Toward the goal of generating a mouse medulloblastoma model with increased tumor incidence, we developed a homozygous version of our ND2:SmoA1 model. Medulloblastomas form in 94% of homozygous Smo/Smo mice by 2 months of age. Tumor formation is, thus, predictable by age, before the symptomatic appearance of larger lesions. This high incidence and early onset of tumors is ideal for preclinical studies because mice can be enrolled before symptom onset and with a greater latency period before late-stage disease. Smo/Smo tumors also display leptomeningeal dissemination of neoplastic cells to the brain and spine, which occurs in many human cases. Despite an extended proliferation of granule neuron precursors (GNP) in the postnatal external granular layer (EGL), the internal granular layer formed normally in Smo/Smo mice and tumor formation occurred only in localized foci on the superficial surface of the molecular layer. Thus, tumor formation is not simply the result of over proliferation of GNPs within the EGL. Moreover, Smo/Smo medulloblastomas were transplantable and serially passaged in vivo, demonstrating the aggressiveness of tumor cells and their transformation beyond a hyperplastic state. The Smo/Smo model is the first mouse medulloblastoma model to show leptomeningeal spread. The adherence to human pathology, high incidence, and early onset of tumors thus make Smo/Smo mice an efficient model for preclinical studies. [Cancer Res 2008;68(6):1768–76]
- leptomeningeal spread
- Sonic hedgehog
- mouse cancer models
Medulloblastomas are the most common nervous system malignancy in children. Medulloblastomas are generally thought to derive, at least in part, from primitive neuroepithelial cells on the roof of the fourth ventricle. During development, these precursor cells migrate over the cerebellar surface to form the external granular layer (EGL) and then migrate inward to form the internal granular layer (IGL). Migration occurs during the formation and growth of the cerebellar folia ( 1), a process that is complete by ∼12 months of age in humans ( 2) and by postnatal day 21 (P21) in mice ( 3). During EGL formation, granule neuron precursors (GNP) undergo a massive expansion, producing an approximate 1000-fold increase of GNPs in the cerebellum ( 4). This proliferation is regulated by the Sonic hedgehog (Shh) and Notch signaling pathways. Shh is secreted from Purkinje cells in the cerebellum and binds to the patched receptor on GNPs, which derepresses the Smoothened (Smo) receptor and activates transcription of Shh targets, such as the Gli transcription factors, N-myc, and the D-type cyclins, to drive GNP proliferation ( 1, 5– 12). The Notch and Wingless (Wnt) pathways balance proliferation and differentiation by regulating neurogenic transcription factor activity and maintaining neural progenitor cells ( 13– 19).
These signaling pathways have also been implicated in the formation of medulloblastoma. Mutations have been identified in components of the Shh and Wnt pathways in sporadic human medulloblastomas, and expression profiling has revealed the elevated expression of genes within the Shh, Wnt, and Notch pathways ( 20– 27). This evidence suggests that medulloblastoma may originate from deregulated signaling in developmental pathways that normally regulate GNP proliferation and migration during cerebellar formation.
Mouse models are essential for understanding the genetic events behind in situ tumor formation and for preclinical evaluation and prioritization of therapeutic agents. Over 15 mouse medulloblastoma models have been generated to examine the necessity and sufficiency of various developmental signals in driving cerebellar tumor formation ( 28, 29). However, three major gaps remain in the utility of these models. First, current models fail to generate a high incidence of medulloblastoma with an early tumor onset for use in preclinical studies. The combination of an unpredictable tumor onset, a low tumor incidence rate, and extended periods of latency before tumor formation produces many complications for preclinical studies aimed at the evaluation of therapeutic agents. Second, many existing models require additional mutant backgrounds to increase tumor incidence. Third, none of the current models develop leptomeningeal metastasis of the brain and spine, which is common in human cases ( 30). To address these issues, we generated a homozygous version of our ND2:SmoA1 model, in which a constitutively activated form of the Smoothened gene (SmoA1) is expressed within cerebellar GNPs through the regulation of the 1-kb neuroD2 (ND2) promoter ( 22, 31). In this new model, now called Smo/Smo, the ND2:SmoA1 transgene has increased expression due to its presence in the homozygous state. We describe here the effect of the homozygous ND2:SmoA1 transgene on cerebellar tumor incidence and show that the Smo/Smo homozygous medulloblastoma model is an efficient model for preclinical studies and strongly adheres to the pathology evident in human medulloblastoma cases.
Materials and Methods
Generation of Smo/Smo transgenic mice. The ND2:SmoA1 transgenic mouse line was previously described ( 22). Smo/Smo homozygous mice were generated on C57Bl/6 background by breeding ND2:SmoA1 hemizygous mice. Mice were genotyped by PCR with primers to the ND2:SmoA1 transgene as previously described, and mice carrying the transgene were analyzed further by fluorescence in situ hybridization (FISH) analysis (see below) to determine if they were hemizygous or homozygous. All animals used to establish the original homozygote breeding colony were analyzed by two independent replicates of FISH to confirm homozygosity. All 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. Animals were given free access to water and feed (PicoLab Rodent Diet 20).
Mapping the ND2:SmoA1 transgene insertion site. To obtain metaphase cells for transgene mapping, splenic lymphocytes from a male mouse carrying the ND2:SmoA1 transgene were cultured at 37°C in RPMI with 10% fetal bovine serum and 40 μg/mL lipopolysaccharide (Sigma) for 48 hours. Cells were harvested using 0.1 μg/mL KaryoMAX Colcemid (Invitrogen Corp.) and 0.075 mol/L KCl and fixed in 3:1 methanol/acetic acid. Plasmid DNA containing the ND2:SmoA1 transgene ( 22) was labeled with Spectrum orange dUTP (Abbott Molecular, Inc.) by nick translation (Nick Translation kit, Abbott Molecular, Inc.) according to the manufacturer's protocol. Labeled probe was hybridized to metaphase cells using a HYBrite denaturation/hybridization system (Vysis, Inc.). Posthybridization washes were performed according to the manufacturer's protocol (Vysis, Inc.). Chromosomes were counterstained using 4′,6-diamidino-2-phenylindole, and probe signals were visualized using a fluorescent microscope. To verify correct chromosome assignment of the transgene insertion, DNA from BAC clone RP23-188H10 that maps to proximal chromosome 14 9 was labeled with Spectrum green dUTP and cohybridized along with the ND2:SmoA1 probe.
Detection of transgene homozygosity. Fibroblasts from 2-mm ear punch biopsies were cultured as previously described with modifications ( 32). One or two ear punches from each mouse were plated into one well of an eight-well chamber slide (Falcon 354108, BD Biosciences) and harvested in situ, such that FISH of eight different mice could be performed on a single slide. These modifications allowed for more efficient screening of a large number of mice. In addition, dual color FISH with both the ND2:SmoA1 plasmid probe and the BAC probe RP23-188H10 was carried out as described above to assess ploidy in occasional chambers, in which no metaphase cells were obtained. The control probe on proximal chromosome 14 allowed us to differentiate true diploid homozygotes from tetraploid hemizygotes in this situation. Twenty interphase nuclei and up to five metaphase cells (when available) were scored for each mouse.
Mouse pathology and immunohistochemistry. Mice were euthanized using CO2 inhalation, the brains were removed, and tissue was fixed in 10% buffered formalin for pathologic examination. Tissue blocks were paraffin-embedded, cut into 4-μm sections, and stained with H&E using standard methods. Spines were also collected from 31 Smo/Smo mice for analysis of tumor spread. After decalcification, spines were paraffin-embedded and processed as above. For immunohistochemical analysis, monoclonal antibodies to cyclin D1 (Lab Vision Corporation), nestin (Chemicon International), NeuN (Chemicon International), p27 (Santa Cruz Biotechnology), RU49 (Affinity BioReagents), and Pax6 (R&D Systems) were used; secondary antibodies were applied according to Vectastain Elite avidin-biotin complex method instructions (Vector Laboratories), and detection was carried out with 3,3′-diaminobenzidine reagent (Vector Laboratories). Cells were visualized with a Zeiss Axioscope 40 microscope, and images were captured with Qimaging MicroImager II digital camera.
Magnetic resonance imaging and chlorotoxin:Cy5.5 imaging analysis. Magnetic resonance imaging (MRI) was performed using a 4.7-T MRI with a Bruker magnet, equipped with an INOVA system (Varian, Inc.). Multislice spin echo acquisition was obtained in the axial plane with a TR/TE of 3,000/20 ms, two averages, FOV of 30 × 30 mm2 (passage 1) and 30 × 55 mm2 (WT and passage 2), with a 256 × 256 matrix, slice thickness of 1 mm, gap of 0 mm, and plane resolution of 0.12 × 0.12 mm2 (passage 1) and 0.12 × 0.21 mm2 (WT and passage 2).
In vivo imaging with chlorotoxin:Cy5.5 (CTX:Cy5.5) bioconjugate was carried out as previously described ( 33). To prevent autofluorescence in the intestine, mice used for CTX:Cy5.5 Xenogen Imaging were fed alfalfa-free food (Teklad Global 18% Protein Rodent Diet 2918) for at least 2 weeks before imaging. Hair was removed from the head of the mice with the over-the-counter product, Nair, 3 days before imaging. Animal experiments were performed according to the NIH Guide for the Care and Use of Experimental Animals and approved by our Institutional Animal Care and Use Committee.
Transplantation experiments. Recipient athymic nude (nu/nu) and C57Bl/6 mice (Charles River) were used in brain xenograph transplant studies. Occipital bone window placement was stereotactically positioned 3 mm posterior from bregma and 3 mm lateral to λ. Recipient mice were anesthetized with isofluorane and 1 mm3 portions of tumor tissue or a suspension of 1 × 106 tumor cells from a symptomatic Smo/Smo mouse were implanted or injected, respectively, into the right cerebellar parenchyma of the recipients. The occipital bone window was covered with Gelfoam (Pharmacia & Upjohn), and the scalp incisions were closed with TISSUMEND II (Veterinary Products Laboratories). Animal experiments were performed according to the NIH Guide for the Care and Use of Experimental Animals and approved by our Institutional Animal Care and Use Committee.
Mice homozygous for the ND2:SmoA1 transgene exhibit an increased tumor incidence. We previously generated transgenic mice that expressed a constitutively active mutant of the Smoothened gene (SmoA1; ref. 34) driven by the 1 kb ND2 promoter, which is expressed primarily in cerebellar GNPs ( 22). Approximately, 50% of ND2:SmoA1 mice develop cerebellar tumors that closely resemble human medulloblastoma. Additionally, granule cell hyperplasia and a persistence of a population of cells resembling the EGL normally found in the developing cerebellum were detected in these mice by 2 months of age, suggesting that increased Shh signaling in the cerebellum might maintain the GNPs in a proliferative or precursor-like state ( 35).
In the current study, we generated mice homozygous for the ND2:SmoA1 transgene on an otherwise wild-type background with no additional engineered mutations to examine the effect on cerebellar tumor incidence. ND2:SmoA1 mice on a C57Bl/6 background were bred together, and fibroblasts from the resulting progeny were examined by FISH to identify mice in which the transgene was present in the homozygous state. The ND2:SmoA1 transgene insertion site was mapped to chromosome 14 using a ND2:SmoA1 plasmid probe that showed a signal on chromosome 14 at band qC1. No signal at any other site could be visualized. Mice homozygous for the ND2:SmoA1 transgene were then bred together and subsequent progeny analyzed by FISH to verify the generation of homozygous Smo/Smo mice ( Fig. 1A ).
A natural history study was carried out with 296 homozygous Smo/Smo mice to examine tumor onset. Two hundred hemizygous ND2:SmoA1 mice were studied in parallel. Mice were grossly examined twice weekly to detect the presence of symptoms indicative of tumor formation. These symptoms included ataxia, indicative of cerebellar abnormalities; a protruded skull, indicative of tumor formation beneath the skull; and a tilted head and hunched posture, indicative of hydrocephalus caused by ventricular obstruction or direct compression of nerves by the tumor. Weight loss was also common in mice with tumors. Animals displaying a protruded skull, head tilt, hunched posture, or weight loss were immediately sacrificed to prevent suffering caused by late-stage disease. The natural history study revealed a tumor incidence of over 80% by 5 months of age in the Smo/Smo mice ( Fig. 1B). This is a substantial increase from the incidence in the ND2:SmoA1 hemizygous mice, of which 50% developed medulloblastomas by 5 months of age. The increase in hemizygous ND2:SmoA1 tumor incidence compared with our previous results ( 22) is primarily due to an improved ability to detect early tumors via clinical manifestations.
Histolopathologic analysis of the brains from mice sacrificed during the natural history study revealed pronounced cerebellar malformations, the majority of which included large exophytic medulloblastomas with cerebellar dysplasia and subsequent cerebellar effacement ( Fig. 2A ). Further analysis revealed that the tumors were composed of polygonal to elongate cells arranged in densely cellular sheets and bundles with cellular palisading, rosette formation, and frequent mitoses ( Fig. 2B). Large tumors had multifocal areas of necrosis and neovascularization, and tumor invasion into the fourth ventricle was also noted ( Fig. 2C). Superficially spreading medulloblastomas with focal invasion into the remaining molecular layer were also observed, as were multifocal granular cell rests in the molecular layer.
Immunohistochemical analysis of frank cerebellar tumors revealed that they consisted of highly proliferative, progenitor-like cells. Tumors had strong positive signals for cyclin D1 and nestin, whereas the remaining IGL with normal cerebellar architecture was negative for both ( Fig. 3A ). Tumors were also characterized by a substantial loss of neuronal differentiation, as indicated by loss of neuN expression, whereas neuN staining was robust in the remaining IGL. Thus, the cells within Smo/Smo tumors maintain the capability for proliferation while hindering neuronal differentiation, but the remaining granule cells differentiate and form the IGL as in the normal cerebellum.
Early onset of Smo/Smo tumors. Further analysis examined 131 homozygous mice at 1-month and 2-month time points to identify tumors not detected by clinical manifestations. Histopathologic analysis of cerebellar sections taken from these mice revealed tumors in asymptomatic animals. This incidence increased from 85% at 1 month to 94% at 2 months ( Fig. 1C). The tumors predominantly localized to the surface of the cerebella, in contrast to the symptomatic tumors described above with cerebellar dysplasia and frequent disruption of the IGL ( Fig. 2D). These superficial spreading medulloblastomas displayed the morphology of tumor cells with abnormal shapes and sizes and pleomorphic nuclei and were detected usually in a subset of cerebellar folia. Similar phenotypes are also seen in human medulloblastoma cases (Supplementary Fig. 1). Immunohistochemical staining with antibodies recognizing cyclin D1, neuN, and nestin indicated that these superficial spreading medulloblastomas displayed a mixed population of cells, with some differentiated, neuN-positive cells interspersed within a larger population of abnormally shaped, nestin-positive and cyclin D1–positive tumor cells ( Fig. 3B). Because these large masses containing atypical cells were microscopically and immunohistochemically consistent with cancer foci, they were scored as such (Supplementary Fig. S2).
In addition to tumors, other abnormalities were noted in the cerebella of mice examined at 1 and 2 months. Ectopic granule cell rests formed in the superficial and midmolecular layer, similar to those sometimes found in human medulloblastomas. Sixty-three percent of mice had ectopic granule cell rests at 1 month and 68% by 2 months (Supplementary Fig. S2). These rests were suggestive of arrested inward migration and consisted of granule cells that underwent subsequent differentiation, as indicated by their regular nuclear morphology. In contrast to the large lesions containing atypical cells, small ectopic granule cell rests at the 1-month and 2-month time points were positive for neuN and negative for cyclin D1, confirming that these cells had differentiated, as expected from their neuronal morphology in H&E-stained sections (Supplementary Fig. S2). These were not scored as cancer foci.
Despite the ability of a fraction of these ectopic cells to differentiate, histologic examination of Smo/Smo cerebella at 1 and 2 months suggests that tumors form in nearly all mice by 1 month of age, yet the mice remain asymptomatic as the tumors progress for a period of time. This timing is ideal for preclinical analyses because mice can be enrolled based on their age and treatments can be assessed for longer periods of time.
Persistent proliferation of the external granular layer in early postnatal Smo/Smo mice. The high incidence of tumors in the mice at 1 and 2 months led us to examine the Smo/Smo cerebella at earlier time points. In wild-type mice, GNP proliferation in the EGL is maximal at approximately P5 and the inward migration to form the terminally differentiated IGL and subsequent depletion of the EGL is complete by P21. Remarkably, all Smo/Smo mice displayed a thickening of the EGL at P14 ( Fig. 4A and B ). There was no evidence of tumor formation at this stage. Interestingly, the expansion of the Smo/Smo EGL did not lead to a subsequent thickening of the IGL, perhaps due to intact downstream processes that regulate cell number and density in the IGL. Further analysis of this expanded EGL through immunohistochemical staining for nestin, neuN, and p27 indicated that the processes of neuronal differentiation and cell cycle withdrawal were still intact in the Smo/Smo cerebellum. Although the thickness of the EGL increased from two to four cells in WT P14 cerebella to 12 to 20 cells in the Smo/Smo P14 cerebella, the boundary between the outer, proliferative EGL, and the inner, arrested EGL was still maintained in the expanded Smo/Smo EGL, as indicated by neuN and p27 staining in the inner EGL ( Fig. 4C and D). Similar analysis was carried out in cerebellar sections from mice sampled at P5. The P5 Smo/Smo EGL was of a similar thickness to that of the wild-type P5 EGL (Supplementary Fig. S3A and B). The foliation pattern seemed normal, with only a minor nodule of hyperplasia detected in one lobule in one of the four P5 Smo/Smo mice examined. Immunohistochemistry with antibodies against nestin, neuN, and p27 displayed similar patterns in the wild-type and Smo/Smo P5 cerebella, with neuronal differentiation and p27 expression indicating the border between the inner and outer EGL (Supplementary Fig. S3C and D). Nestin expression was detected in the radial glia, as well as in some cells of the EGL (Supplementary Fig. S3C and D). Thus, early cerebellar formation is normal in Smo/Smo mice, with the exception of increased GNP proliferation around P14. However, the excess GNPs are not maintained in the IGL and tumor formation occurs instead in localized foci near the surface of the molecular layer, most often near the roof of the fourth ventricle as in human medulloblastoma.
Smo/Smo mice display leptomeningial tumor spread. Further examination of the Smo/Smo mice revealed laterally spreading tumors on the surface of the brain and within the spinal cord. Foci of neoplastic cells were detected in the brain but distant from the primary cerebellar tumor, as well as within the gray matter of the spinal cord ( Fig. 5A ). Immunohistochemical staining of spinal sections indicated that these ectopic cells expressed the cerebellar transcription factors RU49 and Pax6 (Supplementary Fig. S4), indicating that these cells originated from the cerebellar tumors. Upon closer examination, tumor cells could be found within the leptomeningial membranes of the brain ( Fig. 5B), suggesting the means by which tumor cells spread from the cerebellar tumor to distant regions within the brain, as in human medulloblastoma. Reevaluation of ND2:SmoA1 specimens revealed infrequent leptomeningeal spread in the hemizygous mice as well. Vasculature was frequently present near the site of the nodules, providing further support for these separate nodules as metastases ( Fig. 5B, red arrows). In human cases, medulloblastoma rarely metastasizes to other organs or tissues, but frequently spreads through the leptomeninges to the brain and the spine ( Fig. 5C and D). Thus, the spread of Smo/Smo tumor cells within the brain and to the spine closely resemble the metastases seen in human disease, again strengthening the Smo/Smo model in its adherence to the biology of human medulloblastoma.
Intracranial transplantation and engraftment of Smo/Smo tumors. To further show the aggressive capacity of Smo/Smo tumor cells beyond simply an excess of proliferation, cells taken from a tumor in a symptomatic Smo/Smo mouse were transplanted to several wild-type recipient mice to examine whether these tumors could be serially passaged. Fresh tumor was harvested from a symptomatic mouse, and the tumor was minced in PBS. After anesthetization, xenografts (1 mm3) of tumor tissue were implanted onto the right cerebellar parenchyma through an occipital bone window in the recipient nude mice.
Three months after transplantation, one of two recipient mice from passage 1 developed a protruded skull and a head tilt, indicative of tumor formation. MRI showed a mass involving both cerebellar hemispheres, with prominent vasculature within the mass and a putative area of central necrosis ( Fig. 6A; passage 1 ). The mouse was then sacrificed, and a vascular cerebellar tumor was evident upon gross examination of the cerebellum. The brain was then cut in half along the saggital midline, and histopathologic evaluation of sections from one half confirmed the presence of an invasive cerebellar tumor closely resembling tumors arising in the Smo/Smo mice with evident leptomeningeal and ventricular spread in addition to abundant neovasculature ( Fig. 6B). The other half of the tumor was subsequently minced into 1 mm3 xenografts and transplanted to five recipient nude mice as above. Two mice died on postoperative days 1 and 3 due to somnolence, ataxia, and difficulty in eating. The other three recipients tolerated the procedure without complications.
Two and a half months later, one of three mice from the second tumor passage developed a right occipital mass, without evidence of neurologic impairment. The symptomatic mouse and a wild-type nude mouse were each injected with 100 μL of 20 μmol/L CTX:Cy5.5, a bioconjugate that delineates medulloblastoma from adjacent nonneoplastic tissue ( 33). Three days later, biophotonic images of the injected mice were obtained on a Xenogen IVIS-100 system. The biophotonic images indicated the presence of a tumor within the symptomatic mouse so MRI was then performed. Analysis by MRI showed a large extracranial tumor overlying a smaller cerebellar tumor in the symptomatic mouse ( Fig. 6A; passage 2). No tumors were noted in the wild-type mouse. Biophotonic imaging was repeated on the isolated tissues after dissection. The tumors emitted near-IR fluorescence at levels significantly above the background luminescence in the control brain ( Fig. 6C). Histopathologic analysis confirmed the presence of a cerebellar tumor that again resembled the original Smo/Smo tumor. The extracranial mass was a highly cellular tumor that was well demarcated and encapsulated ( Fig. 6D).
Additional experiments showed that Smo/Smo tumors could be transplanted to recipients on either a nude or C57Bl/6 background. Tumors from five separate donors were transplanted by injecting 1 × 106 Smo/Smo tumor cells suspended in PBS into the cerebella of recipient mice. Three of six recipients developed tumors histologically resembling the original Smo/Smo tumors from three separate donors. Two of two nude recipients developed tumors whereas only one of four recipients on the C57B1/6 background developed a tumor; tumor-take may have been impeded on the wild-type background. In total, tumors were successfully transplanted from four distinct donors, thus confirming that Smo/Smo medulloblastomas contain transformed cells capable of initiating tumors.
In this study, we generated a mouse medulloblastoma model that develops tumors earlier and more frequently than previous models, thus making it an efficient model for preclinical studies. The homozygous ND2:SmoA1 transgene induced formation of medulloblastomas and leptomeningeal spread of tumor cells to the brain and spine. This closely models the human disease, wherein tumors form primarily within the cerebellum but spread to other regions of the brain and spine through the leptomeninges ( 30). Such leptomeningeal spread has not previously been shown in any other autochthonous medulloblastoma mouse model nor has increased hedgehog activity alone been shown to be sufficient for driving metastasis. Smo/Smo mice develop medulloblastomas as early as 1 month after birth and with a subclinical tumor incidence of 94% by 2 months of age. Analysis of the Smo/Smo cerebella at 1 and 2 months indicated that nearly all of these early lesions go on to form tumors, which is unlike the preneoplastic lesions reported in the patched mutant mice, in which only 10% to 20% of mice develop medulloblastomas despite over half displaying early lesions by 3 to 6 weeks of age ( 36, 37). Crossing patched heterozygous mice onto the p53-null background increases tumor incidence to nearly 100% with early tumor onset ( 38). Thus, the events triggered by ND2:SmoA1 transgene expression are more potent at driving tumor formation and maintenance than the partial loss of inhibition by patched and similar to those of patched heterozygotes on the p53 null genetic background. The tumors that form by 1 and 2 months in the Smo/Smo model precede the onset of symptoms so mice can be enrolled into preclinical studies based upon their age and not the symptomatic appearance of tumors. This feature is important because symptomatic mice generally require sacrifice within days after the onset of symptoms.
Further analysis of the Smo/Smo mice during early postnatal development revealed an expansion of cells in the EGL at P14. The thickness of the Smo/Smo EGL expanded from the normal width of two to four cells deep to ∼12 to 20 cells deep, which is even greater than the thickness of the WT EGL when proliferation is maximal around P5. This uniform EGL expansion suggested that the increased Shh pathway activity induced by ND2:SmoA1 transgene expression within the GNPs caused them to undergo additional proliferation, which would be expected based on the normal role of Shh activity in driving GNP expansion during cerebellar development. Examination at earlier time points revealed that the initial foliation pattern of the cerebellum was preserved and that the thickening of the Smo/Smo EGL occurred after P5. However, the expansion of the EGL did not translate into an increase in the total thickness of the IGL, indicating that the normal processes of cell differentiation, migration, and apoptosis are still intact within the early postnatal Smo/Smo cerebellum. Thus, tumor formation in this instance is not simply due to an excess of granule neurons generated by their overproliferation within the EGL. Remarkably, the majority of the cerebellum developed normally, with proper foliation and organization of the various cell layers. Instead, tumor formation occurred only in localized foci that initiated on the superficial surface of the molecular layer usually on the roof of the fourth ventricle, which again closely resembles the appearance of human medulloblastomas. This finding shows that despite expression of the ND2:SmoA1 transgene in all GNPs, this tumor-initiating mutation is capable of generating medulloblastomas in only a subset of cells. The mechanism by which this occurs remains to be determined.
Smo/Smo tumors also proved capable of being transplantable and serially passaged in vivo, further supporting the histopathologic data that these tumors are not simply the result of overproliferation of GNPs in the EGL, but instead form from a select population of cells and consist of highly aggressive cells. Recipient mice developed medulloblastomas resembling the original tumors with tumor cells that disseminated through the leptomeningeal membranes in recipient brains similar to the lateral spread observed in Smo/Smo mice.
Another feature of the Smo/Smo model was that the high tumor incidence is not dependent on the addition of other engineered mutant backgrounds. Whereas other mouse medulloblastoma models have shown that genetic loss of DNA repair and apoptosis components increases the aggressiveness of tumors and their incidence ( 38– 44), the sole genetic change in Smo/Smo mice is the introduction of activated hedgehog signaling via the ND2:SmoA1 transgene. Tumor formation in Smo/Smo mice does not require crossing the mice to other genetic backgrounds with mutations that are rare in human medulloblastomas, such as mutations in p53, and the Smo/Smo model may thus more accurately predict the human response to therapeutic agents in preclinical studies. The homozygous transgene also bypasses the need for genotyping, reducing time and costs associated with colony management.
Given the high incidence of tumors in the Smo/Smo model, their aggressive nature, and the abundance of cerebellar abnormalities produced, it is possible that Smo/Smo tumors have accumulated additional mutations, but the presence of other mutations in Smo/Smo tumors has not yet been evaluated. Pilot comparative genomic hybridization studies have revealed gains of whole chromosome 6 and/or chromosome 14 in Smo/Smo tumors, and pilot microarray studies indicate that mRNAs from some genes on these chromosomes are increased. However, it remains to be determined whether these increases contribute to tumor formation.
The Smo/Smo mice are thus an efficient model for the preclinical evaluation of therapeutic agents and for further investigating the series of events leading to medulloblastoma formation and progression. A predictable and high-tumor incidence as early as 1 month allows for enrollment into preclinical studies well before the onset of tumor symptoms, the presence of which requires the mouse to be sacrificed, providing ample time to monitor the response to a given treatment. In situ tumor formation also occurs among the other cells of the cerebellum and allows complex processes, such as angiogenesis, tumor-host interactions, and tumor spread, to be monitored. Preclinical studies with targeted therapies aimed at preventing tumor formation and reducing tumor growth are currently under way. Ideally, these studies will expedite the selection of effective novel agents for use in human clinical trials and lead to improved and better tolerated treatments for medulloblastoma patients.
Grant support: NIH grants CA112350-03, CA114567-02, and CA119408-03, NIH contract NO1-CA037122, Children's Hospital and Regional Medical Center Neurooncology Endowment Support, Damon Runyon Clinical Investigator award (J. Olson), and NIH grant T32 CA093251 (E. Villavicencio).
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 the Histopathology Shared Resource at Fred Hutchinson Cancer Research Center for sectioning and staining the transgenic mouse tissues and Jennifer Stoeck and Keith Strand for assistance with transgenic animals.
Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).
- Received August 28, 2007.
- Revision received December 18, 2007.
- Accepted January 10, 2008.
- ©2008 American Association for Cancer Research.