Introduction
An important goal in designing therapies for patients with the malignant brain tumor medulloblastoma is to reduce the risk of metastasis. A defining characteristic of metastasis in medulloblastoma is leptomeningeal dissemination (LMD), the spread of tumor cells via the cerebrospinal fluid (CSF) to the leptomeningeal spaces of the brain and spinal cord. The need to minimize metastasis risk is critical because survival times are very low once LMD has occurred. Guided by this principle, pediatric oncologists have designed multimodality treatments, which combine surgery, chemotherapy, and craniospinal radiation [
1],[
2]. These aggressive regimens reduce the risk of metastasis, but they are associated with disabling side effects, including neuropsychiatric challenges, stunted body growth, hormonal imbalance, epilepsy, and stroke in long-term survivors [
3]-[
5].
Because the prospect for long-term survival is so poor for patients with LMD, radiation to the entire neuraxis remains an indispensable part of medulloblastoma treatment. Unfortunately, intellectual capacity and academic achievement decline even in children treated with protocols that use reduced radiation doses [
6]. Hope for prolonging disease-free survival and eliminating treatment-related neurotoxicity rests on developing therapies that specifically target the molecular mediators of LMD, our knowledge of which is limited.
Analysis of the medulloblastoma transcriptome in large cohorts of patients has shown that medulloblastomas do not comprise a uniform disease entity, but rather a diverse set of tumors, which have different gene expression profiles, rates of metastatic dissemination, and patient survival times [
7]-[
9]. Despite the identification of molecular signatures with prognostic implications, metastasis stage remains a critical determinant of high-risk tumors [
10]. The expression signature of most common medulloblastoma subtype affecting infants and adults implicates the Sonic Hedgehog (Shh) signaling pathway in tumor pathogenesis. Shh signaling, which governs many aspects of embryogenesis, stimulates proliferation and inhibits differentiation of neural progenitor cells in the cerebellum [
11]. Consistent with this developmental function, genetically engineered mouse models have shown that aberrant activation of the Shh pathway in either multipotent neural stem cells or granule neuron precursor cells (GNPs) in the developing cerebellum can initiate medulloblastoma formation [
12]-[
17].
We reported previously that ectopic expression of genes encoding Eras (embryonic stem cell–expressed Ras), Lhx1 (LIM-class homeobox gene 1), Ccrk (cell cycle–related kinase), and the phosphatidyl inositol 3-kinase signaling protein Akt shifted the growth characteristics of Shh-induced medulloblastoma from a localized pattern to one in which tumor cells seeded the leptomeninges of the brain and spinal cord [
18]. The idea that these proteins might be LMD drivers came from a screen for metastasis genes using transposon-mediated mutagenesis in Patched-deficient mice, which develop medulloblastomas spontaneously [
19].
Extensive evidence from the field of epithelial cancer metastasis indicates that a large number of genes choreograph the multistep process whereby tumor cells breach the basement membrane in the originating tissue, intravasate into the bloodstream, and colonize distant organs [
20],[
21]. Although the chain of events involved in LMD is different from the invasion-metastasis cascade of carcinomas, the changes in cell physiology needed for medulloblastoma cells to detach from the tumor mass, enter the CSF, and proliferate in the subarachnoid spaces no doubt require an equivalently diverse set of genes.
To expand the set of LMD-driving genes, we transferred and expressed Arnt (aryl hydrocarbon receptor nuclear translocator) and Gdi2 (GDP dissociation inhibitor 2), which had been identified previously as common insertion sites for the Sleeping Beauty (SB) transposon, in cerebellar neural progenitor cells in mice by retroviral transfer in combination with Shh. Here we show that ectopic expression of Arnt and Gdi2 promotes spinal LMD in mice bearing Shh-induced medulloblastomas and demonstrate the effects of these genes on the motility, invasiveness, and anchorage-independent growth of medulloblastoma tumor cells and precursor cells in culture.
Materials and methods
Retroviral vector construction
Construction of RCAS-Shh, which contains an in-frame, carboxy-terminal epitope tag consisting of six repeats of the influenza virus hemagglutinin (HA) epitope, was described previously [
14]. The cDNA clones for mouse
Arnt and
Gdi2 were obtained from the American Type Culture Collection (Manassas, VA), where they were deposited by the Integrated Molecular Analysis of Genomes and their Expression (IMAGE) consortium (
http://www.imageconsortium.org). RCAS vectors were prepared by ligating a PCR-generated cDNA corresponding to the complete coding sequence into the parent retroviral vector RCASBP(A) [
22]. RCAS-Gdi2 contained an internal ribosome entry site (IRES) coupled to the
Aequorea coerulescens green fluorescent protein (GFP) for tracking the cellular localization of the expressed proteins. To produce live virus, we transfected plasmid versions of RCAS vectors into immortalized chicken fibroblasts (DF-1 cells) and allowed them to replicate in culture.
In vivo somatic cell gene transfer in transgenic mice
The use of mice in this study was approved by the Institutional Animal Care and Use Committee of the University of Utah. To induce medulloblastomas in mice, we used a version of the RCAS/
tv-a somatic cell gene transfer system to transfer and express the
Shh gene in Nestin-expressing cells in the cerebellum. Nestin, an intermediate filament protein, is a marker for neural progenitor cells prior to neuronal or glial differentiation. The RCAS/
tv-a system uses a replication-competent, avian leukosis virus, splice acceptor (RCAS) vector, derived from the subgroup A avian leukosis virus (ALV-A), and a transgenic mouse line (
Ntv-a) that produces TVA (the cell surface receptor for ALV-A) under control of the
Nestin gene promoter [
23]. After TVA-mediated infection of mammalian cells with RCAS retrovirus, the newly synthesized provirus integrates into the host cell genome where the transferred gene is expressed constitutively. RCAS-transduced mammalian cells do not produce infectious virus because mRNA splicing events remove the retroviral genes necessary for viral replication.
To initiate gene transfer, we injected retrovirus packaging cells (DF-1 cells transfected with and producing recombinant RCAS retrovirus) into the lateral cerebellum of the mouse from an entry point just posterior to the lambdoid suture of the skull (bilateral injections of 105 cells in 1–2 μl of phosphate buffered saline (PBS)). For experiments involving simultaneous transfer of two genes, we prepared cell pellets by mixing equal numbers of both retrovirus-producing cells. We injected mice within 72 hours after birth because the number of Nestin+ cells decreases progressively during the course of neuronal differentiation. The mice were sacrificed when signs of increased intracranial pressure became apparent, indicated by enlarging head circumference (a sign of hydrocephalus), head tilt, gait ataxia, or failure to eat or drink. Asymptomatic mice were sacrificed 4 months after injection. The brains were fixed in formalin, and divided into quarters by parallel incisions in the coronal plane. To identify spinal LMD, we fixed whole spinal column preparations in formalin for 48–72 hours and then removed the spinal cord by microdissection. Brain and spinal cord specimens were embedded in paraffin and sectioned for histochemical analysis.
Immunocytochemistry and microscopy
Methods for immunoperoxidase staining of mouse brain and spinal cord sections were described previously [
18]. We used the following antibodies from the indicated commercial sources: Mab9E10—c-Myc (Santa Cruz Biotechnology, Santa Cruz, CA); Mab3580—GFP (Chemicon, Temecula, CA); ab14545—βIII-tubulin (Abcam, Cambridge, MA); Mab2F11—70 kDa neurofilament protein (Dako, Carpinteria, CA). Tissue sections were visualized using a Zeiss Axiovert 200 microscope, and photomicrographs were captured using an AxioCam high-resolution CCD camera and Axiovision imaging software (Carl Zeiss International, Germany).
Gene transfer by lentivirus
To express
Arnt and
Gdi2 in cultured cells, we used the HIV-ZsGreen lentivirus system [
24]. The gene-transfer plasmid (pHIV-ZsGreen, Addgene plasmid 18121) encodes a GFP from the reef coral
Zoanthus to track expression of the transgene in infected cells. To produce live virus for gene transfer, we ligated
Arnt and
Gdi2 cDNAs into the NotI-XbaI site of pHIV-ZsGreen and transferred the recombinant plasmids to HEK293T packaging cells by transfection in combination with helper plasmids pMDLgRRE, pRSV-Rev, and pCMV-VSV-G. Virus particles were collected by ultracentrifugation and applied to recipient cells in tissue culture plates under biosafety level 2 conditions. Viral titer was determined from the ratio of fluorescent to nonfluorescent cells.
In vitro scratch assay
SHH-NPD or DAOY cells, previously infected with HIV-ZsGreen lentivirus and expressing Arnt or Gdi2, were seeded into 6-well plates (35-mm well diameter) in triplicate and incubated at 37°C under 5% CO2 until the cell density reached 90% confluence. The cells were washed with PBS and incubated overnight in serum-free Dulbecco’s modified Eagle medium (DMEM) containing 10 μg/ml mitomycin C to suppress proliferation. A linear wound was created in the nearly confluent cell monolayer by scratching across the center of the plate using a sterile 200-μl pipette tip in a smooth, even fashion, and the medium was replaced with DMEM containing 10% fetal bovine serum. Digital photomicrographs were taken 0 and 10 hours after making the scratch wound through a 10× phase contrast microscope objective. Prior to each photo session, the medium was replaced with PBS to obtain a clear photographic image. To quantify the rate of cell migration across the wound gap, we measured the distance between the two advancing, converging cell fronts at five points equidistant along the scratch wound. The rate of cell migration (μm/hour) was calculated using the following formula: [average scratch width at 0 hours − average scratch width at 10 hours] ÷ 20. The denominator of 20 is used to account for the ten hours of observation and two advancing cell fronts. Each experiment was repeated on a separate day. Standard deviation (SD) for the mean migration rate was derived from 30 data points.
Matrigel chemoinvasion assay
SHH-NPD or DAOY cells (2 × 105), previously infected with HIV-ZsGreen lentivirus and expressing Arnt or Gdi2, were plated into the upper chamber of an 8-μm, 6-well BD Biocoat Matrigel invasion chamber (BD Biosciences, Bradford, MA) in 500 μl of serum-free DMEM. Prior to plating, the cells were treated with 10 μg/ml mitomycin C to suppress proliferation. Each bottom chamber was filled with 2 ml of DMEM containing 10% fetal bovine serum and then incubated for 24 hours at 37°C under 10% CO2. After incubation, the noninvading cells are removed from the upper chamber with a cotton swab. The invading cells on the lower surface of the membrane were stained with Hema 3 (Fisher Scientific, Loughborough, UK) and counted under a microscope. Five nonoverlapping microscope fields (10× objective) were counted in each well. Each assay was repeated, and the mean number of invading cells per well was calculated. SD was derived from 20 data points.
Cells (104) were suspended in culture medium (2 ml) containing 0.3% agar (Difco Bactoagar, Detroit, MI) and plated over a layer of 0.5% agar in the same culture medium in tissue culture plates (35 mm in diameter). Triplicate cultures were incubated at 37°C under 5% CO2. Colonies were counted 14 days after plating by viewing under a microscope through a 10× objective and a graduated eyepiece reticle. Any colonies larger than 38 μm in diameter (>10 cells) were scored as positive. SD was derived from three data points.
Reverse transcriptase–polymerase chain reaction (RT-PCR)
Cerebella were removed from euthanized mice and frozen immediately in liquid nitrogen. Tissue specimens were homogenized in Trizol, and total RNA was extracted using RNeasy Mini Kit (Qiagen, Valencia, CA) according to the manufacturer’s instructions. RT-PCR was carried out using the SuperScript III One-Step RT-PCR kit (Life Technologies, Grand Island, NY) as described previously [
25]. In brief, cDNA was synthesized from total RNA by reverse transcription and PCR in the presence of oligonucleotide primers specific for Myc epitope–tagged
Arnt. RT-PCR products corresponding to the constitutively expressed, glycolytic enzyme glyceraldehyde-3-phosphate dehydrogenase (GAPDH) served as internal controls for sample variation in mRNA degradation and gel loading. Reaction products were separated by electrophoresis through agarose gels and visualized by UV illumination after immersion in ethidium bromide solution.
Expression profiling and molecular subgrouping of human medulloblastomas
Human primary medulloblastomas were profiled on Affymetrix Genechip Human Exon 1.0ST arrays at The Centre for Applied Genomics in Toronto, Canada (
www.tcag.ca). Expression analysis was performed using Affymetrix Expression Console (Version 1.1) as previously described [
26]. Additional, publically available medulloblastoma expression data sets were obtained from NCBI Gene Expression Omnibus and used to validate our findings [
9],[
27]. Subgrouping of tumors was performed using an 84-gene expression classifier [
28].
Statistics
We assessed statistical significance of intergroup differences in cell migration, invasion, and anchorage-independent growth using ANOVA and Fisher’s protected least significant difference test. The significance of intergroup differences in the incidence of tumor formation and LMD was assessed using the χ2 contingency test.
Discussion
We show here that ectopic expression of Arnt and Gdi2, which were identified in an unbiased genetic screen for metastasis genes using SB transposon mutagenesis, promotes LMD in Shh-induced medulloblastomas in mice. When overexpressed in cell culture models of Shh-driven medulloblastomas and tumor precursors, Arnt stimulated motility, invasiveness, and anchorage-independent growth, cell traits that are closely associated with metastatic competence in carcinomas. Gdi2 had these same effects on tumor precursors and simulated invasiveness in fully transformed medulloblastoma cells.
Although Arnt and Gdi2 have previously been implicated in cancer biology, the molecular mechanisms by which these proteins cause medulloblastoma cells to disseminate remain uncertain. Arnt is most widely known in oncology as the constitutively expressed binding partner of the hypoxia-inducible transcription factors HIF1α and HIF2α, whose downstream target genes play active roles in angiogenesis, proliferation, invasion, and metastasis (reviewed in [
37]). Arnt occupies a second sphere of biology through its interaction with the aryl hydrocarbon receptor (AhR), which binds dioxin and other xenobiotics. After binding a xenobiotic molecule, AhR translocates from its normal location in the cytoplasm to the nucleus, where it dimerizes with Arnt and activates transcription of genes containing dioxin-response elements.
A substantial body of evidence indicates that the role of Arnt/AhR signaling in cell biology is not restricted to the xenobiotic response but rather extends to normal developmental processes and tumorigenesis (reviewed in [
38]). During mouse cerebellar development, for example, both Arnt and AhR are highly expressed in GNPs during their active proliferation phase on postnatal days 5–6, suggesting that Arnt might maintain GNPs and possibly derivative medulloblastoma cells in an undifferentiated state [
39]. We do not know whether the LMD-inducing effects of Arnt are mediated by HIF or AhR signaling. Arguing against HIF as the relevant pathway, we could not detect increased expression of two known HIF transcription targets, vascular endothelial growth factor and glucose transporter-1, by immunostaining sections of Shh + Arnt–induced medulloblastomas (data not shown).
Gdi2 prevents the dissociation of GDP from Rab proteins, a family of membrane-localized GTPases, which switch between active GTP-bound and inactive GDP-bound states. There is extensive literature to support the general concept that Rab proteins coordinate the intracellular movement of membrane vesicles (reviewed in [
40]). The function of individual Rab family members is specified by their subcellular localization. By inhibiting GDP release, Gdi2 maintains Rab in an inactive conformation, an action that seems uncharacteristic for an oncoprotein. Nevertheless, Gdi2 has a second, more oncogenic role as a chaperone protein, facilitating the transport of Rab molecules from inactive reserve stores in the cytosol to active sites in their specific membrane compartments [
41],[
42]. Certainly, Rab GTPases contribute directly to cancer cell physiology. Rab25, for example, promotes invasive, metastasis-like migration of cells by binding α5β1 integrin molecules and directing their localization to the tips of extending pseudopods [
43].
Extrapolation of our cell culture results to the in vivo growth of medulloblastoma supports a mechanistic scheme whereby Arnt and Gdi2 cause shedding of cells from the primary tumor mass into the CSF by increasing cell motility and invasiveness. By making tumor cells competent to proliferate without surface attachment, expression of Arnt and Gdi2 would also favor the formation of suspended colonies of cells capable of surviving in the CSF before reimplanting on a leptomeningeal surface. The ability of medulloblastoma cells to circumvent apoptotic programs normally triggered by detachment from a solid surface (anoikis) is most likely a prerequisite for LMD. This adaptive response is analogous to that of metastasizing carcinoma cells, which must survive the passage through the bloodstream before colonizing a distant organ.
We could not determine in all cases whether the presence of tumor cells in the brain stem and forebrain resulted from direct extension from the primary tumor or dispersal through the CSF. Nevertheless, the cell traits that empower cells to disseminate to the spinal column (motility, invasiveness, and anchorage-independent growth) could also promote extension from the cerebellum to the fourth ventricle or remote sites in the forebrain.
We demonstrated the LMD-inducing effects of
Arnt and
Gdi2 in Shh-induced medulloblastomas in vivo and validated these effects in culture using cell lines in which the Shh signaling pathway is active. Therefore, we cannot conclude that
Arnt and
Gdi2 are LMD drivers in Shh-independent medulloblastomas even though
Arnt and
Gdi2 are expressed in all subgroups. Nevertheless, the fact that medulloblastomas of any molecular subgroup can metastasize, either at initial presentation or relapse, broadens the clinical significance of LMD driver genes discovered in different genetic backgrounds [
44].
The multiple cellular transitions connecting tumor stromal detachment, CSF passage, and finally leptomeningeal spreading indicate that LMD might approach the complexity of the invasion–metastasis model used to describe hematogenous metastasis of epithelial cancers. Transposon mutagenesis revealed hundreds of candidate genes, although we do not know how many of these genes are drivers as opposed to passengers in the LMD process. Through the functional validation studies we have reported here and previously [
18], we know that Shh-induced medulloblastomas can start down a path of disseminated growth by addition of single LMD driver genes.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made.
The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
NCJ – cell culture experiments (migration and invasion assays), analysis and interpretation of data, manuscript writing and preparation. RK – mouse experiments, analysis and interpretation of data. AD – computational analysis of human tumor data sets, analysis and interpretation of data. WS – cell culture assays (proliferation). CAP – mouse experiments, analysis of gene expression by immunocytochemistry. XW – computational analysis of Sleeping Beauty transposon data, analysis and interpretation of data. MDT – analysis and interpretation of data, development of methodology. DWF – analysis and interpretation of data, manuscript writing and preparation. All authors read and approved the final manuscript.