Search

Germline mutations in the NF1 tumor suppressor gene cause neurofibromatosis type 1 (NF-1), the most common human genetic cancer predisposition syndrome. NF1 patients are prone to develop peripheral and central nervous system tumors, with glioma occurring in 15-20% of patients. The molecular profiles of glioma from patients with NF-1 have been previously explored by large scale DNA and RNA bulk analyses, providing the first comprehensive genetic landscape of NF-1 glioma. However, the intra-tumoral architecture and the microenvironment organization remained largely concealed within the global bulk profile. Here, to characterize the complexity of molecular states of both malignant and non-malignant cells, we profiled NF-1 glioma gene expression at individual cell resolution. We performed single-nuclei RNA sequencing (Nucseq) on frozen NF-1 glioma specimens, including high-grade and low-grade tumors. The functional classification of the transcriptional programs activated in the individual tumor cells identified three main molecular states that partially overlapped with the cellular subtypes of sporadic glioblastoma. Tumor cells of low-grade NF-1 glioma exhibited marked enrichment of neuronal functions, suggesting that the gliomagenesis process in the context of NF1 inactivation might be supported by specialized neuronal activities. Conversely, the proliferative state characterized high-grade NF-1 glioma, indicating that cell cycle-associated pathways are drivers of tumor progression. The single-cell deconvolution of immune microenvironment in NF-1 low grade glioma revealed infiltration of CD8+ T lymphocytes and the active crosstalk of these cells with pro-inflammatory macrophages and mature dendritic cells, supporting a possible anti-tumor cell-mediated response. The identification of divergent tumor cell states as well as the characterization of the immune infiltrates provide new insights for accurate stratification of NF-1 glioma patients and to inform personalized therapeutics.

LZTR1, the substrate-specific adaptor of a CUL3-dependent ubiquitin ligase is among the most frequently mutated ubiquitin ligase coding gene in syndromic and sporadic human cancers including glioblastoma multiforme, in which approximately 27% of cases harbor inactivating mutations and copy number loss. However, both the identity of the protein substrates targeted by LZTR1-mediated ubiquitylation and the biological contexts regulated by specific LZTR1-substrate(s) interactions remain uncertain. Here, we combined biochemical and genetic studies to identify LZTR1 substrates and interrogated their tumor-driving function in the context of LZTR1 loss-of-function mutations and in a new conditional Lztr1 knockout mouse. Multiple screens converged on the receptor tyrosine kinases EGFR and AXL as LZTR1 interactors targeted for ubiquitin-dependent degradation in the lysosome by LZTR1-CUL3 complexes. Pathogenic mutations affecting LZTR1 failed to promote degradation of EGFR and AXL in human tumors. Mice harboring conditional deletion of the LZTR1 gene combined with loss of CDKN2A in the neural progenitor compartment generated peripheral nervous system tumors including schwannoma like and malignant peripheral nervous system tumors (MPNST). Tumors from the LZTR1-mutant mouse model accumulated very high levels of EGFR and AXL and exhibited potent and specific vulnerability to the combinatorial inhibition of EGFR and AXL kinases. These findings explain the mechanism of tumorigenesis associated with LZTR1 inactivation and offer a therapeutic strategy to patients affected by tumors carrying mutations of LZTR1.

Glioblastoma multiforme (GBM) is the most common and malignant type of primary brain tumor. Despite intensive multimodal treatment, the prognosis of GBM patients is poor with the median overall survival under 15 months. Defining the genetic profiles of rare long-term GBM survivors may provide insights into the molecular basis of GBM progression. IDH mutations have been identified as a favorable prognostic factors, and define a clinically and molecularly distinct group of GBM. Additional molecular mechanisms underlying the prolonged survival in IDHwt GBM patients remained unclear. Therefore, we explored the molecular profiles of 71 IDHwt GBM homogeneously treated patients exhibiting disproportionate survival. The cohort included 36 short-term survivors that deceased within 1 year after diagnosis (median survival of 7 months) and 35 long-term survivors exhibiting an exceptional survival above 4 years (median survival of 88 months). The differential genetic landscape of short- and long-term IDHwt GBM survivors outlined similar mutation load and few differences in the frequency of recurrent alterations, including chromosomes 19 and 20 duplication significantly overrepresented in long-term survivors. To assess the potential impact of T lymphocyte neoantigen-mediated activation towards extended GBM survival, we inferred the restricted HLA binding affinity of the somatic mutations and applied the immune fitness model to accurately predict high-quality neoantigens. Although the general neoantigen burden remained unvaried between short- and long-term survivors, we found that the immune quality of the predicted neoantigens is significantly higher in the patients with exceptional long survival. Moreover, the immune deconvolution of the tumor microenvironment from the gene expression data indicated the presence of CD8+ T lymphocytes and pro-inflammatory macrophages in samples harboring high-quality neoantigens from long-term survivors. This study identified the neoantigen quality as a main distinctive feature of exceptionally long-term survivors supporting the key role of cellular-mediated immune response to restrict tumor progression in IDHwt GBM.

Current transcriptomic classification of Glioblastoma Multiforme (GBM) has been ineffective to predict survival and therapeutic vulnerabilities. Recently, we proposed a four-group functional classification of GBM that included proliferative/progenitor, neuronal, mitochondrial and glycolytic/plurimetabolic subtypes with prognostic and therapeutic implications as the mitochondrial subtype carries the best survival and exhibits distinct sensitivity to mitochondrial OXPHOS inhibitors. To uncover novel therapeutic targets for each functional GBM subtype, we focused on protein kinases for their attractive features as both drivers and drug targets, with current availability of 62 FDA-approved inhibitors available for cancer precision therapeutics. We designed an unbiased integrative, machine learning-based proteomics/phosphoproteomics network for the identification of Master Kinases (MKs) responsible for effecting key phenotypic hallmarks of each of the four GBM subtypes. Here we report the identification and validation of protein kinase C delta (PRKCd) and DNA-PKcs as MKs that sustain the glycolytic/plurimetabolic and proliferative/progenitor GBM subtypes, respectively. Genetic and pharmacologic inactivation of PKCd in GBM patient-derived organoids of the glycolytic/plurimetabolic subtype blocked glucose uptake and lipid accumulation, resulting in marked anti-tumor effects. We also validated the role of PKCd in oncometabolic processes at the intersection between insulin, IGF, and lipid metabolism. DNA-PKcs was qualified as MK for the proliferative/progenitor GBM subtype, which is characterized by high basal level of replication stress. Biochemical experiments showed activation of DNA-PK in GBM patient-derived organoids of the proliferative/progenitor subgroup. Targeting DNA-PK in proliferative/progenitor GBM organoids with the clinically tested DNA-PKcs inhibitor nedisertib in association with gamma irradiation, the essential component of the standard of care for patients with GBM, led to an unsustainable level of DNA damage and massive GBM cell death selectively in this GBM subtype. As DNA-PKcs inhibitors have been introduced into clinical trials, our findings suggest that pre-selection of patient with PPR tumors is likely to enhance therapeutic success.

Tumor heterogeneity and broad therapeutic resistance are major challenges when treating patients with glioblastoma (GBM). Moreover, different from other cancer types, the absence of clinically useful classifiers for glioblastoma has hampered the translation of genomic and proteomic information to improve diagnosis and design precision therapeutics for patients. We analyzed a large dataset of human GBM to determine the multi-omics features characterizing the four GBM subtypes that we have recently identified through single-cell RNA sequencing and validated in bulk tumor analysis. Each subtype exhibits activation of unique functional traits that confer distinct therapeutic vulnerabilities. The examination of proteomics, phosphoproteomics, metabolomics, lipidomics and acetylomics data revealed that each GBM subtype has a coherent molecular structure driving the dominant function traceable in each analytical platform. Functional classes are not a specific attribute of GBM as we identified the same subtypes in breast and lung cancer. To test the translational impact of proteomic data in GBM, we developed an unbiased protein kinase signaling network approach for the selection of master kinases (MKs) aberrantly activated in each GBM subtype. We identified therapeutically actionable MKs and novel phosphorylation substrates that we experimentally validated. To provide rapid translation of the functional classifier for precision medicine in GBM, we developed a probabilistic classification tool which determines the probability that a patient’s GBM belongs to one of the four subtypes, exhibiting optimal performance when using RNA extracted from either frozen and paraffin-embedded tissues. The algorithm is publicly accessible and can be used to evaluate the association of therapeutic response with GBM subtypes and as tool for selection criteria in prospective clinical trials.

Glioblastoma harboring Fibroblast Growth Factor Receptor 3 (FGFR3)-Transforming Acidic Coiled-Coil containing 3 (TACC-3) (FGFR3-TACC3) fusion is a distinct molecular subclass of tumors that can be targeted by FGFR kinase inhibitors. We generated a knock-in mouse in which the expression of FGFR3-TACC3 is doxycycline inducible and reversible. The tetO-FGFR3-TACC3-IRES-GFP cassette was placed into the Rosa26 locus and the FGFR3-TACC3 knock-in strain was intercrossed with GFAP-rtTA*M2 mice to trigger expression of the tetracycline transactivator (and the fusion gene) from the GFAP promoter. To recapitulate the genetic lesions that are more frequently associated with FGFR3-TACC3 in human glioblastoma, we also intercrossed the tetO-FGFR3-TACC3-IRES-GFP;GFAP-rtTA*M2 mice with TP53+/-or Ink4A+/-mouse strains. Both strains developed aggressive and infiltrating gliomas within 90 days after induction of FGFR3-TACC3 by doxycycline containing diet. Tumors occurred with a prevalence of ~50% and were found almost exclusively at two brain locations: the lateral septum and the neuro-epithelium adjacent to the fimbria of the hippocampus, thus recapitulating the restricted expression of endogenous FGFR3 in the ventral telencephalon and ganglionic eminence during development. The FGFR3-TACC3-positive glioma resembled human glioblastoma as they exhibited marked invasiveness and vascularization, elevated proliferative index and positivity for Nestin and GFAP. Rapid tumor progression caused appearance of neurological symptoms and death within few weeks. Doxycycline withdrawal from the diet of mice diagnosed with brain tumors by MRI resulted in rapid loss of FGFR3-TACC3 from glioma cells, marked inhibition of tumor growth with essentially complete glioma regression in some cases and recovery of general and neurological clinical symptoms. Our findings validate FGFR3-TACC3 as powerful oncogene that drives development of high-grade glioma from distinct areas of the brain. They also provide an accurate pre-clinical model in which glioma maintenance requires continuous expression of the fusion gene, thus supporting therapeutic targeting of the fusion protein in patients with FGFR3-TACC3-positive glioblastoma.

We propose a generic framework for gene regulatory network (GRN) inference approached as a feature selection problem. GRNs obtained using Machine Learning techniques are often dense, whereas real GRNs are rather sparse. We use a Tikonov regularization inspired optimal L-curve criterion that utilizes the edge weight distribution for a given target gene to determine the optimal set of TFs associated with it. Our proposed framework allows to incorporate a mechanistic active biding network based on cis-regulatory motif analysis. We evaluate our regularization framework in conjunction with two non-linear ML techniques, namely gradient boosting machines (GBM) and random-forests (GENIE), resulting in a regularized feature selection based method specifically called RGBM and RGENIE respectively. RGBM has been used to identify the main transcription factors that are causally involved as master regulators of the gene expression signature activated in the FGFR3-TACC3-positive glioblastoma. Here, we illustrate that RGBM identifies the main regulators of the molecular subtypes of brain tumors. Our analysis reveals the identity and corresponding biological activities of the master regulators characterizing the difference between G-CIMP-high and G-CIMP-low subtypes and between PA-like and LGm6-GBM, thus providing a clue to the yet undetermined nature of the transcriptional events among these subtypes.

Gene fusions generated by chromosomal translocations are the best examples of success for targeted cancer therapy. Our work in malignant glioma identified the first examples of highly oncogenic and recurrent gene fusions that result in constitutive activation of receptor tyrosine kinase genes (FGFR, EGFR and NTRK). Among them, FGFR-TACC is the addicting gene fusion with highest therapeutic value. It is also the gene fusion most frequently represented across other cancer types. Thus, elucidating how FGFR-TACC drives oncogenesis and mechanisms of resistance to FGFR inhibitors are crucial questions with fundamental general implications. By using a combination of unbiased mass spectrometry and functional validation experiments, we have recently identified the tyrosine phosphorylation targets of FGFR-TACC fusions in glioblastoma. We have also generated an inducible and reversible mouse model for FGFR-TACC-mediated tumorigenesis. These mice revealed a strong tumor initiation capacity by FGFR-TACC fusions in the nervous system. They also exposed strict dependency of tumor cells to continuous expression of the FGFR-TACC fusion gene. The mouse model has also allowed us to establish the significance of the new signaling pathways and determine the mechanisms of recurrence of FGFR-TACC-initiated tumors following fusion ablation. The integrated availability of FGFR-TACC substrates and the reversible mouse model of FGFR-TACC-driven tumors are elucidating the mechanisms of resistance triggered by chronic targeting of the fusion gene. The model applied to FGFR-TACC is now being extended to EGFR and NTRK gene fusions to determine similarities and differences and assign accurate therapeutic value to individual chromosomal translocations. Together with the screening and personalized validation paradigm recently completed for the identification, characterization and targeting of FGFR-TACC fusions in patients malignant glioma, the new information will allow us to accurately predict strength of evidence of oncogenic activity and size of anticipated targeting effect for the selection of any “druggable” genetic alteration in GBM patients.

Medulloblastoma (MB), a tumor of the cerebellum, is the most common malignant pediatric tumor of the central nervous system. Human MBs have been classified by gene expression profiling into four subgroups. A subgroup of human MBs (∼25%) harbors constitutive activation of the Sonic Hedgehog (SHH) signaling pathway. Atoh1, a proneural basic helix-loop-helix transcription factor is essential for cerebellum development and collaborates with SHH signaling for the formation of SHH MB in vivo. Because Atoh1 acts as a lineage dependency transcription factor in SHH MB, mechanisms controlling Atoh1 expression are keen interest in the field. However, how this master transcription factor is regulated has remained poorly understood. Using a proteomic approach we recently described a functional interaction between Atoh1 and the E3 ubiquitin ligase Huwe1 during cerebellar development. We uncovered that SHH regulates Atoh1 degradation in a phospho- and E3-dependent manner in cerebellar granule neuron progenitors. In human SHH MB, low HUWE1 expression is associated with poor prognosis. These results led us to deepen the relationship between Huwe1 and Atoh1 in SHH MB formation. More specifically, we wish to explore whether Huwe1 could cooperate with the constitutive activation of the SHH pathway for the formation of SHH MB in vivo.

Cerebellar development requires a balance between several biological processes including proliferation, differentiation and migration. The pro-neural basic helix-loop-helix transcription factor Atoh1 was shown to be a key player controlling the formation of granule neuron progenitors (GNPs). Its uncontrolled levels in collaboration with Sonic Hedgehog (SHH) activation are critical for medulloblastoma (MB) formation, a devastating pediatric tumor that originates from the cerebellum. Because Atoh1 acts as a lineage dependency transcription factor in MB, mechanisms controlling Atoh1 expression are keen interest in the field. However, how this master transcription factor is regulated has remained poorly understood. In our study, we dissected the machinery that regulates Atoh1 in GNPs and in MB. We first identified two phosphorylation sites that regulate Atoh1 stability and function in GNPs. Second, combining tandem affinity purification with an Atoh1 phospho-deficient mutant and multidimensional protein identification technology analysis, we identified the E3 ubiquitin ligase Huwe1 as Atoh1 binding partner. We showed that Atoh1 physically interacted with Huwe1 in a phospho-dependent manner in GNPs, mediating its ubiquitylation whereas Cre-mediated deletion of Huwe1 from postnatal GNPs increased Atoh1 protein levels. Next, we uncovered that SHH regulates Atoh1 stability by preventing its phospho-dependent degradation by Huwe1. In the tumoral context, consistent with Atoh1 upregulation in MBs driven by SHH activation, Huwe1 was found downregulated in primary mouse tumors. In human SHH MB, low HUWE1 expression was found specifically associated with poor prognosis. Overall, our results reveal that SHH and Atoh1 contribute to a positive autoregulatory loop promoting neuronal precursor expansion. Consequently, Huwe1 alterations in SHH MB illustrate the disruption of this developmental mechanism in cancer. This crosstalk between SHH signaling and Atoh1 during cerebellar development highlights a new collaborative network that could be further targeted in MB.

Medulloblastoma is the most common malignant pediatric brain tumor and is divided in four subgroups by gene profiling. The well-known subgroup harboring an activation of the Sonic Hedgehog (SHH)/Patched (PTCH) pathway shows an upregulation of the pro-neural basic helix-loop-helix transcription factor Atoh1. Atoh1 is essential for cerebellum development and more specifically for the formation of the granule neuron progenitors (GNPs), which are the cells of origin of SHH induced medulloblastoma. In tumoral context, Atoh1 acts as a pro-tumor factor in cooperation with SHH/PTCH pathway and its inhibition prevents medulloblastoma proliferation in vitro and in vivo. However, to date, mechanisms underlying Atoh1 regulation remained to be elucidated. We recently found that the mitogen SHH, by regulating Atoh1 phosphorylation, can protect Atoh1 against Huwe1-mediated degradation. We also showed that Atoh1 degradation by the E3 ubiquitin ligase Huwe1 was required for proper cerebellum development. Moreover, low expression of HUWE1 in human Sonic Hedgehog medulloblastoma subgroup was associated with a poor survival rate of patients. Together, we uncovered a signaling pathway that controls the balance between neuronal proliferation and differentiation in GNPs and that could be further targeted in order to elaborate new therapeutic strategies.