeks of life, and this tumor progression, the acquisition of S62P/T58P-N-Myc. closely correlated with N-Myc expression and the This is in contrast to the proposed sequence of replicative state. Tumor promotion and progression phosphorylation steps that characterized N-Myc pro- required further increases in N-Myc and, this was teolysis during normal terminal GNP differentiation, as provided by loss of Ptch1 expression from the remaining N-Myc expression is shut down (Oliver et al., 2003; allele, and a reduction in the expression levels of factors Kenney et al., 2004; Sjostrom et al., 2005). Phosphor- regulating the Myc protein degradation pathway. Our ylation at S62 stabilized N-Myc, whereas subsequent

Oncogene Medulloblastoma and N-Myc WD Thomas et al 1613 phosphorylation at T58 was required for N-Myc associated with tumor aggressiveness (Smith et al., degradation (Sears et al., 2000; Kenney et al., 2004; 2006). Sjostrom et al., 2005). Consistent with these observa- In our experiments, tumor cells had complete loss of tions, stabilized Myc through T58 mutation induced Ptch1 expression as a collaborating oncogenic event. oncogenic transformation in human cells (Bahram et al., It is not clear in other Ptch1 þ /À medulloblastoma models 2000; Yeh et al., 2004). Furthermore, in vivo retroviral whether Ptch1 was expressed or silenced (Wetmore studies expressing N-Myc mutated at T58 in GNPs et al., 2000; Romer et al., 2004; Oliver et al., 2005), with significantly enhanced Shh retroviral-induced cerebellar the use of specific Ptch1 primers critical in the tumor incidence and aggressiveness, when compared to determination of Ptch1 expression (Oliver et al., 2005). wild-type N-Myc (Browd et al., 2006). Our data showing Our studies are consistent with the complete loss of high levels of S62/T58 phosphorylation and not T58 Ptch1 expression being a necessary event for tumor phosphorylation at the late tumor stage indicate a progression. The mechanism for silencing PTCH1 gene failure of N-Myc protein degradation at the S62 expression in sporadic medulloblastoma is still undeter- dephosphorylation step in the proteolysis pathway mined (Pritchard and Olson, 2008). (Figure 7c). Our in vivo findings strongly suggest Pin1 As another layer of regulation, Myc can activate a and/or PP2A-B56a as candidate factors permitting death program through senescence or apoptosis in increased N-Myc protein stability in tumor. Pin1 and certain cell types, in addition to stimulating cell PP2A-B56a proteins have been shown to possess tumor proliferation (Evan et al., 1992; Dang, 1999). Our work suppressor properties (Arnold and Sears, 2006; Junttila in perinatal GNPs, and previously in perinatal neuro- et al., 2007). Expression of Fbxw7a and -b protein was blasts (Hansford et al., 2004), indicates that certain found to be unchanged in tumor compared to GNPs. It embryonal cell types are not only susceptible to the Myc is unclear whether the transcriptional repression of these proliferative signal, but that Myc expression in these Myc degradation factors was due to the activated Shh cells conversely inhibits cell death. Consistent with these signal, N-Myc alone or simply arose as a general data Ptch1 þ /À GNPs that have increased phosphory- consequence of genomic instability that occurred during lated S62-N-Myc show increased sensitivity to cell death the course of medulloblastoma tumorigenesis. In vitro when PI-3 kinase is inhibited. Elucidation of the studies on Shh-Light II cells have also indicated that the mechanism transiently repressing Myc-induced apopto- Shh pathway may have a function when re-activated in sis in these specialized embryonal cells will be an adult cancers of inhibiting and stabilizing just not important step toward understanding those factors that N-Myc, but Myc degradation, thereby permitting collaborate with N-Myc to initiate embryonal tumor- increased expression of this oncogene. As most previous igenesis. analyses of N-Myc expression in human medulloblas- toma have assessed only N-Myc mRNA levels and/or amplification, it may also be necessary to reassess Materials and methods N-Myc protein expression levels in human medulloblas- toma tissues (Garson et al., 1989; Moriuchi et al., 1996) Mice and its relevance to disease phenotype. Mice with a conditional allele of homolog 1 (Ptch1flox/ þ ) An expanded EGL was observed in this model, were mated with transgenic mice carrying the Cre recombinase similarly in ND2:SmoA1 P14 mice with a constitutively gene under the control of the rat nestin promoter and enhancer activated Smo gene (Hatton et al., 2008). It may be that (Tronche et al., 1999; Ellis et al., 2003). Progeny were other groups have not previously noted an expanded genotyped by PCR after blinded experiments (Ellis et al., 2003). Cerebella were dissected into four equal sagittal parts, EGL in Ptch1 þ /À mice with analysis carried out by a paraffin-embedded, sectioned and stained with H&E. The total of percentage positive 5-bromo-2-deoxyuridine average width of the EGL was determined by eight measure- cells of the EGL. Several groups have previously noted ments, covering the same areas of the cerebellum for each the presence of precancerous lesions in Shh-driven genotype. The width of the inner and outer EGL was models of medulloblastoma, and that this was N-Myc determined by the cell morphology according to Koppel’s dependent (Goodrich et al., 1997; Corcoran and Scott, method (Koppel and Lewis, 1983). Cell counts from the 2001; Kim et al., 2003; Oliver et al., 2005). Here we have molecular layer were made from 6 to 8 areas (0.01–0.03 mm2) for the first time shown that focal hyperplasias in the per section. A total of 157 mice were used in a blinded, weekly, perinatal period appeared to be selected for tumor histologic examination of postnatal cerebellum. H&E sections promotion and progression on the basis of N-Myc were reviewed under light microscopy by two independent expression. Moreover, a further incremental increase in observers and each section scored for the presence of >30-cell focal hyperplasia, diffuse hyperplasia >5000 cells and tumor. N-Myc expression was required to progress from GNP hyperplasia to medulloblastoma. The higher N-Myc expression level, resulting from increased N-Myc protein Cell line and protein stability experiments stability, seen in medulloblastoma tumor cells correlated Shh-Light II cells derived from NIH/3T3 cell line were obtained from the ATCC (CRL-2795) and cultured in with effects on some known N-Myc transcriptional Dulbecco’s modified Eagle’s medium with G-418 (0.4 mg/ml), target genes: Activin-A, Id2, Tert, MDM2 and Pax3. Zeocin (0.15 mg/ml) and 10% fetal calf serum. Shh-Light II Little is known of the phenotypic or cells were plated into 25 cm2 flasks at 104 cells per cm2, and profile outcomes in cancer cells expressing different stimulated with 40 ng/ml of Shh (R&D Systems, Minneapolis, levels of N-Myc, although increasing Myc gene dosage is MN, USA) for 24–36 h. Protein half-life was determined by

Oncogene Medulloblastoma and N-Myc WD Thomas et al 1614 treating cells with 20 mg/ml of cycloheximide for the designated RT–PCR and immunoblot. Following transfection with either length of time followed by immunoblot analysis. Primary scrambled shRNA or shRNA targeting N-Myc, the cells were GNPs and tumor protein half-life was determined by treat- cultured with complete media plus 20 ng/ml of Shh overnight ment with actinomycin-D (20 mg/ml; Sigma, Australia) and and then separated into two populations, incubated for a cycloheximide followed by immunoblot analysis. further 24 h with and without Shh, and viability was assessed using the alamarBlue assay. Isolation of GNPs and tumor cells for culture GNPs were isolated from the cerebellum of P7 mice, and þ /À Supplementary material tumor cells from 10- to 45-week-old Ptch1 mice. Cells were Materials and methods for Immunohistochemistry, TUNEL, purified by digesting in Hank’s balanced salt solution (HBSS; immunoblot, RT–PCR, real-time PCR and Annexin V are Invitrogen, Mount Waverley, VIC, Australia) containing L- given in Supplementary material. cysteine (0.20 mg/ml) and papain (20 U/ml; Sigma) at 37 1C for 30 min with shaking. Samples were washed, resuspended in HBSS containing 250 U/ml DNase, 4 mM Mg, and were Statistical analysis triturated. Cells were then transferred to HBSS containing Results are expressed as mean values with standard error of 100 U/ml DNase and washed. Dissociated cells were washed the mean (s.e.m). For experiments involving two groups of with complete media: neurobasal media (Invitrogen) supple- mice, comparisons were made using one-way analysis of mented with N2 (1% v/v; Invitrogen) and sodium pyruvate variance. Po0.05 (*) and 0.01 (**) were considered statisti- (1 mM). The cell suspension was centrifuged through a step cally significant. gradient of 30 and 60% Percoll buffer (GE Healthcare, Piscataway, NJ, USA). Cells obtained from the interface were recognized as purified GNPs and were 98% positive for Math- Ethical approval 1 (data not shown). GNPs were cultured in complete media The present study was approved by the animal care and ethics with Shh (20 ng/ml) at 37 1C overnight before cell viability committee of the University of New South Wales, Sydney, assays. GNPs for RNA and protein extraction were immedi- Australia and was conducted under the Animal Research Act ately snap-frozen in liquid nitrogen after purification. 1985 (NSW, Australia) and the Australian Code of Practice for the Care and Use of Animals for Scientific Purposes (1997). Cell viability Purified GNPs were seeded in a 96-well plate with complete Acknowledgements media at 3.0 Â 105 cells per cm2 with Shh (20 ng/ml) or control for 30 h. Cell viability was assessed using the alamarBlue This work was supported by funds from the Cure For Life reagent as previously described (Haber et al., 1999). Foundation (WT), the Cure Cancer Australia Foundation (WT), the National Health and Medical Research Council Transient transfection (Australia; GMM, MDN, BW and MH), the Cancer Institute GNPs (2 Â 106 cells, 5 mg shRNA; SuperArray, Frederick, MD, New South Wales (Australia; GMM, MDN and MH) and USA) were transfected using the Nucleofector and mouse John Trivett Foundation (TE). The Children’s Cancer neuron (Amaxa, Gaithersburg, MD, USA) as per the Institute Australia for Medical Research is affiliated with the manufacturer’s protocol. Primary cells transfected with University of New South Wales and Sydney Children’s shRNA were assessed at 24–48 h after transfection by Hospital.

References

Arnold HK, Sears RC. (2006). Protein phosphatase 2A regulatory Corrales JD, Blaess S, Mahoney EM, Joyner AL. (2006). The level of subunit B56alpha associates with c-myc and negatively regulates sonic hedgehog signaling regulates the complexity of cerebellar c-myc accumulation. Mol Cell Biol 26: 2832–2844. foliation. Development 133: 1811–1821. Arnold HK, Sears RC. (2008). A tumor suppressor role for PP2A- Dahmane N, Ruiz i Altaba A. (1999). Sonic hedgehog regulates the B56alpha through negative regulation of c-Myc and other key growth and patterning of the cerebellum. Development 126: 3089–3100. oncoproteins. Cancer Metastasis Rev 27: 147–158. Dang CV. (1999). c-Myc target genes involved in cell growth, Bahram F, von der Lehr N, Cetinkaya C, Larsson LG. (2000). c-Myc apoptosis, and metabolism. Mol Cell Biol 19: 1–11. hot spot mutations in lymphomas result in inefficient ubiquiti- Eberhart CG, Kratz J, Wang Y, Summers K, Stearns D, Cohen K nation and decreased proteasome-mediated turnover. Blood 95: et al. (2004). Histopathological and molecular prognostic markers in 2104–2110. medulloblastoma: c-myc, N-myc, TrkC, and anaplasia. J Neuro- Barnes EA, Heidtman KJ, Donoghue DJ. (2005). Constitutive pathol Exp Neurol 63: 441–449. activation of the shh–ptc1 pathway by a patched1 mutation Ellis T, Smyth I, Riley E, Graham S, Elliot K, Narang M et al. (2003). identified in BCC. Oncogene 24: 902–915. Patched 1 conditional null allele in mice. Genesis 36: 158–161. Benassi B, Fanciulli M, Fiorentino F, Porrello A, Chiorino G, Loda M Evan GI, Wyllie AH, Gilbert CS, Littlewood TD, Land H, Brooks M et al. (2006). c-Myc phosphorylation is required for cellular response et al. (1992). Induction of apoptosis in fibroblasts by c-myc protein. to oxidative stress. Mol Cell 21: 509–519. Cell 69: 119–128. Blackwood EM, Eisenman RN. (1991). Max: a helix-loop-helix zipper Felsher DW. (2008). Oncogene addiction versus oncogene amnesia: protein that forms a sequence-specific DNA-binding complex with perhaps more than just a bad habit? Cancer Res 68: 3081–3086; Myc. Science 251: 1211–1217. discussion 3086. Browd SR, Kenney AM, Gottfried ON, Yoon JW, Walterhouse D, Ferrer I, Soriano E, del Rio JA, Alcantara S, Auladell C. (1992). Cell Pedone CA et al. (2006). N-myc can substitute for insulin-like death and removal in the cerebral cortex during development. Prog signaling in a mouse model of sonic hedgehog-induced Neurobiol 39: 1–43. medulloblastoma. Cancer Res 66: 2666–2672. Fogarty MP, Kessler JD, Wechsler-Reya RJ. (2005). Morphing into Corcoran RB, Scott MP. (2001). A mouse model for medulloblastoma cancer: the role of developmental signaling pathways in brain tumor and basal cell nevus syndrome. J Neurooncol 53: 307–318. formation. J Neurobiol 64: 458–475.

Oncogene Medulloblastoma and N-Myc WD Thomas et al 1615 Garson JA, Pemberton LF, Sheppard PW, Varndell IM, Coakham Nesbit CE, Tersak JM, Prochownik EV. (1999). MYC and HB, Kemshead JT. (1989). N-myc gene expression and oncoprotein human neoplastic disease. Oncogene 18: 3004–3016. characterisation in medulloblastoma. Br J Cancer 59: 889–894. Oliver TG, Grasfeder LL, Carroll AL, Kaiser C, Gillingham CL, Lin Goodrich LV, Milenkovic L, Higgins KM, Scott MP. (1997). Altered SM et al. (2003). Transcriptional profiling of the Sonic hedgehog neural cell fates and medulloblastoma in mouse patched mutants. response: a critical role for N-myc in proliferation of neuronal Science 277: 1109–1113. precursors. Proc Natl Acad Sci USA 100: 7331–7336. Haber M, Bordow SB, Gilbert J, Madafiglio J, Kavallaris M, Marshall Oliver TG, Read TA, Kessler JD, Mehmeti A, Wells JF, Huynh TT GM et al. (1999). Altered expression of the MYCN oncogene et al. (2005). Loss of patched and disruption of granule cell modulates MRP gene expression and response to cytotoxic drugs in development in a pre-neoplastic stage of medulloblastoma. Devel- neuroblastoma cells. Oncogene 18: 2777–2782. opment 132: 2425–2439. Hahn H, Wojnowski L, Specht K, Kappler R, Calzada-Wack J, Potter Pizzo PA, Poplack DG. (2002). Principles and Practice of Pediatric D et al. (2000). Patched target Igf2 is indispensable for the formation Oncology. Lipincott Williams and Wilkins: Philadelphia, PA. of medulloblastoma and rhabdomyosarcoma. J Biol Chem 275: Pomeroy SL, Tamayo P, Gaasenbeek M, Sturla LM, Angelo M, 28341–28344. McLaughlin ME et al. (2002). Prediction of central nervous system Hansford LM, Thomas WD, Keating JM, Burkhart CA, Peaston AE, embryonal tumour outcome based on gene expression. Nature 415: Norris MD et al. (2004). Mechanisms of embryonal tumor 436–442. initiation: distinct roles for MycN expression and MYCN amplifica- Popov N, Wanzel M, Madiredjo M, Zhang D, Beijersbergen R, tion. Proc Natl Acad Sci USA 101: 12664–12669. Bernards R et al. (2007). The ubiquitin-specific protease USP28 is Hatton BA, Knoepfler PS, Kenney AM, Rowitch DH, de Alboran IM, required for MYC stability. Nat Cell Biol 9: 765–774. Olson JM et al. (2006). N-myc is an essential downstream effector of Pritchard JI, Olson JM. (2008). Methylation of PTCH1, the Patched-1 Shh signaling during both normal and neoplastic cerebellar growth. gene, in a panel of primary medulloblastomas. Cancer Genet Cancer Res 66: 8655–8661. Cytogenet 180: 47–50. Hatton BA, Villavicencio EH, Tsuchiya KD, Pritchard JI, Ditzler S, Romer JT, Kimura H, Magdaleno S, Sasai K, Fuller C, Baines H et al. Pullar B et al. (2008). The Smo/Smo model: hedgehog-induced (2004). Suppression of the Shh pathway using a small molecule medulloblastoma with 90% incidence and leptomeningeal spread. inhibitor eliminates medulloblastoma in Ptc1(+/À)(À/À) mice. Cancer Res 68: 1768–1776. Cancer Cell 6: 229–240. Junttila MR, Puustinen P, Niemela M, Ahola R, Arnold H, Bottzauw Ruiz i Altaba A, Stecca B, Sanchez P. (2004). Hedgehog—Gli signaling T et al. (2007). CIP2A inhibits PP2A in human malignancies. Cell in brain tumors: stem cells and paradevelopmental programs in 130: 51–62. cancer. Cancer Lett 204: 145–157. Kadin ME, Rubinstein LJ, Nelson JS. (1970). Neonatal cerebellar Sasai K, Romer JT, Lee Y, Finkelstein D, Fuller C, McKinnon PJ medulloblastoma originating from the fetal external granular layer. et al. (2006). Shh pathway activity is down-regulated in cultured J Neuropathol Exp Neurol 29: 583–600. medulloblastoma cells: implications for preclinical studies. Cancer Kenney AM, Cole MD, Rowitch DH. (2003). Nmyc upregulation by Res 66: 4215–4222. sonic hedgehog signaling promotes proliferation in developing Sears R, Nuckolls F, Haura E, Taya Y, Tamai K, Nevins JR. (2000). cerebellar granule neuron precursors. Development 130: 15–28. Multiple Ras-dependent phosphorylation pathways regulate Myc Kenney AM, Widlund HR, Rowitch DH. (2004). Hedgehog and PI-3 protein stability. Genes Dev 14: 2501–2514. kinase signaling converge on Nmyc1 to promote cell cycle progression Sears RC. (2004). The life cycle of C-myc: from synthesis to in cerebellar neuronal precursors. Development 131: 217–228. degradation. Cell Cycle 3: 1133–1137. Kim JY, Nelson AL, Algon SA, Graves O, Sturla LM, Goumnerova Sjostrom SK, Finn G, Hahn WC, Rowitch DH, Kenney AM. (2005). LC et al. (2003). Medulloblastoma tumorigenesis diverges from The Cdk1 complex plays a prime role in regulating N-myc cerebellar granule cell differentiation in patched heterozygous mice. phosphorylation and turnover in neural precursors. Dev Cell 9: Dev Biol 263: 50–66. 327–338. Knoepfler PS, Kenney AM. (2006). Neural precursor cycling at sonic Smith DP, Bath ML, Metcalf D, Harris AW, Cory S. (2006). MYC speed: N-Myc pedals, GSK-3 brakes. Cell Cycle 5: 47–52. levels govern hematopoietic tumor type and latency in transgenic Knudson Jr AG. (1971). Mutation and cancer: statistical study of mice. Blood 108: 653–661. retinoblastoma. Proc Natl Acad Sci USA 68: 820–823. Tronche F, Kellendonk C, Kretz O, Gass P, Anlag K, Orban PC et al. Koppel H, Lewis PD. (1983). Cell proliferation and DNA-synthesis in (1999). Disruption of the glucocorticoid gene in the nervous the external granular layer of the postnatal rat cerebellum: a system results in reduced anxiety. Nat Genet 23: 99–103. quantitative study. Neuropathol Appl Neurobiol 9: 207–214. Wechsler-Reya RJ. (2003). Analysis of gene expression in the normal Lelievre V, Seksenyan A, Nobuta H, Yong WH, Chhith S, and malignant cerebellum. Recent Prog Horm Res 58: 227–248. Niewiadomski P et al. (2008). Disruption of the PACAP gene Wechsler-Reya RJ, Scott MP. (1999). Control of neuronal precursor promotes medulloblastoma in ptc1 mutant mice. Dev Biol 313: 359–370. proliferation in the cerebellum by Sonic Hedgehog. Neuron 22: 103–114. Lossi L, Merighi A. (2003). In vivo cellular and molecular mechanisms Wetmore C, Eberhart DE, Curran T. (2000). The normal patched of neuronal apoptosis in the mammalian CNS. Prog Neurobiol 69: allele is expressed in medulloblastomas from mice with heterozygous 287–312. germ-line mutation of patched. Cancer Res 60: 2239–2246. Lum L, Beachy PA. (2004). The Hedgehog response network: sensors, Yeh E, Cunningham M, Arnold H, Chasse D, Monteith T, Ivaldi G switches, and routers. Science 304: 1755–1759. et al. (2004). A signalling pathway controlling c-Myc degradation Moriuchi S, Shimizu K, Miyao Y, Hayakawa T. (1996). An that impacts oncogenic transformation of human cells. Nat Cell Biol immunohistochemical analysis of medulloblastoma and PNET with 6: 308–318. emphasis on N-myc protein expression. Anticancer Res 16: 2687–2692. Zakhary RK, Keles GE, Aldape K, Berger MS. (2001) Medulloblasto- Murone M, Rosenthal A, de Sauvage FJ. (1999). Sonic hedgehog ma and primitive neuroectodermal tumors. In: Kaye AH, Law ER signaling by the patched-smoothened receptor complex. Curr Biol 9: (eds). Brain Tumours: An Encyclopedic Approach. Churchill Livingstone: 76–84. London, pp 605–615.

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

Oncogene