Oncogene (2010) 29, 3465–3476 & 2010 Macmillan Publishers Limited All rights reserved 0950-9232/10 www.nature.com/onc ORIGINAL ARTICLE Canonical Notch signaling is not required for the growth of Hedgehog pathway-induced medulloblastoma

E Julian1, RK Dave1, JP Robson1, AR Hallahan2 and BJ Wainwright1

1Institute for Molecular Bioscience, University of Queensland, Brisbane, Australia and 2Department of Paediatrics and Child Health, University of Queensland, Brisbane, Australia

Current treatment for medulloblastoma is successful in deficiencies, renal failure, hearing loss and psychiatric more than half of all cases but results in substantial and social difficulties. The elucidation of signaling disability in survivors. Accordingly, there is considerable pathways involved in medulloblastoma pathogenesis interest in drugs that may target specific signaling could substantially improve the clinical management of pathways activated in the tumors, with inhibitors of both these neoplasms, a more accurate prediction of the the Hedgehog and Notch pathways currently proposed as disease risk could be achieved and new targeted possible therapeutics. Here, we tested the hypothesis that treatments developed. Notch pathway inhibition in vivo may block the formation A number of studies have identified candidate- of Hedgehog-dependent medulloblastoma. We took the signaling pathways, which may have a role in the general approach of using a cre recombinase under the genesis of medulloblastoma (Marino, 2005). The poten- control of the GFAP promoter to generate medulloblas- tial complexity of the interaction of signaling pathways toma in mice carrying a conditional Ptc1 allele and that give rise to medulloblastoma is reflected in the introduced a conditional RBP-J allele to ablate canonical diversity of knockout and transgenic mouse models that Notch signaling. Loss of RBP-J from the developing have been found to develop medulloblastoma. These cerebellum led to a modest loss of stem cells and an overall include cell cycle regulators such as Ink4c (Zindy et al., developmental delay. These phenotypes could be partially 2007), p53/Rb (Marino et al., 2000; Wetmore et al., compensated by activation of the Hedgehog pathway. 2001), Bmi1 (Leung et al., 2004), RenKCTD11 (Di Hedgehog-dependent medulloblastoma were not blocked Marcotullio et al., 2004), Brca2 (Frappart et al., 2007) by loss of RBP-J, indicating that canonical Notch signaling and Parp (Tanori et al., 2008), chemokine receptor is not required for tumor initiation and growth in this model. Cxcr6 (Sasai et al., 2007), growth factors Hgf (Binning Oncogene (2010) 29, 3465–3476; doi:10.1038/onc.2010.101; et al., 2008), Igf1 (Rao et al., 2004) and Ifna (Wang published online 26 April 2010 et al., 2003), a regulator of PKA activity, Pacap (Lelievre et al., 2008), and the transcription factor Keywords: medulloblastoma; Notch; Hedgehog; therapy; Hic1 (Briggs et al., 2008). In the mouse at least most of RBP-J these manipulations appear to either directly modulate the Hedgehog-signaling pathway or cooperate with it to generate the medulloblastoma. Insight into the molecular basis of medulloblastoma Introduction has come from studies of the Hedgehog-signaling pathway. Sonic hedgehog (Shh) is a secreted protein Medulloblastoma are tumors of the cerebellum that that has a critical role in patterning of the nervous most commonly occur within the first few years of life. system, the limbs, the skin and other tissues (Wicking Children with the disease face a 3-year event-free et al., 1999). Patched1 (Ptc1) is a transmembrane protein survival rate of less than 35% if the tumor is metastatic that can bind Shh and appears to serve as a receptor. or recurrent or if the child is o3 years old (Packer, The first evidence that Hedgehog signaling might be 1994). The approach of clinical trial cooperative groups involved in medulloblastoma came from studies of the toward these patients has been the intensification of human PATCHED (PTCH1) gene in an inherited post-surgical chemotherapeutic and brain radiation cancer disposition syndrome, Naevoid Basal Cell regimens, but recent advances have been limited. Carcinoma Syndrome (NBCCS), which is characterized Even when successful, current therapies cause major toxi- by widespread skin tumors, craniofacial and skeletal cities, including severe cognitive impairment, endocrine abnormalities, and an increased incidence of medullo- blastoma. PTCH1 was found to be mutated in both the Correspondence: Professor BJ Wainwright, Institute for Molecular inherited and sporadic form of basal cell carcinoma Bioscience, The University of Queensland, Building 80, St Lucia (Hahn et al., 1996) and subsequently it was discovered Campus, Brisbane, Queensland 4072, Australia. that in some studies up to 20% of sporadic medullo- E-mail: [email protected] Received 27 September 2009; revised 15 February 2010; accepted blastoma have causative mutations at the PTCH1 locus 4 March 2010; published online 26 April 2010 (Pietsch et al., 1997; Vorechovsky et al., 1997; Dong RBP-J is not required for Hedgehog-induced medulloblastoma E Julian et al 3466 et al., 2000). Mutation of the Hedgehog pathway media- accepted and relevant model for human medullo- tor suppressor of fused has also been observed in blastoma (Pazzaglia et al., 2006; Yang et al., 2008; medulloblastoma and provides a potential mechanistic Thomas et al., 2009), and here we addressed the link between Hedgehog signaling and the Wnt signaling intersection of Hedgehog and Notch signaling in vivo pathway in this tumor (Taylor et al., 2002). Although using a genetic approach to ablate Notch signaling in mutation of Hedgehog pathway components is present this model. The developing cerebellum expresses multi- in a minority of medulloblastoma (approximately 25%), ple components of the Notch-signaling pathway, in- the pathway is detected as perturbed by unknown cluding at least 4 ligands, 4 receptors and multiple mechanisms in most tumors (Hallahan et al., 2004; putative downstream effectors, and the rhombic lip/ Thompson et al., 2006). Small molecule inhibitors of the EGL itself expresses receptors Notch1, Notch2, Notch4 Hedgehog pathway show significant activity in Ptc1 and ligands Jag1, Dll1, at least. Although the combina- medulloblastoma animal models and on that basis are torial complexity of Notch ligand/receptor is high, all currently in clinical trials (Romer and Curran, 2005; ‘canonical’ Notch signaling requires the CSL/Rbpj Rudin et al., 2009). (RBP-J) transcription co-factor so in practice, inhibition The Notch receptor-signaling pathway has a key role of RBP-J function will negate Notch receptor signaling in many different cell fate decisions and has a particular (Komine et al., 2007). There is some emerging genetic prominence due to its function as a key determinant of evidence to suggest the existence of an RBP-J-indepen- asymmetric/symmetric cell division (Corbin et al., 2008). dent (‘non-canonical’) pathway through which Notch Notch signaling uses a membrane bound ligand that receptor signaling may exert some influence (Talora contacts an appropriate receptor leading to the release et al., 2008). However, the key effectors implicated in of a cytoplasmic fragment of the receptor, which then medulloblastoma growth, Hes1 and Hes5 are clearly binds to a multi-unit complex capable of transcriptional both ‘canonical’ Notch targets (Jarriault et al., 1995, regulation, transforming it from a repressor to an 1998; Sweeney et al., 2004; Ong et al., 2006). Given the activator complex. In mammals, ligands include mem- number of ligands and receptors potentially having a bers of the Delta-like (Dll1, Dll3, Dll4) and Jagged role in the cerebellum, here we took the approach of (Jag1, Jag2) families, and receptors include Notch1–4. blocking Notch signaling by deleting the common Given the array of Notch ligands and receptors it is not pathway effector RBP-J, which has proved to be a surprising that this pathway forms a complex network, highly effective strategy in analyses of Notch signaling which interacts with many other pathways such as (Fujikura et al., 2006; Komine et al., 2007). Our results Hedgehog, Wnt/wingless and TGFb (Hurlbut et al., indicate that although loss of RBP-J function results in a 2007). The Notch pathway likely has a significant role in developmental delay both within the stem cell niche and the genesis of medulloblastoma and it clearly has a role consequently in granule neuron precursor development, in normal cerebellar development (Tanaka et al., 1999; this leaves cerebellar development largely intact and Klein et al., 2004; Komine et al., 2007; Koo et al., 2007). these effects can be moderated by Hedgehog pathway Perturbation of Notch1 receptor signaling negatively activation. Medulloblastoma formed at high frequency regulates growth of the external germinal layer (EGL) following Hedgehog pathway activation and loss of (Lutolf et al., 2002) whereas activation of Notch2 Notch signaling failed to prevent either expansion of receptor promotes granule cell proliferation (Solecki granule neuron precursors (GNPs) or the subsequent et al., 2001). Other ligands, receptors and targets (Jag1, development of medulloblastoma. Jag2, Dll1, Dll3, Notch3, Notch4, Hes1, Hes5) are also expressed in the developing cerebellum (Stump et al., 2002; Gazit et al., 2004; Irvin et al., 2004; Weller et al., 2006). A number of studies have noted dysregulation of Results Notch pathway components in both human and murine medulloblastoma (Berman et al., 2002; Hallahan et al., The RBP-J gene is efficiently inactivated in the external 2004; Yokota et al., 2004; Dakubo et al., 2006) including germinal layer of RBP-Jlox/lox;GFAP-Cre mice the transcription factor targets Hes1 and Hes5, and the Through a combination of expression analyses and expression of Hes1 is associated with poor clinical lineage tracing, the function of the GFAP-Cre allele outcome (Fan et al., 2004). Medulloblastoma cell lines used in this study has been previously defined at key treated with g-secretase inhibitors that block Notch stages of cerebellar development from E12.5 (Yang receptor endoproteolysis display reduced growth, clono- et al., 2008). GFAP-Cre directed recombinase function genicity and tumorigenicity (Hallahan et al., 2004; Fan is found in the ventricular zone (VZ) and cells derived et al., 2006). Accordingly, g-secretase inhibitors have from that layer with the exception of Purkinje neurons, been proposed as a chemotherapeutic approach to which appear to originate from a GFAP-negative treating medulloblastoma (Purow, 2009). population of VZ cells. In addition, the GFAP-Cre Although it has been suggested that removing Notch allele directs a high level (495%) of excision of a function in medulloblastoma may be a powerful conditional Ptc1 allele leading to activation of the therapeutic approach, particularly in those in which Hedgehog pathway, expansion of granule neuron the Hedgehog pathway is activated (Hallahan et al., precursors and to medulloblastoma that are fully 2004), in vivo evidence supporting this strategy has yet to penetrant by p21 (Yang et al., 2008; Figure 5). As the be described. The Ptc1 mutant mouse is a widely approach taken here depends upon a high level of

Oncogene RBP-J is not required for Hedgehog-induced medulloblastoma E Julian et al 3467

Figure 1 GFAP-Cre deletes RBP-J with high efficiency. RNA from EGL of RBP-Jlox/lox;GFAP-Cre and control embryos at E18.5 was harvested by laser-capture microdissection. (a) levels of floxing were tested by PCR on cDNA with primers flanking the excised exons. In RBP-Jlox/lox;GFAP-Cre individuals (lane 1–2) only a 250 bp band is visible, representing the floxed allele. In RBP-Jlox/lox controls (lane 3) the 498 bp band of the unfloxed allele has been amplified. (b) cresyl violet stained section of E18.5 cerebellum before and after laser microdissection at  10 magnification. (c) In situ hybridization for RBP-J on E18.5 brains shows expression of RBP-J in RBP-Jlox/lox controls and loss of RBP-J in the EGL of RBP-Jlox/lox;GFAP-Cre embryos. Scale bar represents 500 mm. excision of a conditional RBP-J allele concomitant with after birth most likely due to severe disruption of the Ptc1 allele, we examined the level of RBP-J deletion structures of the dorsal brain (data not shown), in cells derived from the VZ of RBP-Jlox/lox;GFAP-Cre particularly because cerebellar development is only mice and controls at E18.5. mildly affected. We analyzed the developmental con- We have previously shown that laser-capture micro- sequences of the Notch/Hedgehog interaction up to dissection followed by RT-PCR is a sensitive method for E18.5, with a focus on the development of granule testing the level of cre-mediated recombination of neurons and their precursors since these cells constitute conditional alleles (Yang et al., 2008). Here, we used the origin of medulloblastoma in this model. this approach to examine the EGL from E18.5 embryos Hematoxylin and eosin staining of embryonic brains of the genotypes RBP-Jlox/lox;GFAP-Cre and RBP-Jlox/lox shows no apparent phenotype of any of the mutants at controls. The RBP-J conditional allele used here has E14.5, which is 2 days after activation of Cre been used in many studies and, through excision of recombinase (data not shown). At E18.5 the Ptc1lox/lox; exons 6 and 7, leads to an abrogation of canonical GFAP-Cre mutants have a thickened EGL (Figure 2b) Notch signaling (Yamamoto et al., 2003). We chose the compared with controls (Figure 2a), as has been EGL because medulloblastoma originate from this cell previously reported (Yang et al., 2008) whereas layer and since the cells comprising it are derived from RBP-Jlox/lox;GFAP-Cre mice show no apparent pheno- the GFAP þ cells of the VZ, a high level of RBP-J gene type in the EGL at this point (Figure 2c). However, in excision in the VZ would be reflected in the EGL. cDNA Ptc1lox/lox;RBP-Jlox/lox;GFAP-Cre embryos at E18.5 the from isolated RNA was used as template for a PCR with proliferative phenotype of the Ptc1 inactivation appears primers flanking the deleted region of the RBP-J gene to be diminished (Figure 2d) suggesting that loss of and results confirm a high efficiency of RBP-J gene RBP-J may be acting to restrain GNP development excision in all mutants tested (Figures 1a and b). In situ when cell autonomous Hedgehog pathway activation is hybridization of RBP-Jlox/lox;GFAP-Cre E18.5 cerebel- present. lum confirmed loss of RBP-J expression from the EGL To investigate the developmental consequences of (Figure 1c). Notch ablation and Hedgehog activation in more detail we next examined the expression of bIII-tubulin, a marker for granule neurons that have committed to the Loss of Notch signaling in GFAP þ cells of the cerebellum lineage but are not yet fully mature (Kempermann et al., promotes a developmental delay and asymmetric stem cell 2003). Compared with control E18.5 cerebellum division but is rescued by cell autonomous Hedgehog (Figure 2e), the EGL from Ptc1lox/lox;GFAP-Cre mice pathway activation not only shows an expanded pool of GNPs but also Next, we addressed the consequences of ablation of strong bIII-tubulin staining. This observation is con- Notch signaling and the activation of the Hedgehog sistent with our previous data that indicated that despite pathway from E12.5 onwards in GFAP þ cells, particu- the mitogenic Hedgehog signal, a significant proportion larly in medulloblastoma formation. RBP-Jlox/lox;GFAP- of GNPs can exit the cell cycle and begin to differentiate Cre and Ptc1lox/lox;RBP-Jlox/lox;GFAP mice die shortly into granule neurons (Yang et al., 2008). Figure 2f

Oncogene RBP-J is not required for Hedgehog-induced medulloblastoma E Julian et al 3468

Figure 2 Hedgehog and Notch signaling interact during cerebellar development. Hematoxylin and Eosin staining of E18.5 brain sections shows thickening of the EGL in Ptc1lox/lox;GFAP-Cre brains (b, *) compared with controls (a), which appears incompletely rescued in Ptc1lox/lox;RBP-Jlox/lox;GFAP-Cre individuals (d, #). Staining of E18.5 brains with the neuronal marker bIII-tubulin shows normal neuronal differentiation in Ptc1-deleted cerebella (f) compared with controls (e), whereas a loss of neurons in the developing molecular and granular layers of RBP-Jlox/lox;GFAP-Cre cerebella (g) is rescued in Ptc1lox/lox;RBP-Jlox/lox;GFAP-Cre mice (h). Math1 expression as assayed by in situ hybridization appears decreased in the EGL of RBP-J deleted embryos (k) compared with controls (i) and Hedgehog-activated embryos (j). This decrease is rescued in most but not all cells of Ptc1lox/lox;RBP-Jlox/lox;GFAP-Cre EGL (l). Gli1 mRNA expression is lost in the EGL of RBP-Jlox/lox;GFAP-Cre brains and partly rescued in Ptc1lox/lox;RBP-Jlox/lox;GFAP-Cre EGL (m–p). All pictures per panel were taken at the same magnification, scale bars represent 200 mm for hematoxylin and eosin, 50 mm for bIII-tubulin staining and 50 mm for in situ hybridization images. Nuclei in fluorescent images were counterstained with DAPI.

indicates that the differentiating GNPs in Ptc1lox/lox; Supplementary Figure S1) suggesting that the cells are GFAP-Cre mice are being retained in the EGL possibly not responding to the endogenous Shh signal emanating as a consequence of the expansion of the EGL with the from Purkinje neurons, but this activity is partially more differentiated cells being physically hindered from restored when cell autonomous Hedgehog pathway migration. The RBP-Jlox/lox;GFAP-Cre cerebellum shows activation is present through the loss of Ptc1 (Figures a modest decrease in bIII-tubulin staining (Figures 2g 2o, p and Supplementary Figure S1). Further investiga- and Supplementary Figure S1). Consistent with these tion of the expression of Notch and Hedgehog pathway observations, expression of Math1, a transcription targets revealed that expression of the canonical Notch- factor required for the maintenance of GNPs and a signaling target gene Hes5 is lost from the EGL at E18.5 marker for the early stages of GNP development, is in the absence of RBP-J as expected (Figures 3a and c) diminished in RBP-Jlox/lox;GFAP-Cre mice (Figures 2k and is not restored after Hedgehog pathway activation and Supplementary Figure S1) and restored upon (Figure 3d). The Hes1 gene is a context-dependent target activation of the Hedgehog pathway in the Ptc1lox/lox; gene of both the Notch and Hedgehog-signaling path- RBP-Jlox/lox;GFAP-Cre EGL (Figures 2l and Supple- ways (Ingram et al., 2008; Wall and Wallace, 2009) and mentary Figure S1), as is bIII-tubulin expression is expressed at a low level in the E18.5 EGL of both (Figures 2h and Supplementary Figure S1). control and RBP-Jlox/lox;GFAP-Cre mice (Figures 3e Interestingly, Gli1 activity appears substantially and g), but is significantly upregulated after Hedgehog reduced in the EGL of RBP-Jlox/lox;GFAP-Cre mice pathway activation in both Ptc1lox/lox;GFAP-Cre and compared with wild type control (Figures 2m, o and Ptc1lox/lox;RBP-Jlox/lox;GFAP-Cre mice (Figures 3f and h).

Oncogene RBP-J is not required for Hedgehog-induced medulloblastoma E Julian et al 3469

Figure 3 Hes1 and Hes5 in cerebellar development. In situ hybridization indicates loss of Hes5 mRNA expression in the EGL of E18.5 RBP-J deleted (c) and Ptc1lox/lox;RBP-Jlox/lox;GFAP-Cre brains (d) compared with control (a) and Ptc1-deleted brains (b). Immunofluorescence staining for Hes1 shows an increase in staining in the EGL of Ptc1-deleted as well as Ptc1 and RBP-J-deleted cerebella (f, h) compared with low levels of Hes1 in controls (e) and RBP-Jlox/lox;GFAP-Cre cerebella (g). All images per panel were taken at the same magnification, inlays for Hes1 are  100, scale bars represent 50 mm for Hes5 and 200 mm for Hes1.

We have previously defined that Hedgehog signaling and 3). We next examined this possibility by continuous can regulate stem cell numbers early in cerebellar culture of neurospheres from all genotypes and measur- development, and we and others have previously noted ing the rate of clonal expansion. Under these conditions that medulloblastoma can be initiated in the stem like the slope of the line reflects the frequency of symmetric cells of the VZ (Schuller et al., 2008; Yang et al., 2008). division such that a lower slope indicates an increasing Accordingly, we next examined the interaction between probability of a stem cell diving to form a stem Notch and Hedgehog signaling in the development of and a progenitor cell (asymmetry) (Galli et al., 2002) the proliferative compartment of the cerebellum. Sox2 as opposed to symmetric division into two daughter marks stem cells of the VZ and also Bergmann glia in stem cells and it is clear that loss of Notch signaling in the developing cortex of the cerebellum (Sottile et al., the VZ results in a significant bias towards asymmetric 2006 and Figure 4a). The VZ of Ptc1lox/lox;GFAP-Cre division (Figure 4f). In contrast to the in vivo and ex vivo mice shows a slight expansion and the Rhombic lip (RL) data (Figures 4a–e), the loss of Ptc1 had no effect on and EGL have increased numbers of Sox2 þ cells the growth of neurospheres in long-term cultures compared with wild-type (Figures 4a and b) consistent (Figure 4f), reflecting the loss of overall Hedgehog with an increase in stem cells, whereas in the absence of pathway responsiveness upon long-term culture irres- Notch signaling stem cell numbers are reduced pective of genotype (data not shown)—a phenomenon (Figure 4c). It is interesting to note that cell autonomous which has been shown previously but remains unex- Hedgehog pathway activation in the VZ and EGL plained (Sasai et al., 2006). appears to promote a stem cell fate even in the absence Overall, these data indicate that while loss of Notch of Notch signaling (Figure 4d). We then analyzed the signaling from the VZ/EGL promotes some modest cerebellar proliferative compartment using a primary effects on stem cells and a possible developmental delay neurosphere assay as we have described previously at E18.5, cerebellar development remains largely intact. (Yang et al., 2008). In this assay, the pool of total stem In addition, a population of Hedgehog-activated GNPs and progenitor cells can be enumerated following persists in the EGL of the Ptc1lox/lox;RBP-Jlox/lox;GFAP- plating at clonal density but it does not distinguish Cre cerebellum, which have the appearance of early between the two populations. Analysis of RBP-Jlox/lox; medulloblastoma formation, as defined previously in GFAP-Cre cerebellum at E18.5 indicates that in the Ptc1lox/lox;GFAP-Cre mice (Yang et al., 2008). absence of Notch signaling these mutants have a significant increase in sphere numbers compared with controls (Figure 4e). As the loss of Notch signaling is Loss of Notch signaling does not block medulloblastoma thought to promote differentiation the increase in sphere formation initiated by Ptc1 deletion numbers likely reflects an increased promotion of stem We have shown that medulloblastoma form in 100% of cell to progenitor cell division in combination with the Ptc1lox/lox;GFAP-Cre mice by p21, and that these tumors possible developmental delay indicated above (Figures 2 can be propagated through orthotopic (Yang et al.,

Oncogene RBP-J is not required for Hedgehog-induced medulloblastoma E Julian et al 3470

Figure 4 Influence of Hedgehog and Notch signaling on cerebellar stem and progenitor cells. Sox2 marks stem cells in the VZ but also Bergmann glia. In Ptc1-deleted cerebella staining appears slightly increased in the proliferating cells of the VZ and EGL as well as in Bergmann glia (b) compared with controls (a). In RBP-Jlox/lox;GFAP-Cre cerebella Sox2 reactivity is reduced (c) although in Ptc1lox/lox; RBP-Jlox/lox;GFAP-Cre brains there is substantial compensation for the loss of Notch signaling (d). All images were taken at identical magnifications with the scale bar representing 200 mm. Tissues were counterstained with DAPI. (e) for sphere forming unit assays number of spheres formed from E18.5 cerebellar cells were counted 5 days after plating. RBP-J deletion causes a significant increase in spheres (P ¼ 0.0012) compared with controls, which is ameliorated by additional Ptc1 deletion (P ¼ 0.0003). (control: n ¼ 11, Ptc1lox/lox; GFAP-Cre: n ¼ 9, RBP-Jlox/lox;GFAP-Cre: n ¼ 7, Ptc1lox/lox;RBP-Jlox/lox;GFAP-Cre: n ¼ 9). (f) for analysis of clonal expansion neurospheres were passaged every 6 days. Cell numbers were transformed into LOG and linear regression analysis was performed on resulting curves. Slopes represent the number of stem cells per population. There is no significant difference between the slopes of control (1.183 þ /À 0.01322) and Ptc1lox/lox;GFAP-Cre-derived cells (slope: 1.186 þ /À 0.03162). However, in RBP-Jlox/lox;GFAP-Cre and Ptc1lox/lox;RBP-Jlox/lox;GFAP-Cre neurospheres we show a significant reduction in stem cells compared with controls (slope: 0.83 þ /À 0.0202, P-value o0.0001 and slope: 0.7773 þ /À 0.04409, P-value compared to control o0.0001, respectively).

2008) or subcutaneous transplantation into immuno- intra-cranial tumors rather than subcutaneous trans- compromised mice (data not shown, and Figure 5) with plants might be improved in the absence of Notch equal efficiency and fidelity. As the homozygous loss of signaling. Nonetheless, in our transplant system, tumors the RBP-J allele under the control of GFAP-Cre results initiated and grew in the absence of RBP-J. in perinatal lethality likely due to severe neocortical To address the possibility that tumors from Ptc1lox/lox; defects we adopted a transplantation approach using RBP-Jlox/lox;GFAP-Cre EGL were derived from cells cells purified from the E18.5 cerebellum to investigate with incomplete inactivation of RBP-J we confirmed whether Notch inhibition can block the formation and deletion of the Ptc1 and RBP-J alleles by testing floxing growth of Hedgehog-induced medulloblastoma. When levels of RNA isolated from tumors. All tumors show a injecting a population of 1 Â 106 cells, neither cells from high level of loss of both Ptc1 and RBP-J (Figure 5b) controls nor those from RBP-Jlox/lox;GFAP-Cre embryos confirming that they are unlikely to result from an RBP- led to tumor formation. However, cells from Ptc1lox/lox; J þ cell population and are truly null for canonical GFAP-Cre as well as those from Ptc1lox/lox;RBP-Jlox/lox; Notch signaling. In addition, real-time quantitative RT- GFAP-Cre cerebella gave rise to tumors within 5–8 PCR confirmed loss of expression of the Notch target weeks post injection (Figure 5a). Tumors from both Hes5 in these tumors (data not shown). genotypes display similar phenotypes with small round granular cells comparable with GNPs interspersed with blood vessels (Figure 5c) and expressed Math1 and Gli1, consistent with our previous data characterizing murine Discussion medulloblastoma (Figure 5c and, (Yang et al., 2008)). As previously noted, medulloblastoma generated by It has been suggested by a number of studies that Notch orthotopic injection of 1 Â 106 cells will develop tumors signaling may have a role in the genesis of medullo- by p21 (Yang et al., 2008) and we cannot completely blastoma and clinical trials using Notch inhibitors have exclude the possibility that survival of mice with been proposed (Berman et al., 2002; Hallahan et al.,

Oncogene RBP-J is not required for Hedgehog-induced medulloblastoma E Julian et al 3471

Figure 5 RBP-J deletion does not block tumor formation initiated by Ptc1 deletion. (a)1Â 106 cells isolated from E18.5 cerebella were injected subcutaneously into SCID mice and resulting tumors were collected when a diameter of approximately 1 cm was reached. Kaplan–Meyer analysis shows similar tumor formation patterns arising from Ptc1lox/lox;GFAP-Cre and Ptc1lox/lox;RBP-Jlox/lox;GFAP-Cre cells (P ¼ 0.2875) whereas no tumor formation is observed in controls and RBP-Jlox/lox;GFAP-Cre cells (control or RBP-Jlox/lox;GFAP- Cre compared with Ptc1lox/lox;GFAP-Cre or Ptc1lox/lox;RBP-Jlox/lox;GFAP-Cre P ¼ 0.08). (b) floxing analysis by PCR on cDNA from collected tumors shows a high percentage of floxing of the Ptc1 allele (lanes 1–3; 1 and 2 tumors from injected Ptc1lox/lox;GFAP-Cre cells, three GNPs isolated from P8 Ptc1lox/lox cerebella) as well as the RBP-J allele (4–8, 4–7 tumors from injected Ptc1lox/lox; RBP-Jlox/lox;GFAP-Cre cells, eight tumor from E18.5 Ptc1lox/lox;GFAP-Cre cells). (c) Hematoxylin and eosin staining of tumors from Ptc1lox/lox;GFAP-Cre and Ptc1lox/lox;RBP-Jlox/lox;GFAP-Cre cells shows similar morphology with small round granulated cells interspersed with blood vessels (arrows). Math1 in situ hybridization confirms a GNP origin and Gli1 in situ hybridization confirms Hedgehog pathway activation in both Ptc1lox/lox;GFAP-Cre and Ptc1lox/lox;RBP-Jlox/lox;GFAP-Cre-derived tumors. Scale bars represent 200 mm.

Oncogene RBP-J is not required for Hedgehog-induced medulloblastoma E Julian et al 3472 2004; Yokota et al., 2004; Dakubo et al., 2006). Notch We have previously shown that the Hedgehog path- signaling is well recognized to influence neuronal and way can act to regulate the proliferative compartment of glial differentiation in the cerebellum. For example, loss the cerebellum (Yang et al., 2008) and the pathway also of the ligand Jag1 or the Notch1 receptor from E8.5 in appears to have a role in controlling stem cell numbers neuroepithelial cells leads to premature differentiation in dorsal brain and other organs such as skin (Adolphe of GNPs and migration defects (Lutolf et al., 2002), et al., 2004). Similarly, the Notch pathway is important (Weller et al., 2006) and formation of Bergmann glial in controlling stem cell numbers in a number of organ cells is regulated by RBP-J (Komine et al., 2007). systems including the CNS and the experiments Conversely, activation of the Notch2 receptor in described here enabled us to examine the interaction cultured granule neuron precursors acts as a mitogenic between Notch and Hedgehog at E18.5 in the cerebel- signal and inhibits their differentiation (Solecki et al., lum. On the basis of Sox2 staining, a marker for 2001). Previously, we have shown that the loss of Ptc1 neuronal stem cells, we noted that ablation of RBP-J under the control of GFAP-Cre leads to medulloblas- promoted loss of stem cells but activation of Hedgehog toma following lineage restriction of the Hedgehog- signaling could promote a stem cell fate, even in the activated GNPs (Yang et al., 2008) and here we tested absence of Notch signaling. Analysis of the proliferative the hypothesis that Notch signaling has a key role in compartment using a neurosphere assay showed an medulloblastoma development. increase in sphere numbers in the absence of RBP-J Unlike Ptc1lox/lox;GFAP-Cre mice, RBP-Jlox/lox;GFAP- compared with both wild type and Hedgehog-activated Cre mice show perinatal lethality and this is not cells but this was corrected in cells derived from rescued in Ptc1lox/lox;RBP-Jlox/lox;GFAP-Cre animals. In Ptc1lox/lox;RBP-Jlox/lox;GFAP-Cre mice. The neurosphere addition to the relatively subtle cerebellar phenotype assay enables the enumeration of stem and progenitor we describe here, RBP-Jlox/lox;GFAP-Cre and Ptc1lox/lox; cells but cannot differentiate between them. The increase RBP-Jlox/lox;GFAP-Cre animals have a severe neocortical in colony number we observed with the loss of RBP-J defect such that the patterning of this tissue is perturbed likely occurs since there is not only a shifted balance and there are vascular defects, which lead to substantial between stem and progenitors but also an associated bleeding into the dorsal brain in the perinatal period. developmental delay that results in a reduced rate of exit Accordingly, we were unable to follow the development from the stem/progenitor cell niche and therefore there of medulloblastoma in vivo through to p21 and beyond is an expansion in the number of ‘colony forming cells’. when using the RBP-Jlox allele in combination with In the E18.5 cerebellum from Ptc1lox/lox;RBP-Jlox/lox; GFAP-Cre. However, we have previously shown that GFAP-Cre mice the number of spheres is decreased to medulloblastoma initiated late in embryogenesis can be control levels (Figure 4e) suggesting that Hedgehog subsequently transplanted into immunocompromised pathway activation was sufficient to restore the balance mice to give rise to tumors. Consequently, in this study between stem and progenitor cells and ameliorate the we used a transplant approach to overcome the perinatal developmental delay, as indicated by the Sox2 expres- lethality of inactivating the RBP-J alleles. The advantage sion data. This interpretation of the data is consistent of using the GFAP-Cre allele in these experiments is that with the conclusions of Gao et al. (2009), who noted that it is active in the ventricular zone and enables the loss of RBP-J under the control of Nestin-Cre in the consequences of manipulating both the Hedgehog and stem cell niche of the developing dorsal forebrain Notch pathways on the stem cell niche to be examined. promoted early differentiation of stem cells to progeni- The Notch pathway has been implicated in control- tors (‘intermediate neural progenitors’, inps), and this ling the balance between symmetric and asymmetric then led to an increase in the number of colonies in a stem cell division in a number of tissues (Louvi and neurosphere assay. Although the activation of the Artavanis-Tsakonas, 2006). We considered that one of Hedgehog pathway appears to have promoted the the possible consequences of loss of Notch signaling by appearance of some Sox2 þ staining in the EGL, we deletion of RBP-J under the control of GFAP-Cre could observed no overall increase in the number of neuro- be substantial cerebellar defects, including a significant spheres compared with controls, indicating that diminution of the GNP pool from which medulloblas- although the balance between stem and progenitors toma arise. We observed a reduced staining with bIII- could have shifted, there was no overall change in the tubulin, Math1 and Gli1 suggesting that loss of RBP-J total number of stem and progenitors. It is interesting to in the ventricular zone was inducing a developmental note that when Gao et al. (2009) performed analyses on delay. However, activation of the Hedgehog pathway dorsal brain at late neurogenic stages such as E17.5 was sufficient to reverse this phenomenon and the E18.5 rather than E11.5, loss of RBP-J led to a large decrease EGL of Ptc1lox/lox;RBP-Jlox/lox;GFAP-Cre mice contained in neurospheres, consistent with the notion that early a substantial population of mitotically active Math1 þ promotion of the progenitor compartment results in a granule neuron precursors with the potential to become premature exhaustion of stem cells at later time points. medulloblastoma. The molecular basis of this interac- On that basis, one would predict that the stem cells of tion of the two pathways is not known but the long-term the cerebellum would behave in a similar fashion in consequences of the Hedgehog/Notch interaction in RBP-J null mice at later time points such as p7, which is cerebellar development could be studied in future using one of the latest stages in wild-type mice where the models we have developed here in association with cerebellar neurospheres can be generated. Owing to the other Cre alleles such as Math1. lethality of the RBP-Jlox/lox;GFAP-Cre model, such an

Oncogene RBP-J is not required for Hedgehog-induced medulloblastoma E Julian et al 3473 investigation would likely require an inducible cre would be warranted, although they may be complicated approach in future studies. Nonetheless, our combined by the functional redundancy between Hes family immunohistochemistry and neurosphere data indicate members and others such as Hey1 (Imayoshi et al., that despite some dysregulation of both the stem cell 2008). compartment and neurogenesis from loss of Notch signaling, cerebellar development does not appear to be fundamentally disrupted at E18.5, and activation of the Materials and methods Hedgehog pathway can largely overcome this pheno- type, leading to a cerebellum with a substantial pool of Mouse models cycling GNPs. Although the mechanism of the interac- All work involving mice was performed with approval and tion between Notch and Hedgehog signaling is not clear, according to guidelines of the University of Queensland it is interesting to speculate there may be some ultimate Animal Ethics Committee. Mouse models used were Ptc1 therapeutic benefit in activating Hedgehog signaling in conditional mice (Ellis et al., 2003) and RBP-J conditional neuronal stem cells as we have shown in this study that mice (Tanigaki et al., 2002) crossed with the GFAP-Cre line activating the pathway can overcome a strong differ- (Zhuo et al., 2001). SCID mice were used as tumor transplant recipients. All staining analyses was performed on 3–5 entiating signal such as loss of Notch signaling. individuals per genotype and representative images for each The data presented here indicate that even in the genotype were included in Figures. absence of canonical Notch signaling, Hedgehog-driven medulloblastoma will still initiate and develop. At E18.5 þ Laser capture microdissection we did observe a diminution in the number of Math1 Fresh frozen tissue in OCT compound (Tissue Tek by Sakura þ and Gli1 cells in the Hh-activated EGL after loss of Finetek Co., Tokyo, Japan) was sectioned (10 mm) onto PEN- RBP-J possibly indicating a reduced pool of tumor- membrane slides (PALM, Bernried, Germany) and stained initiating cells, although interestingly overall PCNA with cresyl violet (Ambion/Applied Biosystems, Austin, TX, staining appeared unchanged (data not shown). USA). Dissection was performed using a Zeiss Axiovert 200 Although in our transplant model this was not reflected Inverted microscope (Carl Zeiss Pty Ltd, North Ryde, NSW, in a significant difference in tumor latency (Figure 5, Australia) and the PALM Robomover software. RNA was P ¼ 0.2875), it remains a possibility that in the setting extracted from captured tissue using the RNeasy Micro of human disease Notch inhibition could have a role kit (Qiagen, Hilden, Germany). Reverse transcription was in restricting growth of an existing tumor, perhaps by performed using the Superscript III system by Invitrogen (Carlsbad, CA, USA). The following primers were used for reducing the pool of tumor-initiating cells. PCR detection of the floxed allele : forward 50-CATCTCCAAAC Although Hedgehog pathway components appear to CCTCCAAAA-30, reverse 50-GTCCAGGAAGCTCCATCGT-30. be mutated in a minority of medulloblastoma, the potential interaction with other genetic pathways such Immunofluorescence as Wnt signaling (Saran, 2009) make it possible that Brain samples were fixed in 4% paraformaldehyde over night Hedgehog signaling is having a wider role in the and either embedded in paraffin or cryoprotected in 30% development of the tumor without itself being mutated sucrose before embedding in OCT compound. Antigen and significant Hedgehog pathway activation can be retrieval of deparaffinized wax tissue sections or defrosted detected in the majority of medulloblastoma (Hallahan cryosections was performed by boiling in antigen unmasking et al., 2004), particularly using sensitive RT-PCR-based solution (Vector Laboratories, Burlingame, CA, USA). Sec- methods rather than microarrays. On the basis of the tions were blocked in 4% horse serum, 1% BSA and 0.2% experiments reported here, we suggest that in any Triton-X in PBS before primary antibody incubation over 1 clinical trial of Notch pathway inhibitors for the night at 4 C. Slides were incubated with secondary antibodies for 1 h at room temperature. For immunofluorescence a DAPI treatment of medulloblastoma it will be of paramount counterstain (Sigma Aldrich, St Louis, MO, USA, 1:10000) importance to determine the genetic signature of the was performed for 5 min before mounting with Fluorescence tumor, particularly around the Hedgehog pathway. Mounting Media (Dako, Carpentaria, CA, USA). For A recent report notes that targeting Hes1 function histological analysis deparaffinized and rehydrated sections may reduce medulloblastoma growth (Garzia et al., were stained in Hematoxylin (Vector Laboratories) and Eosin 2009). Rather than regard this as an indication of Notch Y (Sigma Aldrich) and mounted in Entellan. Antibodies used pathway modulation, it is worth noting that we and were bIII-tubulin (1:2000, Promega Corporation, Madison, others have described that Hes1 can also be a Notch- WI, USA) and Sox2 (1:200, R&D Systems, Minneapolis, MN, independent transcriptional target of the Hedgehog USA), both of them on cryo sections, and Hes1 (1:400, a gift pathway (Solecki et al., 2001; Ingram et al., 2008; Wall from R. Kageyama, Kyoto, Japan) on paraffin sections. Fluorescent secondary antibodies used were anti-rabbit et al., 2009). The results from this study support this Alexa555 (1:250, Invitrogen), anti-mouse Alexa488 (1:250, conclusion (Figure 3) because it is clear that Hes5 Invitrogen) and anti-goat Cy3 (1:250, Abacus ALS Pty Ltd, expression is lost in RBP-J null tissue during cerebellar Brisbane, Australia). development at E18.5 but Hes1 is not. In Hedgehog pathway-activated cells Hes1 is significantly upregulated In situ hybridization and Hes5 is not. As this observation extends to In situ hybridization was performed as previously published medulloblastoma generated from Ptc1lox/lox;RBP-Jlox/lox; (Adolphe et al., 2006). In summary, probes were prepared GFAP-Cre mice (data not shown) future studies looking using DIG-labeled probe amplification followed by phenol/ more specifically at the role of Hes1 in medulloblastoma chloroform extraction and precipitation. 6 mm paraffin-

Oncogene RBP-J is not required for Hedgehog-induced medulloblastoma E Julian et al 3474 embedded sections were treated with 2 mg/ml ProK (Roche Technologies, Tullamarine, VIC, Australia), 5% BSA (Sigma), Diagnostics, Mannheim, Germany) in TE buffer, fixed in 4% 1% Penicillin/Streptomycin) containing EGF (20ng/ml) in a paraformaldehyde and acetylated. Hybridization was per- 96-well plate. For each individual the assay was set up in formed in hybridization buffer at 64 1C for Gli1 and 65 1C triplicates. The number of spheres per well was counted 5 days for Math1, Hes5 and RBP-J overnight. A series of SSC washes after plating. Statistical analysis was performed using Graph- was followed by blocking and washing (DIG block and wash pad Prism 4 for unpaired t-tests. buffer set, Roche) before incubation with anti-DIG-AP For passaging cells/neurospheres were harvested and dis- antibody (Roche). Color reaction was performed using sociated in 0.05% trypsin (Gibco by Invitrogen) followed by 3.5 mg/ml nitroblue tetrazolium (NBT) and 1.75 mg/ml 5- addition of trypsin inhibitor. Resuspended cell pellets were Bromo-4-chloro-3-indolyl-phosphate (BCIP) (both Roche) in then replated at a density of 0.5 Â 105 cells per ml. Cell counts 10% Polyvinylalcohol (PVA) (Sigma). Color formation was at every passage were used for growth curve analysis. For stopped in TE solution followed by counter staining with analysis of neurosphere-replating assays in Graphpad Prism 4 nuclear fast red (Vector Labs), post-fixation and mounting. expansion numbers were transformed into LOG and linear Probes used were Math1 (a gift from J Johnson, Dallas, regression analysis was performed on resulting curves. TX, USA), Gli1 (a gift from A Joyner, New York, NY, USA), Student’s t-test analysis was used to determine significant RBP-J (a gift from T Honjo, Kyoto, Japan) and Hes5 differences between slopes. (a gift from R Kageyama). Subcutaneous injection of cell suspensions into SCID mice Microscopy 1 Â 106 cells were injected in a total volume of 200 mlin Light and general fluorescence microscopy was performed Dulbecco’s phosphate-buffered saline (Invitrogen) subcuta- using an Olympus BX-51 upright microscope. Confocal images neously into the flanks of SCID mice. Isoflurane was used as were taken on a Zeiss LSM 510 META. anaesthetic. Tumors were collected when a cross sectional diameter of approximately 1 cm was reached. Quantification of immunofluorescence and in situ hybridization data To quantitate immunofluorescence staining for bIII-tubulin Conflict of interest the number of bIII-tubulin-expressing cells relative to the number of cells (by DAPI staining) was counted in three fields adjacent to the EGL per individual and three individuals were The authors declare no conflict of interest. quantitated per genotype. For in situ hybridization for Math1 and Gli1 three fields within the EGL were quantitated per individual, with three individuals per genotype. The total cell Acknowledgements number per field was determined via nuclear fast red counter- stain. Statistical analysis was performed using unpaired t-test E Julian is an ANZ Trustees Research Scholar. This work was on the nine values per genotype in Graphpad Prism 4. supported by funds from the National Health and Medical Research Council of Australia, The John Trivett Foundation, Stem/progenitor cell analysis the ARC Special Research Centre for Functional and Applied E18.5 cerebellum cells were harvested and subsequently Genomics and the Australian Cancer Research Fund. We dissociated. Cells were plated at a density of 1 Â 105 cells per thank Professor Tasuku Honjo for RBP-J mice and Professor ml in 200 ml NSA media (Neurosphere media containing 10% Ryoichiro Kageyama and Dr Zeng-jie Yang for helpful Neurocult neural stem cell proliferation supplement (Stem Cell discussions.

References

Adolphe C, Hetherington R, Ellis T, Wainwright B. (2006). Patched1 Dakubo GD, Mazerolle CJ, Wallace VA. (2006). Expression of Notch functions as a gatekeeper by promoting cell cycle progression. and Wnt pathway components and activation of Notch signaling in Cancer Res 66: 2081–2088. medulloblastomas from heterozygous patched mice. J Neurooncol Adolphe C, Narang M, Ellis T, Wicking C, Kaur P, Wainwright B. 79: 221–227. (2004). An in vivo comparative study of sonic, desert and Indian Di Marcotullio L, Ferretti E, De Smaele E, Argenti B, Mincione C, hedgehog reveals that hedgehog pathway activity regulates Zazzeroni F et al. (2004). REN(KCTD11) is a suppressor of epidermal stem cell homeostasis. Development 131: 5009–5019. Hedgehog signaling and is deleted in human medulloblastoma. Proc Berman DM, Karhadkar SS, Hallahan AR, Pritchard JI, Eberhart Natl Acad Sci USA 101: 10833–10838. CG, Watkins DN et al. (2002). Medulloblastoma growth inhibition Dong J, Gailani MR, Pomeroy SL, Reardon D, Bale AE. (2000). by hedgehog pathway blockade. Science 297: 1559–1561. Identification of PATCHED mutations in medulloblastomas by Binning MJ, Niazi T, Pedone CA, Lal B, Eberhart CG, Kim KJ et al. direct sequencing. Hum Mutat 16: 89–90. (2008). Hepatocyte growth factor and sonic Hedgehog expression in Ellis T, Smyth I, Riley E, Graham S, Elliot K, Narang M et al. cerebellar neural progenitor cells costimulate medulloblastoma (2003). Patched 1 conditional null allele in mice. Genesis 36: initiation and growth. Cancer Res 68: 7838–7845. 158–161. Briggs KJ, Corcoran-Schwartz IM, Zhang W, Harcke T, Devereux Fan X, Matsui W, Khaki L, Stearns D, Chun J, Li YM et al. (2006). WL, Baylin SB et al. (2008). Cooperation between the Hic1 and Notch pathway inhibition depletes stem-like cells and blocks Ptch1 tumor suppressors in medulloblastoma. Genes Dev 22: engraftment in embryonal brain tumors. Cancer Res 66: 770–785. 7445–7452. Corbin JG, Gaiano N, Juliano SL, Poluch S, Stancik E, Haydar TF. Fan X, Mikolaenko I, Elhassan I, Ni X, Wang Y, Ball D et al. (2004). (2008). Regulation of neural progenitor cell development in the Notch1 and notch2 have opposite effects on embryonal brain tumor nervous system. J Neurochem 106: 2272–2287. growth. Cancer Res 64: 7787–7793.

Oncogene RBP-J is not required for Hedgehog-induced medulloblastoma E Julian et al 3475 Frappart PO, Lee Y, Lamont J, McKinnon PJ. (2007). BRCA2 is Lutolf S, Radtke F, Aguet M, Suter U, Taylor V. (2002). Notch1 is required for neurogenesis and suppression of medulloblastoma. required for neuronal and glial differentiation in the cerebellum. EMBO J 26: 2732–2742. Development 129: 373–385. Fujikura J, Hosoda K, Iwakura H, Tomita T, Noguchi M, Masuzaki Marino S. (2005). Medulloblastoma: developmental mechanisms out H et al. (2006). Notch/Rbp-j signaling prevents premature endocrine of control. Trends Mol Med 11: 17–22. and ductal cell differentiation in the pancreas. Cell Metab 3: 59–65. Marino S, Vooijs M, van Der Gulden H, Jonkers J, Berns A. (2000). Galli R, Fiocco R, De Filippis L, Muzio L, Gritti A, Mercurio S et al. Induction of medulloblastomas in p53-null mutant mice by somatic (2002). Emx2 regulates the proliferation of stem cells of the adult inactivation of Rb in the external granular layer cells of the mammalian central nervous system. Development 129: 1633–1644. cerebellum. Genes Dev 14: 994–1004. Gao F, Zhang Q, Zheng MH, Liu HL, Hu YY, Zhang P et al. (2009). Ong CT, Cheng HT, Chang LW, Ohtsuka T, Kageyama R, Stormo Transcription factor RBP-J-mediated signaling represses the differ- GD et al. (2006). Target selectivity of vertebrate notch proteins. entiation of neural stem cells into intermediate neural progenitors. Collaboration between discrete domains and CSL-binding site Mol Cell Neurosci 40: 442–450. architecture determines activation probability. J Biol Chem 281: Garzia L, Andolfo I, Cusanelli E, Marino N, Petrosino G, De Martino 5106–5119. D et al. (2009). MicroRNA-199b-5p impairs cancer stem cells Packer RJ. (1994). Diagnosis, treatment, and outcome of primary through negative regulation of HES1 in medulloblastoma. PLoS central nervous system tumors of childhood. Curr Opin Oncol 6: One 4: e4998. 240–246. Gazit R, Krizhanovsky V, Ben-Arie N. (2004). Math1 controls Pazzaglia S, Tanori M, Mancuso M, Gessi M, Pasquali E, Leonardi S cerebellar granule cell differentiation by regulating multiple et al. (2006). Two-hit model for progression of medulloblastoma components of the Notch signaling pathway. Development 131: preneoplasia in Patched heterozygous mice. Oncogene 25: 903–913. 5575–5580. Hahn H, Wicking C, Zaphiropoulous PG, Gailani MR, Shanley S, Pietsch T, Waha A, Koch A, Kraus J, Albrecht S, Tonn J et al. (1997). Chidambaram A et al. (1996). Mutations of the human homolog of Medulloblastomas of the desmoplastic variant carry mutations Drosophila patched in the nevoid basal cell carcinoma syndrome. of the human homologue of Drosophila patched. Cancer Res 57: Cell 85: 841–851. 2085–2088. Hallahan AR, Pritchard JI, Hansen S, Benson M, Stoeck J, Hatton BA Purow B. (2009). Notch inhibitors as a new tool in the war on cancer: a et al. (2004). The SmoA1 mouse model reveals that notch signaling pathway to watch. Curr Pharm Biotechnol 10: 154–160. is critical for the growth and survival of sonic hedgehog-induced Rao G, Pedone CA, Del Valle L, Reiss K, Holland EC, Fults DW. medulloblastomas. Cancer Res 64: 7794–7800. (2004). Sonic hedgehog and insulin-like growth factor signaling Hurlbut GD, Kankel MW, Lake RJ, Artavanis-Tsakonas S. (2007). synergize to induce medulloblastoma formation from nestin- Crossing paths with Notch in the hyper-network. Curr Opin Cell expressing neural progenitors in mice. Oncogene 23: 6156–6162. Biol 19: 166–175. Romer J, Curran T. (2005). Targeting medulloblastoma: small- Imayoshi I, Shimogori T, Ohtsuka T, Kageyama R. (2008). Hes genes molecule inhibitors of the Sonic Hedgehog pathway as potential and neurogenin regulate non-neural versus neural fate specification cancer therapeutics. Cancer Res 65: 4975–4978. in the dorsal telencephalic midline. Development 135: 2531–2541. Rudin CM, Hann CL, Laterra J, Yauch RL, Callahan CA, Fu L et al. Ingram WJ, McCue KI, Tran TH, Hallahan AR, Wainwright BJ. (2009). Treatment of medulloblastoma with hedgehog pathway (2008). Sonic Hedgehog regulates Hes1 through a novel mechanism inhibitor GDC-0449. N Engl J Med 361: 1173–1178. that is independent of canonical Notch pathway signalling. Saran A. (2009). Medulloblastoma: role of developmental pathways, Oncogene 27: 1489–1500. dna repair signaling, and other players. Curr Mol Med 9: 1046–1057. Irvin DK, Nakano I, Paucar A, Kornblum HI. (2004). Patterns of Sasai K, Romer JT, Kimura H, Eberhart DE, Rice DS, Curran T. Jagged1, Jagged2, Delta-like 1 and Delta-like 3 expression during (2007). Medulloblastomas derived from Cxcr6 mutant mice late embryonic and postnatal brain development suggest multiple respond to treatment with a smoothened inhibitor. Cancer Res 67: functional roles in progenitors and differentiated cells. J Neurosci 3871–3877. Res 75: 330–343. Sasai K, Romer JT, Lee Y, Finkelstein D, Fuller C, McKinnon PJ Jarriault S, Brou C, Logeat F, Schroeter EH, Kopan R, Israel A. et al. (2006). Shh pathway activity is down-regulated in cultured (1995). Signalling downstream of activated mammalian Notch. medulloblastoma cells: implications for preclinical studies. Cancer Nature 377: 355–358. Res 66: 4215–4222. Jarriault S, Le Bail O, Hirsinger E, Pourquie O, Logeat F, Strong CF Schuller U, Heine VM, Mao J, Kho AT, Dillon AK, Han YG et al. et al. (1998). Delta-1 activation of notch-1 signaling results in HES-1 (2008). Acquisition of granule neuron precursor identity is a critical transactivation. Mol Cell Biol 18: 7423–7431. determinant of progenitor cell competence to form Shh-induced Kempermann G, Gast D, Kronenberg G, Yamaguchi M, Gage FH. medulloblastoma. Cancer Cell 14: 123–134. (2003). Early determination and long-term persistence of adult- Solecki DJ, Liu XL, Tomoda T, Fang Y, Hatten ME. (2001). generated new neurons in the hippocampus of mice. Development Activated Notch2 signaling inhibits differentiation of cerebellar 130: 391–399. granule neuron precursors by maintaining proliferation. Neuron 31: Klein AL, Zilian O, Suter U, Taylor V. (2004). Murine numb regulates 557–568. granule cell maturation in the cerebellum. Dev Biol 266: 161–177. Sottile V, Li M, Scotting PJ. (2006). Stem cell marker expression in the Komine O, Nagaoka M, Watase K, Gutmann DH, Tanigaki K, Honjo Bergmann glia population of the adult mouse brain. Brain Res 1099: T et al. (2007). The monolayer formation of Bergmann glial cells is 8–17. regulated by Notch/RBP-J signaling. Dev Biol 311: 238–250. Stump G, Durrer A, Klein AL, Lutolf S, Suter U, Taylor V. (2002). Koo BK, Yoon MJ, Yoon KJ, Im SK, Kim YY, Kim CH et al. (2007). Notch1 and its ligands Delta-like and Jagged are expressed and An obligatory role of mind bomb-1 in notch signaling of active in distinct cell populations in the postnatal mouse brain. mammalian development. PLoS One 2: e1221. Mech Dev 114: 153–159. Lelievre V, Seksenyan A, Nobuta H, Yong WH, Chhith S, Sweeney C, Morrow D, Birney YA, Coyle S, Hennessy C, Scheller A Niewiadomski P et al. (2008). Disruption of the PACAP gene et al. (2004). Notch 1 and 3 receptor signaling modulates vascular promotes medulloblastoma in ptc1 mutant mice. Dev Biol 313: 359–370. smooth muscle cell growth, apoptosis, and migration via a CBF-1/ Leung C, Lingbeek M, Shakhova O, Liu J, Tanger E, Saremaslani P RBP-Jk dependent pathway. FASEB J 18: 1421–1423. et al. (2004). Bmi1 is essential for cerebellar development and is Talora C, Campese AF, Bellavia D, Felli MP, Vacca A, Gulino A et al. overexpressed in human medulloblastomas. Nature 428: 337–341. (2008). Notch signaling and diseases: an evolutionary journey from Louvi A, Artavanis-Tsakonas S. (2006). Notch signalling in vertebrate a simple beginning to complex outcomes. Biochim Biophys Acta neural development. Nat Rev Neurosci 7: 93–102. 1782: 489–497.

Oncogene RBP-J is not required for Hedgehog-induced medulloblastoma E Julian et al 3476 Tanaka M, Kadokawa Y, Hamada Y, Marunouchi T. (1999). Notch2 Wang J, Pham-Mitchell N, Schindler C, Campbell IL. (2003). expression negatively correlates with glial differentiation in the Dysregulated Sonic hedgehog signaling and medulloblastoma postnatal mouse brain. J Neurobiol 41: 524–539. consequent to IFN-alpha-stimulated STAT2-independent produc- Tanigaki K, Han H, Yamamoto N, Tashiro K, Ikegawa M, Kuroda K tion of IFN-gamma in the brain. J Clin Invest 112: 535–543. et al. (2002). Notch-RBP-J signaling is involved in cell fate Weller M, Krautler N, Mantei N, Suter U, Taylor V. (2006). Jagged1 determination of marginal zone B cells. Nat Immunol 3: 443–450. ablation results in cerebellar granule cell migration defects and Tanori M, Mancuso M, Pasquali E, Leonardi S, Rebessi S, Di Majo V depletion of Bergmann glia. Dev Neurosci 28: 70–80. et al. (2008). PARP-1 cooperates with Ptc1 to suppress medullo- Wetmore C, Eberhart DE, Curran T. (2001). Loss of p53 but not ARF blastoma and basal cell carcinoma. Carcinogenesis 29: 1911–1919. accelerates medulloblastoma in mice heterozygous for patched. Taylor MD, Liu L, Raffel C, Hui CC, Mainprize TG, Zhang X et al. Cancer Res 61: 513–516. (2002). Mutations in SUFU predispose to medulloblastoma. Wicking C, Smyth I, Bale A. (1999). The hedgehog signalling pathway Nat Genet 31: 306–310. in tumorigenesis and development. Oncogene 18: 7844–7851. Thomas WD, Chen J, Gao YR, Cheung B, Koach J, Sekyere E et al. Yamamoto N, Tanigaki K, Han H, Hiai H, Honjo T. (2003). Notch/ (2009). Patched1 deletion increases N-Myc protein stability RBP-J signaling regulates epidermis/hair fate determination of hair as a mechanism of medulloblastoma initiation and progression. follicular stem cells. Curr Biol 13: 333–338. Oncogene 28: 1605–1615. Yang ZJ, Ellis T, Markant SL, Read TA, Kessler JD, Bourboulas M Thompson MC, Fuller C, Hogg TL, Dalton J, Finkelstein D, Lau CC et al. (2008). Medulloblastoma can be initiated by deletion of et al. (2006). Genomics identifies medulloblastoma subgroups Patched in lineage-restricted progenitors or stem cells. Cancer Cell that are enriched for specific genetic alterations. J Clin Oncol 24: 14: 135–145. 1924–1931. Yokota N, Mainprize TG, Taylor MD, Kohata T, Loreto M, Vorechovsky I, Tingby O, Hartman M, Stromberg B, Nister M, Ueda S et al. (2004). Identification of differentially expressed Collins VP et al. (1997). Somatic mutations in the human and developmentally regulated genes in medulloblastoma using homologue of Drosophila patched in primitive neuroectodermal suppression subtraction hybridization. Oncogene 23: 3444–3453. tumours. Oncogene 15: 361–366. Zhuo L, Theis M, Alvarez-Maya I, Brenner M, Willecke K, Messing Wall DS, Mears AJ, McNeill B, Mazerolle C, Thurig S, Wang Y et al. A. (2001). hGFAP-cre transgenic mice for manipulation of glial and (2009). Progenitor cell proliferation in the retina is dependent on neuronal function in vivo. Genesis 31: 85–94. Notch-independent Sonic hedgehog/Hes1 activity. J Cell Biol 184: Zindy F, Uziel T, Ayrault O, Calabrese C, Valentine M, Rehg JE et al. 101–112. (2007). Genetic alterations in mouse medulloblastomas and genera- Wall DS, Wallace VA. (2009). Hedgehog to Hes1: the heist of a Notch tion of tumors de novo from primary cerebellar granule neuron target. Cell Cycle 8: 1301–1302. precursors. Cancer Res 67: 2676–2684.

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

Oncogene