We examined the genetic requirements for the Myc family of oncogenes in normal Sonic hedgehog (Shh)–mediated cerebellar granule neuronal precursor (GNP) expansion and in Shh pathway–induced medulloblastoma formation. In GNP-enriched cultures derived from N-mycFl/Fl and c-mycFl/Fl mice, disruption of N-myc, but not c-myc, inhibited the proliferative response to Shh. Conditional deletion of c-myc revealed that, although it is necessary for the general regulation of brain growth, it is less important for cerebellar development and GNP expansion than N-myc. In vivo analysis of compound mutants carrying the conditional N-myc null and the activated Smoothened (ND2:SmoA1) alleles showed, that although granule cells expressing the ND2:SmoA1 transgene are present in the N-myc null cerebellum, no hyperproliferation or tumor formation was detected. Taken together, these findings provide in vivo evidence that N-myc acts downstream of Shh/Smo signaling during GNP proliferation and that N-myc is required for medulloblastoma genesis even in the presence of constitutively active signaling from the Shh pathway. (Cancer Res 2006; 66(17): 8655-61)
- granule neuron precursors
- Sonic hedgehog
Medulloblastoma is the most common childhood malignancy of the central nervous system and, despite aggressive multimodal therapy with surgery, radiation, and chemotherapy, 5-year survival rates have only approached recently >60% ( 1). Medulloblastomas are thought to derive from granule precursor cells of the cerebellum and comprise several histologic and prognostic subgroups. The desmosplastic subtype is distinguished from “classic” medulloblastoma by the presence of relatively acellular nodules within the dense sheets of malignant cells ( 2).
The mitogenic activity of the Sonic hedgehog (Shh) pathway is well characterized in the development of the cerebellum, where Shh drives the proliferation of granule neuronal precursors (GNP) in the external granular layer (EGL; refs. 3, 4). GNPs originate from the rhombic lip during embryonic development and then migrate out to cover the outer surface of the developing cerebellum during both late embryogenesis and the postnatal period, where they undergo significant expansion to form the EGL ( 5). Shh is secreted from Purkinje cells and binds to the Patched (PTCH) receptor on the GNPs, which relieves inhibition of the G-coupled transmembrane receptor Smoothened (Smo). Smo then activates the Gli family of zinc finger transcription factors ( 6– 8) and the proto-oncogene N-myc ( 9), as well as expression of cyclin D and cyclin E ( 10), which together are thought to promote the proliferation of GNPs in the EGL ( 11).
Numerous studies have shown that deregulated Shh signaling is critical to the genesis and survival of both classic and desmoplastic subtypes of sporadic medulloblastoma ( 12– 16). Genetic alterations associated with medulloblastoma include activating mutations in Smo and inactivating mutations in negative regulators of Shh signaling, such as Ptch1 and suppressor of fused (SUFU). Thus, Shh signaling regulates normal cerebellar development, and deregulation of this signaling plays a critical role in the formation of medulloblastomas.
In addition to mutations in the Ptch1, SUFU, and Smo genes, amplifications of the proto-oncogenes N-myc and c-myc have also been observed in 5% to 8% of medulloblastoma cases ( 17). The Myc family of basic helix-loop-helix leucine zipper proteins act as transcriptional regulators that direct cell growth and proliferation, and Myc family members play a key role in promoting the progression of the cell cycle in proliferating cells (for reviews, see refs. 18, 19). In vitro studies with cultures enriched for GNPs have shown that treatment with soluble Shh protein up-regulates N-myc expression and that N-myc overexpression can promote GNP proliferation even in the absence of Shh treatment ( 9). Overexpression of c-myc in cultures of nestin-positive neural progenitors promoted their proliferation ( 20), but treatment of GNP-enriched cultures with Shh did not induce c-myc expression ( 9), suggesting that c-myc may not be transcriptionally regulated by Shh. Furthermore, expression of dominant-negative mutants of myc block the proliferative response induced by Shh ( 9, 21), and conditional knockout of N-myc within the brain severely disrupts the expansion of GNPs within the cerebellum ( 11), linking N-myc to normal granule cell development.
We reported previously a transgenic mouse model, in which the expression of a constitutively activated form of Smoothened (SmoA1) was driven in cerebellar GNPs through the use of the 1-kb neuroD2 promoter (ND2:SmoA1; ref. 16), whose expression begins during embryonic development. The SmoA1 mutation was originally identified in cases of sporadic basal cell carcinoma, and the mutated Smo protein was shown to transform rat embryonic fibroblast cells in culture ( 22). Mutations that activate Smo have also been associated with sporadic brain tumors ( 23– 25). The ND2:SmoA1 mice exhibit a high penetrance (48%) of cerebellar tumors that histologically resemble human medulloblastoma with a median tumor onset of 6 months ( 16).
Whereas the role of Shh in driving GNPs expansion within the developing cerebellum and the link between deregulated Shh signaling and medulloblastoma formation have been clearly established, the requirement for Myc family function has not been as well defined in Shh-induced tumorigenesis. In particular, previous studies have not examined the role of N-myc by genetic loss-of-function analysis in the context of medulloblastoma tumorigenesis. Furthermore, potential roles for the other major myc family member, c-myc, in GNP expansion are unknown. To address these important questions, we ablated N-myc or c-myc using Cre/lox conditional targeting both in vitro with cultured GNPs and in vivo with conditional knockout mice.
Materials and Methods
Cerebellar GNP cultures. GNPs were isolated from wild-type (WT) N-mycFl/Fl and c-mycFl/Fl mice at postnatal day 5 (P5) and cultured overnight in the presence of Shh (3 μg/mL) and serum and then with Shh and without serum for 3 to 5 days as described previously ( 11, 26). For infection, Shh-conditioned medium was removed and cells were incubated with retroviral supernatants for 2 hours. Viral supernatants were then removed and Shh-conditioned medium was replaced. GNPs were infected on three sequential days leading to an infection rate of >90% as verified by green fluorescent protein (GFP) status. The biologically active, modified NH2-terminal peptide of Shh was used for Shh-conditioned medium and provided by Curis, Inc. (Cambridge, MA).
Retroviral constructs and virus production. Amphotropic retrovirus was produced using the VSV-G system ( 27). Briefly, GFP or GFP-Cre retroviral MSCV plasmids were cotransfected with VSV-G helper plasmids into 293T cells. Two days later, conditioned medium containing virus was harvested and concentrated to 30× by centrifugation (Sorvall 20K G-30 minutes) and the viral pellet was then resuspended in GNP medium.
Immunocytochemical labeling and analysis. For analysis of proliferative capacity, GNP cultures were pulsed with 25 μg/mL bromodeoxyuridine (BrdUrd) for 2 hours before fixation in 4% paraformaldehyde. Cells were washed in PBS and treated with 2 N HCl. Fresh 2 N HCl was added again and cells were incubated for 30 minutes at room temperature. Cells were then immunocytochemically labeled using standard methods. Primary antibodies included mouse anti-BrdUrd and rabbit anti-N-myc (Santa Cruz Biotechnology, Santa Cruz, CA). Fluorochrome-conjugated secondary antibodies included tetramethyl rhodamine isothiocyanate–conjugated anti-mouse and anti-rabbit IgG (Jackson Immunoresearch Laboratories, West Grove, PA). A 4′,6-diamidino-2-phenylindole (DAPI) costain was used to label cell nuclei. Staining was visualized with a Zeiss Axioskop 40 microscope and images were captured with an Axiocam MRm digital camera. For all BrdUrd quantification studies, 100 to 200 cells from 3 independent fields were counted for each sample.
Generation of conditional c-myc null mice. Transgenic mice were maintained in accordance with the NIH Guide for the Care and Use of Experimental Animals with approval from our Institutional Animal Care and Use Committee (IR# 1064). c-mycFl/Fl mice ( 28) were crossed to nestin-Cre mice ( 29) to create c-mycFl/WT nestin-Cre mice. These heterozygous mice were then crossed to c-mycFl/Fl mice to generate mice null for c-myc in neural stem and progenitor cells. The number of mice null for c-myc matched that predicted by Mendelian inheritance, showing that this combination was not embryonic lethal. Animals were generated on a mixed C57Bl/6 and 129 background.
Generation of combined N-myc conditional knockout and ND2:SmoA1 mice. Transgenic mice were maintained in accordance with the NIH Guide for the Care and Use of Experimental Animals with approval from our Institutional Animal Care and Use Committee (IR# 1457). A two-step breeding strategy was used to combine the two transgenic mouse models, which resulted in 12 different possible genotypes. These 12 genotypes were then further subdivided (see Table 1 ) for phenotypic and histologic analysis. Animals were generated on a mixed C57Bl/6 and 129 background.
Mouse pathology. Mice were euthanized using CO2 inhalation; the brains were removed and tissue was fixed in 4% paraformaldehyde for pathologic examination. Tissue blocks were paraffin embedded, cut into 4-μm sections, and stained with H&E using standard methods. For examination of the four groups of mice from the combined N-myc conditional knockout and ND2:SmoA1 models, persistent EGL was defined as pockets of granule cells remaining in the molecular layer; hyperplasia was defined as morphologically abnormal populations of cerebellar granule cells; a tumor was defined as sheets of small round blue cells without identifiable cerebellar architecture. For analysis of ND2:SmoA1 expression, sections were processed with standard immunohistochemical methods. A monoclonal anti-PentaHis antibody (Molecular Probes, Eugene, OR) was used with a FITC-conjugated anti-mouse IgG antibody (Jackson Immunoresearch Laboratories) and a DAPI costain was used to label cell nuclei. For Ki-67 and nesting staining, a polyclonal Ki-67 antibody (Santa Cruz Biotechnology, Santa Cruz, CA) and a monoclonal nestin antibody (Chemicon International, Temecula, CA) were used; secondary antibodies applied according to Vectastain Elite avidin-biotin complex method instructions (Vector Laboratories, Burlingame, CA) and detection were carried out with 3,3′-diaminobenzidine reagent (Vector Laboratories). Cells were visualized with a Zeiss Axioscope 40 microscope and images were captured with a Qimaging MicroImager II digital camera.
Results and Discussion
N-myc is required for the proliferative response of cultured GNPs to Shh. GNPs were isolated from N-mycFl/Fl mice and WT control mice at the height of their postnatal expansion (P5), cultured as described previously ( 26), and infected with retroviruses encoding bicistronic Cre recombinase and GFP. A retrovirus expressing GFP alone was used as a control. GNPs were cultured in the presence of Shh (3 μg/mL) throughout the experiment. At 5 days after infection, GFP-Cre-infected cells were verified to be N-myc deficient by immunofluorescent staining ( Fig. 1C ). We have found previously that such N-myc null GNPs display no change in apoptosis or in specification of granule cell identity as determined by cleaved caspase-3 and Zic1 immunofluorescence, respectively ( 30). At 3 and 5 days following infection, control and null GNPs were pulsed for 2 hours with BrdUrd to label G1 and S phase cells and then fixed and stained to examine BrdUrd incorporation. In both WT and GFP-infected control cultures by day 5, ∼80% of cells incorporated BrdUrd, indicating a highly proliferative population. In contrast, only 10% of the N-myc-deficient cells incorporated BrdUrd, an 8-fold reduction ( Fig. 1A and B). A similar level of reduced proliferation is observed in knockout GNPs derived from N-mycFl/Fl nestin-Cre P5 pups. 8 The markedly reduced levels of proliferation in N-myc null cells were similar to those seen in GNPs cultured in the absence of Shh ( Fig. 1B), suggesting that N-myc is an essential downstream effector of Shh signaling.
c-myc function is generally required for normal brain growth but has no specific roles in cerebellar development. N-myc is expressed in the proliferative region of the EGL, where Shh activity has been shown to drive GNP proliferation ( 4, 9, 31). These studies indicated that N-myc is the only myc family member in the cerebellum that is regulated by Shh and raised the possibility that N-myc may have a unique role in the proliferative expansion of GNPs, a notion supported by studies showing that the conditional knockout of N-myc within the developing nervous system prevented the full expansion of GNPs ( 11). Nonetheless, c-myc is also expressed in the developing brain ( 32) and low levels or transient expression of c-myc could potentially contribute to GNP growth.
To further address the role of c-myc in the development of the cerebellum, we conditionally knocked out c-myc in neural stem and progenitor cells in vivo using nestin-Cre. As shown in Fig. 2A , mice conditionally null for c-myc did not display a cerebellum-specific phenotype as was seen in the conditional N-myc knockout mice ( 11). Although the overall size of the c-myc null brain was diminished (18% reduction; Fig. 2B) compared with that of control mice, the expansion of GNPs and the development of the cerebellum were not specifically impaired by the loss of c-myc ( Fig. 2A). The cerebella of animals lacking c-myc resembled those of control animals in the thickness of the inner granular layer (IGL) as well as in the extent of cerebellar foliation ( Fig. 2C). In contrast, animals lacking N-myc in neural stem and progenitor cells displayed a reduction in total brain mass ( Fig. 2B) and a failure in full expansion of the GNPs ( Fig. 2C) as was shown previously ( 11).
To examine whether the acute loss of c-myc would have a similar effect on the proliferative response of GNPs to Shh in vitro, Cre/lox conditional targeting was used to knockout c-myc in primary GNP cultures. Cultures were infected with the GFP-Cre retrovirus or the control GFP virus, pulsed with BrdUrd 3 days after infection, and fixed and stained to examine BrdUrd incorporation. The infection of c-mycFl/Fl GNP cultures with the GFP-Cre or GFP control retroviruses did not affect their proliferative response to Shh in either case. Approximately equal fractions of both control and GFP-Cre infected cells incorporated BrdUrd ( Fig. 1B), indicating that c-myc is dispensable for the proliferative response of GNP cultures to Shh. The levels of BrdUrd incorporation were lower in the control c-mycFl/Fl cultures infected with GFP alone compared with the control GFP-infected N-mycFl/Fl or WT cultures due to experimental variability in the preparation of the cultures. Note that N-mycFl/Fl cultures infected with GFP-Cre had incorporation levels below both c-mycFl/Fl and WT cultures infected with either GFP-Cre or GFP alone. Thus, it seems that both c-myc and N-myc have important roles in the overall growth of the brain but that N-myc has a more critical role in cerebellar development and specifically in the expansion of GNPs. Because the function of N-myc was clearly critical to the proliferation of GNP cultures in response to Shh, we chose to examine the requirement for N-myc in the context of ND2:SmoA1-driven cerebellar hyperplasia and tumor formation.
N-myc functions downstream of Smo and is necessary for the hyperproliferation induced by constitutive activation of the Shh pathway. In addition to the formation of medulloblastomas, 55% of mice carrying the ND2:SmoA1 transgene display hyperproliferation of the granular layer by 3 months of age ( Table 1). Expanded pockets of granule cells are present in the IGL of these mice, indicative of the excess proliferation of these cells. Furthermore, 80% of ND2:SmoA1 mice have a persistent EGL ( Table 1), which is normally absent by postnatal day 20 in WT mice ( 17). These populations of cells also express the nuclear antigen Ki-67 ( Fig. 3A and B ), which marks proliferative cells, and the neural stem cell marker nestin ( Fig. 3E and F), indicating that the presence of the ND2:SmoA1 transgene may sustain granule neural precursors in a progenitor state. Ki-67 and nestin expression was not seen in WT controls ( Fig. 3C and G) or in ND2:SmoA1 cerebella lacking hyperplastic or persistent EGL phenotypes. This granule cell hyperplasia and the persistent EGL observed at 3 months may thus represent early stages in the formation of cerebellar tumors in the ND2:SmoA1 mice. Real-time PCR analysis showed that N-myc expression is elevated in cerebellar tumors generated in the ND2:SmoA1 medulloblastoma model ( 16), suggesting that the overexpression of N-myc may be involved in the initiation of hyperplasia and progression to tumor formation in SmoA1-induced medulloblastoma.
To determine whether N-myc is necessary for the hyperproliferation of GNPs seen in the ND2:SmoA1 transgenic mice, we crossed the ND2:SmoA1 mice with conditional (Nestin-Cre) N-myc knockout mice ( 11) to generate mice that possess persistently elevated Shh pathway activity in the cerebellum while lacking N-myc in the same region. These crosses produced four general groups of mice that we subsequently analyzed ( Table 1). The number of mice that expressed ND2:SmoA1 and was null for N-myc in the cerebellum matched the number predicted by Mendelian inheritance, showing that this combination was not embryonic lethal.
The ND2:SmoA1 × N-myc conditional knockout mice were initially analyzed at 3 months of age. From birth onward, mice nullizygous for N-myc displayed moderate growth retardation and, by 3 weeks of age, began to exhibit tremors, an ataxic gait and raised tails, regardless of the presence or absence of the ND2:SmoA1 transgene. These phenotypes are also characteristic of the N-myc knockout mice ( 11). Forty-one mice were sacrificed at 3 months of age and it was noted that, compared with WT controls ( Fig. 4A ), nullizygous N-myc mice had significantly diminished cerebella in both the presence ( Fig. 4C) and absence ( Fig. 4D) of the ND2:SmoA1 transgene. Expression of the ND2:SmoA1 transgene was confirmed by immunofluorescent staining against the His tag present on the Smo protein encoded by the ND2:SmoA1 transgene. Staining with an anti-His antibody verified the expression of the ND2:SmoA1 transgene within granule cells of the N-myc null, ND2:SmoA1+ (N-myc?, SmoA1+) cerebella ( Fig. 5A ) as was seen in N-mycWT/WT, ND2:SmoA1+ controls ( Fig. 5C). Therefore, ND2:SmoA1 expression was not disrupted by the loss of N-myc, and expression occurs throughout the remaining GNP population.
Histologic examination revealed a strong reduction in cerebellar foliation in mice lacking N-myc, regardless of the presence ( Fig. 4G) or absence ( Fig. 4H) of the ND2:SmoA1 transgene. Note that the granular, molecular, and Purkinje cell layers are present. Additionally, the expression of Ki-67 and nestin was not observed in cerebella lacking N-myc ( Fig. 3D and H), suggesting that N-myc may be required for maintaining a proliferative, precursor-like population of cells during Shh-induced tumorigenesis. That constitutive activation of Smo fails to reverse the nullizygous N-myc overall brain and cerebellar-specific phenotypes suggests that N-myc is a critical downstream effector of Smo in the Shh signaling pathway. These findings also showed the requirement for N-myc in Shh-induced hyperplasia and in the persistent EGL observed in the majority of mice carrying the ND2:SmoA1 transgene.
N-myc is required for tumorigenesis induced by deregulated Shh signaling. Further analysis was carried out in mice at 6 months of age to examine the requirement for N-myc in ND2:SmoA1-driven medulloblastoma formation. Fifty-five mice from our study were sacrificed at 6 months. Mice nullizygous for N-myc and positive for the ND2:SmoA1 transgene phenotypically resembled the N-myc conditional knockout mice with significantly diminished cerebella ( Fig. 4C). Histologic analysis at this time point revealed that mice lacking N-myc ( Fig. 4G and H) have defects in cerebellar foliation compared with that of WT controls ( Fig. 4E), with a reduction of the granular layer despite the presence of the ND2:SmoA1 transgene. No evidence of hyperproliferation or tumor formation was detected in mice lacking N-myc or in the WT controls ( Table 1). Mice WT for N-myc and positive for the ND2:SmoA1 transgene developed hyperplasia and tumors ( Fig. 4B and F) with an incidence similar to that seen in the study of the ND2:SmoA1 model ( Table 1). These results indicate that N-myc is necessary for the hyperproliferation and tumor formation induced by the ND2:SmoA1 transgene. Furthermore, the hyperactivity of the Shh signaling pathway induced by the ND2:SmoA1 transgene is unable to compensate for the loss of N-myc in both normal GNP expansion as well as in hyperproliferation and tumor formation. Taken together, these findings provide in vivo evidence that N-myc acts downstream of Shh and Smo signaling during GNP proliferation and that N-myc is required for medulloblastoma genesis even in the presence of constitutively active signaling from the Shh pathway.
Although the loss of N-myc limits the expansion of the GNP population during normal cerebellar development, its loss does not ablate these cells entirely. GNPs are still specified and the granular layer does not fail to form. Thus, the cells thought to be the originators of tumor formation are still present within cerebella lacking N-myc. The phenotype of the N-myc null cerebellum was attributed previously to the inhibition of GNP proliferation. In addition, precocious differentiation of precursor cells in the cerebral cortex in the absence of N-myc was hypothesized to contribute to the failure of overall brain growth ( 11), indicating that N-myc may be functioning to maintain cells in a progenitor-like state. This role is consistent with the function of the Shh signaling pathway, which has been shown to prevent the terminal differentiation of GNPs ( 3, 4, 26), and whose deregulation might contribute to maintaining a subpopulation of GNPs in a progenitor-like state during tumorigenesis. The results of this study also suggest that other components of the Shh signaling pathway are unable or insufficient to activate the proliferative machinery of cell in the absence of N-myc despite the constitutive activation of the Smoothened signal beginning in early embryonic development, showing that N-myc plays a central role in Shh-mediated proliferation and tumorigenesis.
The requirement for N-myc during Shh-induced hyperplasia and tumorigenesis makes it an appealing target for therapeutic intervention in the treatment of medulloblastoma. Studies using GNP cultures have shown that the overexpression of N-myc can rescue the inhibition induced by treatment of these cultures with the Shh/Smo antagonist cyclopamine ( 21). Taken together with the results of our current study, these findings suggest that the prevention of overactive Shh signaling through the inhibition of N-myc activity may be beneficial. Our study shows that N-myc is a necessary downstream effector of Shh/Smo signals in vitro and in vivo and that N-myc may be an efficient target for the prevention of tumor growth because it may act as an integrator of multiple signaling pathways that act to drive GNP proliferation, in addition to being a key effector of the Shh signaling pathway. N-myc levels in GNPs are regulated in part by glycogen synthase kinase-3β phosphorylation, which targets N-myc for proteasome degradation ( 33). Therefore, antagonism of N-myc activity by phosphatidylinositol 3-kinase blockade or other targeted therapies could lead to a clinically relevant reduction in the pool of proliferating cells that progress to malignant tumors.
Grant support: NIH grant CA112350-01, Children's Brain Tumor Foundation, and Damon Runyon Clinical Investigator Award (B.A. Hatton and J.M. Olson); NIH grant CA114400-01 (P.S. Knoepfler); Medical Foundation, Inc. and Sontag Foundation (A.M. Kenney); National Institute of Neurological Disorders and Stroke and James McDonnel Foundation (D.H. Rowitch); Spanish Ministry of Education and Science (I.M. de Alborán); and NIH grant CA20525 (R.N. Eisenman).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
We thank Linda Cherepow in the Histopathology Shared Resource at Fred Hutchinson Cancer Research Center (Seattle, WA) for sectioning and H&E staining the transgenic mouse tissues, Jennifer Stoeck for assistance with transgenic animals, and Tina Xu for excellent technical assistance.
Note: B.A. Hatton and P.S. Knoepfler contributed equally to this work.
R.N. Eisenman is an American Cancer Society Research Professor.
↵8 P.S. Knoepfler and Robert N. Eisenman, unpublished data.
- Received May 3, 2006.
- Revision received June 19, 2006.
- Accepted June 23, 2006.
- ©2006 American Association for Cancer Research.