Medulloblastoma tumorigenesis diverges from cerebellar ... - Core

Available online at www.sciencedirect.com R

Developmental Biology 263 (2003) 50 – 66

www.elsevier.com/locate/ydbio

Medulloblastoma tumorigenesis diverges from cerebellar granule cell differentiation in patched heterozygous mice John Y.H. Kim,a,b,1 Aaron L. Nelson,a Sibel A. Algon,c Ondrea Graves,a Lisa Marie Sturla,a Liliana C. Goumnerova,c David H. Rowitch,b Rosalind A. Segal,b and Scott L. Pomeroya,* a

Department of Neurology, Division of Neuroscience, Children’s Hospital, Harvard Medical School, Boston, MA 02115, USA b Department of Pediatric Oncology, Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA 02115, USA c Department of Neurosurgery, Division of Neuroscience, Children’s Hospital, Harvard Medical School, Boston, MA 02115, USA Received for publication 22 October 2002, revised 1 July 2003, accepted 1 July 2003

Abstract Medulloblastoma is a cerebellar tumor that can arise through aberrant activation of Sonic hedgehog (Shh) signaling, which normally regulates cerebellar granule cell proliferation. Mutations of the Shh receptor PATCHED (PTCH) are associated with medulloblastomas, which have not been found to have loss of PTCH heterozygosity. We address whether patched (Ptc) heterozygosity fundamentally alters granule cell differentiation and contributes to tumorigenesis by increasing proliferation and/or decreasing apoptosis in Ptc⫹/⫺ mice. Our data show that postnatal Ptc⫹/⫺ mouse granule cell precursor growth is not globally altered. However, many older Ptc⫹/⫺ mice display abnormal cerebellar regions containing persistently proliferating granule cell precursors. Since fewer Ptc⫹/⫺ mice form medulloblastomas, these granule cell rests represent a developmentally disrupted, but uncommitted stage of tumorigenesis. Although Ptc⫹/⫺ mouse medulloblastomas express neurodevelopmental genes, they diverge from granule cell differentiation in their discordant coexpression of postmitotic markers despite their ongoing growth. Like human medulloblastomas, mouse tumors with reduced levels of the neurotrophin-3 receptor, trkC/Ntrk3, display decreased apoptosis in vivo, illustrating the role of TrkC in regulating tumor cell survival. These results indicate that Ptc heterozygosity contributes to tumorigenesis by predisposing a subset of granule cell precursors to the formation of proliferative rests and subsequent dysregulation of developmental gene expression. © 2003 Elsevier Inc. All rights reserved. Keywords: Apoptosis; Cerebellum; Granule cell; Medulloblastoma; Mouse; patched; Proliferation; trkC

Introduction Recent evidence links brain tumor oncogenesis to mechanisms of normal brain development. Medulloblastoma is the most common malignant brain tumor of childhood (Chintagumpala et al., 2001). The histogenesis of human medulloblastoma in the cerebellar granule cell lineage is supported by their expression of developmentally regulated granule cell-specific markers (Giangaspero, et al., 2000; Kadin, et al., 1970; Kozmik et al., 1995; Pomeroy et al.,

* Corresponding author. Fax: ⫹1-617-730-0242. E-mail address: [email protected] (S.L. Pomeroy). 1 Current address: Texas Children’s Cancer Center, 6621 Fannin Street, MC 3-3320, Houston, TX 77030, USA. 0012-1606/$ – see front matter © 2003 Elsevier Inc. All rights reserved. doi:10.1016/S0012-1606(03)00434-2

2002; Yokota et al., 1996). Unlike most neuronal populations, granule cell precursors remain mitotically active even after birth and constitute potential targets of transforming insults (Wang and Zoghbi, 2001). Activation of Sonic hedgehog (Shh) signaling normally regulates granule cell proliferation and differentiation but can also contribute to medulloblastoma formation (Dahmane et al., 2001; Wechsler-Reya and Scott, 1999). Humans with Gorlin’s or nevoid basal cell carcinoma syndrome (NBCCS) and germline mutations of the Shh receptor gene, PATCHED (PTCH), are predisposed to medulloblastoma tumorigenesis (Gorlin, 1995; Hahn et al., 1996; Johnson et al., 1996). Mice with heterozygous patched (Ptc/Ptc1) mutation are also susceptible to medulloblastoma formation (Goodrich et al., 1997; Hahn et al., 2000). However, only a

J.Y.H. Kim et al. / Developmental Biology 263 (2003) 50 – 66

minority of either NBCCS patients or Ptc⫹/⫺ mice go on to develop medulloblastomas, which do not feature loss of heterozygosity of the remaining wild-type PTCH or Ptc allele, respectively (Dong et al., 2000; Wetmore et al., 2000). Since Ptc does not appear to act as a classical tumor suppressor, we investigated whether Ptc heterozygosity itself contributes to medulloblastoma tumorigenesis by disrupting granule cell development. Normally, granule cells arise from committed precursors that migrate during late embryonic development over the surface of the cerebellar anlage to form a superficial germinal layer known as the external granule cell or external granular layer (EGL) (Altman and Bayer, 1997; Rakic, 1973; Wang and Zoghbi, 2001). Perinatally, the EGL expands dramatically as immature granule cell precursors in the outermost regions proliferate under the mitogenic influence of Shh (Wechsler-Reya and Scott 1999). In the deeper region EGL, granule cells withdraw from the cell cycle and migrate inward, terminally differentiating in the internal granule cell layer (IGL). Our data indicate that, rather than globally increasing proliferation or constitutively decreasing apoptosis in granule cell precursors of the EGL, Ptc heterozygosity instead perturbs differentiation in only a subset of EGL cells by promoting the abnormal proliferation in a significant fraction of Ptc⫹/⫺ mice. Since only a smaller minority of Ptc⫹/⫺ mice eventually form medulloblastomas, the greater incidence of these EGL rests indicates that they represent potential but not committed premalignant lesions. We also find that Ptc⫹/⫺ mouse medulloblastomas paradoxically express markers of postmitotic granule cells that belie their ongoing cell division. Murine medulloblastomas express a wide range of neuronal and granule cell-specific markers previously described in primary human tumors and appear similar to the desmoplastic subtype of human medulloblastoma. We addressed whether other signaling molecules critical for granule cell differentiation regulate tumor progression by testing the role of the neurotrophin-3 receptor, TrkC, a clinically significant prognostic indicator associated with apoptosis in human medulloblastoma (Segal et al., 1994; Kim et al., 1999; Grotzer et al., 2000). We crossbred the Ptc⫹/⫺ mouse with the heterozygous trkC⫹/⫺ strain containing a mutant trkC (Ntrk3) allele that lacks the tyrosine kinase-signaling domain to generate double heterozygous trkC⫹/⫺Ptc⫹/⫺ mice (Klein et al., 1994). Medulloblastomas from trkC⫹/⫺Ptc⫹/⫺ mice express lower levels of trkC and display decreased apoptosis compared with tumors from parental Ptc⫹/⫺ mice, supporting the in vivo role of TrkC in modulating tumor cell survival. We conclude that germline Ptc heterozygosity does not globally alter growth, but instead predisposes a subpopulation of granule cell precursors to form proliferative rests that represent potential targets of additional mutations and subsequent dysregulation of developmental gene expression that contribute to medulloblastoma tumorigenesis.

51

Materials and methods Animals Heterozygous patched (Ptc⫹/⫺) mice (kindly provided by Matthew P. Scott (Stanford University, Stanford, CA) and heterozygous trkC (trkC⫹/⫺) mice [B6, 129-Ntrk3tm1 kindly provided by Immaculada Silos-Santiago (Millennium Pharmaceuticals, Cambridge, MA)] were maintained on a mixed C57B1/6 ⫻ 129Sv background, including 573 Ptc⫹/⫺ and 163 trkC⫹/⫺ animals (Goodrich et al., 1997; Klein et al., 1994). Both heterozygous parental strains were crossed to generate compound heterozygous trkC⫹/ ⫺Ptc⫹/⫺ offspring (n ⫽ 82). All mice were housed in a barrier facility according to N.I.H., U.S.D.A., and institutional (Children’s Hospital, Boston, MA) guidelines and regulations. Animals were monitored daily until at least 100 weeks old for signs of illness and behavioral evidence of tumor, such as lethargy, decreased grooming, weight loss, or ataxia. Symptomatic mice were euthanized and their brain tissue processed as described below. All animals with confirmed genotypes were included in statistical analyses regardless of cause of death. Polymerase chain reaction Genotyping was performed on genomic DNA from tail biopsies obtained at weaning, based on previously described methods (Goodrich et al., 1997; Schimmang et al., 1995). For Ptc genotyping, the primer pair WT3 (5⬘-CTGCGGCAAGTTTTTGGTTG-3⬘) and WT4 (5⬘-AGGGCTTCTCGTTGGCTACAAG-3⬘) yield target sequences from the wild-type (wt) Ptc allele, and the primer pair NeoF2 (5⬘-GGATGATCTGGACGAAGAGC-3⬘) and NeoB2 (5⬘AGAAGGCGATAGAAGGCGAT-3⬘) yield target sequences from the neoR transgene. For trkC genotyping, the common 5⬘ primer, P093-1 (5⬘-CTGAAGTCACTGGCTAGAG-TCTGGG-3⬘), with the 3⬘ wt primer, P092-1 (5⬘GTCCCATCTTGCTTA-CCCTGAGG-3⬘), yields a 380-bp target from the wt trkC allele, or with the 3⬘ mutant primer P098-0 (5⬘-CCAGCCTCTGAGCCCAGAAAGC-3⬘) yields a 500-bp target from the pgk-1 promoter of the neoR cassette. Tissue processing and RNA preparation Brain and tumor tissue was obtained from euthanized mice for either RNA isolation or postfixation and subsequent sectioning, as previously described (Kim et al., 1999). For brain and tissue sections, mice were euthanized by intraperitoneal Avertin overdose before intracardiac perfusion with ice-cold paraformaldehyde [PFA, 4% (w/v) in PBS; Sigma, St. Louis, MO]. Tumor tissue was dissected from surrounding normal brain tissue, postfixed overnight for 12–18 h in 4% PFA/PBS at 4°C. For frozen sections, postfixed tissue was cryoprotected in 20% sucrose/PBS

52


(w/v) before flash freezing in liquid nitrogen, embedding in Tissue-Tek O.C.T. compound (Sakura Finetek, Torrance, CA), and storage at ⫺80°C. Glass slides were used for all tissue sections (Superfrost, Fisher Scientific, Pittsburgh, PA). Frozen 7.5-and 10-␮m cryostat sections (Leica Jung Frigicut 2800E) were prepared and stored at ⫺80°C for in situ hybridization, immunofluorescent staining, proliferation, and apoptosis assays. To confirm the presence of neoplastic cells, paraffinized 5-␮m sections were prepared from postfixed tissue by using a rotary microtome (Reichert-Jung 2050) and rehydrated through xylene/ethanol series for staining with cresyl violet or hematoxylin and eosin according to standard methods or for reticulin using Gomori’s method employed for human desmoplastic medulloblastoma (Luna, 1968). Immunophenotyping Antibodies used to characterize tumor sections include mouse antibodies that recognize NeuN (MAB377; Chemicon International, Temecula, CA), the cyclin-dependent kinase inhibitor, p27Kip1 (BD Transduction Laboratories, Lexington, KY), and rabbit antibody against the proliferation marker, Ki-67 (Vector Laboratories, Burlingame, CA). Rabbit polyclonal anti-Zic antibody was raised against a carboxy-terminal peptide (AVHHTAGHSALSSNFEWYV) (Borghesani et al., 2002). Fluorescently labeled secondary antibodies included Cy2-and Cy3-conjugates of goat-antimouse IgG and goat-anti-rabbit IgG (Jackson ImmunoResearch Laboratories, West Grove, PA). Negative tissue controls include adjacent and separate brain sections as well as staining controls with only secondary antibodies. For each genotype, Ptc⫹/⫺ and trkC⫹/⫺Ptc⫹/⫺, frozen tissue sections from at least eight mice with tumors and eight unaffected age-matched mice were analyzed separately for phenotypic marker expression by using standard immunocytochemical methods. Sections were rehydrated and blocked with 4% normal goat serum (NGS; GibCoBRL, Grand Island, NY) and 0.1% NP-40 (Sigma) in PBS for 60 min at 25°C. Normal mouse serum (NMS; 2%) was included for use with monoclonal antibodies. For additional permeabilization, sections were treated in 0.5% Triton-X (Sigma) in PBS for 5–10 min at 25°C. For nuclear antigens such as NeuN and Ki-67, antigen retrieval by microwaving three times for 5 min in 0.01 M sodium citrate, pH 6.0, improved sensitivity. Primary antibody incubations were performed at the appropriate dilution in blocking buffer for 12–18 h at 4°C, followed by PBS washes and secondary antibody incubations for 60 min at 25°C. Sections were mounted after nuclear counterstaining with bisbenzimide (Hoechst 33342, 10 ␮M in PBS; Sigma) (Darzynkiewicz et al., 1994) and examined with a Nikon E800 microscope. Images were analyzed by using a SPOT RT color camera and software (version 3.0, Diagnostic Instruments, Sterling Heights, MI) and Metamorph software (v.4.6, Universal Imaging, Downingtown, PA).

In situ hybridization and Northern blot analyses Northern blot analysis of total cellular RNA using a random oligonucleotide-primed (32P)-labeled trkC cDNA probe prepared from a plasmid containing 513 bp of the extracellular domain of human trkC cDNA (courtesy of David Shelton, Genentech) was performed as previously described (Kim et al., 1999). Target mRNA signals are measured by using a PhosphorImager (Molecular Dynamics, Sunnyvale, CA) and corrected for variations in sample loading by comparison to internal 28S rRNA expression, as previously described (Kim et al., 1999). For in situ hybridization, digoxigenin-labeled antisense and control sense riboprobes were prepared from linearized plasmids encoding Ptc, cyclinD1, Nmyc, NeuroD1, and Math1 (courtesy of Allen Bale, Yale University, New Haven, CT, and Qui Fu Ma, Dana-Farber Cancer Institute, Boston, MA) (Kenney and Rowitch, 2000). Frozen tissue sections were prepared from a minimum of five mice as previously described (Zhao et al., 2002). Proliferation analysis To quantitate proliferation of murine medulloblastoma cells and granule cell precursors by in vivo bromodeoxyuridine (BrdU) incorporation, animals were injected intraperitoneally with BrdU in PBS (50 ␮g per gm body weight; Roche Molecular Biochemicals), as previously described (Campana et al., 1988). After 60 min, mice were euthanized and their tissues processed as described above. For immunostaining of incorporated BrdU, frozen sections were permeabilized with 1% Triton-X in PBS at 25°C for 15 min, treated with 0.1 M NaBH4, followed by denaturation in 2 N hydrochloric acid (Fisher Scientific) for 45 min at 37°C, and neutralization in two changes of 0.2 M sodium borate (Sigma), then washed in PBS at 25°C. Primary incubation with anti-BrdU monoclonal antibody (clone BMC9318 at 6 ␮g/ml in 4% NGS/PBS; Roche Molecular Biochemicals) for 60 min at 25°C was followed by washes and secondary incubation with Cy2-or Cy3-conjugated secondary goatanti-mouse IgG antibodies (Jackson ImmunoResearch) for 30 min at 25°C. After nuclear counterstaining with bisbenzimide (Hoechst 33342; 10 ␮M in PBS), sections were examined as above. The proportion of BrdU-positive cells was determined by counting BrdU-positive nuclei and total (bisbenzimidestained) nuclei in randomly selected high-power fields (hpf). For each tumor specimen, 6 –12 hpf were counted for each of 6 –12 sections by at least 2 blinded observers. For normal granule cell analysis, representative cerebellar sections from midline rostral to caudal levels (e.g., lobules II, VI, and IX) as well as lateral regions (lobule VI) from 8 –12 mice were similarly processed and examined for each age time point and genotype. Negative immunostaining controls included BrdU-negative tumor sections and staining with secondary antibody only. Positive controls included immu-


53

nostaining with rabbit antibody against the Ki-67 proliferation marker as well as confirmatory staining of BrdUlabeled neonatal mouse cerebellar sections containing proliferating granule cell precursors. Apoptosis analysis Apoptotic cells were detected in frozen tissue sections by terminal deoxynucleotide transferase-mediated dUTP nick end-labeling (TUNEL) with digoxigenin-labeled nucleotides using ApopTag Red reagents (Intergen, Purchase, NY) including rhodamine (TRITC)-conjugated anti-digoxigenin antibody, according to manufacturer’s instructions. The proportion of TUNEL-positive cells was calculated by counting TUNEL-positive nuclei divided by the number of total (bisbenzimide-stained) nuclei in randomly selected high-powered fields. Since apoptotic nuclei were often fragmented, foci of positive staining clustered less than 5 ␮m apart were counted as a single positive nucleus. Eight to twelve hpf were counted for each of at least four sections from each of six specimens by at least three blinded observers.

Results The developmental effects of PTCH mutation have been well described in humans with NBCCS (Gorlin, 1995). As in NBCCS patients, mice with Ptc mutations can develop spontaneous cerebellar tumors (Goodrich et al., 1997; Hahn et al., 2000). Although homozygous null mutation of Ptc (Ptc⫺/⫺) results in embryonic lethality, description of the neurodevelopmental effects of Ptc heterozygosity on cerebellar development has been limited (Goodrich et al., 1997). Heterozygous Ptc⫹/⫺ mice display markedly decreased survival relative to their wt littermates primarily because of cerebellar tumors that closely resemble human medulloblastomas (P ⬍ 0.0001, log-rank test). The Ptc⫹/⫺ mouse tumors cause obvious symptoms of ataxia, dehydration, and diminished activity that subsequently progress to death within hours to days. Of Ptc⫹/⫺ mice observed between 5 and 54 weeks (wks) of age, approximately 9.8% developed medulloblastomas (n ⫽ 56) that presented with symptoms at a median age of 26.9 wks. Since human medulloblastomas may arise in the cerebellar granule cell lineage, we tested whether Ptc mutation contributes to the initial stages of tumorigenesis by globally deranging granule cell differentiation and increasing the proliferation and/or decreasing the apoptosis of granule cell precursors. Only a subset of Ptc⫹/⫺ mouse granule cell precursors display elevated proliferation To address whether granule cell development proceeds normally throughout the cerebellum in Ptc⫹/⫺ compared with wt mice, we evaluated animals at postnatal day 4 (P4), P8, P15, and 15 weeks of age. During the first postnatal

Fig. 1. Increased in vivo BrdU-uptake by external granule cell layer (EGL) rests/nodules and medulloblastomas (MB) from Ptc⫹/⫺ mice indicate elevated proliferation compared with granule cell precursors from control wild-type mice. Values expressed as mean of % BrdU positive nuclei in EGL as determined by counting BrdU positive (by immunostaining) and Hoechst 33342-stained cells in 6 –12 high-powered fields of 6 –12 cerebellar sections from each of 8 –12 mice. Error bars express standard error of the mean (S.E.M.).

week, normal wt granule cell precursors undergo a burst of proliferation causing marked expansion of the EGL. Direct measurements of the proliferation by in vivo BrdU incorporation reveal 13.5 ⫾ 1.3% (mean ⫾ S.E.M.) of normal granule cell precursors are in S-phase in the developing EGL of P4 wt mouse pups (Fig. 1). Brisk proliferation continues in P8 wt pups with 11.2 ⫾ 0.7% BrdU-positive granule cell precursors, when the EGL achieves its maximum dimension of 8 –12 cells thick. By the end of the second postnatal week, the EGL shrinks to only 1–2 cells thick with only 5.5 ⫾ 1.0% BrdU-positive nuclei in P15 wt mice (Fig. 1). Thereafter, the EGL diminishes so that by early adulthood (15 weeks of age) no BrdU uptake is detectable in wt mice Immunostaining for BrdU revealed similar overall labeling indices in the EGL of Ptc⫹/⫺ mice compared with their wt littermates (Fig. 1). In Ptc⫹/⫺ mice, 15.2 ⫾ 1.0% and 10.1 ⫾ 0.6% of EGL granule cell precursors are BrdUpositive at P4 and P8, respectively, like their wt littermates (Figs. 1 and 2A). By P15, the shrinking EGL of Ptc⫹/⫺ mice contain 7.5 ⫾ 0.8% BrdU-positive granule cell precursors (Figs. 1 and 2B), and by 15 weeks, no appreciable BrdU uptake is observed in the EGL. Similar trends in immunostaining for the proliferation marker, Ki-67, were also observed and correlate with BrdU incorporation data (Fig. 2C and D) (Onda et al., 1996). Although the overall proliferation rates of granule cell precursors are similar to their wt littermates, Ptc⫹/⫺ mice are susceptible to increased proliferation in a significant subset of granule cell precursors. In 6 of 10 asymptomatic P15 Ptc⫹/⫺ mice examined, we detected patches of intense superficial cerebellar X-gal staining, as previously reported (Goodrich et al., 1997). Microscopic examination reveal that these X-gal-staining regions are thickened superficial remnants of the EGL, typically measuring 10 –20 cells thick (Fig. 3A, C, and E) and significantly larger than adjacent normal EGL regions (Fig. 3B and D). In 2 P15 Ptc⫹/⫺ animals, the hyperplastic EGL remnants formed overt nod-

54


Fig. 2. Expression of the cerebellar granule cell-specific transcription factor, Zic, and the neuronal nuclear protein, NeuN, peaks in postmitotic granule cells. Double immunostaining in (A-B) with monoclonal ␣-BrdU primary with Cy2-conjugated goat ␣-mouse IgG secondary antibodies and primary rabbit polyclonal ␣-Zic with Cy3-conjugated goat ␣-rabbit IgG secondary antibodies. (A) Zic expression increases in normal postmitotic premigratory granule cells. Immunostaining of a normal P8 Ptc⫹/⫺ cerebellar section reveals BrdU uptake (Cy2; green), indicating proliferation of superficial granule cell precursors in the EGL. Zic expression (Cy3; red) increases as granule cell precursors withdraw from the cell cycle in the inner EGL and migrate through the molecular layer (ML) to the internal granule cell layer (IGL) (40⫻). (B) Proliferation of granule cell precursors decreases as EGL shrinks after second postnatal week. Immunostaining of a normal P15 Ptc⫹/⫺ cerebellar section shows decreased BrdU uptake (Cy2) in thinning EGL and continued Zic expression (Cy3) in IGL (40⫻). Double immunostaining in (C-D) with rabbit ␣-Ki-67 and Cy3-conjugated goat ␣-rabbit IgG antibodies with mouse ␣-NeuN and Cy2-conjugated goat ␣-mouse IgG antibodies. (C) Normal P8 Ptc⫹/⫺ proliferating granule cell precursors do not express NeuN. Immunostaining of a normal P8 Ptc⫹/⫺ cerebellar section reveals Ki-67-positive (Cy3) proliferating superficial granule cell precursors in the EGL, in contrast to NeuN expression (Cy2) that is limited to postmitotic granule cells of the deeper EGL and IGL (40 ⫻). (D) Proliferation decreases as EGL shrinks after second postnatal week. Normal P15 Ptc⫹/⫺ cerebellar section shows decreased Ki-67 (Cy3) immunostaining in EGL remnant and continued NeuN (Cy2) immunostaining in IGL (40⫻).

ules (Fig. 3F–G) that morphologically resemble small tumors, but otherwise share the characteristics of smaller EGL rests (Fig. 3C and E). The cerebellar sites of these Ptc⫹/⫺ EGL rests and nodules varied widely from discrete patches

or regions within a single sulcus or at the tip of a single folium to more diffuse involvement extending over multiple folia, ranging from midline to lateral locations. When analyzed for proliferation, these P15 Ptc⫹/⫺ EGL


55

56


Fig. 4. Neurodevelopmental gene expression pattern of Ptc⫹/⫺ mouse medulloblastomas is consistent with histogenesis in the granule cell lineage but dysregulated relative to proliferation. In situ hybridization of Ptc⫹/⫺ mouse medulloblastomas with digoxigenin-conjugated antisense riboprobes reveal diffuse expression of: (A) the cerebellar granule cell-specific bHLH transcription factor, Math1, normally limited to proliferating granule cell precursors (20 ⫻); (B) the neurogenic bHLH transcription factor, NeuroDl, at levels exceeding those in mature postmitotic granule cells of the IGL (20⫻); (C) the Shh-signaling target, cyclin D1, indicating ongoing proliferation of medulloblastoma cells (20 ⫻); (D) the oncogene, Nmyc, that is overexpressed in human desmoplastic medulloblastomas (20⫻).

Fig. 3. EGL rests/nodules in Ptc⫹/⫺ cerebella display a pattern of marker expression and proliferation similar to normal granule cell precursors. Double immunostaining in (A-B) with monoclonal ␣-BrdU primary with Cy2-conjugated goat ␣-mouse IgG secondary antibodies and primary rabbit polyclonal ␣-Zic with Cy3-conjugated goat ␣-rabbit IgG secondary antibodies. (A) EGL rests from P15 Ptc⫹/⫺ contain granule cell precursors that are either BrdU-positive and proliferating or Zic-positive and postmitotic. Double BrdU (Cy2) and Zic (Cy3) immunostaining reveals that proliferating cells are confined to the outer lamina of the EGL, and that Zic expression increases in granule cell precursors as they withdraw from the cell cycle in the inner lamina of the EGL and in the IGL (40 ⫻). (B) Normal P15 Ptc⫹/⫺ EGL shows fewer proliferating BrdU-positive granule cell precursors in the superficial EGL and increased Zic expression in postmitotic granule cells of the deeper EGL and mature granule cells of the IGL. Field shown adjacent to that in (A) shows decreased cell number in the normally thinner EGL and mutually exclusive BrdU (Cy2) and Zic (Cy3) immunostaining (40 ⫻). Double immunostaining in Panels C-D, and F with rabbit ␣-Ki-67 and Cy3-conjugated goat ␣-rabbit IgG antibodies with mouse ␣-NeuN and Cy2-conjugated goat ␣-mouse IgG antibodies. (C) P15 Ptc⫹/⫺ EGL rests contain granule cell precursors that express either the proliferation marker, Ki-67, or the postmitotic marker, NeuN. Double Ki-67 (Cy3) and NeuN (Cy2) immunostaining is parallel to BrdU uptake and Zic (Cy3) immunostaining, respectively (shown above). Expression of the postmitotic neuronal marker, NeuN, is appropriately limited to nondividing granule cells of deeper EGL and mature granule cells of the IGL (40⫻). (D) Normal postmitotic P15 Ptc⫹/⫺ granule cell precursors and mature granule cells express NeuN. Field shown adjacent to that in (C) shows decreased cell number in the normal thinner EGL and mutually exclusive pattern of Ki-67 (Cy3) and NeuN (Cy2) immunostaining (20⫻). (E) P15 Ptc⫹/⫺ EGL rests contain granule cell precursors that express either Ki-67 or p27Kip1. Double immunostaining for the cyclin-dependent kinase inhibitor, p27Kip1 (with mouse ␣-p27Kip1 and Cy2-conjugated goat ␣-mouse IgG antibodies) and proliferation marker, Ki-67 (with rabbit ␣-Ki-67 and Cy3-conjugated goat ␣-rabbit IgG antibodies), also shows mutually exclusive expression in the EGL and appropriate p27Kip1 expression in postmitotic granule cells of deeper EGL and mature granule cells of IGL (40⫻). Proliferating granule cell precursors in P15 Ptc⫹/⫺ EGL also form nodules that heterogeneously express NeuN or p27Kip1. (F) Double NeuN (Cy2) and Ki-67 (Cy3) immunostaining of EGL nodule resembles pattern noted in EGL rests, shown in (C) (20⫻). (G) Double p27Kip1 (Cy2) and Ki-67 (Cy3) immunostaining of EGL nodule also reveals mutually exclusive expression pattern noted in EGL rests, shown in (E) (20⫻). In situ hybridizations of thickened P15 Ptc⫹/⫺ EGL rests with digoxigenin-conjugated antisense riboprobes reveal increased expression as indicated by dark staining: (H) the Shh receptor, Ptc (20⫻); (I) the cerebellar granule cells-specific bHLH transcription factor, Math1, (20 ⫻); and (J) the oncogene, Nmyc (20⫻) expressed in proliferating granule cell precursors in the outer lamina of the EGL.


rests/nodules display a higher percentage of BrdU-positive nuclei (13.1 ⫾ 1.2%; P ⬍ 0.004, ANOVA) (Figs. 1 and 3A) compared with adjacent normal P15 Ptc⫹/⫺ EGL (7.5 ⫾ 0.8%; Fig. 3B) or P15 wt EGL (5.5 ⫾ 1.0%). Immunostaining for Ki-67 also labels EGL rests/nodules predominantly in the outermost superficial regions (Fig. 3C and E) similar to BrdU labeling and reminiscent of normal P4 to P8 EGL. When analyzed separately, these Ptc⫹/⫺ EGL rests/nodules display a clear growth advantage over neighboring granule cell precursors and approximate the BrdU-incorporation rate of granule cell precursors in the P4 to P8 EGL (Fig. 1). EGL rests/nodules present at earlier postnatal ages are not readily distinguishable from normal neighboring granule cell precursors, and are not numerous enough to inflate the granule cell proliferation indices of Ptc⫹/⫺ mice at P4 or P8 relative to their wt littermates In addition to increased and protracted proliferation, the focal accumulation of granule cell precursors in EGL rests/ nodules might involve decreased programmed cell death (Krueger et al., 1995). However, our in situ TUNEL assays of developing cerebellum do not reveal significant differences within the EGL of P4, P8, or P15 Ptc⫹/⫺ mice compared with their wt littermates, indicating that the Ptc⫹/⫺ EGL rests/nodules do not result from decreased apoptosis of granule cell precursors (data not shown). External granule cell layer rests/nodules express markers of normal granule cell precursors The EGL rests/nodules may arise in ptc⫹/⫺ mice because of disrupted cerebellar differentiation that results in delayed proliferation arrest. The X-gal staining of P15 Ptc⫹/⫺ EGL rests/nodules reflects their expression of the mutant lacZ-containing transgene, and parallels that of their remaining intact Ptc allele (Fig. 3H) (Goodrich et al., 1997). In addition to lacZ and Ptc, the EGL rests/nodules express the oncogene, Nmyc, which is expressed by granule cell precursors stimulated by Shh in vitro (Fig. 3J) (Kenney et al., 2003; Zhao et al., 2002). The Ptc⫹/⫺ mouse EGL rests/nodules also appropriately express transcription factors such as Math1 (Atoh-1), the mammalian atonal homolog expressed by proliferating granule cell precursors in the EGL and required for granule cell neurogenesis (Fig. 3I) (Ben-Arie et al., 1997), and NeuroD (NeuroDI), the neurogenic basic helix–loop– helix (bHLH) transcription factor, which becomes maximally expressed by postmitotic granule cells during terminal differentiation. The Ptc⫹/⫺ EGL rests/nodules also resemble normal proliferating granule cell precursors in their minimal expression of mature granule cell markers. The dividing cells in the EGL rests/nodules express the proliferation marker, Ki-67, in addition to incorporating BrdU, but neither Ki-67 nor BrdU labeling overlaps with the expression of the postmitotic granule cell markers, NeuN or Zic, respectively (Fig. 3A, C, and F). The Zic transcription factors are zincfinger-containing odd-paired homologs, including Zic

57

(Zic1), which is normally expressed at very low levels in granule cell precursors of the superficial proliferative layer of the EGL and achieves maximal expression in postmitotic granule cells of the IGL (Figs. 2A, B, and 3B) (Aruga et al., 1994). The EGL rests/nodules display lower levels of Zic and only in those nonproliferating cells that do not incorporate BrdU (Fig. 3A). Like Zic, the neuron-specific nuclear marker, NeuN, is also maximally expressed in mature postmitotic granule cells. NeuN upregulation normally coincides with cell cycle withdrawal and/or terminal neuronal differentiation (Mullen et al., 1992). In 15-day-old mouse cerebellum, only rare cells in the normal EGL express NeuN in contrast to mature postmitotic granule cells in the IGL (Fig. 2D) (Sarnat et al., 1998). EGL rests/nodules express NeuN only in Ki-67-negative or nonproliferating cells (Fig. 3C and F), comparable to mature IGL granule cells (Fig. 3C, D, and F). Similarly, the cyclin-dependent kinase inhibitor, p27Kip1, is upregulated in postmitotic granule cells and found only in Ki-67-negative postmitotic cells in EGL rests/ nodules (Fig. 3E and G). These data indicate that the granule cell precursors comprising the Ptc⫹/⫺ EGL rests/nodules observed in P15 Ptc⫹/⫺ mice appropriately express neuronal and immature granule cell markers relative to their persistent proliferation like those in normal P4 and P8 EGL, but fail to cease dividing. This developmental defect represents an intermediate stage of tumorigenesis relative to the Ptc⫹/⫺ mouse medulloblastoma phenotype. Dysregulation of proliferation relative to granule cell differentiation in murine medulloblastoma In parallel analyses, murine medulloblastomas exhibit evidence of aberrant differentiation compared with normal granule cell precursors and Ptc⫹/⫺ EGL rests/nodules. The Ptc⫹/⫺ mouse tumors appear similar to human medulloblastomas as densely cellular cerebellar mass lesions that compress neighboring normal-appearing cerebellar tissue without significant invasion across cortical layers. Murine medulloblastomas heterogeneously display regions of nodularity surrounded by thin acellular rims, areas of increased vascularity, or smaller focal areas of hemorrhage or necrosis. Some murine tumors involve superficial regions corresponding to EGL remnants, but are distinct from normal granule cells in their microscopic appearance. Immunostaining for in vivo BrdU incorporation by cerebellar tumors in Ptc⫹/⫺ mice reveals 17.5 ⫾ 5.0% BrdUpositive nuclei (mean ⫾ S.E.M.; Fig. 1). These proliferation data indicate a markedly elevated fraction of murine medulloblastoma cells labeled in S-phase. Values for murine tumors are higher than those obtained from normal granule cell precursors in the developing EGL of P15 wt and Ptc⫹/⫺ mouse pups. Disruption of developmental programs in Ptc⫹/⫺ mouse medulloblastomas is apparent in their misexpression of neurogenic transcription factors associated with postmitotic

58


Fig. 5. Proliferating Ptc⫹/⫺ medulloblastomas inappropriately coexpress markers of proliferation and of postmitotic granule cells. (A) Double immunostaining for NeuN (with Cy2-conjugated secondary antibody) and Ki-67 (with Cy3-conjugated secondary antibody) shows heterogeneous NeuN expression on a background of proliferating Ki-67-positive tumor nuclei (20⫻). (B) Double immunostaining reveals diffuse BrdU uptake (with Cy2 conjugated secondary antibody) and tumor expression of Zic (with Cy3-conjugated secondary antibody) (20⫻). (C) Higher power view of section shown in (A) reveals mutually exclusive expression of NeuN (Cy2) and Ki-67 (Cy3), indicating that NeuN is expressed in nonproliferating tumor cells (arrows indicate NeuN/Cy2-positive nuclei; 60⫻). (D) Higher power view of section shown in (B) reveals colabeling of BrdU and Zic, which is minimally expressed in normal proliferating granule cell precursors [arrows indicate BrdU (Cy2) and Zic (Cy3) double-stained nuclei; 60⫻].

granule cell maturation. Like proliferating granule cell precursors and P15 Ptc⫹/⫺ EGL rests/nodules, murine medulloblastomas diffusely express Math1 RNA (Fig. 4A). However, the diffuse expression of NeuroD, NeuN, and Zic by murine tumor cells contrasts with their ongoing growth. In situ hybridizations reveal that NeuroD is expressed throughout proliferating Ptc⫹/⫺ mouse tumors at levels exceeding postmitotic differentiated granule cells in the IGL (Fig. 4B). NeuN, a nuclear marker for normal postmitotic neurons, is

expressed by Ptc⫹/⫺ mouse medulloblastomas but not in the same tumor cells that express the proliferation marker, Ki-67 (Fig. 5A and C). Zic expression has been documented in human medulloblastomas, and immunostaining reveals elevated nuclear Zic expression throughout Ptc⫹/⫺ mouse tumors at higher levels than even mature postmitotic granule cells (Fig. 5B) (Yokota et al., 1996). Zic is detected predominantly in BrdU-negative cells, but double labeling also reveals a Zic-positive subpopulation of proliferating


59

Fig. 6. Ptc⫹/⫺ mouse medulloblastomas resemble human medulloblastomas. (A) X-gal-staining of Ptc⫹/⫺ mouse medulloblastoma indicates upregulated expression of lacZ- containing transgene that parallels expression of the intact Ptc allele (sagittal view). (B) Ptc⫹/⫺ mouse medulloblastoma reveal bands of reticulin by histochemical staining (20⫻). (C) Micrograph of hematoxylin-stained Ptch⫹/⫺ mouse medulloblastoma (20⫻).

60


tumors cells (Fig. 5D), indicating dysregulation of proliferation relative to Zic expression. Murine medulloblastoma resemble desmoplastic human medulloblastoma Like human medulloblastomas, Ptc⫹/⫺ mouse tumor cells are uniformly small with scant cytoplasm and high nuclear:cytoplasmic ratios (Fig. 6C) (Goodrich et al., 1997). Murine Ptc⫹/⫺ medulloblastomas are also similar to human tumors in their expression of homologous neuronal markers and granule cell-specific transcription factors that normally regulate granule cell precursor proliferation and differentiation, which supports their origin in the granule cell lineage: NeuroD, Math1, and Zic family members (Aruga et al., 1994; Ben-Arie et al., 1997; Miyata et al., 1999; Mullen et al., 1992; Rostomily et al., 1997). The in situ hybridizations of Ptc⫹/⫺ mouse medulloblastomas reveal the upregulation of the Shh/Ptc-signaling targets, cyclinDI (Fig. 4C) and Nmyc, which are upregulated in Shhstimulated granule cell precursors and overexpressed in human desmoplastic medulloblastoma (Fig. 4D) (DumanScheel et al., 2002; Kenney and Rowitch, 2000; Kenney et al., 2003; Pomeroy et al., 2002; Zhao et al., 2002). The classic histologic subtype of human medulloblastoma appears densely cellular throughout, whereas acellular reticulin-rich bands separating cellular nodules characterize the desmoplastic variant. Although several investigators have noted an association between desmoplastic histology and Ptc mutation, the histologic appearance of Ptc⫹/⫺ mouse cerebellar tumors is not overtly desmoplastic (Dong et al., 2000; Pietsch et al., 1997; Raffel et al., 1997). Nonetheless, areas of decreased cellularity with tumor nodularity are often observed in Ptc⫹/⫺ mouse medulloblastomas, with surrounding acellular reticulin-containing bands, reminiscent of desmoplastic human tumors (Fig. 6B). These parallels support the molecular similarity of Ptc⫹/⫺ mouse tumors in addition to their anatomic and histologic resemblance to human desmoplastic medulloblastomas. Reducing trkC expression decreases apoptosis but not the incidence of trkC⫹/⫺Ptc⫹/⫺ mouse medulloblastomas One of the neuronal markers associated with the desmoplastic histologic subtype is TRKC, the transmembrane receptor for neurotrophin-3 (NT-3) that is increasingly expressed during normal granule cell maturation, as well as in Ptc⫹/⫺ EGL rests/nodules and in primary human medulloblastomas (Pomeroy et al., 2002; Segal et al., 1992). In order to test how misexpressed neurodevelopmental genes modulate medulloblastoma growth, we examined tumors in Ptc⫹/⫺ mice bearing additional mutation of trkC. To generate Ptc⫹/⫺ mice with trkC mutation, we crossbred Ptc⫹/⫺ mice. Of the nine theoretically possible geno-

types of F1 offspring of trkC⫹/⫺ ⫻ Ptc⫹/⫺ crosses, only four genotypes were viable to reproductive age: wt, parental Ptc⫹/⫺ and trkC⫹/⫺ strains, and double heterozygous trkC⫹/⫺ Ptc⫹/⫺ mice. The absence of homozygous trkC null (trkC⫺/⫺) offspring to survive to weaning, after the previously reported survival limit of trkC⫺/⫺ pups, was not surprising (Klein et al., 1994; Tessarollo et al., 1997). Double heterozygous trkC⫹/⫺Ptc⫹/⫺ mice bred from Ptc⫹/⫺ and trkC⫹/⫺ parents show significantly diminished survival compared with wt littermates because of cerebellar tumors (P ⬍ 0.0001; log-rank test), like Ptc⫹/⫺ littermates (P ⫽ 0.91, log-rank test). Of double heterozygous trkC⫹/ ⫺Ptc⫹/⫺ mice between the ages of 5 and 54 weeks, 13.4% develop cerebellar tumors (n ⫽ 11) like parental heterozygous Ptc⫹/⫺ mice (9.8%; P ⫽ 0.25; log-rank test), with a similar median age of incidence at 30.6 weeks (P ⬎ 0.73; ANOVA) and mean of 27.0 ⫾ 14.1 wks (S.D.). These data indicate that mutation of trkC does not alter the formation and emergence of medulloblastoma on a Ptc⫹/⫺ background. Medulloblastomas in trkC⫹/⫺Ptc⫹/⫺ mice are anatomically and microscopically indistinguishable from those in Ptc⫹/⫺ mice. Immunophenotypically, trkC⫹/ ⫺Ptc⫹/⫺ mouse medulloblastomas are similar to Ptc⫹/⫺ mouse tumors. Northern and in situ hybridization analyses of cerebella and tumors from trkC⫹/ ⫺Ptc⫹/⫺ mice reveal a similar pattern of granule cell marker and neurodevelopmental gene expression (Table 1). In situ hybridizations reveal that trkC⫹/⫺Ptc⫹/⫺ mouse medulloblastomas express less trkC than Ptc⫹/⫺ mouse tumors (Fig. 7A and B) despite the variability among different tumors. As in parental trkC⫹/⫺ mice, Northern blot analysis of cerebellar RNA shows that compound heterozygous trkC⫹/⫺Ptc⫹/⫺ mice express less than two-thirds of the trkC levels found in wt or Ptc⫹/⫺ littermates (data not shown). Proliferation analysis of trkC⫹/⫺Ptc⫹/⫺ mouse medulloblastomas reveals similar in vivo BrdU incorporation in 19.6 ⫾ 5.4% of tumor cell nuclei, compared with Ptc⫹/⫺ mouse tumors (17.5 ⫾ 5%; P ⬎ 0.63, ANOVA; Fig. 1). However, examination of spontaneous apoptosis in murine medulloblastomas in vivo reveals differences between Ptc⫹/⫺ and trkC⫹/⫺Ptc⫹/⫺ tumor cells consistent with the proapoptotic role of TrkC. When examined microscopically for spontaneous apoptosis in randomly selected regions, the trkC⫹/⫺Ptc⫹/⫺ mouse medulloblastomas displayed a lower proportion of TUNEL-positive nuclei, 4.8 ⫾ 0.6% (mean ⫾ S.E.M.) compared with 7.6 ⫾ 1.8% in Ptc⫹/⫺ mouse tumors (Fig. 7C–E; P ⬍ 0.015, ANOVA). These differences indicate that diminished trkC expression in murine medulloblastomas is associated with decreased spontaneous apoptosis, analogous to low TRKC/NTRK3expressing human medulloblastomas (Kim et al., 1999; Segal et al., 1994).


61

Table 1 Mouse medulloblastoma resemble human medulloblastoma and granule cell precursors Marker

Granule cell lineage EGL

Synaptophysin NSE NeuroD1 NF-H (200 kDa) TrkC GLI Math1 NeuN Zic BrdU uptake Ki-67

IGL

outer

inner

⫹ ⫹ ⴚ ⫹ ⴚ ⫹ ⫹ ⴚ ⴞ ⴙ ⴙ

⫹ ⫹ ⴙ ⫹ ⴙ ⫹ ⫺ ⴙ ⴙ ⴚ ⴚ

⫹ ⫹ ⴙⴙ ⫹ ⴙⴙ ⫺ ⫺ ⴙⴙ ⴙⴙ ⴚ ⴚ

Ptc⫹/⫺ mouse medulloblastoma

trkC⫹/⫺Ptc⫹/⫺ mouse medulloblastoma

Human medulloblastoma

⫹ ⫾ ⴙ/ⴙⴙ ⫹ ⴙ/ⴙⴙ ⫹ ⫹ ⴞ ⴙⴙ/ⴙⴙⴙ ⴙⴙ ⴙⴙ

⫹ ⫾ ⴙ/ⴙⴙ ⫹ ⴞ ⫹ ⫹ ⴞ ⴙⴙ/ⴙⴙⴙ ⴙⴙ ⴙⴙ

⫹/⫹⫹ ⫹ ⫹ ⫹ ⴙ/ⴙⴙⴙ variable ND ND ⫹ ⫹⫹ ⫹⫹

Note. Bold entries indicate discordances between murine medulloblastomas and the granule cell lineage. ⫹, Uniformly positive expression; ⫾, minimal to heterogeneous expression; ⫺, uniformly negative expression.

Discussion Persistent proliferation marks a subset of Ptc⫹/⫺ granule cell precursors Because Ptc negatively regulates Shh signaling, investigators have proposed that heterozygous Ptc mutation contributes to medulloblastoma tumorigenesis by phenocopying the proliferative effects of Shh signaling (Wetmore et al., 2000; Zurawel et al., 2000). Our results, however, indicate that Ptc haploinsufficiency does not cause globally increased proliferation of granule cell precursors during early postnatal cerebellar development. Instead, we discovered a subset of granule cell precursors that continue to proliferate at rates comparable to normal development even as proliferation effectively ceases for the remainder of the EGL. Although Ptc haploinsufficiency could potentially decrease programmed cell death to promote the development of these rests of persistently proliferating progenitors, we found that early postnatal Ptc⫹/⫺ mice do not display decreased granule cell apoptosis compared with their wt littermates. We calculated the proportion of cells in S-phase in each high-powered microscopic field by dividing the number of BrdU-positive nuclei by the total number of cells comprising the entire thickness of the EGL. As a result, our counts from the entire EGL indicate peak proliferation shortly after birth and a steadily declining percentage of BrdU-positive cells. This contrasts with previous reports of peak BrdU or tritiated thymidine labeling at P8 that were based on counts of cells only in the mitotically active outermost EGL sublamina (Fujita et al., 1966). When similarly normalized, our data agree closely with established figures and show diminishing proliferation during the second postnatal week as the EGL markedly shrinks. We observed mitotically active EGL rests and nodules in a majority of the asymptomatic P15 Ptc⫹/⫺ mice examined. These abnormally proliferat-

ing granule cell precursors display a similar pattern of marker expression as normal immature granule cell precursors found in the EGL of P4 and P8 mice. Expression of differentiation markers by Ptc⫹/⫺ murine medulloblastoma supports histogenesis in the granule cell lineage but it is dysregulated In contrast to granule cell precursors in EGL rests/nodules, murine medulloblastomas reveal dysregulated expression of neuronal markers and neurogenic transcription factors relative to normal granule cell proliferation and differentiation. The proportion of Ptc⫹/⫺ mouse medulloblastoma cells that are actively proliferating as measured by BrdU incorporation is comparable to figures reported for primary human medulloblastoma (Gilbertson et al., 1997; Hoshino et al., 1985; Ito et al., 1992; Murovic et al., 1986; Onda et al., 1996; Yoshii et al., 1986). Their expression of Ki-67 and cyclinD2 and the absence of p27Kip1 are consistent with ongoing proliferation (Miyazawa et al., 2000; Zhao et al., 2002). Our results show that murine medulloblastomas express not only high levels of MathI, cyclinD2, and Ki-67 (normally restricted to proliferating granule cell precursors), but they also express NeuN, NeuroD, and Zic at levels normally associated with postmitotic and mature granule cells. The aberrant expression of these mature granule cell markers in proliferating Ptc⫹/⫺ mouse medulloblastomas demonstrates that the tumor cells have taken another step in the process of oncogenesis, clearly diverging from the normal developmental pattern displayed by the granule cell rests. Moreover, the expression of postmitotic markers in proliferating cells is indicative of the emerging pathological state as granule cell progenitors evolve into the cells of a growing tumor. Normal immature granule cells in the inner EGL withdraw from the cell cycle and cease to express Math1 and cyclinD2. Concurrently, the expression of trkC/Ntrk3, neu-

62


Fig. 7. trkC⫹/⫺Ptc⫹/⫺ mouse medulloblastomas express less trkC and display less apoptosis compared with Ptc⫹/⫺ medulloblastomas. (A, D, and show Ptc⫹/⫺ medulloblastomas, and B, E, and G show trkC⫹/⫺Ptc⫹/⫺ medulloblastomas.) trkC⫹/⫺Ptc⫹/⫺ mouse medulloblastomas express less trkC mRNA compared with Ptc⫹/⫺ medulloblastomas. (A) In situ hybridization with digoxigenin-conjugated antisense riboprobe reveals diffuse expression of trkC in a Ptc⫹/⫺ mouse medulloblastoma and adjacent normal cerebellum (20⫻). (B) In situ hybridization with digoxigenin-conjugated antisense riboprobe reveals diffusely decreased expression of trkC in a trkC⫹/⫺Ptc⫹/⫺ medulloblastoma (20⫻). trkC mutation decreases apoptosis in trkC⫹/⫺Ptc⫹/⫺ medulloblas-


rofilament (NF) subunits, p27Kip1, Nscl-1, NeuroD, and Zic increases as postmitotic granule cells mature (Alder et al., 1996; Katoh-Semba et al., 2000; Millen et al., 1995; Miyata et al., 1999; Segal et al., 1992). NeuN is a nuclear protein widely expressed in mature postmitotic neurons including granule cells (Mullen et al., 1992). The heterogeneous expression of NeuN suggests that this subpopulation of NeuNpositive medulloblastoma cells is not actively proliferating as indicated by the lack of Ki-67 coexpression and that they may instead represent asymmetric mitoses or differentiation among tumor cells. In contrast, NeuroD is expressed in every cell throughout murine medulloblastomas that continue to proliferate. Like NeuroD, Zic is found in primary human and Ptc⫹/⫺ mouse medulloblastomas (Rostomily et al., 1997; Yokota et al., 1996). Normally, Zic expression increases dramatically in granule cells as they undergo proliferation arrest and terminal differentiation (Aruga et al., 1994). The overall pattern of marker expression in Ptc⫹/⫺ mouse medulloblastomas resembles but is distinct from the persistently proliferating granule cell precursors found in EGL rests/nodules. Our results suggest that the Ptc⫹/⫺ EGL rests represent granule cell precursors at an early stage of departure from normal neurogenesis. A developmental defect occurring in the setting of germline Ptc mutation arises only in a subset of granule cell precursors of the EGL, rendering them susceptible to persistent proliferation. Although more than half of P15 Ptc⫹/⫺ pups display EGL rests or nodules, only 10 –15% of adult Ptc⫹/⫺ mice eventually develop medulloblastoma, indicating that the persistently proliferating granule cell precursors are not committed to becoming overtly malignant. Since most of these proliferating Ptc⫹/⫺ EGL rests do not progress to become medulloblastomas, they instead represent an abnormally persistent pool of vulnerable granule cell precursors capable of regression via delayed differentiation or cell death. We hypothesize that subsequent mutation(s) or other genetic dysregulation in some of these susceptible precursors may confer a further proliferative advantage, which in turn contributes to frank tumor formation. While our evidence indicates a multistep process in the ontogeny of murine Ptc⫹/⫺ medulloblastomas, insights about the molecular mechanisms that promote these steps are just beginning to emerge. Wetmore et al. (2001) demonstrated that TP53 mutation accelerates medulloblastoma formation in a TP53⫺/⫺Ptc⫹/⫺ compound mutant mouse, leading to complete penetrance of medulloblastoma formation. While loss of p53 function may directly promote deregulation of cell cycle control, it also is possible that p53 loss promotes the accumulation of other genetic mutations

63

that act in concert with Ptc haploinsufficiency to promote the malignant phenotype. Since loss of Ptc heterozygosity has not been found in murine Ptc⫹/⫺ medulloblastomas, mutations of other genes may be the “second hit” that promotes tumorigenesis. Recent studies have also identified epigenetic mechanisms that alter gene expression to promote tumorigenesis. Igf2 is normally imprinted so that only the paternal allele is expressed, and has emerged as a candidate gene for epigenetic regulation in cancer (Reik et al., 2000). Hahn et al., (2000) found that Igf2 is required for Ptc⫹/⫺ mouse tumorigenesis since compound Ptc⫹/⫺Igf2⫹/⫺ mutant mice fail to develop either medulloblastomas or rhabdomyosarcomas. However, Hahn et al. did not find evidence for epigenetic mechanisms, such as loss of imprinting or uniparental disomy, nor did they detect Igf2 amplification or polyploidy in Ptc⫹/⫺Igf2⫹/⫺ mouse rhabdomyosarcomas. They did not report similar studies of Ptc⫹/⫺Igf2⫹/⫺ mouse medulloblastoma. Genetic modifiers, often strain-dependent, are also influential in the phenotype of mutant mouse models of development, and strain-specific effects were initially described in the Ptc⫹/⫺ model (Goodrich et al., 1997; Hide et al., 2002). Our experiments involved mutant mice with a mixed background of C57B1/6 and 129Sv strains, and the overall incidence and phenotype of tumors observed were in close agreement with published results (Goodrich et al., 1997; Wetmore et al., 2000). To summarize, our results indicate that murine medulloblastoma tumorigenesis is a multi-stage process in Ptc⫹/⫺ mice. Our findings are consistent with general models of carcinogenesis, in which transformed subclones emerge from a susceptible population of cells through progressive alterations in gene expression (Corcoran and Scott, 2001; Hahn and Weinberg, 2002; Kinzler and Vogelstein, 1996; Wu and Pandolfi, 2001). Further investigation will be needed to determine whether these steps involve genetic or epigenetic deregulation of gene expression. Mouse medulloblastoma phenotype resembles human medulloblastoma The inappropriate coexpression of postmitotic granule cell markers by dividing murine medulloblastomas indicates disruption of the coordinated regulation of differentiation with proliferation, and belies the simple model of arrested development in a mitotically active granule cell precursor that has been suggested for human medulloblastoma. Like primary human medulloblastomas, Ptc⫹/⫺ mouse tumors appear morphologically undifferentiated, but can also ex-

tomas compared with Ptc⫹/⫺ medulloblastomas. (C) Values expressed as mean of % TUNEL-positive nuclei determined from counting TUNEL/TRITCpositive nuclei and Hoechst 33342-positive cells in 8 –12 high-powered fields of at least 4 tumor sections from each of 6 mice. Error bars express S.E.M. (D) TRITC-conjugated TUNEL assay of Ptc⫹/⫺ medulloblastoma (arrows indicate apoptotic nuclei, 40⫻). (E) TRITC-conjugated TUNEL assays of trkC⫹/⫺Ptc⫹/⫺ medulloblastoma (arrows indicate apoptotic nuclei, 40⫻). (F) Hoechst 33342 nuclear counterstaining of same high-powered field as in (D) (40⫻). (G) Hoechst 33342 nuclear counterstaining of same field as in (E) (40⫻).

64


press markers of divergent phenotypes. Murine medulloblastomas also reveal regions of extracellular reticulin staining in patterns resembling the histologic hallmark of desmoplastic medulloblastomas (McLendon et al., 1998; Rorke, 1983). Medulloblastomas in Ptc⫹/⫺ mice are representative of the desmoplastic subset of human tumors, and provide an experimental system for exploring issues of diagnostic and therapeutic significance. We have reported the association of clinical outcome with TRKC/NTRK3 expression (Segal et al., 1994). The greater the level of TRKC/NTRK3, the better the observed overall and progression-free survival. We also previously reported that TRKC/NTRK3 expression in primary human medulloblastomas correlates with apoptosis in situ, and that NT-3 activation of TrkC induces apoptosis of tumor cells in vitro and in intracerebral xenografts in vivo (Kim et al., 1999). Although studies of apoptosis in human medulloblastoma have not revealed an association with prognosis, the regulation of tumor growth by TrkC-mediated apoptosis may underlie the prognostic value of TRKC/ NTRK3 expression or contribute to treatment-responsiveness. In order to demonstrate the role of TrKC-mediated apoptosis, we generated tumor-forming double heterozygous trkC⫹/⫺Ptc⫹/⫺ mice, by crossbreeding Ptc⫹/⫺ animals with trkC⫹/⫺ heterozygous mice. While the Ptc⫹/⫺ mouse medulloblastomas express trkC at variable levels comparable to normal mouse brain, the trkC⫹/⫺Ptc⫹/⫺ offspring resemble parental trkC⫹/⫺ mice and express decreased levels of trkC in their brains and cerebellar tumors (Klein et al., 1994; Wetmore et al., 2000). The trkC⫹/⫺Ptc⫹/⫺ mouse medulloblastomas are relatively low trkC-expressing tumors, morphologically similar to Ptc⫹/⫺ mouse tumors. The incidence and age distribution of trkC⫹/⫺Ptc⫹/⫺ medulloblastomas are also similar to Ptc⫹/⫺ tumors, indicating that trkC mutation does not alter early tumorigenesis. In addition to comparable proliferation rates, trkC⫹/⫺Ptc⫹/⫺ mouse medulloblastomas also express similar neuronal and granule cell markers. However, decreasing trkC expression in trkC⫹/ ⫺Ptc⫹/⫺ mouse medulloblastomas, relative to Ptc⫹/⫺ mouse tumors, diminishes their spontaneous apoptosis in vivo analogous to low trkC-expressing human tumors. The similarities in medulloblastoma incidence and phenotype among trkC⫹/⫺Ptc⫹/⫺ and Ptc⫹/⫺ mice are consistent with the limitation of medulloblastoma growth by TrkCmediated apoptosis after oncogenic events have transpired rather than contributing to early tumorigenesis. Our characterization of murine medulloblastoma phenotypes reveals similarities to primary human medulloblastoma, supporting both their origin in and significant divergence from the granule cell lineage. We demonstrate that Ptc heterozygosity does not globally disrupt postnatal granule cell differentiation. Our findings instead support a model for medulloblastoma tumorigenesis as a multistage process beginning with Ptc mutation that predisposes a subset of granule cell precursors to persistent proliferation, but not

decreased apoptosis, as seen in the EGL rests/nodules as an intermediate stage of tumorigenesis. Subsequent transforming events or mutations, may act as additional “second hits” en route to medulloblastoma. The transformation from the intermediate stage of a mitotically active but nonmalignant granule cell rest to a frank medulloblastoma does not involve TrkC-dependent apoptosis as shown by the similar incidence rate and age distribution of medulloblastomas in trkC⫹/⫺Ptc⫹/⫺ compound heterozygotes. Instead, the medulloblastomas in trkC⫹/⫺Ptc⫹/⫺ mice display decreased tumor cell apoptosis further supporting the role of TrkC signaling, once the tumor is already established, in regulating later tumor growth. These results further clarify the relationship between normal granule cell development and medulloblastoma tumorigenesis, and establish the validity of the Ptc⫹/⫺ mouse as an experimental model system for human medulloblastoma.

Acknowledgments We thank Pieter Dikkes and Dong-In Yuk for their technical assistance, and Charles D. Stiles and Ching C. Lau for helpful discussions. This work is supported by NIH grants [NS043517-01 (J.Y.H.K), NS35701 (S.L.P.), NS37757 (R.A.S.), and the Children’s Hospital Mental Retardation Research Center (HD18655)]; the Emily Dorfman Foundation and the American Brain Tumor Association (J.Y.H.K); Kyle Mullarkey Medulloblastoma Research Fund (S.L.P.); the David S. Mahoney Center for Neuro-oncology (J.Y.H.K), the Claudia Adams Barr Program (R.A.S.) of the Dana-Farber Cancer Institute, and a generous gift from Dr. and Mrs. Sidney Feldman.

References Alder, J., Cho, N.K., Hatten, M.E., 1996. Embryonic precursor cells from the rhombic lip are specified to a cerebellar granule neuron identity. Neuron 17, 389 –399. Altman, J., Bayer, S.A., 1997. The generation, movements, and settling of cerebellar granule cells and the formation of parallel fibers, in: Development of the Cerebellar System: In Relation to Its Evolution Structure and Functions. CRC Press, New York, pp. 334 –361. Aruga, J., Yokota, N., Hashimoto, M., Furuichi, T., Fukuda, M., Mikoshiba, K., 1994. A novel zinc finger protein, Zic, is involved in neurogenesis, especially in the cell lineage of cerebellar granule cells. J. Neurochem. 63, 1880 –1890. Ben-Arie, N., Bellen, H.J., Armstrong, D.L., McCall, A.E., Gordadze, P.R., Guo, Q., Matzuk, M.M., Zoghbi, H.Y., 1997. Math1 is essential for genesis of cerebellar granule neurons. Nature 390, 169 –172. Borghesani, P.R., Peyrin, J.M., Klein, R., Rubin, J., Carter, A.R., Schwartz, P.M., Luster, A., Corfas, G., Segal, R.A., 2002. BDNF stimulates migration of cerebellar granule cells. Development 129, 1435–1442. Campana, D., Coustan-Smith, E., Janossy, G., 1988. Double and triple staining methods for studying the proliferative activity of human B and T lymphoid cells. J. Immunol. Methods 107, 79 – 88. Chintagumpala, M., Berg, S., Blaney, S.M., 2001. Treatment controversies in medulloblastoma. Curr. Opin. Oncol. 13, 154 –159.

J.Y.H. Kim et al. / Developmental Biology 263 (2003) 50 – 66 Corcoran, R.B., Scott, M.P., 2001. A mouse model for medulloblastoma and basal cell nevus syndrome. J. Neurooncol. 53, 307–318. Dahmane, N., Sa´ nchez, P., Gitton, Y., Palma, V., Sun, T., Beyna, M., Weiner, H., Ruiz i Atalba, A., 2001. The Sonic Hedgehog-Gli pathway regulates dorsal brain growth and tumorigenesis. Development 128, 5201–5212. Darzynkiewicz, Z., Li, X., Gong, J., 1994. Assays of cell viability. Discrimination of cells dying in apoptosis; in: Darzynkiewicz, Z., Crissman, H.A., Robinson, J.R. (Eds.), Cell Biology: Flow Cytometry, 2nded. Academic Press, Burlington, MA. Dong, J., Gailani, M.R., Pomeroy, S.L., Reardon, D., Bale, A.E., 2000. Identification of PATCHED mutations in medulloblastomas by direct sequencing. Hum. Mutat. 16, 89 –90. Duman Scheel, M., Weng, L., Xin, S., Du, W., 2002. Hedgehog regulates cell growth and proliferation by inducing Cyclin D and Cyclin E. Nature 417, 299 –304. Fujita, S., Shimada, M., Nakamura, T., 1966. H3-thymidine autoradiographic studies on the cell proliferation and differentiation in the external and the internal granular layers of the mouse cerebellum. J. Comp. Neurol. 128, 191–208. Giangaspero, F., Bigner, S.H., Giordana, M.T., Kleihues, P., Trojanowski, J.Q., 2000. Medulloblastoma, in: Kleihues, P., Cavence, W.K. (Eds.), Pathology and Genetics: Tumours of the Nervous System. World Health Organization Classification of Tumours. International Agency for Research of Cancer, Lyon, France, pp. 96 –103. Gilbertson, R.J., Jaros, E., Perry, R.H., Kelly, P.J., Lunec, J., Pearson, A.D.J., 1997. Mitotic percentage index: a new prognostic factor for childhood medulloblastoma. Eur. J. Cancer 33, 609 – 615. Goodrich, L.V., Milenkovic, L., Higgins, K.M., Scott, M.P., 1997. Altered neural cell fates and medulloblastoma in mouse patched mutants. Science 277, 1109 –1113. Gorlin, R.J., 1995. Nevoid basal cell carcinoma syndrome. Dermatol. Clin. 13, 113–125. Grotzer, M.A., Janss, A.J., Fung, K.M., Biegel, J.A., Sutton, L.N., Rorke, L.B., Zhao, H., Cnaan, A., Phillips, P.C., Lee, V.M., Trojanowski, J.Q., 2000. TrkC expression predicts good clinical outcome in primitive neuroectodermal brain tumors. J. Clin. Oncol. 18, 1027–1035. Grotzer, M.A., Geoerger, B., Janss, A.J., Zhao, H., Rorke, L.B., Phillips, P.C., 2001. Prognostic significance of Ki-67 (MIB-1) proliferation index in childhood primitive neuroectodermal tumors of the central nervous system. Med. Pediatr. Oncol. 36, 268 –273. Hahn, H., Wicking, C., Zaphiropoulus, P.G., Gailani, M.R., Shanley, S., Chidambaram, A., Vorechovsky, I., Holmberg, E., Unden, A.B., Gillies, S., Negus, K., Smyth, I., Pressman, C., Leffell, D.J., Gerrard, B., Goldstein, A.M., Dean, M., Toftgard, R., Chenevix-Trench, G., Wainwright, B., Bale, A.E., 1996. Mutations of the human homolog of Drosophila patched in the nevoid basal cell carcinoma syndrome. Cell 85, 841– 851. Hahn, H., Wojnowski, L., Specht, K., Kappler, R., Calzada-Wack, J., Potter, D., Zimmer, A., Muller, U., Samson, E., Quintanilla-Martinez, L., Zimmer, A., 2000. Patched target Igf2 is indispensable for the formation of medulloblastoma and rhabdomyosarcoma. J. Biol. Chem. 275, 28341–28334. Hahn, W.C., Weinberg, R.A., 2002. Rules for making human tumor cells. N. Engl. J. Med. 347, 1593–1603. Hide, T., Hatakeyama, J., Kimura-Yoshida, C., Tian, E., Takeda, N., Ushio, Y., Shiroishi, T., Aizawa, S., Matsuo, I., 2002. Genetic modifiers of otoceophalic phenotypes in Otx2 heterozygous mutant mice. Development 129, 4347– 4357. Hoshino, T., Kobayashi, S., Townsend, J.J., Wilson, C.B., 1985. A cell kinetic study on medulloblastomas. Cancer 55, 1711–1713. Ito, S., Hoshino, T., Prados, M.D., Edwards, M.S., 1992. Cell kinetics of medulloblastomas. Cancer 70, 671– 678. Johnson, R.L., Rothman, A.L., Xie, J., Goodrich, L.V., Gare, J.W., Bonifas, J.M., Quinn, A.G., Myers, R.M., Cox, D.R., Epstein, E.H., Scott, M.P., 1996. Human homolog of patched, a candidate gene for the basal cell nevus syndrome. Science 272, 1668 –1671.

65

Kadin, M.E., Rubenstein, L.J., Nelson, J.S., 1970. Neonatal cerebellar medulloblastoma originating from the fetal external granular layer. J. Neuropathol. Exp. Neurol. 29, 583– 600. Katoh-Semba, R., Takeuchi, I.K., Semba, R., Kato, K., 2000. Neurotrophin-3 controls proliferation of granular precursors as well as survival of mature granule neurons in the developing rat cerebellum. J. Neurochem. 74, 1923–1930. Kenney, A.M., Rowitch, D.H., 2000. Sonic hedgehog promotes G(1) cyclin expression and sustained cell cycle progression in mammalian neuronal precursors. Mol. Cell. Biol. 20, 9055–9067. Kenney, A.M., Cole, M.D., Rowitch, D.H., 2003. Nmyc upregulation by sonic hedgehog signaling promotes proliferation in developing cerebellar granule neuron precursors. Development 130, 15–28. Kim, J.Y.H., Sutton, M.E., Lu, D.J., Cho, T.A., Goumnerova, L.C., Goritchenko, L., Kaufman, J.R., Lam, K.K., Billett, A.L., Tarbell, N.J., Wu, J., Allen, J.C., Stiles, C.D., Segal, R.A., Pomeroy, S.L., 1999. Activation of neurotrophin-3 receptor TrkC induces apoptosis in medulloblastomas. Cancer Res. 59, 711–719. Kinzler, K.W., Vogelstein, B., 1996. Lessons from hereditary colorectal cancer. Cell 87, 159 –170. Klein, R., Silos-Santiago, I., Smeyne, R.J., Lira, S.A., Brambilla, R., Bryant, S., Zhang, L., Snider, W.D., Barbacid, M., 1994. Disruption of the neurotrophin-3 receptor gene trkC eliminates Ia muscle afferents and results in abnormal movements. Nature 368, 249 –251. Kozmik, Z., Sure, U., Ru¨ edi, D., Busslinger, M., Aguzzi, A., 1995. Deregulated expression of PAX5 in medulloblastoma. Proc. Natl. Acad. Sci. USA 92, 5709 –5713. Krueger, B.K., Burne, J.F., Raff, M.C., 1995. Evidence for large-scale astrocyte death in the developing cerebellum. J. Neurosci. 15, 3366 – 3374. Luna, L.G., 1968. Manual of Histologic Staining Methods of the Armed Forces Institute of Pathology, 3rd ed. McGraw-Hill Book Company, new york. McLendon, R.E., Enterline, D.S., Tien, R.D., Thorstad, W.L., Bruner, J.M., 1998. Tumors of central neuroepithelial origin, in: Russell, D.S., Rubenstein, L.J. (Eds.), Pathology of Tumours of the Nervous System, 6th ed. Williams and Wilkins, Baltimore, pp. 456 – 479. Millen, K.J., Hui, C.-c., Joyner, A.L., 1995. A role for En-2 and other murine homologues of Drosophila segment polarity genes in regulating positional information in the developing cerebellum. Development 121, 3935–3945. Miyazawa, K., Himi, T., Garcia, V., Yamagishi, H., Sato, S., Ishizaki, Y., 2000. A role for p27/Kip1 in the control of cerebellar granule cell precursor proliferation. J. Neurosci. 20, 5756 –5763. Miyata, T., Maeda, T., Lee, J.E., 1999. NeuroD is required for differentiation of the granule cells in the cerebellum and hippocampus. Genes Dev. 13, 1647–1652. Mullen, R.J., Buck, C.R., Smith, A.M., 1992. NeuN, a neuronal specific nuclear protein in vertebrates. Development 116, 201–211. Murovic, J.A., Nagashima, T., Hoshino, T., Edwards, M.S., Davis, R.L., 1986. Pediatric central nervous system tumors: a cell kinetic study with bromodeoxyuridine. Neurosurgery 19, 900 –904. Onda, K., Davis, R.L., Edwards, M.S., 1996. Comparison of bromodeoxyuridine uptake and MIB1 immunoreactivity in medulloblastomas determined with single and double immunohistochemical staining methods. J Neuro-Oncol. 29, 129 –136. Pietsch, T., Waha, A., Koch, A., Kraus, J., Albrecht, S., Tonn, J., Sorensen, N., Berthold, F., Henk, B., Schmandt, N., Wolf, H.K., von Deimling, A., Wainwright, B., Chenevix-Trench, G., Wiestler, O.D., Wicking, C., 1997. Medulloblastomas of the desmoplastic variant carry mutations of the human homologue of Drosophila patched. Cancer Res. 57, 2085– 2088. Pomeroy, S.L., Tamayo, P., Gaasenbeek, M., Sturla, L.M., Angelo, M., McLaughlin, M.E., Kim, J.Y.H., Goumnerova, L.C., Black, P.M., Lau, C.C., Allen, J.C., Zagzag, D., Olson, J., Curran, T., Wetmore, C., Biegel, J.A., Poggio, T., Mukherjee, S., Califano, A., Stolovistzky, G., Louis, D.N., Mesirov, J.P., Lander, E.S., Golub, T.R., 2002. Gene

66


expression-based classification and outcome prediction of central nervous system embryonal tumors. Nature 415, 436 – 442. Raffel, C., Jenkins, R.B., Frederick, L., Hebrink, D., Alderete, B., Fults, D.W., James, C.D., 1997. Sporadic medulloblastomas contain PTCH mutations. Cancer Res. 57, 842– 845. Rakic, P., 1973. Kinetics of proliferation and latency between final cell division and onset of differentiation of cerebellar stellate and basket neurons. J. Comp. Neurol. 147, 523–546. Reik, W., Constancia, M., Dean, W., Davies, K., Bowden, L., Murrell, A., Feil, R., Walter, J., Kelsey, G., 2000. Igf2 imprinting in development and disease. Int. J. Dey. Biol. 44, 145–150. Rorke, L.B., 1983. The cerebellar medulloblastoma and its relationship to primitive neuroectodermal tumors. J. Neuropathol. Exp. Neurol. 42, 1–15. Rostomily, R.C., Bermingham-McDonogh, O., Berger, M.S., Tapscott, S.J., Reh, T.A., Olson, J.M., 1997. Expression of neurogenic basic helix-loop-helix genes in primitive neuroectodermal tumors. Cancer Res. 57, 3526 –3531. Sarnat, H.B., Nochlin, D., Born, D.E., 1998. Neuronal nuclear antigen (NeuN): a marker of neuronal maturation in early human fetal nervous system. Brain Dev. 20, 88 –94. Schimmang, R., Minichiello, L., Vazquez, E., San Jose, I., Giraldez, F., Klein, R., Represa, J., 1995. Developing inner ear sensory neurons require TrkB and TrkC receptors for innervation of their peripheral targets. Development 121, 3381–3391. Segal, R.A., Goumnerova, L.C., Kwon, Y.K., Stiles, C.D., Pomeroy, S.L., 1994. Expression of the neurotrophin receptor TrkC is linked to a favorable outcome in medulloblastoma. Proc. Natl. Acad. Sci. USA 91, 12867–12871. Segal, R.A., Takahashi, H., McKay, R.D., 1992. Changes in neurotrophin responsiveness during the development of cerebellar granule neurons. Neuron 9, 1041–1052.

Tessarollo, L., Tsoulfas, P., Donovan, M.J., Palko, M.E., Blair-Flynn, J., Hempstead, B.L., Parada, L.F., 1997. Targeted deletion of all isoforms of the trkC gene suggests the use of alternate receptors by its ligand neurotrophin-3 in neuronal development and implicates trkC in normal cardiogenesis. Proc. Natl. Acad. Sci. USA 94, 14776 –14781. Wang, V.Y., Zoghbi, H.Y., 2001. Genetic regulation of cerebellar development. Nat. Rev. Neurosci. 2, 484 – 491. Wechsler-Reya, R.J., Scott, M.P., 1999. Control of neuronal precursor proliferation in the cerebellum by Sonic Hedgehog. Neuron 22, 103– 114. Wetmore, C., Eberhart, D.E., Curran, T., 2000. The normal patched allele is expressed in medulloblastomas from mice with heterozygous germline mutation of patched. Cancer Res. 60, 2239 –2246. Wetmore, C., Eberhart, D.E., Curran, T., 2001. Loss of p53 but not ARF accelerates medulloblastoma in mice heterozygous for patched. Cancer Res. 61, 513–516. Wu, X., Pandolfi, P.P., 2001. Mouse models for multistep tumorigenesis. Trends Cell Biol. 11, S2–S9. Yokota, N., Aruga, J., Takai, S., Yamada, K., Hamazaki, M., Iwase, T., Sugimura, H., Mikoshiba, K., 1996. Predominant expression of human Zic in cerebellar granule cell lineage and medulloblastoma. Cancer Res. 56, 377–383. Yoshii, T., Maki, Y., Tsuboi, K., Tomono, Y., Nakagawa, K., Hoshino, T., 1986. Estimation of growth fraction with bromodeoxyuridine in human central nervous system tumors. J. Neurosurg. 65, 659 – 663. Zhao, Q., Kho, A., Kenney, A.M., Yuk, D.I., Kohane, I., Rowitch, D.H., 2002. Identification of genes expressed with temporal-spatial restriction to developing cerebellar neuron precursors by a functional genomic approach. Proc. Natl. Acad. Sci. USA 99, 5704 –5709. Zurawel, R.H., Allen, C., Wechsler-Reya, R., Scott, M.P., Raffel, C., 2000. Evidence that haploinsufficiency of Ptch leads to medulloblastoma in mice. Genes Chrom. Cancer 28, 77– 81.