Downloaded from genesdev.cshlp.org on October 1, 2021 - Published by Cold Spring Harbor Laboratory Press

RESEARCH COMMUNICATION

induce differentiation of GNPs by binding to their recep- Post-transcriptional tors, BMPR1a, BMPR1b, and BMPR2, to activate down-regulation of Atoh1/ Smad1,5,8 phosphorylation and regulation and to trigger the transcription of two basic helix–loop–helix Math1 by bone morphogenic (bHLH) , Id1 and Id2 (Angley et al. 2003; Rios et proteins suppresses al. 2004). Here, we demonstrate that BMPs similarly block the proliferation of MB cells in vitro and in vivo medulloblastoma development and provide evidence that down-regulation of the Shh- induced , Atoh1, is required for these Haotian Zhao,1,3 Olivier Ayrault,1,3 effects. Frederique Zindy,1 Jee-Hae Kim,2 and Martine F. Roussel1,4 Results and Discussion BMPs antagonize Shh-dependent proliferation and 1Departments of Genetics and Tumor Cell Biology, St. Jude Children’s Research Hospital, Memphis, Tennessee 38105, induce differentiation of GNPs and GNP-like MB cells USA; 2Laboratory of Developmental Neurobiology, Primary GNPs isolated from postnatal day 7 (P7) mouse Rockefeller University, New York, New York 10021, USA cerebella, a time at which their proliferation is maximal, were enriched by equilibrium Percoll density gradient Bone morphogenic proteins 2 and 4 (BMP2 and BMP4) centrifugation and cultured in vitro (Uziel et al. 2005). inhibit proliferation and induce differentiation of cer- Treatment of GNPs with recombinant human BMP2 or ebellar granule neuron progenitors (GNPs) and primary BMP4 in the presence of Shh reduced their incorporation GNP-like medulloblastoma (MB) cells. This occurs of BrdU so that after3dofBMPtreatment, only ∼5% of through rapid proteasome-mediated degradation of GNPs remained in cycle (Fig. 1A; Supplemental Fig. 1A). Math1 (Atoh1), a transcription factor expressed in pro- Inhibition of proliferation was confirmed by analysis of liferating GNPs. Ectopic expression of Atoh1, but not of the cells’ DNA content and mimicked effects of Shh Sonic hedgehog (Shh)-regulated Gli1 or Mycn, cancels withdrawal or of the cells’ response to cyclopamine, an inhibitor of Shh signaling (Fig. 1B). these BMP-mediated effects and restores Shh-dependent To evaluate whether BMPs also inhibit proliferation of proliferation of GNPs and MB cells in vitro and in vivo. tumor cells, we isolated and cultured primary GNP-like regulating the BMP signaling pathway are down- tumor cells from MBs arising in predisposed Cdkn2c−/−, regulated in mouse MBs. Thus, BMPs are potent inhibi- Trp53Fl/Fl, Nes-cre+, and Cdkn2c−/−, Ptch1+/− mice (Uziel tors of MB and should be considered as novel therapeutic et al. 2005; Zindy et al. 2007). Because these tumors ex- agents. press a constitutively activated Shh signaling pathway (Lee et al. 2003, Zindy et al. 2007), they no longer depend Supplemental material is available at http://www.genesdev.org. on Shh addition to the culture medium to proliferate. Received November 21, 2007; revised version accepted The number of MB cells approximately doubled after 72 January 16, 2008. h of culture but did not expand in number when treated either with BMP4 or cyclopamine (Fig. 1C; Supplemental Fig. 1B). Like primary GNPs, only 8% of GNP-like tumor Balancing the proliferation, migration, and differentia- cells remained in S phase after 72 h of culture in the tion of cerebellar granule neuron progenitors (GNPs) is presence of BMP2, BMP4, BMP7, or cyclopamine (Fig. essential for proper development of the cerebellum and 1D; Supplemental Fig. 1C, left panel). FACS analysis of for suppression of medulloblastoma (MB), the most com- propidium iodide-stained cells indicated that they had mon malignant pediatric brain tumor. In the external arrested in G1 phase with a 2N DNA content, but unlike germinal layer (EGL) of the developing postnatal cerebel- previous reports (Hallahan et al. 2003), Annexin V stain- lum, proliferation of GNPs is stimulated through the ing of tumor cells did not demonstrate increased apopto- Sonic hedgehog (Shh) signaling pathway (Wallace 1999; sis (Supplemental Fig. 1C, right panel). Immunostaining Wechsler-Reya and Scott 2003), which is deregulated by of GNPs (Fig. 1E) and GNP-like tumor cells (Fig. 1F) mutations in ∼30% of human MB (Dahmane et al. 2001; treated for 72 h with BMP2 or BMP4 revealed increased Kenney et al. 2003; Oliver et al. 2003; Marino 2005). expression of Tag1 (Cntn2) (Fig. 1E [panels b,c vs. a], F Bone morphogenic proteins (BMPs), a subgroup of the [panel b vs. a]), Class III ␤-tubulin/Tuj1 (Tubb1) (Fig. 1E transforming growth factor-␤ (Tgfb1) superfamily, also [panels h,i vs. g], F [panel f vs. e]), NeuN (Neuna60) and play critical roles in fate determination, patterning, dif- NF200 (Nefh) (Supplemental Fig. 1D) and Cdkn1b ferentiation, and cell survival during cerebellar develop- (p27Kip1) (Supplemental Fig. 1E), several markers of neu- ment (Angley et al. 2003; Rios et al. 2004; Fogarty et al. ronal differentiation. Thus, primary GNPs and MB cells 2005). BMPs antagonize Shh-dependent proliferation and exit the division cycle and differentiate in response to BMP treatment without evidence of apoptosis. Although basic fibroblast growth factor (bFGF) was previously [Keywords: Medulloblastoma; Sonic hedgehog; bone morphogenic pro- shown to block Shh-dependent proliferation in GNPs tein; Atoh1/Math1; Mycn] and MB cells (Fogarty et al. 2007), bFGF did not mimic 3 These authors contributed equally to this work. the effects of BMPs under our conditions of cell purifi- 4Corresponding author. E-MAIL [email protected]; FAX (901) 495-2381. cation and culture. Comparison of pro- Article is online at http://www.genesdev.org/cgi/doi/10.1101/gad.1636408. files of GNPs purified from P6 cerebella of wild-type and

722 GENES & DEVELOPMENT 22:722–727 © 2008 by Cold Spring Harbor Laboratory Press ISSN 0890-9369/08; www.genesdev.org Downloaded from genesdev.cshlp.org on October 1, 2021 - Published by Cold Spring Harbor Laboratory Press

BMPs suppress medulloblastoma

modestly reduced in cells treated for 24 h with Shh in the presence of BMPs (Fig. 2A). However, after 72-h treat- ment with Shh and BMPs, Mycn, Ccnd1, and Cdk2 pro- tein levels, together with Gli1, Gli2, and Mycn mRNAs, were markedly diminished (Fig. 2B; Supplemental Fig. 2A) (Alvarez-Rodriguez et al. 2007). In turn, the levels of three transcription factors—Neurod1, Zic1, and Pax6— expressed in granule neurons (Aruga et al. 1998; Miyata et al. 1999; Yamasaki et al. 2001) were unchanged after 24 h (Fig. 2A), while after3dofBMPtreatment, Neurod1 and Zic1 levels were slightly decreased (Fig. 2B). Again, Cntn2 expression was increased after 72 h of BMP treat- ment as cells ceased proliferating (Fig. 1E). Thus, while Shh and Bmp signaling converge in regulating the cell division cycle, they do so in a different manner. During cerebellar development, the bHLH transcrip- tion factor Atoh1 is detected in proliferating GNPs, but not in their post-mitotic derivatives (Akazawa et al. 1995; Ben-Arie et al. 1996, 1997; Lumpkin et al. 2003; Machold and Fishell 2005). In Atoh1-null mice, the specification of GNPs is intact and granule neuron iden- tity is maintained, but proliferation of GNPs and their subsequent differentiation and migration are compro- Figure 1. BMP induces cell cycle arrest and differentiation of GNPs mised (Ben-Arie et al. 1997; Gazit et al. 2004). Unlike the and GNP-like MB cells. (A) GNPs were treated for 24, 48, and 72 h many other regulators whose alterations in expression in the presence of Shh (black bars), Shh and BMP2 (gray bars), or Shh temporally mirrored the rate of BMP-induced cell cycle and BMP4 (diagonal shaded bars). Cell proliferation was measured withdrawal, Atoh1 expression became undetect- by BrdU incorporation (n = 3). (B) GNPs were treated for 72 h with- out Shh (blank bar), with Shh (black bar), with Shh and BMP4 (di- able as early as 24 h after BMP treatment (Fig. 2A) and agonal shaded bar), or with Shh and cyclopamine (vertical shaded remained down-regulated after 72 h in the presence of bar). Cell proliferation was determined by FACS analysis of DNA BMP (Fig. 2B) as confirmed by immunofluorescence content. The percentage of cells in S phase is shown. (C) Prolifera- staining (Supplemental Fig. 2B). tion of tumor cells grown in the absence (Blank, —Ⅵ—, n =5)or Like primary GNPs, BMP-treated MB cells showed presence (—ᮡ—, n = 5) of BMP4 or cyclopamine (—ᮢ—, n = 5) for 72 h was assessed by counting cell number. (D) GNP-like tumor comparably increased levels of phosphorylated Smad1,5,8 cells were treated without (black bars) or with (diagonal shaded bars) and ID2 (Fig. 2C), and ID1 (data not shown), indicating BMP4 or with cyclopamine (blank bars) for 48 and 72 h. The per- that this signaling pathway remained intact in tumor centage of cells in S phase was determined as in B. (E) Expression of Cntn2 and Tubb1 in GNPs after 72-h treatment with Shh (panels a,d,g,j), Shh and BMP2 (panels b,e,h,k), or Shh and BMP4 (panels c,f,i,l) was determined by immunofluorescence. Red fluorescent staining indicates the expression of Cntn2 (panels a–c) or Tubb1 (panels g–i). (Panels d–f,j–l) DAPI staining was used to confirm cell viability. (F) Tumor cells treated for 72 h without (panels a,c,e,g)or with (panels b,d,f,h) BMP4 were stained with antibodies to Cntn2 (panels a,b) or Tubb1 (panels e,f). (Panels c,d,g,h) DAPI staining con- firmed cell viability. Error bars indicate SEM and asterisks indicate P < 0.05 (Student’s t-test). tumor-prone mice (Supplemental Fig. 1F, panel a) with those of MBs arising in genetically predisposed mice (Supplemental Fig. 1F, panel b) revealed that many effec- tors of BMP signaling were down-regulated in MBs, sug- gesting that BMP signaling might normally play a role in tumor suppression.

BMP treatment leads to rapid down-regulation of Atoh1 protein When immunoblotting (Fig. 2A,B) and quantitative RT– PCR (q-RT–PCR) (Supplemental Fig. 2A) were used to survey gene expression in GNPs treated with Shh alone or together with BMP, Smad1,5,8 phosphorylation, and protein levels of Id1 and Id2, were greatly increased after Figure 2. BMP treatment results in rapid loss of Atoh1 in primary BMP treatment, but not by Shh alone (Fig. 2A,B). Con- GNPs and MB cells. Immunoblotting was used to analyze protein versely, expression of Shh-responsive targets, Gli1, expression in GNPs treated 24 h (A)or72h(B) with (+) or without (−) BMP2, BMP4, and/or Shh, using antibodies to the indicated pro- Mycn, and Ccnd1 (cyclin D1), was induced by Shh (Ken- teins. ␤-Actin was used as a loading control. (C) Kinetic analysis of ney and Rowitch 2000; Kenney et al. 2003; Corrales et al. protein expression in MB cells treated with control medium (Blank), 2004), but their protein levels were either unchanged or BMP4 or cyclopamine (Cyclo) for different times.

GENES & DEVELOPMENT 723 Downloaded from genesdev.cshlp.org on October 1, 2021 - Published by Cold Spring Harbor Laboratory Press

Zhao et al. cells. Atoh1 expression was maintained in cultured MB cells during at least3dofculture consistent with con- stitutive activation of the Shh signaling pathway. The relative levels of Atoh1 mRNA were also higher in MBs than in primary GNPs (Supplemental Fig. 1F, bottom lane, panel b vs. a). Yet, Atoh1 protein levels decreased rapidly within 12 h and became undetectable by 24 h after BMP addition (Fig. 2C). Whereas cyclopamine treat- ment down-regulated Mycn expression within 12 h, it did not reduce Atoh1 protein levels as quickly (Fig. 2C). Conversely, BMP treatment did not affect the levels of Mycn within the first 24 h of culture but reduced the levels of Mycn and Cdk2 only after 3 d, concomitant with the exit of the tumor cells from the cell division cycle and their differentiation (Fig. 2C). Thus, as in nor- mal GNPs, activation of BMP signaling in tumor cells resulted in rapid disappearance of Atoh1 protein without affecting Shh activity.

BMP-dependent Atoh1 protein down-regulation occurs via a post-transcriptional mechanism Atoh1 protein levels were maintained when proliferating GNPs were cultured with Shh but decreased rapidly in its absence (Fig. 3A). Atoh1 protein and mRNA levels were similarly reduced when GNPs, cultured in the pres- ence of Shh, were treated with cyclopamine (Supplemen- Figure 3. BMP4 inhibits Shh-dependent GNP proliferation via tal Fig. 2C,D, respectively), again highlighting the fact post-transcriptional down-regulation of Atoh1. Kinetic analysis of that Atoh1 expression in proliferating GNPs depends on Atoh1 protein expression in GNPs untreated (−) or treated (+) with Shh pathway activation (Berman et al. 2002; Kenney et Shh (A) or Shh and BMP4 (B) for the indicated times. (C) Q-RT–PCR al. 2003). However, Atoh1 protein levels were no longer analysis of Gli1, Gli2, and Atoh1 expression in GNPs untreated (blank bars) or treated with Shh (black bars), BMP4 (gray bars), or Shh detected after only 12 h of BMP treatment (Fig. 3B). In and BMP4 (diagonal shaded bars) for 18 h. (D) Analysis of Atoh1 contrast, even after 18 h of BMP4 exposure, Atoh1 RNA protein expression in GNP-like tumor cells treated with (+) or with- levels, as well as those of Gli1 and Gli2, remained simi- out (−) BMP4 or MG-132 for 10 h. (E, panel a) Analysis of Atoh1 lar to those quantified in GNPs treated with Shh alone protein expression after transfection with (+) or without (−) Tcfe2a (Fig. 3C). Similarly, BMP treatment of MB cells did not into 293T cells. After 48 h, cells were then treated with (+) or with- out (−) MG-132 for 1.5 h. Kinetic analysis of Atoh1 protein expres- affect Atoh1 RNA levels (data not shown). Thus, BMP sion after cycloheximide treatment of 293T cells transfected with treatment appeared to trigger a rapid loss of Atoh1 pro- Atoh1 alone (panel b) or Atoh1 and Tcfe2a (panel c). (Panel d) 293T tein expression via a post-transcriptional mechanism. cells were transfected with Atoh1, Tcfe2a, and Id2 either alone or in To determine whether BMP treatment accelerated different combinations and grown for 48 h. Atoh1 protein expression Atoh1 protein turnover, GNP-like MB cells were treated was analyzed by immunoblotting. (F) Proliferation analysis of retro- with the proteasome inhibitor MG-132 or with solvent virally infected GNPs in response to Shh (black bars) or Shh plus BMP4 (diagonal shaded bars). After infection with retroviruses that (DMSO) alone for 10 h, with or without BMP4 (Fig. 3D). express different cDNAs indicated in the X-axis and GFP, GNPs MG-132, but not DMSO treatment, prevented Atoh1 were cultured for 24 h with Shh before addition of BMP4. After 72 h, protein down-regulation even in the presence of BMP4. proliferation of infected GNPs was determined by BrdU incorpora- Atoh1 forms complexes with the ubiquitous bHLH tran- tion assays. The Y-axis represents BrdU incorporation rate of GFP- scription factor Tcfe2a (E47) to bind to and target genes labeled infected cells. Error bars indicate SEM and asterisks indicate P < 0.05 (Student’s t-test). ␤-Actin was used as a loading control in through E-box sites. Because GNPs cannot be transfected all immunoblots. and retroviral infections lead to few integrated copies per cell, we examined whether an interaction of Tcfe2a with Atoh1 affected its stability in 293T cells. MG-132 treat- Tcfe2a. Interestingly, when Id1 (data not shown) or Id2 ment greatly enhanced Atoh1 protein levels in trans- was transfected together with Atoh1 and Tcfe2a into fected 293T cells (Fig. 3E, panel a), but when Tcfe2a and 293T cells, the levels of stabilized Atoh1 protein were Atoh-1 were coexpressed, Atoh1 was stabilized and MG- significantly reduced (Fig. 3E, panel d). Thus, by shifting 132 treatment was without further effect (Fig. 3E, panel Atoh1 from transcriptionally active complexes contain- a). When expressed alone, Atoh1 protein levels fell rap- ing Tcfe2a to inactive Id-containing complexes, BMP idly and disappeared within 2.5 h following inhibition of treatment may also enhance Atoh1 protein turnover. In new protein synthesis with cycloheximide (Fig. 3E, panel this respect, Atoh1 regulation may mirror that of Mash1 b), but when Tcfe2a was also present, Atoh1 protein lev- (Ascl1), another bHLH protein expressed during CNS de- els remained largely unaffected during the same interval velopment (Shou et al. 1999; Vinals et al. 2004). Like (Fig. 3E, panel c). Thus, Tcfe2a can stabilize Atoh1 and Ascl1, Ser 193 phosphorylation of Atoh1 is expected to prevent its degradation by the proteasome. regulate its binding to Tcfe2a and its stability. These In GNPs and GNP-like MB cells treated with BMP, the findings suggested that Atoh1 down-regulation was levels of Id1 and Id2 were greatly increased (Fig. 2A–C) likely to be required for cell cycle exit in response to raising the possibility that they might also affect Atoh1 BMP treatment. Indeed, ectopic overexpression of protein turnover by competing with Atoh1 for binding to Atoh1, but not of two Atoh1 DNA-binding mutants

724 GENES & DEVELOPMENT Downloaded from genesdev.cshlp.org on October 1, 2021 - Published by Cold Spring Harbor Laboratory Press

BMPs suppress medulloblastoma

(E165G and R158G) or the Shh-responsive proteins Mycn and GFP were much smaller and almost devoid of GFP- and Gli1, abolished the BMP-induced cell cycle arrest of positive cells (2.4% ± 1.51, n = 8) (Fig. 4A [panel b], B). primary GNPs (Fig. 3F). Thus, down-regulation of Atoh1 All allografts maintained pathological and molecular fea- expression preceded BMP-induced cell cycle arrest, tures of the original MB (Supplemental Fig. 3). Because whereas expression of functional Atoh1 protein main- the efficiency of retroviral infection of donor cells typi- tained Shh-mediated proliferation of GNPs treated with cally ranged from 40% to 60%, the small tumors lacking BMP. GFP expression that arose in BMP-infected tumor cells might have originated from uninfected tumor cells. In- deed, tumor cells infected with control virus and sorted BMPs inhibit tumor growth in vivo for GFP expression grew back quickly, but no tumors To directly assess whether BMP inhibits MB develop- developed when cells infected with virus coexpressing ment in vivo, GNP-like tumor cells purified from the BMP4 and GFP were first sorted for GFP expression prior aforementioned murine MBs were infected with retrovi- to injection (Fig. 4A, panel d). Importantly, tumors failed ral vectors coexpressing human BMP4 and green fluores- to grow when cells were pretreated with BMP4 for3din cent protein (GFP), or GFP alone, and injected subcuta- vitro before injection, whereas allografts grew back neously into the flanks of athymic mice. Allografts in quickly from untreated tumor cells (Supplemental Fig. the flank of recipient mice derived from 5 × 105 tumor 3). Because cultured GNPs overexpressing Atoh1 were cells infected with the control vector grew rapidly into resistant to BMP-induced effects, we assessed whether large tumors containing many GFP-positive cells enforced expression of Atoh1 in tumor cells would ren- (37.8% ± 14.1, n = 8) quantified by flow cytometric der them resistant to BMP treatment. Tumor cells puri- analysis of dissociated tumor cells (Fig. 4A [panel a], B). fied from spontaneously derived MBs from Cdkn2c−/−, Typically, tumors were palpable 1 wk after injection and Ptch1+/− mice were infected with retroviral vectors ex- were harvested 2–4 wk later. In contrast, allografts from pressing Atoh1 and GFP or GFP alone, and were trans- tumor cells infected with the vector coexpressing BMP4 planted into the flanks of immunocompromised recipi- ent mice. Allografts from tumor cells infected with GFP alone grew back with the same time of onset and growth rate as noninfected tumor cells, and retained the same pathological features of the original MB (data not shown). Remarkably, allografts from MB cells infected with viruses expressing Atoh1 and GFP were brightly fluorescent (Fig. 4A, panel c), and virtually all tumor cells purified from these transplants overexpressed Atoh1 (98% ± 1.15, n = 3) (Fig. 4B). GNP-like tumor cells isolated from both types of allografts were treated with BMP4 or cyclopamine for 72 h in culture. As expected, cyclopamine effectively blocked the proliferation of tu- mor cells from both types of transplants (Fig. 4C), indi- cating that tumor cells depend on the constitutively ac- tivated Shh pathway activity for their proliferation. BMP4 treatment of GNP-like tumor cells that expressed GFP alone significantly reduced the proportion of cells in S phase (Fig. 4C), but in contrast, failed to do so in tumor cells overexpressing Atoh1 (Fig. 4C). The importance of the Shh pathway as a therapeutic target in MB was demonstrated recently by the use of Smoothened inhibitors (cyclopamine and HhAntag) that cause regression of MBs in vivo (Berman et al. 2002; Romer et al. 2004; Sanchez and Ruiz i Altalba 2005). BMP2 induces apoptosis in human MB cells (Hallahan et Figure 4. BMP4 suppresses mouse MB proliferation in vivo. (A) al. 2003), while BMP2 and BMP4 suppress human glio- Representative image of allografts from tumor cells infected with blastoma as well (Piccirillo et al. 2007). Although BMP GFP control virus (panel a), or with virus expressing BMP4 and GFP (panel b) or Atoh1 and GFP (panel c). (Panel d) Athymic mice bearing and cyclopamine each arrest the proliferation of MB allograft from FACS-sorted GFP-positive tumor cells infected with cells, they elicit unique responses, raising the possibility GFP control (arrow) and GFP and BMP4 (circle). (B) FACS analysis of that BMP agonists and SHH antagonists might be thera- GFP expression from allografts derived from tumor cells infected peutically combined to reduce the potential deleterious with GFP control virus (black bar, 37.8% ± 14.1 positive cells, n = 8), effect of either agent. Indeed, we found that BMP and virus encoding BMP4 and GFP (gray bar, 2.4% ± 1.51 positive cells, n = 8), or Atoh1 and GFP (dotted bar, 98% ± 1.15 positive cells, cyclopamine acted additively to suppress proliferation n = 3). (C) GNP-like tumor cells infected with control GFP virus or and to induce differentiation of GNP-like MB cells. virus encoding Atoh1 and GFP were treated with control medium While BMP4 or cyclopamine alone functioned optimally (black bars), BMP4 (diagonal shaded bars), or cyclopamine (blank at 100 ng/mL or 10 µM, respectively, to arrest tumor cell bars) for 72 h. FACS analysis for DNA content shows the percentage proliferation, a similar level of inhibition was achieved of cells in S phase (n = 3). (D) GNP-like tumor cells were untreated with a combination of only 2.5 µM cyclopamine and 25 (−) or treated with BMP4 and/or cyclopamine (the number in the X-axis represents the concentration of BMP4 [in nanograms per mil- ng/mL BMP4 (Fig. 4D). Nonetheless, caution should be liliter] and of cyclopamine [in micromolar]. Cell proliferation was exercised in evaluating the effectiveness and safety of assessed by counting cell number after 72 h. Error bars indicate SEM such compounds, particularly in the treatment of pedi- and asterisks indicate P < 0.05 (Student’s t-test). atric malignancies, since compounds of this type may

GENES & DEVELOPMENT 725 Downloaded from genesdev.cshlp.org on October 1, 2021 - Published by Cold Spring Harbor Laboratory Press

Zhao et al. well have adverse effects on normal developmental pro- suggestions, Richard A. Ashmun and Ann-Marie Hamilton-Easton for cesses. flow cytometric analysis, and the Developmental Studies Hybridoma Bank for Atoh1 (Dr. Jane Johnson) and Cntn2 (Dr. Miyuki Yamamoto) antibodies. This study was supported by NIH grant CA-096832 and Can- Materials and methods cer Core Grant CA-21765 (to M.F.R.) and NS-15429 (to M.E.H.), La Fon- dation pour la Recherche Medicale and the Gephardt Endowed Fellow- Mouse husbandry ship Signal Transduction (to O.A.), and the American Lebanese Syrian Mice (Cdkn2c−/−, Trp53Fl/Fl, Nes-cre+) and (Cdkn2c−/−, Ptch1+/−) were Associated Charities (ALSAC) of St. Jude Children’s Research Hospital. derived and maintained as described (Uziel et al. 2005; Zindy et al. 2007). H.Z., O.A., and F.Z. performed the experiments. M.F.R. supervised and Immunocompromised athymic mice (CD1; Jackson Laboratory) were helped plan the work and wrote the paper. We all discussed the results used as allograft recipients. and helped formulate the manuscript.

Cell culture Purification of GNPs and GNP-like tumor cells was performed as de- References scribed (Zindy et al. 2007). GNPs were maintained in Neurobasal me- Akazawa, C., Ishibashi, M., Shimizu, C., Nakanishi, S., and Kageyama, R. dium containing B27 supplement, 2 mM glutamine, 100 U/mL penicil- 1995. A mammalian helix–loop–helix factor structurally related to lin/streptomycin (all from Invitrogen), 0.45% D-glucose, 1× SPITE the product of Drosophila proneural gene atonal is a positive tran- medium supplement, and 1× linoleic acid–oleic acid (all from Sigma- scriptional regulator expressed in the developing nervous system. J. Aldrich). GNP-like tumor cells were grown in medium containing N2 Biol. Chem. 270: 8730–8738. supplement and 4 mg/mL bovine serum albumin (Invitrogen) instead of Alvarez-Rodriguez, R., Barzi, M., Berenguer, J., and Pons, S. 2007. Bone SPITE and linoleic acid–oleic acid. Human recombinant BMP2, BMP4, or morphogenic protein 2 oposes Shh-mediated proliferation in cerebel- BMP7 (R&D) were used at 100 ng/mL and cyclopamine (LC Laboratories) lar granule cells through a TIEG-1-based regulation of NMyc. J. Biol. was used at 10 µM. 293T cells were maintained in Dulbecco’s Modified Chem. 282: 37170–37180. Eagle’s Medium with 10% fetal calf serum, and 100 U/mL penicillin/ Angley, C., Kumar, M., Dinsio, K.J., Hall, A.K., and Siegel, R.E. 2003. streptomycin. DNA expression constructs were transfected using Lipo- Signaling by bone morphogenetic proteins and Smad1 modulates the fectamine 2000 (Invitrogen). Where indicated, MG-132 (Calbiochem) postnatal differentiation of cerebellar cells. J. Neurosci. 23: 260–268. was used at 10 µM and cycloheximide (Sigma-Aldrich) was used at Aruga, J., Minowa, O., Yaginuma, H., Kuno, J., Nagai, T., Noda, T., and 100 µg/mL. Mikoshiba, K. 1998. Mouse Zic1 is involved in cerebellar develop- ment. J. Neurosci. 18: 284–293. DNA constructs, retrovirus production, and infection Ben-Arie, N., McCall, A.E., Berkman, S., Eichele, G., Bellen, H.J., and Human BMP2/4 hybrid cDNA was generated as described previously Zoghbi, H.Y. 1996. Evolutionary conservation of sequence and ex- (Peng et al. 2001). Generation of point mutations (E165G and R158G) in pression of the bHLH protein Atonal suggests a conserved role in the Atoh1 DNA-binding domain was performed by PCR site-directed neurogenesis. Hum. Mol. Genet. 5: 1207–1216. mutagenesis. cDNAs encoding a BMP2/4 hybrid, wild type, or mutant Ben-Arie, N., Bellen, H.J., Armstrong, D.L.M.A.E., Gordadze, P.R., Guo, Atoh1, Gli1, Mycn, Id1, Id2, and Tcfe2a were cloned into an MSCV- Q., Matzuk, M.M., and Zoghbi, H.Y. 1997. Math1 is essential for IRES-GFP vector, verified by DNA sequencing, and used to generate ret- genesis of cerebellar granule neurons. Nature 390: 169–172. roviruses as described (Zindy et al. 2007). Enriched GNPs and GNP-like Berman, D.M., Karhadkar, S.S., Hallahan, A.R., Pritchard, J.I., Eberhart, tumor cells were infected during the preplating stage. C.G., Watkins, D.N., Chen, J.K., Cooper, M.K., Taipale, J., Olson, J.M., et al. 2002. MB growth inhibition by hedgehog pathway block- Other analytical procedures ade. Science 297: 1559–1561. Immunoblotting and immunohistochemistry were performed (Zindy et Corrales, J.D., Rocco, G.L., Blaess, S., Guo, Q., and Joyner, A.L. 2004. al. 2007) using antibodies raised against Cdk2, Ccnd1, Id1, Id2, Neurod1, Spatial pattern of sonic-hedgehog signaling through Gli genes during ␤ and -actin (Santa Cruz Biotechnology); phosphorylated Smad1,5,8 (Cell cerebellum development. Development 131: 5581–5590. Signaling); Zic1 and Gli1 (Rockland); Mycn (Calbiochem); Atoh1 (Devel- Dahmane, N., Sanchez, P., Gitton, Y., Palma, V., Sun, T., Beyna, M., opmental Studies Hybridoma Bank, DSHB); Pax6 and Tubb1 (Covance); Weiner, H., and Ruiz i Altalba, A. 2001. The Sonic Hedgehog–Gli glial fibrillary acidic protein (GFAP); and synaptophysin (DAKO). For pathway regulates dorsal brain growth and tumorigenesis. Develop- immunofluorescence, cells fixed on slides were blocked with 3% normal ment 128: 5201–5212. goat serum and then incubated with primary antibodies raised against Fogarty, M.P., Kessler, J.D., and Wechsler-Reya, R.J. 2005. Morphing into Atoh1 and Tubb1 (Covance); Cntn2 (DSHB); Cdkn1b (Beckman Dickin- cancer: The role of developmental signaling pathways in brain tumor son); BrdU (Santa Cruz Biotechnology); Neuna60 (Millipore); Nefh formation. J. Neurobiol. 64: 458–475. (Sigma-Aldrich); and GFP (Invitrogen) (Uziel et al. 2005). Cultured GNPs Fogarty, M.P., Emmenegger, B.A., Grasfeder, L.L., Oliver, T.G., and were incubated with 10 µM BrdU (BD Biosciences) for 1.5 h and stained Wechsler-Reya, R.J. 2007. Fibroblast growth factor blocks Sonic with antibodies to BrdU as described (Uziel et al. 2005). BrdU-positive hedgehog signaling in neuronal precursors and tumor cells. Proc. nuclei within at least 600 DAPI- or GFP-labeled cells were counted under Natl. Acad. Sci. 104: 2973–2978. a fluorescent microscope, and the percentage of BrdU-positive cells Gazit, R., Krizhanovsky, V., and Ben-Arie, N. 2004. Math1 controls cer- within the total number of DAPI- or GFP-labeled stained nuclei are in- ebellar granule cell differentiation by regulating multiple compo- dicated. Each treatment was repeated independently at least three times. nents of the . Development 131: 903–913. Q-RT–PCR on RNA extracted from cultured cells was performed with Hallahan, A.R., Pritchard, J.I., Chandraratna, R.A., Ellenbogen, R.G., primers and probes described previously (Lee et al. 2003; Zindy et al. Geyer, J.R., Overland, R.P., Strand, A.D., Tapscott, S.J., and Olson, 2007). J.M. 2003. BMP-2 mediates retinoid-induced apoptosis in medullo- blastoma cells through a paracrine effect. Nat. Med. 9: 1033–1038. Acknowledgments Kenney, A.M. and Rowitch, D.H. 2000. Sonic hedgehog promotes G(1) cyclin expression and sustained cell cycle progression in mammalian We thank Charles J. Sherr for support, helpful criticisms, and assistance neuronal precursors. Mol. Cell. Biol. 20: 9055–9067. in formulating the manuscript and all members of the laboratory for Kenney, A.M., Cole, M.D., and Rowitch, D.H. 2003. N- upregulation helpful discussions. We are indebted to Dr. Marie-Elizabeth Hatten for by sonic hedgehog signaling promotes proliferation in developing cer- sharing her unpublished data. We thank Dr. Chunxu Qu for microarray ebellar granule neurons. Development 130: 15–28. data and Drs. Jerold Rehg and David Elisson for pathology analysis. We Lee, Y., Miller, H.L., Jensen, P., Hernan, R., Connelly, M., Wetmore, C., thank Drs. Peter McKinnon, Tom Curran, John Cleveland, and Barbara Zindy, F., Roussel, M.F., Curran, T., Gilbertson, R., et al. 2003. A Christy for Atoh1, Gli1, Mycn, Id1, and Id2 cDNAs, respectively. We are molecular fingerprint for medulloblastoma. Cancer Res. 63: 5428– indebted to Deborah Yons, Robert Jenson, and Shelly Wilkerson for ex- 5437. cellent technical assistance, and Suqing Xie for help with q-RT–PCR. We Lumpkin, E.A., Collisson, T., Parab, P., Omer-Abdalla, A., Haeberle, H., thank Dr. Kimura Hiromichi for sharing unpublished data and helpful Chen, P., Doetzlhofer, A., White, P., Groves, A., Segil, N., et al. 2003.

726 GENES & DEVELOPMENT Downloaded from genesdev.cshlp.org on October 1, 2021 - Published by Cold Spring Harbor Laboratory Press

BMPs suppress medulloblastoma

Math1-driven GFP expression in the developing nervous system of transgenic mice. Brain Res. Gene Expr. Patterns 3: 389–395. Machold, R. and Fishell, G. 2005. Math1 is expressed in temporally dis- crete pools of cerebellar rhombic-lip neural progenitors. Neuron 48: 17–24. Marino, S. 2005. Medulloblastoma: Developmental mechanisms out of control. Trends Mol. Med. 11: 17–22. Miyata, T., Maeda, T., and Lee, J.E. 1999. NeuroD is required for differ- entiation of the granule cells in the cerebellum and hippocampus. Genes & Dev. 13: 1647–1652. Oliver, T.G., Grasfeder, L.L., Carroll, A.L., Kaiser, C., Gillingham, C.L., Lin, S.M., Wickramasinghe, R., Scott, M.P., and Wechsler-Reya, R.J. 2003. Transcriptional profiling of the Sonic hedgehog response: A critical role for N-myc in proliferation of neuronal precursors. Proc. Natl. Acad. Sci. 100: 7331–7336. Peng, H., Chen, S.T., Wergedal, J.E., Polo, J.M., Yee, J.K., Lau, K.H., and Baylink, D.J. 2001. Development of an MFG-based retroviral vector system for secretion of high levels of functionally active human BMP4. Mol. Ther. 4: 95–104. Piccirillo, S.G., Reynolds, B.A., Zanetti, N., Lamorte, G., Binda, E., Broggi, G., Brem, H., Olivi, A., Dimeco, F., and Vescovi, A.L. 2007. Bone morphogenetic proteins inhibit the tumorigenic potential of human brain tumour-initiating cells. Nature 444: 761–765. Rios, I., Alvarez-Rodriguez, R., Marti, E., and Pons, S. 2004. Bmp2 an- tagonizes sonic hedgehog-mediated proliferation of cerebellar granule neurons through Smad5 signaling. Development 131: 3159–3168. Romer, J.T., Kimura, H., Magdaleno, S., Sasai, K., Fuller, C., Baines, H., Connelly, M., Stewart, C.F., Gould, S., Rubin, L.L., et al. 2004. Sup- pression of the Shh pathway using a small molecule inhibitor elimi- nates medulloblastoma in Ptc1+/−−/− mice. Cancer Cell 6: 229– 240. Sanchez, P. and Ruiz i Altalba, A. 2005. In vivo inhibition of endogenous brain tumors through systemic interference of Hedgehog signaling in mice. Mech. Dev. 122: 223–230. Shou, J., Rim, P.C., and Calof, A.L. 1999. BMPs inhibit neurogenesis by a mechanism involving degradation of a transcription factor. Nat. Neurosci. 2: 339–345. Uziel, T., Zindy, F., Xie, S., Lee, Y., Forget, A., Magdaleno, S., Rehg, J.E., Calabrese, C., Solecki, D., Eberhart, C.G., et al. 2005. The tumor suppressors Ink4c and p53 collaborate independently with Patched to suppress medulloblastoma formation. Genes & Dev. 19: 2656–2667. Vinals, F., Reiriz, J., Ambrosio, S., Bartrons, R., Rosa, J.L., and Ventura, F. 2004. BMP-2 decreases Mash1 stability by increasing Id1expression. EMBO J. 23: 3527–3537. Wallace, V.A. 1999. Purkinje cell-derived Sonic hedgehog regulates gran- ule neuron precursor cell proliferation in the developing mouse cer- ebellum. Curr. Biol. 9: 445–448. Wechsler-Reya, R.J. and Scott, M.P. 2003. Control of neuronal precursor proliferation in the cerebellum by Sonic hedgehog. Neuron 22: 103– 114. Yamasaki, T., Kawaji, K., Ono, K., Bito, H., Hirano, T., Osumi, N., and Kengaku, M. 2001. Pax6 regulates granule cell polarization during parallel fiber formation in the developing cerebellum. Development 128: 3133–3144. Zindy, F., Uziel, T., Ayrault, O., Calabrese, C., Valentine, M., Rehg, J.E., Gilbertson, R.J., Sherr, C.J., and Roussel, M.F. 2007. Genetic alter- ations in mouse medulloblastomas and generation of tumors de novo from primary cerebellar granule neuron precursors. Cancer Res. 67: 2676–2684.

GENES & DEVELOPMENT 727 Downloaded from genesdev.cshlp.org on October 1, 2021 - Published by Cold Spring Harbor Laboratory Press

Post-transcriptional down-regulation of Atoh1/Math1 by bone morphogenic proteins suppresses medulloblastoma development

Haotian Zhao, Olivier Ayrault, Frederique Zindy, et al.

Genes Dev. 2008, 22: Access the most recent version at doi:10.1101/gad.1636408

Supplemental http://genesdev.cshlp.org/content/suppl/2008/03/03/22.6.722.DC1 Material

Related Content Hh and BMP Converge on Atoh1 Annalisa VanHook Sci. Signal. March , 2008 1: ec111

References This article cites 33 articles, 18 of which can be accessed free at: http://genesdev.cshlp.org/content/22/6/722.full.html#ref-list-1

Articles cited in: http://genesdev.cshlp.org/content/22/6/722.full.html#related-urls

License Freely available online through the Genes & Development Open Access option.

Email Alerting Receive free email alerts when new articles cite this article - sign up in the box at the top Service right corner of the article or click here.

Copyright © 2008, Cold Spring Harbor Laboratory Press