Skip to main content
NCBI home page
Neuro-Oncology logoLink to Neuro-Oncology
. 2013 Nov;15(Suppl 3):iii235–iii241. doi: 10.1093/neuonc/not193

TUMOR MODELS (IN VIVO/IN VITRO)

PMCID: PMC3823908
Neuro Oncol. 2013 Nov;15(Suppl 3):iii235–iii241.

TM-001. THE DEVELOPMENT TUMOR MODEL: THE BIOLOGICAL, DYNAMIC, IN VIVO MODEL TO UNDERSTAND AND TRY TO DEFEAT GLIOBLASTOMA

Enrico Brognaro 1

Abstract

Science is based on the experimental method since the time of Galileo and Descartes. Even the most brilliant theories and hypothesis must be experimentally verified. The general theory of relativity gained its magnificence only after the solar eclipse of 1919 confirmed the deflection of light experimentally. From that moment, the theory that the space-time bends became reality. To understand and try to defeat glioblastoma, an experimental in vivo model which can allow us to biologically follow and monitor the entire dynamic progression of the tumor both in the bulk and in the microinfiltrated brain parenchyma is indispensable. “The development tumor model” is a xenogeneic orthotopic transplantation model with human glioblastoma-derived tumorspheres of the pre-hypoxic phase cultured in neurobasal serum-free medium as transplanted material. The model posits the transplantation of several animals of an inbred strain of immunodeficient mice or rats with the same material at the same time (time zero). Thus, the model creates a pool of twin immunodeficient transplanted animals in the same condition. By sacrificing one animal a week (but other intervals can be used as needed), we can obtain cells, biopsies (at least four: in the bulk, ipsilateral hemisphere, corpus callosum, contralateral hemisphere) and multiple stainings on section throughout tumor development. This way, the glioblastoma progression can be monitored biologically in time as if it were a single animal both in the bulk and, most importantly, in the microinfiltrated brain parenchyma. This should finally allow us to recognize the three phases of glioblastoma development (one pre-hypoxic and two post-hypoxic) and shed light on which specific cells and peculiar markers to focus on in order to develop innovative therapeutic strategies. Indeed, advanced therapy will have to target both the CSCs after surgery of the primary bulk as well as the CSCs in the microinfiltrated brain parenchyma.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii235–iii241.

TM-002. ROLE OF MICROVASCULAR STRUCTURES IN MIGRATION AND INVASION OF GLIOBLASTOMA-DERIVED STEM CELLS

Soyoung Chang 1,2, Junghwa Cha 1, Kyungsun Choi 1,2, Chulhee Choi 1,2

Abstract

Glioblastoma mutiforme (GBM), the most aggressive primary malignant brain tumor, has highly heterogeneous cell populations. Cancer stem cells (CSCs), a subpopulation with self-renewal, pluripotency and tumorigenesis, have known to play an important role in growth, invasion, angiogenesis and immune evasion of GBM, thus conferring anti-cancer drug resistance. Although CSCs are highly similar to the normal stem cells, growth of these cells are tightly regulated by the tumor microenvironment such as tumor vasculature (called as perivascular niche). It has currently been reported that tumor-associated endothelial cells are vital for maintenance of stemness of CSCs. It has also been shown that CSCs as well as GBM cells are prone to invade along the microvessels, which is termed as co-option. In this vascular niche, CSCs might be affected by chemical cues such as secreted factors by endothelial cells or mechanical cues such as microvascular structure. Indeed, we observed that short-term culture of CSCs on the grooved substrate induced adhesion of CSCs onto the substrate and enhanced migration along the grooved pattern; while CSCs cultured on the flat substrate formed aggregates and neurospheres. To investigate the effect of microvessels on CSCs in vivo, we transplanted spheroids of green fluorescence protein-tagged CSCs originated from GBM onto mouse brain cortex and monitored the behavior of CSCs using in-vivo two-photon microscopy. We found that CSCs at the core of spheroids rapidly died out probably by hypoxia, and CSCs at the periphery moved to and migrated along nearby microvessels. These results collectively indicate that CSCs might be attracted to microvessels due to chemical factors and migrate along the vessel structure mainly via mechanical cues. Therefore, we propose that brain CSCs has specific mechanism to recognize mechanical and chemical cues from tumor vasculature, potential target of tumor therapy.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii235–iii241.

TM-003. AN IN VITRO MOUSE MODEL OF GBM WITH ABRUPT AND PREDICTABLE ONSET

Jessica DePetro 1,2,4, Carmen Binding 1,2,4, Michael Blough 1,2,4, John Kelly 1,2,4, Sam Lawn 1,2,4, Jennifer Chan 1,2,4, Sam Weiss 3,4, Gregory Cairncross 1,2,3,4

Abstract

BACKGROUND: IDH wild-type glioblastoma (GBM) harbors multiple genetic alterations at diagnosis; yet often begins abruptly in older persons. Although many of these alterations are ‘passengers’, those that involve receptor tyrosine kinase signaling and the p53 and RB pathways are found in most newly diagnosed GBMs and are thought to be ‘drivers’ of this cancer. Here we report a PDGFα-linked in vitro mouse model of GBM in which malignant transformation appears to occur abruptly, and the responsible genetic events can be studied. METHODS: Cells from the subventricular zone (SVZ) of p53-null, six-week old mice were dissected and cultured as spheres in serum-free media supplemented with either EGF/FGF or PDGFα. The expression of Nestin, NG2, Olig2 and GFAP was assessed in PDGFα cultures at early and late passages (< P2 and >P10) by Western blot. RESULTS: SVZ cells cultured continuously in EGF/FGF (for up to 18 months) proliferated rapidly but remained growth factor dependent and non-tumorigenic. In contrast, SVZ cells in media containing PDGFα initially grew poorly; these cultures were debris-filled with small spheres. However, at P8, after 4-6 months in vitro, sphere formation and size accelerated abruptly in multiple independent cultures. These transformed cells proliferated rapidly in the absence of PDGFα, and unfailingly, generated tumors with a striking resemblance to GBMs when implanted into the striatum of immune-competent, p53 wild-type mice. In early and late passage PDGFα cultures, Olig2 was continuously expressed, whereas Nestin and NG2 expression increased, and GFAP expression decreased. Transformed SVZ cells and oligodendrocyte progenitor cells have a similar marker profile. CONCLUSION: This model recapitulates other systems in which PDGFα-driven glioma formation has been achieved in vivo in p53-null mice, but may have these advantages: low cost, easy accessibility to sequential molecular interrogation, and suitability for rapid screening of libraries of potential inhibitors of gliomagenesis.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii235–iii241.

TM-004. EVALUATION OF EFFICACY AND FEASIBILITY OF LOCAL TUMOR THERAPY USING CONVECTION ENHANCED DELIVERY WITH TEMOZOLOMIDE AS POTENTIAL TREATMENT STRATEGY FOR GLIOBLASTOMA IN A RAT TUMOR MODEL

Andrea Eisenbeis 1, Roland Goldbrunner 1, Marco Timmer 1

Abstract

Despite extensive treatment options involving surgical resection as well as radio- and chemotherapy, glioblastoma remains the most common and most aggressive intrinsic brain tumor. The current gold standard treatment includes systemic application of the alkylating chemotherapeutic agent temozolomide (TMZ). Side effects of systemic therapy are common and limit the treatment efficacy. Local tumor control with targeted strategies offer higher intratumoral concentration and homogenous distribution of agents. The recent disposability of soluable TMZ allows local treatment of brain tumors. In this study we hypothesize that local, peritumoral therapy with temozolomide leads to a significant longer survival compared to systemic therapy and to a significant reduction of tumor volume, metabolism, as well as side effects. We investigated feasibility and safety of convection enhanced application in a murine brain tumor model. Three groups of 8 male nude rats were implanted with human U87 glioma cells. Cells were stereotactically administered into the caudate nucleus via a burr hole at day one. One control group received no treatment. The second group was treated with intraperitoneal TMZ for systemic chemotherapy on day 8 after tumor implantation. The third group received stereotactic guided peritumoral administration of TMZ. Initial proof of tumor growth and follow up controls before and under treatment were gained with invasive and noninvasive micro-in vivo imaging. Animals meeting the abort criteria were perfusion fixed and histopathologically evaluated. Our results indicate a significantly longer overall survival of animals in the local peritumoral treatment group compared to systemic intraperitoneal treatment. The control group without treatment showed a significantly lower survival. Tumor volume and metabolism was markedly lower in the local treatment group. Preliminary results enhance the safety and efficacy of convention enhanced delivery of soluable TMZ. These results confirm the importance of local treatment in human glioma and may be a promising local treatment opportunity.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii235–iii241.

TM-005. MODELING TUMOR MICROENVIRONMENT IN VIVO: IMPACT OF CONDITIONAL MACROPHAGE ABLATION IN GLIOMA PROGRESSION

Konrad Gabrusiewicz 1, Nahir Cortes-Santiago 1, Xuejun Fan 1, Mohammad B Hossain 1, Bozena Kaminska 2, Amy Heimberger 1, Ganesh Rao 1, WK Alfred Yung 1, Frank Marini 3, Juan Fueyo 1, Candelaria Gomez-Manzano 1

Abstract

Modeling tumor microenvironment is one of the most challenging areas of research due to the paucity of models to study its inherent complexity. Monocyte/macrophages are an influential component of glioma microenvironment characterized by their diversity and plasticity, which can undergo two polarization states representing extremes of a continuum: the classically activated M1 (proinflammatory) and the alternatively activated M2 (tumor progression) phenotypes. To characterize and study the role of monocytes/macrophages in glioma growth, we modeled the disease using a GFP+ macrophage Fas-induced apoptosis (MAFIA) transgenic mice. In this model, conditional macrophage ablation may be achieved by exposure to the AP20187 dimerizer. Analyses of the kinetics of a syngeneic intracranial glioma using this transgenic model, revealed the presence of GFP+ cells as early as 5d after cell implantation and, within 14d, these cells constitute a major component of the stroma. We identified this GFP+ population as M2 macrophages (Arg1 +), with a low representation of M1 macrophages (iNOS+). The genetic-guided depletion of macrophages showed effective reduction in the number of GFP+ cells (50-60%), targeting mainly Arg1+/GFP+ cells in the tumor periphery. Of interest, we observed agglomerations of iNOS+/GFP+ cells in the central tumor areas, following repopulation of GFP+ cells. Quantification of iNOS+/GFP+ cells revealed a ∼20-fold increase in their number when compared to control mice, suggesting an M2-M1 shift. Macrophage-depleted mice had significantly smaller tumors exhibiting lower mitotic index, when compared with controls. These results show the possibility of studying in vivo the role and phenotype of macrophages in gliomas, and suggest that transitory depletion of the macrophage population influences the phenotype of these cells, ultimately suppressing tumor growth. Furthermore, the MAFIA model constitutes a much-needed advance to define the role of macrophages in gliomas.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii235–iii241.

TM-006. EFFICIENT microRNA INHIBITION IN A HIGHLY INVASIVE ORTHOTOPIC GLIOBLASTOMA MODEL

Bo Halle 2,1, Eric Marcusson 5, Charlotte Aaberg-Jessen 1, Stine Skov Jensen 1,3, Morten Meyer 4,3, Mette Katrine Schulz 2, Claus Andersen 2, Bjarne, Winther Kristensen 1,3

Abstract

BACKGROUND: Up-regulated microRNA (miR) is a promising new target in glioblastoma therapy. Both in vitro and in vivo inhibition have been shown to diminish proliferation and invasion. However, most often the tumor models used are non-orthotopic and poorly resemble the highly invasive glioblastoma found in patients. Moreover the miR inhibition is done in a way that is not translational into clinical use. The purpose of this study was to determine if convection enhanced delivery (CED) of an antisense oligonucleotide could efficiently inhibit a miR in a highly translational tumor model. METHODS: Cells from a patient-derived glioblastoma stem cell line were transfected by free uptake for 1 and 5 days to test transfection efficacy and stability. Cells were also inoculated into the brains of nude rats. Post-tumor formation, a micro infusion pump was implanted and anti-miR-let-7 or saline was infused for 31 days. Then tumors were removed and tumor mRNA level of the well-validated miR-let-7 target High-mobility group AT-hook 2 (HMGA2) was determined by PCR. RESULTS: Five days free uptake transfection in vitro was more efficient than 1 day. The transfection efficacy dropped at 48 hours. HMGA-2 was efficiently de-repressed in vivo. Residual anti-miR from the pump reservoirs was non-degraded, the animals showed no signs of adverse effects and their weight gain was stable. CONCLUSION: Intratumoral CED of anti-miR-let-7 in a highly translational glioblastoma model is well-tolerated, preserves the integrity of the anti-miR and causes efficient target de-repression. The result holds promise for future miR oligonucleotide anti-sensing in clinical neuro-oncology.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii235–iii241.

TM-007. NF2 + CDKN2A DEFICIENCY IS A COMMON AND ACTIONABLE GENOTYPE OF TUMORIGENIC MENINGIOMAS

Rintaro Hashizume 1, Yuichiro Ihara 1, Tomoko Ozawa 1, Andrew Parsa 1, Jennifer Clarke 1, Nicholas Butowski 1, Michael Prados 1, Arie Perry 1, Michael McDermott 1, David James 1

Abstract

We have developed a meningioma xenograft panel based on engrafting atypical and anaplastic meningioma surgical specimens. Molecular analysis performed to date shows that 3 of 4 tumorigenic specimens are both NF2 and p16 (CDKN2A) deficient. For two additional tumorigenic meningioma cell sources developed by others, IOMM-LEE and CH-157, CH-157 also shows NF2 and p16 deficiency. Thus, 4 of 6 meningioma xenografts we've examined have this combination of gene alterations, which has been described in meningiomas having undergone malignant progression (Clin Cancer Res. 2010;16:4155-64), and that have been used for developing genetically engineered mouse models of malignant meningioma (Brain Pathol. 2008;18:62-70). Thus, our results associated with the characterization of tumorigenic human meningioma cell sources are consistent with previously published results supporting NF2 + p16 deficiency as a powerful combination that promotes meningioma malignant behavior. Our primary interest in developing a meningioma xenograft panel is for testing therapeutic hypotheses, based on tumor molecular characteristics, and the observation of p16 deficiency presents opportunity to test meningioma xenografts for differential response to cdk4/6 inhibition. We have previously shown that Palbociclib (PD0332991), a highly-specific inhibitor of cdk4/6 enzymatic activity, is effective in delaying growth and extending survival of animal subjects with intracranial glioblastoma xenografts that are p16 deficient (Cancer Res. 2010;70:3228-38), and here we have conducted preliminary analysis of the in vivo response of NF2 + p16 deficient cell source SF8867 for response to Palbociclib following subcutaneous implantation. Our results show that daily oral dosing of Palbociclib significantly reduces the growth rate of subcutaneous SF8867 xenografts. An experiment for testing meningioma response to Palbociclib following skull base implantation of tumor cells is ongoing. Our results to date support molecular characterization of tumorigenic meningioma xenografts as providing information for rational preclinical testing of therapeutics for treating malignant meningioma in patients.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii235–iii241.

TM-008. PRELIMINARY STUDIES TOWARD DEVELOPMENT OF A RODENT MODEL OF POST-TREATMENT RADIOGRAPHIC IMAGING CHANGES, SO CALLED “PSEUDOPROGRESSION”

Randy Jensen 1, David Gillespie 1

Abstract

INTRODUCTION: The lack of a definitive imaging modality to distinguish post-treatment radiographic imaging changes (PTRIC), including pseudoprogression and radiation necrosis, from true tumor progression presents a major unmet clinical need in the management of GBM patients. At present there are no animal models of PTRIC. METHODS: We developed a rodent model of PTRIC by intracranially implanting rats with either preirradiated (2-10 Gy) or nonirradiated glioma cells transfected with a constitutively active luciferase reporter. The rats were treated with or without temozolomide (TMZ) chemotherapy and rats implanted with nonirradiated cells were treated with focused tumor irradiation (2-10 Gy) to the implanted tumor cells and surrounding brain. Cell growth was monitored by bioluminescence imaging of the luciferase reporter and brain MRI measurement of tumor growth. Tumor histology was confirmed by light microscopy and immunohistochemistry. RESULTS: Animals implanted with irradiated cells did not produce PTRIC even when co-treated with TMZ. Rats considered to have tumor progression demonstrated steadily increasing luciferase activity coupled with increasing gadolinium-enhanced tumor volume and were found in animals receiving radiation doses less than 2Gy X3 without TMZ chemotherapy. Conversely, pseudoprogression was manifest by decreasing luciferase activity with increasing gadolinium volume over time found in tumors treated with 2Gy X3 and concurrent TMZ. Tumor histology was correlated with luciferase activity. CONCLUSIONS: Our preliminary data suggests a potential model of pseudoprogression and/or radiation necrosis. This model needs to be further characterized with advanced imaging techniques including perfusion MR and multitracer PET imaging. We plan to use this model for future molecular and cellular studies to understand the cause and potential exploitation of these effects for better patient care.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii235–iii241.

TM-009. IN VIVO GROWTH OF HIGHLY INVASIVE GLIOBLASTOMAS DERIVED FROM STEM-LIKE CULTURES IS INHIBITED BY TARGETING THE IGF-1 RECEPTOR

Tobias Martens 1, Martin Zamykal 1, Manfred Westphal 1, Katrin Lamszus 1

Abstract

The insulin-like growth factor (IGF) signaling pathway is known to play an important role in many human cancers. In glioblastoma, tumor stem cells are believed to be involved in tumor growth and recurrence after treatment. IGF-1 was shown to stimulate proliferation and inhibit apoptosis in glioma cell lines. We used an IGF-1 receptor (IGF-1R) antibody (IMC-A12) to inhibit growth of glioma stem-like cells (GSC) in vitro and in vivo. In order to identify a glioma cell line with high expression of IGF-1R, flow cytometric analyses were performed on a panel of GSCs, as well as on conventional serum-cultured glioma cell lines. GSC lines displayed the highest expression levels of IGF-1R and thus were examined in the following functional experiments. IMC-A12 treatment reduced proliferation of GSCs in vitro. To evaluate IGF-1R targetability in vivo, xenograft tumors were established in the brains of nude mice using the GS-12 cell line, which forms highly invasive tumors resembling human glioblastomas. Eight weeks after GS-12 xenotransplantation, osmotic minipumps were implanted, in order to administer the IGF-1R antibody directly into the tumor inoculation site over a period of four weeks, after which the animals were sacrificed. Histological analyses and quantifications revealed that tumor growth was inhibited by 50% compared to vehicle-treated control tumors (p < 0.001). Immunohistochemistry showed that expression of the proliferative marker Ki-67 was reduced by 27% (p < 0.05). Apoptotic signals and blood vessel densities were not significantly different in treated tumors compared to controls. In conclusion, IGF-1R blockade in GSCs inhibits proliferation and tumor growth in vitro and in vivo. These results indicate that the IGF signaling pathway is relevant for brain tumor growth.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii235–iii241.

TM-010. IDENTIFICATION OF THERAPEUTIC TARGETS IN AN INTRACRANIAL XENOGRAFT MODEL OF PITUITARY ADENOMA

Eric Monsalves 1,2, Shahrzad Jalali 1,2, Toru Tateno 1,2, Shereen Ezzat 1,2, Gelareh Zadeh 1,2

Abstract

INTRODUCTION: Pituitary adenomas (PA) are common brain tumors yet the molecular mechanisms of PA pathogenesis are not well understood. Fibroblast growth factor receptor 4 (FGFR4) is a transmembrane kinase which can harbor a single nucleotide polymorphism (SNP) substituting a glycine (G) with an arginine (R) residue. This SNP is associated with aggressive and treatment-resistant cancers. Currently, no robust in vivo PA models harboring the FGFR4 SNP exist. Therefore, the aim of our study was to establish an intracranial xenograft model of PA and use this model to identify novel therapies. METHODS: We generated stable cell lines using mammosomatotroph cells transfected with either FGFR4-G388 or FGFR4-R388 and used these cells to establish an intracranial mouse model of PA. Longitudinal MRI imaging of the mice post-injection was performed to obtain tumor volumes and growth rates. Histological analysis was conducted to elucidate expression patterns of hormones and mTOR pathway signalling in response to FGFR4-G388 or FGFR4-R388. RESULTS: Volumetric analysis revealed that the tumor growth pattern was exponential in all groups. Specifically, controls had a shorter latency of tumor growth while G388 and R388 had a longer latency and faster growth rate thereafter. Histologically, we found that the FGFR4-G388 tumors express predominately prolactin, while FGFR4-R388 tumors express growth hormone. We next looked at downstream effectors of the FGFR signaling pathway, specifically, we observed an upregulation of the mTOR pathway as demonstrated by increased p-s6 and p-4EBP1 expression. CONCLUSION: Our findings highlight the importance of the FGFR4 allele in PA and identify a possible therapeutic target for this tumor. Our data suggests that an upregulation of mTOR signalling may explain differences in tumor growth and hormone production. Our ongoing studies are evaluating the therapeutic potential of drugs targeting the mTOR pathway.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii235–iii241.

TM-011. LONGITUDINAL 18F-FET MICROPET IMAGING TO EVALUATE TREATMENT RESPONSE IN HUMAN GLIOBLASTOMA MULTIFORME XENOGRAFTS

Mette Kjølhede Nedergaard 1,3, Karina Kristoffersen 3, Hans Skovgaard Poulsen 3,4, Marie-Thérése Stockhausen 3, Ulrik Lassen 2, Andreas Kjær 1

Abstract

BACKGROUND AND AIM: Clinical and pre-clinical brain tumor imaging is challenging. Although 18F-FET PET is already widely used in the clinic the role of 18F-FET to evaluate brain tumor growth and treatment response in glioblastoma multiforme (GBM) xenografts has not been assessed to date. The aim of this study was to evaluate the performance of in vivo 18F-FET MicroPET imaging in monitoring tumor growth and in detecting a response to chemotherapy in GBM xenografts. METHODS: Human GBM cells (NGBM_CPH048p6, 105 cells) were injected orthotopically in athymic nude mice. In vivo uptake of 18F-FET was followed by weekly serial MicroPET imaging to monitor tumor growth as a function of time. Ten minutes (min) static images were obtained at 20-30 min postinjection of approximately 10 MBq 18F-FET in the tail vein. Tracer uptake was expressed as a tumor-to-brain ratio (T/B) of SUVmax. Mice were included in the treatment study at tumor take, which was defined as a T/B ratio of 1.3 or above. At tumor take mice were treated with irinotecan or saline weekly. Mice were sacrificed when weight loss was above 20% or after 2 weeks of treatment. RESULTS: A statistically significant difference in relative T/B ratio between the irinotecan (n = 8) and the saline group (n = 6) was detected after one week (mean T/B ratio 117% ± 1.6 vs. 136% ± 5.1, p = 0.009) and after two weeks of treatment (mean T/B ratio 142% ± 7.6 vs. 188% ± 18.8, p = 0.04). CONCLUSIONS: 18F-FET MicroPET can be used to follow tumor growth and to detect a treatment response to irinotecan in GBM xenografts. Investigations are ongoing to elucidate the performance of 18F-FET to detect a response to anti-angiogenic treatment and updated date will be presented. 18F-FET MicroPET can possibly be used to test new anticancer agents in GBM models in the future.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii235–iii241.

TM-012. INTERPLAY BETWEEN GENETIC LOSS OF p53, Nf1 AND HISTONE MODIFICATIONS IN TUMORIGENESIS OF GLIOBLASTOMA

Fumiharu Ohka 1,2, Atsushi Natsume 2, Hui Zong 3, Chong Liu 3, Akira Hatanaka 1, Keisuke Katsushima 1, Keiko Shinjo 1, Toshihiko Wakabayashi 2, Yutaka Kondo 1

Abstract

Recently, interplay/crosstalk between genetic alterations such as IDH-1, ATRX, H3F3A mutations and epigenetic alterations has been demonstrated. Here we investigate interplay between genetic loss of p53, Nf1 and histone modifications during glioblastomagenesis. MADM (Mosaic Analysis with Double Markers)-based glioma model, initiating concurrent p53/Nf1 loss sporadically in mouse neural stem cell, generates glioblastoma (GBM) which displays high similarity to proneural GBM in TCGA sample set (Zong et al. Cell, 2011). MADM system enables us to trace pre-transformed cells during tumor formation by GFP (Green Fluorescent Protein). Recently many evidence revealed that the accumulation of epigenetic alterations such as aberrant DNA methylation (G-CIMP) and histone modifications may play important roles in a stepwise proneural GBM formation. Using MADM mice, we analyzed alterations of histone modifications in pre-transformed cells (Postnatal 8 days, P8), tumor cells (P150). Tumor cells showed high expression of Ezh2 (Enhancer of zeste homolog 2), histone H3 lysine27 (H3K27) methyltransferase. We performed genome-wide ChIP-chip (Chromatin Immunoprecipitation and microarray based analysis in total 25,840 genes) for histone H3 trimethyl lysine4 (H3K4me3), active marks, and histone H3 trimethyl lysine27 (H3K27me3), repressive marks. This analysis revealed that a large number of genes are repressed by H3K27me3 (2,561 genes) and bivalent modifications (1,126 genes) in tumor cells. Most of these repressed genes are specific in tumor (60.9%), not seen in P8 and included many genes, regulating neural differentiation (p < 0.001). Almost one third of genes with tumor specific H3K27me3 are also the targets of p53 binding (31.8%). These data suggested that upregulated Ezh2 and loss of p53 binding, in some coordinated manner, accumulate aberrant H3K27me3 modifications and play important roles in proneural GBM formation via dysregulation of differentiation. Our data may provide novel insights into the interplay between genetic alterations and aberrant histone modifications during tumorigenesis of proneural GBM.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii235–iii241.

TM-013. DEVELOPMENT OF AN INDUCIBLE MURINE MODEL OF MEDULLOBLASTOMA

Katie Picotte 1, Li Li 1, Brian Westerhuis 1, Haotian Zhao 1

Abstract

INTRODUCTION: Medulloblastoma is comprised of four main subgroups, one involving mutations of the sonic hedgehog (Shh) pathway. We are developing a novel mouse model to induce overexpression of the transcription factor Atoh1, which enhances Shh-dependent proliferation of granule neuron progenitors and has shown to promote medulloblastoma formation when overexpressed. In this study, we generated three different models of transgenic mice to explore which would provide a strong model to pair with additional, established strains to study tumorigenesis. METHOD: Three different transgenic models were explored to produce mice with overexpression of Atoh1 using the Cre-lox system. All models were bred to a Math1-Cre or Atoh1-CreER transgenic strain. The first model used the CMV-LSL-Atoh1-IRES-LacZ (A1Z) construct. The second construct was modified only at the promoter region to CAG-LSL-Atoh1-IRES-LacZ (CAG-A1Z). A third construct involved an HA-tagged Atoh1 cDNA inserted into the Rosa26 locus; LacZ was also changed to GFP (A1G). Analysis involved whole mount X-gal staining of the brain, histological examination, and immunoblotting. RESULTS: Using the CMV promoter, LacZ expression was not detected in the cerebellum of the A1Z strains. Modification to the CAG promoter in the CAG-A1Z line, however, has resulted in strong LacZ expression in the cerebellum. Using immunoblotting, differences in LacZ expression levels are seen between the five strains. One strain using the third construct (A1G) was able to produce offspring and has shown positive GFP expression in the cerebellum. CONCLUSIONS: Out of the three transgenic constructs, the CAG-A1Z and A1G lines have shown positive results. To analyze Atoh1 transgene function in these two models, tumor cell proliferation, differentiation, and GNP survival will be followed at different time points using immunoblotting and immunohistochemistry. Whichever model provides optimal Atoh1 overexpression will be further analyzed by breeding to additional strains to follow tumor growth.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii235–iii241.

TM-014. CRITICAL ROLE FOR NF-κB INHIBITORS IN MENINGIOMA CELL DEATH

Scott Plotkin 1, Marianne James 1, Michel Kalamarides 2, Wen-ning Zhao 1, James Kim 1, Anat Stemmer-Rachamimov 1, Stephen Haggarty 1, James Gusella 1, Vijaya Ramesh 1, Fabio Nunes 1

Abstract

BACKGROUND: Meningiomas are common brain tumors accounting for over 33% of all human CNS neoplasms. Meningiomas are caused by loss of function of the NF2 gene in 50-60% of cases; in non-NF2 meningiomas, mutations in TRAF7 and KLF4 occurs in almost 25% of meningiomas and mutations in SMO and AKT occur in an additional 5%. Currently, there are no medical treatments for meningiomas that progress despite surgery or radiation. METHODS: Using a primary human meningioma cell line, we performed a chemical screen of over 2000 FDA-approved compounds to identify drugs capable of promoting meningioma cell death. Lead compounds were selected based on their ability to decrease cell viability luminescence signals by ≥70% compared to untreated controls. Human and MGS3 murine meningioma cell lines were then treated with lead compounds for 24 hours, and growth inhibition was determined 72 h after treatment. For in vivo studies, adult FVB wild-type mice were injected with 7-10 µL of MGS3 cells in the subdural space. Drug treatment was started 3 weeks after MSG3 implantation. RESULTS: We identified two compounds that were able to promote cell death in benign, atypical, and malignant meningioma cell lines in vitro: BAY 11-7082, an inhibitor of the nuclear factor-kappaB (NF-κB) pathway (EC50, 2-6 mM), and bortezomib (Velcade, Millennium Pharmaceuticals), an FDA-approved proteasome inhibitor (IC50, 2-80 nM). Incubation of MGS3 cells with bortezomib resulted in a dose-dependent growth inhibition (IC50, 10.9 nmol/L). Mice with meningioma implants were treated either with bortezomib (1mg/kg, intraperitoneal twice a week, n = 7) or saline (n = 7). Quantitative scoring of microscopic meningioma volume showed that bortezomib-treated mice had reduced tumor volume compared with controls (median tumor volume: 168 vs 406.6 mm2 for bortezomib and control, respectively; p = 0.0095, Mann-Whitney test). CONCLUSIONS: Our results suggest that NF-κB and proteasome inhibitors are a reasonable option for clinical trials of meningiomas.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii235–iii241.

TM-015. SURVIVIN TRANSCRIPT VARIANT 2 DRIVES ANGIOGENESIS AND MALIGNANT PROGRESSION IN PRONEURAL GLIOMAS

Ganesh Rao 1, Tiffany Doucette 1, Yuhui Yang 1, Gregory Fuller 1, Arvind Rao 1

Abstract

INTRODUCTION: Survivin expression correlates with poor prognosis in high-grade glioma. The influence of Survivin isoforms on clinical outcome is less well understood. We analyzed the dominant anti-apoptotic transcript variants of Survivin using expression data and modeled them in vivo. METHODS: TCGA mRNA expression data was used to study Survivin in low- and high-grade gliomas including proneural(PN) and mesenchymal(Mes) subtypes. The anti-apoptotic isoforms of Survivin (transcript variants 1 and 2) were modeled in vivo using the PDGFB-dependent RCAS/Ntv-a system. RESULTS: In low-grade glioma, increased expression of Survivin occurred in 22/167(13.2%) cases, and was associated with shortened survival (log rank test, p = 0.005). In high-grade gliomas, increased expression occurred in 258/547(47%) cases. Variant 1 was expressed in 215/258(83%) cases, and variant 2 was expressed in 26/258(10%) cases. Median survival was 451 days and 382 days for variant 1 and 2 expressors respectively. Survivin was preferentially expressed in PN relative to Mes gliomas(chi squared, p < 0.0001). In PN gliomas, Survivin was expressed in 94/141(67%) cases and correlated with shorter progression free survival(log rank test, p = 0.04). In PN gliomas, median survival was 447 days and 394 days for variant 1 and variant 2 expressors respectively. In vivo ectopic expression of variant 1 yielded tumors in 28/30(93%) mice and only 7/28(25%) were high-grade. Ectopic expression of variant 2 yielded tumors in 27/28(96%) of mice and the majority (22/27,(81%)) were high grade(chi squared, p < 0.0001). CD31 expression (to quantify angiogenesis) was significantly higher in variant 2-derived tumors than variant 1(t test, p < 0.0001). Tumor-free survival was shorter in mice with variant 2-derived tumors(log rank test, p = 0.01). CONCLUSIONS: Survivin expression in low-grade gliomas is associated with poor survival and is significantly more common in PN gliomas. Poorer survival correlates with Survivin variant 2, relative to the common isoform. In a PN murine glioma model, variant 2 promotes malignant progression, angiogenesis and shortened tumor-free survival.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii235–iii241.

TM-016. A CLINICAL, BIOLOGICAL AND TRANSCRIPTIONAL ANALYSIS OF 12 CASES OF INTRAVENTRICULAR NEUROCYTOMAS

Nils Ole Schmidt 1, Niklas Humke 1, Hildegard Meissner 1, Franz-Josef Mueller 2, Manfred Westphal 1

Abstract

Neurocytomas are rare tumors of the CNS usually arising in the ventricular system. Although surgically resectable and normally benign, these tumors appear to have a higher incidence of recurrence during long-term follow-up than previously thought. Adjuvant therapeutic concepts based on a thorough understanding of their biology and clinical course are clearly needed. Clinical data of 12 patients with intraventricular neurocytoma treated and followed at our center were analyzed with regard to presentation, treatment, clinical course and outcome (mRS, SQLI). We developed an isolation and culture protocol to generate stable neurocytoma cell cultures (n = 5). Cultures were assessed for growth characteristics, transcriptional profile (whole genome microarrays) and comparative immunohistochemical expression of 12 lineage markers before and after differentiation-inducing conditions. Sensitivity of cells to chemotherapeutic agents temozolomide, BCNU, topotecan, carboplatin and ifosfamide was assessed using cytotoxicity assays. We performed surgical resections in all patient with tumors localized in the lateral (n = 14) or third ventricle (n = 2). Surgical morbidity was low and overall outcome was good (mean mRS of 1.4, SQLI of 8) with only one death due to rapid tumor progress. Recurrence occurred in 4 (33%) patients independent of the extent of resection and 5 (42%) patients received postoperative radiation but no chemotherapy. Neurocytoma cultures allowed us to detect a significant chemosensitivity for topotecan, ifosfamide and its combinations but resistance to the other agents. Transcriptionally different from other human neural stem cells lines lineage analysis in vitro revealed that neurocytomas display a preferential neuronal antigenicity but were able to differentiate in astroglial and unexpectedly into oligodendroglial cells demonstrating multipotentiality. Surgical resection is safe and with a favorable outcome but the high recurrence rate warrants long term follow-up and the establishment of adjuvant therapeutic protocols. This is the first report of established long-term cultures that can be used to study the biology of neurocytomas.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii235–iii241.

TM-017. CELL-CELL INTERACTIONS BETWEEN GLIOMA CELLS AND BONE MARROW MESENCHYMAL STEM CELLS DETERMINE EFFECTS OF INTEGRIN INHIBITORS IN VITRO

Oliver Schnell 1, Irene Jaehnert 1, Valerie Albrecht 1, Peng Fu 1, Jörg-Christian Tonn 1, Christian Schichor 1

Abstract

OBJECTIVE: Integrins are involved in a magnitude of cell-cell interactions. Here, we analyzed expression of integrins as well as the functional impact of integrin inhibition on glial tumor cells and possible recruited host cells like mesenchymal stem like cells. METHODS: Tumor cells from malignant glioma (U87-MG) as well as bone marrow derived mesenchymal stem cells (bmMSC) of healthy donors were analyzed for expression of integrin αvβ3. To determine if this integrin is involved in cell adhesion or proliferation of U87-MG alone or in combination with bmMSC, integrin-inhibitors (INIH) were administered simultaneously (before seeding) or delayed (after seeding and adhesion) to cell cultures of U87-MG and bmMSC with or without conditioned tumor medium. RESULTS: Whereas integrin subunit αv showed higher expression in bmMSC compared to U87-MG, no significant difference could be detected for subunits β3 and β5. Regarding functional impact, U87-MG cell cultures displayed almost complete loss of adhesion with small dosages of INIH. Moreover, almost complete loss of adhesion could be found in U87-MG independently, if INIH were administered simultaneously with or delayed after seeding. Simultaneous addition of INIH with cell seeding led to a dose-dependent loss of adhesion in bmMSC. Yet, delayed addition of INIH did not lead to a significant loss of adhesion in these cells. Addition of tumor-conditioned medium completely reversed loss of adhesion in bmMSC treated simultaneously with INIH. For bmMSC cell cultures, reduced proliferation rates could be detected already at 24h after addition of INIH, whereas INIH displayed no influence to proliferation rates in U87-MG cells. CONCLUSION: Functional impact of integrin αvβ3 can not be predicted sufficiently by integrin expression alone since inhibition of integrin αvβ3 can be overcome by glioma secreted factors. Differential response of glioblastomas to INIH treatment should be interpreted on the background of intercellular crosstalk between tumor and recruited host cells.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii235–iii241.

TM-018. BarTeL, A GENETICALLY VERSATILE AND EFFICIENT BIOLUMINESCENT MOUSE MODEL FOR MEDULLOBLASTOMA

Gregory Shackleford 1,2, Kimberly Swanson 1, Xiang-He Shi 1, Massimo D'Apuzzo 3, Ignacio Gonzalez-Gomez 1,2, Richard Sposto 1,2, Robert Seeger 1,2, Anat Erdreich-Epstein 1,2, Rex Moats 1,2

Abstract

We report a new transgenic mouse model for medulloblastoma – BarTeL – designed with a unique set of desirable characteristics and capabilities. The transgene utilizes a Barhl1 promoter, which is transiently expressed primarily in cerebellar granule neuron precursor cells (cGNPs) of neonatal mice. This promoter drives the bicistronic expression of the avian Tva gene, which encodes the receptor for avian sarcoma-leukosis viruses including RCAS retroviral vectors, together with an eGFP-luciferase fusion protein. Thus, tumors are initiated by the neonatal infection of BarTeL mouse cGNPs with RCAS vectors carrying genes-of-choice. As proof of concept, we produced bioluminescent tumors with a preprocessed version of sonic hedgehog (Shh) together with a stabilized Mycn mutant. These tumors continue to express the transgene and develop on a decreasing – and eventually negative – bioluminescent background, thus allowing early identification of tumor-bearing animals and monitoring of tumor growth and regression in oncogene evaluation studies and in preclinical studies of therapeutics. The retroviral delivery of genes into the model allows modeling of somatically acquired mutations in a context of a normal cellular environment. We further developed a bicistronic, IRES-containing, RCAS vector for introduction of the Shh and Mycn genes. This vector produced medulloblastomas at 100% efficiency in the model. The expression of eGFP in normal neonatal cGNPs and in tumor cells enables FACS purification of cGNPs and primitive tumor cells. We modeled the Shh subgroup of medulloblastomas here, but recent studies have shown that cGNPs are a cell-of-origin for Group 3 tumors and that BARHL1 is highly expressed in Group 4 tumors, suggesting that BarTeL mice may be able to model these subgroups as well. Thus, this single BarTeL mouse model provides a relatively wide array of opportunities for studying medulloblastoma biology and therapeutics.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii235–iii241.

TM-019. TISSUE ENGINEERED SCAFFOLDS FOR STUDYING GLIOBLASTOMA INVASION IN 3D

Rachael W Sirianni 1,2, John M Heffernan 1,2, Derek J Overstreet 2

Abstract

Standard invasion assays (e.g., Boyden chamber) present cells with 2D surfaces that are highly dissimilar to native tissue. The necessity of culturing cells in 2D to analyze their movement poses a particular challenge to studying patient-derived brain tumor cells, since critical molecular phenotypes are lost in extended 2D culture. Here we present two biomaterial platforms suitable for high-content analysis of invading cells in 3D. In the first system, hyaluronic acid (HA) and gelatin are cross-linked with poly(ethylene glycol) divinyl sulfone. By altering the total polymer density and the ratio of HA to gelatin, we have made and characterized a library of scaffolds with tunable properties (stiffness, porosity, density of cellular adhesion sites) to emulate key features of the native tumor microenvironment. GFP-expressing glioblastoma spheroids were cultured on the surface of the scaffold, and invading cells were readily observed by automated confocal microscopy. Invasion phenotype depended on microenvironmental cues, including scaffold properties and the presence of chemokinetic factors. In the second system, poly(N-isopropylacrylamide) modified with RGD is thermally gelled to form a scaffold that supports cell invasion at 37°C. Once the desired period of time has elapsed, the spheroid is plucked from the surface of the gel, and the scaffold rapidly re-liquifies upon mild cooling. This reversibly gelable scaffold enables us to enrich and collect invading tumor cells for molecular analysis in direct comparison to the spheroid (parent) population. Both sets of materials possess working times in the range of minutes and can be cast in well-plate format. Importantly, the semi-synthetic schemes permit long-term assay of cell invasion with precise control over batch to batch properties. One novel feature of our new set of materials is that they are hydrolytically stable, forming scaffolds that supported cell invasion for long periods of time (i.e., weeks).

Neuro Oncol. 2013 Nov;15(Suppl 3):iii235–iii241.

TM-020. SULFASALAZINE SENSITIZES GLIOBLASTOMA CELLS TO RADIATION TREATMENT

Linda Sleire 1, Bente Sanvei Skeie 1,3, Inger Anne Netland 1, Jan Heggdal 2, Paal-Henning Pedersen 3, Per Øyvind Enger 1

Abstract

Glioblastoma (GBM) is a lethal cancer with a limited response to ionizing radiation. Recent studies suggest that Sulfasalazine (SAS), a drug used to treat inflammatory bowel disease, inhibits the Xc− antiporter system in glioma cells, thereby blocking their uptake of cystein. This blocking decrease intracellular antioxidant levels, and we wanted to investigate whether sulfasalazine sensitizes glioma cells to radiation. Expression of system Xc− was found in 28 patient GBM biopsies. All glioma cell lines showed altered growth in response to SAS treatment. With increasing doses of SAS, glutathione levels decreased in a dose response manner. In addition, cysteine was added to the medium to see if the toxic effects of SAS could be counteracted. Glioma cells were also treated with escalating doses of SAS, alone or in combination with radiation (4 and 8 Gy). The presence of double stranded breaks increased markedly in the irradiated samples and also with increasing doses with SAS. In addition, cell death, viability and clonogenicity were investigated using live/dead staining, the MTS assay and the clonogenic assay. All treatment groups exhibited increased rates of cell death compared to untreated controls. A combination of SAS and radiation resulted in higher levels of cell death than radiation or SAS administered alone. Furthermore we continued with implantation of human GBMs into the brain of Nude rats. These animals were treated with Gamma Knife Radiosurgery alone or in combination with SAS. SAS were administered as a pre-treatment before Radiosurgery. The rats receiving the combination treatment lived significantly longer compared to either treatment alone. Interestingly, the animals only receiving pre-treatment with SAS lived longer compared to the untreated controls, although this was not statistically significant. We are currently preparing a clinical trial for patients with GBM recurrences combining pre-treatment with Sulfasalazine and Gamma Knife Radiosurgery.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii235–iii241.

TM-021. HDAC INHIBITORS: A SYNTHETIC LETHAL TOOL FOR TARGETED THERAPY OF PLGAs?

Charles Stiles 1,2, Yu Sun 1,2, Shwetal Mehta 1,2, Christopher Taylor 1,2, John Alberta 1,2

Abstract

The p53 tumor suppressor mediates cellular responses (growth arrest and/or apoptosis) in response to ionizing radiation or genotoxic drugs. In previous studies, we have shown that the neurogenic transcription factor Olig2 opposes these p53-mediated responses to DNA-damaging agents in neurosphere cultures from high-grade adult gliomas. However, a second function p53 is to induce growth arrest or apoptosis in response to inappropriate mitogenic cues (i.e. oncogenes). We show here that this second function of p53 is also suppressed by Olig2. The findings are relevant to pediatric low-grade astrocytomas (PLGAs) characterized by activating mutations in BRAF. METHOD: A high percentage of juvenile pilocytic astrocytomas express the KIAA 1549 truncation fusion form of BRAF. Other PLGAs (e.g. gangliogliomas, diffuse astrocytomas, pleomorphic xanthoastrocytomas) express either the V600E point mutation of BRAF or the KIAA 1549 mutant. Notably these two versions of the BRAF oncogene are mutually exclusive in PLGAs. We used retroviral expression vectors to introduce these two oncogenic versions of BRAF into either neural progenitor cells isolated from wild type or Olig2 null mouse embryos. RESULTS: Neural progenitors from wild type mouse embryos (Olig2-positive) tolerated transformation with either form of the BRAF oncogene. However, both forms of the BRAF oncogene were toxic to Olig2 null progenitors. CONCLUSIONS: Genetic ablation of Olig2 in murine neural progenitors is synthetic lethal to transformation by oncogenic mutants of BRAF. The synthetic lethal relationship between Olig2 and BRAF oncogenes could have practical overtones for targeted therapy of BRAF-transformed PLGAs - the majority of which are wild type with respect to p53 status and are Olig2-positive.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii235–iii241.

TM-022. QUANTITATIVE MRI TRACKING OF MELANOMA BRAIN METASTASIS IMPROVES MOUSE MODEL PREDICTIVITY

Terje Sundstrøm 1,2, Ingvild Wendelbo 1, Inderjit Daphu 1, Erlend Hodneland 1, Arvid Lundervold 1, Heike Immervoll 3, Kai Ove Skaftnesmo 1, Michal Babic 4, Pavla Jendelova 4,5, Eva Sykova 4,5, Morten Lund-Johansen 1,2, Rolf Bjerkvig 1,6, Frits Thorsen 1

Abstract

Reproducible and predictive animal models are needed to unravel the biologic conundrum of melanoma brain metastasis, to functionally validate new therapeutic targets, and to bridge the gap between preclinical and clinical efficacy studies. Notably, most existing data on drug efficacy against brain metastases are generated from mouse models where it is impossible to know with certainty whether or not the mice have had an adequate tumor cell exposure to the brain. With this perspective, we developed a robust model of brain metastasis that enables tracking of cellular distribution and tumor development (Cancer Res 2013; 73; 2445–56). Human melanoma cells were labeled with superparamagnetic iron oxide nanoparticles (SPIONs) and injected intracardially in immunodeficient mice. Short T2*-weighted MRI sequences were used to visualize the labeled cells as hypointensive signals within the brain parenchyma. SPION exposure did not impose changes in cellular viability, directional migration, or spheroid growth across multiple cell lines; neither did it affect cell-cycle distribution or apoptosis. Furthermore, we showed in vivo that several labeled cell lines exhibited similar characteristics on brain MRI and histology. Fully automated MRI-based quantification of SPION-positive cells in the mouse brain showed a sigmoid relationship between metastasis frequency and the number of injected cells (104, 105, 5x105, and 106). Validation of this quantitative approach showed a strong concordance with the frequency of brain metastases (p < 0.001) and manually identified signals (p < 0.001). Metastasis development reflected the growth pattern observed in humans and was unaffected by SPION exposure (histology; tumor count, p = 0.686; survival, p = 0.547). Taken together, we present here a unique animal model that can improve upon the predictive value of biologic and therapeutic data gained from preclinical studies of melanoma brain metastasis, perhaps ultimately helping reduce the number of clinical trials that fail to show patient benefit.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii235–iii241.

TM-023. GDNF MEDIATES GLIOBLASTOMA-INDUCED MICROGLIA ATTRACTION BUT NOT ASTROGLIOSIS

Michael Synowitz 1,2, Min-Chi Ku 2, Susanne A Wolf 2, Dorota Respondek 2, Vitali Matyash 2, Andreas Pohlmann 2, Sonia Waiczies 2, Helmar Waiczies 2, Thoralf Niendorf 2, Rainer Glass 3, Helmut Kettenmann 2

Abstract

High-grade gliomas are the most common primary brain tumors. Their malignancy is promoted by the complex crosstalk between different cell types in the central nervous system. Microglia/brain macrophages infiltrate high-grade gliomas and contribute to their progression. To identify factors that mediate the attraction of microglia/ macrophages to malignant brain tumors, we established a glioma cell encapsulation model that was applied in vivo. Mouse GL261 glioma cell line and human high-grade glioma cells were seeded into hollow fibers (HF) that allow the passage of soluble molecules but not cells. The glioma cell containing HF were implanted into one brain hemisphere and simultaneously HF with non-transformed fibroblasts (controls) were introduced into the contralateral hemisphere. Implanted mouse and human glioma- but not fibroblast-containing HF attracted microglia and up-regulated immunoreactivity for GFAP, which is a marker of astrogliosis. In this study, we identified GDNF as an important factor for microglial attraction: (1) GL261 and human glioma cells secret GDNF, (2) reduced GDNF production by siRNA in GL261 in mouse glioma cells diminished attraction of microglia, (3) over-expression of GDNF in fibroblasts promoted microglia attraction in our HF assay. In vitro migration assays also showed that GDNF is a strong chemoattractant for microglia. While GDNF release from human or mouse glioma had a profound effect on microglial attraction, the glioma-induced astrogliosis was not affected. Finally, we could show that injection of GL261 mouse glioma cells with GDNF knockdown by shRNA into mouse brains resulted in reduced tumor expansion and improved survival as compared to injection of control cells.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii235–iii241.

TM-024. DEVELOPMENT OF APPROPRIATE MODELS FOR ASSESSING EFFECTS OF PROTEIN KINASE INHIBITON ON THE TYROSINE KINASE SIGNALLING NETWORK IN GLIOBLASTOMA

Nicola Thompson 1, Douglas Elder 1, Kirsten Hopkins 2, Venkat Iyer 3, Nicki Cohen 4, Jeremy Tavare 1

Abstract

Tyrosine kinase signalling networks are centrally involved in regulating an array of cellular activities including proliferation, protein synthesis and apoptosis. The up regulation of activity of some downstream components of these networks, including MAPK and STAT3, is frequently observed in a wide range of cancers, including glioblastoma multiforme, and is associated with treatment resistance and poor prognosis. Novel inhibitors targeted at several proteins lying within these networks are in clinical development, however have had mixed success to date in glioblastoma. We have devised a novel system for assessing the effect of these inhibitors on tyrosine kinase signalling networks. A tumour explant model, where small biopsies of glioblastoma tissue are placed in culture within 24 hours of theatre has been established and used to assess the effect of protein kinase inhibition on tyrosine kinase signalling networks through western blot analysis. We have compared this system with the response to inhibitors to primary cell lines derived from the same patients. We have found that a MEK inhibitor (Selumetinib) promotes consistent up-regulation of STAT3 phosphorylation in both the explant model and corresponding primary cell lines from the same patient. This has high clinical relevance given the role of STAT3 in glioblastoma in mediating angiogenesis and immune evasion. Our experiments further suggest specific combinations of protein kinase inhibitors that may be beneficial for use in glioblastoma. Our work shows that explant model provides a valuable method for rapidly assessing the effect of protein kinase inhibitors on signalling networks in patient biopsy tissue, and may have the potential to guide clinical treatment based on the biochemical response of the tumour implant model to drug treatment.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii235–iii241.

TM-025. EARLY TUMOR DETECTION AND CHARACTERIZATION OF CHANGES IN TUMOR PERMEABILITY OF BRAIN METASTASES USING MULTIMODAL IMAGING

Frits Thorsen 1, Brett Fite 2, Lisa M Mahakian 2, Jai W Seo 2, Shengping Qin 2, Victoria Harrison 3, Terje Sundstrøm 1,4, Patrick N Harter 5, Sarah Johnson 2, Elizabeth Ingham 2, Charles Caskey 2, Thomas Meade 3, Kai Ove Skaftnesmo 1, Katherine W Ferrara 2

Abstract

INTRODUCTION: Clinically, brain metastases are detected late. The prognosis is poor, suggesting that early detection of brain metastasis may improve therapy and survival. However, most of the chemotherapeutics do not cross the blood-brain barrier (BBB). Further, little is known regarding at which time point in tumor development the BBB is disrupted, and to what extent larger molecules are able to penetrate tumor tissue. We addressed these issues by using multimodal imaging to study early tumor detection and tumor permeability, after injecting various sized contrast agents into a melanoma brain metastasis model. MATERIALS AND METHODS: We injected tumor cells intracardially into nod/scid mice. All animals were followed by bioluminescence imaging (BLI) weekly for 6 weeks. T1 weighted (T1w) MRI was performed at weeks 3-6 after tumor cell inoculation, before and after injections of either Gd-HPDO3A (MW 559D) or a newly synthetized Gadolinium contrast agent with 3 Gd(III) chelates (MW 2.066kD). 18F-FLT PET (MW 244D) was performed at week 4, and 64Cu-Albumin PET (MW 66.5kD) at week 4 and 6. RESULTS: BLI confirmed tumor spread to animal brains after 1 week, and gradual increase in tumor burden until week 6. T1w MRI at week 4 detected non-leaky tumors down to 100 mm in diameter, while permeable tumors down to 200 mm sizes were detected at week 5 by T1w MRI after contrast injection. The two MRI tracers extravasated after 5 weeks. PET imaging showed that 18F-FLT accumulated in the brain at week 4, and demonstrated tumor permeability to 64Cu-BSA at week 6. CONCLUSIONS: High field T1w MRI without contrast may improve the detection limit of small brain metastases, allowing for earlier diagnosis of patients. Imaging contrast agents of various sizes can be used to demonstrate changes in tumor permeability during metastatic progression, which could guide clinicians in choosing more tailored treatment strategies.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii235–iii241.

TM-026. A REVERSE GENETIC SCREEN FOR GBM DRIVERS

Barbara R Tschida 1, Anita R Lowy 1,2, Courtney A Marek 1, Thomas Ringstrom 1, Thomas J Beadnell 1, Stephen M Wiesner 1, David A Largaespada 1

Abstract

INTRODUCTION: Glioblastoma multiforme (GBM) is the most common malignant primary brain tumor. It is highly invasive, greatly limiting safety and efficacy of surgery and radiation therapy, and no optimal chemotherapeutic regimen has been defined. Identifying molecular GBM drivers is needed to develop effective therapies to directly inhibit tumor growth or limit invasion, improving surgical efficacy. METHODS: By mining publically available data from comprehensive sequencing and gene expression analysis by The Cancer Genome Atlas (TCGA) Research Network and the Howard Hughes Medical Institute (Parsons et al 2008), we selected 50 genes overexpressed in a high proportion of GBM. We hypothesize a subset of these directly enforce tumor initiation or maintenance in most GBM. We screened these genes in vivo using Sleeping Beauty (SB) transposon-based gene delivery. Plasmids carrying SB transposon expression vectors for each of these cDNAs and SV40 large T antigen (TAg) were generated using Gateway® cloning (Invitrogen). These cDNA vectors were stereotactically injected into the lateral ventricles of neonatal mice in batches of 9, along with the TAg vector to facilitate tumor formation, and a SB expression plasmid containing a firefly luciferase transposon. Tumor development was monitored by luciferase imaging. RESULTS: A subset of mice developed tumors closely resembling human GBM. The majority of tumors were highly infiltrative with vascular proliferation, histologically resembling grade 3 GBM. HIF1a, a known infiltration driver, was associated with infiltrative tumors in our model. Of the 50 cDNAs tested, we identified 4 strong candidate GBM drivers, SOX2, RAB13, SHCBP1, and CCNE2. Over 10% of animals injected with them developed tumors containing detectable levels of these cDNAs. CONCLUSIONS: We have successfully modeled human GBM in a screen for common drivers. Of the 50 genes screened, we identified 4 candidates which we are testing singly and in combinations to further elucidate their roles in GBM.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii235–iii241.

TM-028. ELECTRIC FIELDS FOR THE TREATMENT OF GLIOBLASTOMAS: A MODELING STUDY

Cornelia Wenger 1, Pedro Cavaleiro Miranda 1, Abeye Mekonnen 1, Ricardo Salvador 1, Peter Basser 2

Abstract

Oscillating electric fields are being investigated as an adjunct and even an alternative to chemotherapy in the treatment of glioblastoma multiforme. According to in vitro studies, an electric field strength greater than 1 V/cm is necessary to achieve inhibition of tumor growth. Here we used a realistic head model constructed from MRI data to calculate the electric field distribution in the brain during the application of tumor treating fields (TTF). A realistically shaped head model was created from MRIs with a voxel size of 1x1x1 mm3. Images were segmented into five different tissue types: scalp, skull, cerebrospinal fluid (CSF), gray matter (GM), and white matter (WM). The electrode arrays and the current intensity used in both models mimicked as closely as possible a commercial device specifically designed for the treatment of tumors (www.novocure.com). The electric field was calculated for two electrode configurations using the finite element method (FEM). In both cases, the magnitude of the electric field exceeded 1 V/cm over large areas of the brain. The magnitude of the electric field is higher in the WM because its impedance is higher than that of GM. The low impedance of the CSF in the ventricles also affects the electric field in the nearby brain tissue. These calculations indicate that the electric field magnitude in some parts of the brain may be sufficiently high to arrest cell proliferation. However, the electric field is not uniform as it is affected by the distribution of tissue types, the location and orientation of interfaces between them, and their individual electrical properties. Patient specific FEM models based upon MRI data could provide a means to estimate the electric field in the tumor and to optimize its delivery. This new tool could be used in treatment planning, and for understanding outcomes of TTF therapy.

Neuro Oncol. 2013 Nov;15(Suppl 3):iii235–iii241.

TM-029. MESENCHYMAL STEM CELLS WITHIN BRAIN TUMOR PROVIDE NICHE FOR CANCER STEM CELLS

Jonghee Yoon 1,2, Honggeun Shin 1,2, Kyungsun Choi 1, Chulhee Choi 1,2

Abstract

Glioblastoma multiforme (GBM), the most common malignant brain tumor, is composed of highly heterogeneous population such as cancer stem cell (CSC) and cancer-associated cells. Recent data have showed that CSC is major component of tumor initiation, relapse, and progression. CSCs have similar property of normal stem cell such as pluripotency and self-renewal and can also initiate and produce bulk of tumors. Like normal stem cell, environments of CSCs called ‘cancer stem cell niche' play critical roles in maintenance of stemness. It has been reported that cancer-associated cells such as endothelial cells and mesenchymal stem cells (MSCs) have a vital role in maintenance of CSC niche and tumorigenesis. It has also been reported that GBM cells recruit MSCs into peritumoral region, and also have MSC-like property. However, relationship between CSCs and MSCs is still unclear. To investigate function of MSCs in GBM and the relation with CSCs, we established a xenograft model by injecting mixture of red fluorescence protein-tagged MSCs and green fluorescence protein-tagged CSCs originated from GBM into dorsal skinfold chamber, and followed up tumor progression using in vivo two-photon microscopy. We found that MSCs showed endothelial phenotype by forming microvasculature-associated structures and CSCs were located adjacent to organized MSCs. On the contrary, CSCs predominantly formed spheres in the avascular region in the absence of MSCs. Collectively; these results suggest that GBM-associated MSCs are responsible for maintaining tumorigenesis and stemness of CSCs by providing a CSC niche.

Articles from Neuro-Oncology are provided here courtesy of Society for Neuro-Oncology and Oxford University Press

RESOURCES