Year: 2020
Type: article
Abstract: In a recent paper in Cell, Yao et al. used the Mosaic Analysis with Double Markers (MADM) genetic system for granule neuron precursor (GNP) cell-specific simultaneous ablation of the tumor suppressor Tp53 and the expression of fluorescent label in a few cells in a Ptch1-heterozygote, Tp53-heterozygote mouse to track these clonally related populations during development of medulloblastoma.1 Lineage tracing revealed that tumor-associated astrocytes, as well as medulloblastoma cells, are derived from GNPs. Transcriptome profiling performed by Yao et al. demonstrated that GNP-derived, tumor-associated astrocytes clustered closer to normal astrocytes than GNP-derived tumor cells. Consistent with trans-differentiation of tumor precursors to tumor-associated astrocytes, serial orthotopic xenograft of human medulloblastoma cells into NOD/SCID mice revealed a subset of tumor-associated cells expressing human GFAP. Tumor-associated astrocytes from 6 of 21 human medulloblastoma samples with 9q deletion also carried this karyotype. Importantly, the GFAP+ cells were not stem cells, but differentiated astrocytes. The authors propose that this tumor-associated astrocyte population supports GNP proliferation and the progression of the pre-neoplastic lesion. In concordance with this hypothesis, the authors were able to demonstrate that tumors were much smaller in GNP-specific, IGFR1-knockout mice. Interestingly, the tumor-associated astrocyte population was not the direct source of IGF1, but rather the tumor-associated microglia produced the IGF1. These microglial cells were not derived from GNPs; however, the GNP-derived tumor-associated astrocytes produced interleukin 4 (IL-4), which in turn induced the tumor-associated microglia to produce IGF1. Taken together, the study by Yao et al. provides evidence for an intriguing concept that tumor cells and their microenvironment co-evolve throughout tumor development and progression. Albeit that GNPs, under physiological conditions, are thought to exclusively give rise to cerebellar granule neurons, it is important to note that GNPs from the external germinal layer have been demonstrated to express both glial as well as neuronal markers before losing their neuronal markers and expressing the differentiation marker S100β characteristic of differentiated astroglial cells, in cultures treated with Sonic hedgehog (Shh) and Bone Morphogenetic Proteins (BMPs).2 Thus, the program for deriving GNP from astrocytes may be a latent property of the tissue, perhaps to be activated in the context of damage. IL-4 is well known for its role in driving alternative activation of macrophages and tissue repair, especially in the context of helminth or toxin-induced tissue damage.3 In this sense, medulloblastoma may co-opt this highly regulated circuit. Notwithstanding, the study by Yao et al. might force a rethink on the identity of medulloblastoma specifically, but also of cancer in general—medulloblastoma and at least few other cancers may not just be an unorganized proliferative mass of undifferentiated or de-differentiated cells where every cell is in competition to outgrow the other, but rather a highly organized collective of tumor cells together with their support cells that manipulate the microenvironment to support their growth. One of the most common early questions from patients and families after a diagnosis of glioma or other brain tumor is whether diet, and in particular sugar intake, can affect the future growth of the tumor. Diabetes has been associated with increasing growth in some cancers, but not yet in glioma. Effects of glycemia on gliomagenesis and growth remain fairly murky, even in the extreme case of hyperglycemia with diabetes. The interactions of glioma with the immune system, both locally and systemically, are the subject of even more intense study in recent years; this has become increasingly urgent with efforts to boost the success of immunotherapy against glioma. A new report by Zhang et al. sheds new light on both areas and demonstrates interesting and complex interactions of hyperglycemia with the anti-glioma immune response.1 The authors investigated the effects of hyperglycemia with the use of two syngeneic, immunocompetent mouse glioma models in both the intracranial (orthotopic) and subcutaneous (heterotopic) settings and with streptozotocin-induced diabetes. With the GL261 glioma model, originally derived through chemical exposure, hyperglycemia increased the growth for both orthotopic and heterotopic tumors. Further study showed this to be mediated at least in part through the RAGE receptor, or Receptor for Advanced Glycation End products, which increase with hyperglycemia. It was also mediated by the RAGE ligand HMGB1 (High-Mobility Group Box 1), which, when released by cells undergoing an immunogenic cell death, acts as a strong immune danger signal (Damage-Associated Molecular Pattern, or DAMP). The second tumor model tested, K-luc, was derived from a genetically engineered mouse model with defined TP53 and NF1 defects, and its gene expression pattern matches that of the mesenchymal glioblastoma subtype. Interestingly, the effects of hyperglycemia diverged for K-luc in the orthotopic and heterotopic settings; growth was slower for orthotopic K-luc and faster for heterotopic K-luc. Importantly, the authors showed that the divergent responses were due at least in part to hyperglycemia decreasing the anti-glioma innate inflammatory response based in innate immune cells such as microglia and macrophages; suppressing these cells with a CSF1R inhibitor or clodronate liposomes also slowed intracranial growth of K-luc but not GL261. This study indicates that hyperglycemia and the innate anti-glioma inflammatory response can have divergent roles in different locations and in different glioma models—perhaps explaining the lack of clear outcomes in epidemiologic studies of hyperglycemia and glioma. The findings suggest that hyperglycemia may increase the growth of gliomas in some settings, but possibly not in those that get a boost from infiltrating innate immune cells that are suppressed by hyperglycemia. These intriguing results deserve further study, including expanding the focus to the ubiquitously used corticosteroids, which can drive hyperglycemia as well as suppress inflammation and immunity. As in other cancers, more effective disease models will be absolutely crucial for future investigations into the targetable aspects of glioma biology. Genetically engineered mouse models (GEMMs) have yielded considerable progress in their functional validation of oncogenic events, but the extent to which GEMMs sufficiently recapitulate the intratumoral heterogeneity characteristic of malignant glioma has been called into question. Patient-derived xenografts have provided an alternative approach that more broadly reflects the molecular intricacies of human glioma. However, such systems are inherently limited in their capacity for extensive genetic modification in the context of adequate experimental control. As described in their recent Nature Communications article, Koga et al. developed an innovative glioma model system based on induced pluripotent stem cells (iPSCs).1 A CRISPR/Cas9 approach was used to introduce glioma-associated molecular abnormalities—either combined NF1 and PTEN loss (NF1-/-; PTEN-/-) or TP53 inactivation coupled with the activating PDGFRA delta 8–9 deletion mutant (TP53-/-; PDGFRAΔ8–9)—into iPSCs, followed by differentiation into neural progenitor cells (NPCs). Subsequent orthotopic transplantation of these cells into immunosuppressed mice yielded malignant gliomas that recapitulated many of the histopathological features of their human counterparts. Moreover, iPSC-derived gliomas retained their respective foundational molecular alterations, could be cultured upon extraction from mouse brain, and could be re-transplanted. Consistent with fundamentally distinct underlying biology, NF1-/-; PTEN-/- gliomas demonstrated resistance to in vivo temozolomide treatment relative to their TP53-/-; PDGFRAΔ8–9 counterparts. Single-cell RNA-sequencing of the iPSC-derived gliomas and their NPC parental lines revealed transcriptional distinctions driven primarily by genotype, reflecting dichotomous mesenchymal and proneural signatures. However, intratumoral heterogeneity with regard to precise transcriptional profiles was notable, recapitulating a key aspect of the human disease. Interestingly, the gene expression patterns of TP53-/-; PDGFRAΔ8–9 gliomas exhibited broader diversity than those of NF1-/-; PTEN-/- gliomas coupled with increased evidence of genomic instability and extrachromosomal DNA. These latter findings underscore the genetic and biological plasticity of the iPSC-based modeling approach. The significance of these findings is considerable, as they lay the groundwork for improved preclinical models of malignant glioma that are both genetically tractable and biologically relevant. Group 3 medulloblastoma, one of the most common malignant pediatric tumors, often presents in early childhood and infancy. These tumors possess the worst prognosis of the medulloblastoma subtypes, are associated with MYC amplification, and often present with concomitant metastatic disease in the central nervous system. Surgery remains the first line of therapy followed by adjuvant chemotherapy. Due to the often young age of these patients, clinicians avoid recommending cranial spinal irradiation following surgery due to the devastating neurocognitive sequelae and therefore depend upon chemotherapy for disease control. The prognosis with current chemotherapy regimens, however, remains extremely poor and, hence, we continue to seek out new treatments. Consider this disease against the backdrop of current drug screening practices, which are complicated by the heterogeneity of the disease entity being studied and often depend upon conventional established cancer cell lines in two-dimensional culture. Additionally, the most common reason for clinical trial failure is lack of efficacy. Hence, drug screening utilizing the cancer organoid model that recapitulates the physiological features of an actual patient’s cancer are being sought for research, which is the topic of a recent paper by Ballabio and colleagues as it relates to modeling group 3 medulloblastoma.1 The authors performed an in vivo, patient-derived screen, identifying Otx2 and c-MYC as strong inducers of Group 3 medulloblastoma to better understand the disease and provide potential treatment options. These findings were then validated in human cerebellar organoids electroporated with Otx2/c-MYC. The resultant DNA methylation signature of these medulloblastoma-like organoids aligned with those of human Group 3 tumors. The authors went on to demonstrate reduction of Otx2/c-MYC oncogenesis in vivo and in human cerebellar organoids by SMARCA4. SMARCA4 and histone methyltransferase enzyme EZH2 possess an antagonistic relationship required for the growth and survival of SMARCA4-deficient tumors. Treatment with a non-specific EZH2 inhibitor, 3-dezaneplanocin, and an EZH2-specific inhibitor, tazemetostat, the latter of which is already in pediatric clinical trials, reduced Otx2/c-MYC-driven tumor formation in ex vivo culture and human cerebellar organoids. Such agents may provide a more promising therapeutic strategy for patients with Group 3 medulloblastoma, while the clinical use of cancer organoids will allow further advancement of precision medicine for difficult-to-treat tumors such as this.
Source: Neuro-Oncology
Language: English
Subfield: Genetics (clinical)
Field: Medicine
Domain: Health Sciences
Open Access status: bronze
OpenAlex ID: W4232461284