The RNA editor ADAR2 promotes immune cell trafficking by enhancing endothelial responses to interleukin-6 during sterile inflammation.

Immune cell trafficking constitutes a fundamental component of immunological response to tissue injury, but the contribution of intrinsic RNA nucleotide modifications to this response remains elusive. We report that RNA editor ADAR2 exerts a tissue- and stress-specific regulation of endothelial responses to interleukin-6 (IL-6), which tightly controls leukocyte trafficking in IL-6-inflamed and ischemic tissues. Genetic ablation of ADAR2 from vascular endothelial cells diminished myeloid cell rolling and adhesion on vascular walls and reduced immune cell infiltration within ischemic tissues. ADAR2 was required in the endothelium for the expression of the IL-6 receptor subunit, IL-6 signal transducer (IL6ST; gp130), and subsequently, for IL-6 trans-signaling responses. ADAR2-induced adenosine-to-inosine RNA editing suppressed the Drosha-dependent primary microRNA processing, thereby overwriting the default endothelial transcriptional program to safeguard gp130 expression. This work demonstrates a role for ADAR2 epitranscriptional activity as a checkpoint in IL-6 trans-signaling and immune cell trafficking to sites of tissue injury.

SARS-CoV-2 infection of thymus induces loss of function that correlates with disease severity.

BACKGROUND: Lymphopenia, particularly when restricted to the T-cell compartment, has been described as one of the major clinical hallmarks in patients with coronavirus disease 2019 (COVID-19) and proposed as an indicator of disease severity. Although several mechanisms fostering COVID-19-related lymphopenia have been described, including cell apoptosis and tissue homing, the underlying causes of the decline in T-cell count and function are still not completely understood. OBJECTIVE: Given that viral infections can directly target thymic microenvironment and impair the process of T-cell generation, we sought to investigate the impact of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) on thymic function. METHODS: We performed molecular quantification of T-cell receptor excision circles and κ-deleting recombination excision circles to assess, respectively, T- and B-cell neogenesis in SARS-CoV-2-infected patients. We developed a system for in vitro culture of primary human thymic epithelial cells (TECs) to mechanistically investigate the impact of SARS-CoV-2 on TEC function. RESULTS: We showed that patients with COVID-19 had reduced thymic function that was inversely associated with the severity of the disease. We found that angiotensin-converting enzyme 2, through which SARS-CoV-2 enters the host cells, was expressed by thymic epithelium, and in particular by medullary TECs. We also demonstrated that SARS-CoV-2 can target TECs and downregulate critical genes and pathways associated with epithelial cell adhesion and survival. CONCLUSIONS: Our data demonstrate that the human thymus is a target of SARS-CoV-2 and thymic function is altered following infection. These findings expand our current knowledge of the effects of SARS-CoV-2 infection on T-cell homeostasis and suggest that monitoring thymic activity may be a useful marker to predict disease severity and progression.

Genome-wide profiling of patient-derived glioblastoma stem-like cells reveals recurrent genetic and transcriptomic signatures associated with brain tumors.

PURPOSE: Patient-derived cancer cell lines can be very useful to investigate genetic as well as epigenetic mechanisms of transformation and to test new drugs. In this multi-centric study, we performed genomic and transcriptomic characterization of a large set of patient-derived glioblastoma (GBM) stem-like cells (GSCs). METHODS: 94 (80 I surgery/14 II surgery) and 53 (42 I surgery/11 II surgery) GSCs lines underwent whole exome and trascriptome analysis, respectively. RESULTS: Exome sequencing revealed TP53 as the main mutated gene (41/94 samples, 44%), followed by PTEN (33/94, 35%), RB1 (16/94, 17%) and NF1 (15/94, 16%), among other genes associated to brain tumors. One GSC sample bearing a BRAF p.V600E mutation showed sensitivity in vitro to a BRAF inhibitor. Gene Ontology and Reactome analysis uncovered several biological processes mostly associated to gliogenesis and glial cell differentiation, S - adenosylmethionine metabolic process, mismatch repair and methylation. Comparison of I and II surgery samples disclosed a similar distribution of mutated genes, with an overrepresentation of mutations in mismatch repair, cell cycle, p53 and methylation pathways in I surgery samples, and of mutations in receptor tyrosine kinase and MAPK signaling pathways in II surgery samples. Unsupervised hierarchical clustering of RNA-seq data produced 3 clusters characterized by distinctive sets of up-regulated genes and signaling pathways. CONCLUSION: The availability of a large set of fully molecularly characterized GCSs represents a valuable public resource to support the advancement of precision oncology for the treatment of GBM.

Increased adenosine-to-inosine RNA editing in rheumatoid arthritis.

OBJECTIVE: Adenosine-to-inosine (A-to-I) RNA editing of Alu retroelements is a primate-specific mechanism mediated by adenosine deaminases acting on RNA (ADARs) that diversifies transcriptome by changing selected nucleotides in RNA molecules. We tested the hypothesis that A-to-I RNA editing is altered in rheumatoid arthritis (RA). METHODS: Synovium expression analysis of ADAR1 was investigated in 152 RA patients and 50 controls. Peripheral blood mononuclear cells derived from 14 healthy subjects and 19 patients with active RA at baseline and after 12-week treatment were examined for ADAR1p150 and ADAR1p110 isoform expression by RT-qPCR. RNA editing activity was analysed by AluSx+ Sanger-sequencing of cathepsin S, an extracellular matrix degradation enzyme involved in antigen presentation. RESULTS: ADAR1 was significantly over-expressed in RA synovium regardless of disease duration. Similarly, ADAR1p150 isoform expression was significantly increased in the blood of active RA patients. Individual nucleotide analysis revealed that A-to-I RNA editing rate was also significantly increased in RA patients. Both baseline ADAR1p150 expression and individual adenosine RNA editing rate of cathepsin S AluSx+ decreased after treatment only in those patients with good clinical response. Upregulation of the expression and/or activity of the RNA editing machinery were associated with a higher expression of edited Alu-enriched genes including cathepsin S and TNF receptor-associated factors 1,2,3 and 5. CONCLUSION: A previously unrecognized regulation and role of ADAR1p150-mediated A-to-I RNA editing in post-transcriptional control in RA underpins therapeutic response and fuels inflammatory gene expression, thus representing an interesting therapeutic target.

Aptamer-Based In Vivo Therapeutic Targeting of Glioblastoma.

Glioblastoma (GBM) is the most aggressive, infiltrative, and lethal brain tumor in humans. Despite the extensive advancement in the knowledge about tumor progression and treatment over the last few years, the prognosis of GBM is still very poor due to the difficulty of targeting drugs or anticancer molecules to GBM cells. The major challenge in improving GBM treatment implicates the development of a targeted drug delivery system, capable of crossing the blood-brain barrier (BBB) and specifically targeting GBM cells. Aptamers possess many characteristics that make them ideal novel therapeutic agents for the treatment of GBM. They are short single-stranded nucleic acids (RNA or ssDNA) able to bind to a molecular target with high affinity and specificity. Several GBM-targeting aptamers have been developed for imaging, tumor cell isolation from biopsies, and drug/anticancer molecule delivery to the tumor cells. Due to their properties (low immunogenicity, long stability, and toxicity), a large number of aptamers have been selected against GBM biomarkers and tested in GBM cell lines, while only a few of them have also been tested in in vivo models of GBM. Herein, we specifically focus on aptamers tested in GBM in vivo models that can be considered as new diagnostic and/or therapeutic tools for GBM patients' treatment.

Differential RNA editing landscapes in host cell versus the SARS-CoV-2 genome.

The SARS-CoV-2 pandemic was defined by the emergence of new variants formed through virus mutation originating from random errors not corrected by viral proofreading and/or the host antiviral response introducing mutations into the viral genome. While sequencing information hints at cellular RNA editing pathways playing a role in viral evolution, here, we use an in vitro human cell infection model to assess RNA mutation types in two SARS-CoV-2 strains representing the original and the alpha variants. The variants showed both different cellular responses and mutation patterns with alpha showing higher mutation frequency with most substitutions observed being C-U, indicating an important role for apolipoprotein B mRNA editing catalytic polypeptide-like editing. Knockdown of select APOBEC3s through RNAi increased virus production in the original virus, but not in alpha. Overall, these data suggest a deaminase-independent anti-viral function of APOBECs in SARS-CoV-2 while the C-U editing itself might function to enhance genetic diversity enabling evolutionary adaptation.

Insight into immune profile associated with vitiligo onset and anti-tumoral response in melanoma patients receiving anti-PD-1 immunotherapy.

Introduction: Immunotherapy with checkpoint inhibitors is an efficient treatment for metastatic melanoma. Development of vitiligo upon immunotherapy represents a specific immune-related adverse event (irAE) diagnosed in 15% of patients and associated with a positive clinical response. Therefore, a detailed characterization of immune cells during vitiligo onset in melanoma patients would give insight into the immune mechanisms mediating both the irAE and the anti-tumor response. Methods: To better understand these aspects, we analyzed T cell subsets from peripheral blood of metastatic melanoma patients undergoing treatment with anti-programmed cell death protein (PD)-1 antibodies. To deeply characterize the antitumoral T cell response concomitant to vitiligo onset, we analyzed T cell content in skin biopsies collected from melanoma patients who developed vitiligo. Moreover, to further characterize T cells in vitiligo skin lesion of melanoma patients, we sequenced T cell receptor (TCR) of cells derived from biopsies of vitiligo and primary melanoma of the same patient. Results and discussion: Stratification of patients for developing or not developing vitiligo during anti-PD-1 therapy revealed an association between blood reduction of CD8-mucosal associated invariant T (MAIT), T helper (h) 17, natural killer (NK) CD56bright, and T regulatory (T-reg) cells and vitiligo onset. Consistently with the observed blood reduction of Th17 cells in melanoma patients developing vitiligo during immunotherapy, we found high amount of IL-17A expressing cells in the vitiligo skin biopsy, suggesting a possible migration of Th17 cells from the blood into the autoimmune lesion. Interestingly, except for a few cases, we found different TCR sequences between vitiligo and primary melanoma lesions. In contrast, shared TCR sequences were identified between vitiligo and metastatic tissues of the same patient. These data indicate that T cell response against normal melanocytes, which is involved in vitiligo onset, is not typically mediated by reactivation of specific T cell clones infiltrating primary melanoma but may be elicited by T cell clones targeting metastatic tissues. Altogether, our data indicate that anti-PD-1 therapy induces a de novo immune response, stimulated by the presence of metastatic cells, and composed of different T cell subtypes, which may trigger the development of vitiligo and the response against metastatic tumor.

Neutralizing IFNγ improves safety without compromising efficacy of CAR-T cell therapy in B-cell malignancies.

Chimeric antigen receptor T (CAR-T) cell therapy may achieve long-lasting remission in patients with B-cell malignancies not responding to conventional therapies. However, potentially severe and hard-to-manage side effects, including cytokine release syndrome (CRS), neurotoxicity and macrophage activation syndrome, and the lack of pathophysiological experimental models limit the applicability and development of this form of therapy. Here we present a comprehensive humanized mouse model, by which we show that IFNγ neutralization by the clinically approved monoclonal antibody, emapalumab, mitigates severe toxicity related to CAR-T cell therapy. We demonstrate that emapalumab reduces the pro-inflammatory environment in the model, thus allowing control of severe CRS and preventing brain damage, characterized by multifocal hemorrhages. Importantly, our in vitro and in vivo experiments show that IFNγ inhibition does not affect the ability of CD19-targeting CAR-T (CAR.CD19-T) cells to eradicate CD19+ lymphoma cells. Thus, our study provides evidence that anti-IFNγ treatment might reduce immune related adverse effect without compromising therapeutic success and provides rationale for an emapalumab-CAR.CD19-T cell combination therapy in humans.

ADAR2 Protein Is Associated with Overall Survival in GBM Patients and Its Decrease Triggers the Anchorage-Independent Cell Growth Signature.

BACKGROUND: Epitranscriptomic mechanisms, such as A-to-I RNA editing mediated by ADAR deaminases, contribute to cancer heterogeneity and patients' stratification. ADAR enzymes can change the sequence, structure, and expression of several RNAs, affecting cancer cell behavior. In glioblastoma, an overall decrease in ADAR2 RNA level/activity has been reported. However, no data on ADAR2 protein levels in GBM patient tissues are available; and most data are based on ADARs overexpression experiments. METHODS: We performed IHC analysis on GBM tissues and correlated ADAR2 levels and patients' overall survival. We silenced ADAR2 in GBM cells, studied cell behavior, and performed a gene expression/editing analysis. RESULTS: GBM tissues do not all show a low/no ADAR2 level, as expected by previous studies. Although, different amounts of ADAR2 protein were observed in different patients, with a low level correlating with a poor patient outcome. Indeed, reducing the endogenous ADAR2 protein in GBM cells promotes cell proliferation and migration and changes the cell's program to an anchorage-independent growth mode. In addition, deep-seq data and bioinformatics analysis indicated multiple RNAs are differently expressed/edited upon siADAR2. CONCLUSION: ADAR2 protein is an important deaminase in GBM and its amount correlates with patient prognosis.

dDAVP Downregulates the AQP3-Mediated Glycerol Transport via V1aR in Human Colon HCT8 Cells.

Vasopressin (AVP) plays a key function in controlling body water and salt balance through the activation of the vasopressin receptors V1aR and V2R. Abnormal secretion of AVP can cause the syndrome of inappropriate antidiuresis that leads to hyponatremia, which is an electrolyte disorder often observed in the elderly hospitalized and oncologic patients. Beyond kidneys, the colonic epithelium modulates water and salt homeostasis. The water channel AQP3, expressed in villus epithelial cells is implicated in water absorption across human colonic surface cells. Here, the action of dDAVP, a stable vasopressin analog, was evaluated on the AQP3 expression and function using human colon HCT8 cells as an experimental model. Confocal and Western Blotting analysis revealed that HCT8 cells express both V1aR and V2R. Long-term (72 h) treatment with dDAVP reduced glycerol uptake and cell viability. These effects were prevented by SR49059, a synthetic antagonist of V1aR, but not by tolvaptan, a specific V2R antagonist. Of note, the SR49059 action was impaired by DFP00173, a selective inhibitor of AQP3. Interestingly, compared to the normal colonic mucosa, in the colon of patients with adenocarcinoma, the expression of V1aR was significantly decreased. These findings were confirmed by gene expression analysis with RNA-Seq data. Overall, data suggest that dDAVP, through the V1aR dependent pathway, reduces AQP3 mediated glycerol uptake, a process that is reversed in adenocarcinoma, suggesting that the AVP-dependent AQP3 pathway may represent a novel target in colon diseases associated with abnormal cell growth.

Safe and effective off-the-shelf immunotherapy based on CAR.CD123-NK cells for the treatment of acute myeloid leukaemia.

BACKGROUND: Paediatric acute myeloid leukaemia (AML) is characterized by poor outcomes in patients with relapsed/refractory disease, despite the improvements in intensive standard therapy. The leukaemic cells of paediatric AML patients show high expression of the CD123 antigen, and this finding provides the biological basis to target CD123 with the chimeric antigen receptor (CAR). However, CAR.CD123 therapy in AML is hampered by on-target off-tumour toxicity and a long "vein-to-vein" time. METHODS: We developed an off-the-shelf product based on allogeneic natural killer (NK) cells derived from the peripheral blood of healthy donors and engineered them to express a second-generation CAR targeting CD123 (CAR.CD123). RESULTS: CAR.CD123-NK cells showed significant anti-leukaemia activity not only in vitro against CD123+ AML cell lines and CD123+ primary blasts but also in two animal models of human AML-bearing immune-deficient mice. Data on anti-leukaemia activity were also corroborated by the quantification of inflammatory cytokines, namely granzyme B (Granz B), interferon gamma (IFN-γ) and tumour necrosis factor alpha (TNF-α), both in vitro and in the plasma of mice treated with CAR.CD123-NK cells. To evaluate and compare the on-target off-tumour effects of CAR.CD123-T and NK cells, we engrafted human haematopoietic cells (hHCs) in an immune-deficient mouse model. All mice infused with CAR.CD123-T cells died by Day 5, developing toxicity against primary human bone marrow (BM) cells with a decreased number of total hCD45+ cells and, in particular, of hCD34+CD38- stem cells. In contrast, treatment with CAR.CD123-NK cells was not associated with toxicity, and all mice were alive at the end of the experiments. Finally, in a mouse model engrafted with human endothelial tissues, we demonstrated that CAR.CD123-NK cells were characterized by negligible endothelial toxicity when compared to CAR.CD123-T cells. CONCLUSIONS: Our data indicate the feasibility of an innovative off-the-shelf therapeutic strategy based on CAR.CD123-NK cells, characterized by remarkable efficacy and an improved safety profile compared to CAR.CD123-T cells. These findings open a novel intriguing scenario not only for the treatment of refractory/resistant AML patients but also to further investigate the use of CAR-NK cells in other cancers characterized by highly difficult targeting with the most conventional T effector cells.

MicroRNA Editing Detection and Function: A Combined In Silico and Experimental Approach for the Identification and Validation of Putative Oncogenic Targets.

MicroRNAs (miRNAs) are a class of ~22 nt noncoding RNAs playing essential roles in the post-transcriptional regulation of gene expression, cell proliferation, and cell differentiation and are often found deregulated in several diseases including cancer.The A-to-I RNA editing, mediated by ADAR enzymes, is a diffuse post-transcriptional mechanism that converts the genetically coded adenosine (A) into inosine (I) at the RNA level. Among different RNA targets, the ADAR enzymes can also edit miRNA precursors. Specifically, a single nucleotide change (A/I) lying within the mature miRNA can alter the miRNA binding specificity and redirect the edited miRNA to a different mRNA target. In several cancer types a consistent deregulation of A-to-I RNA editing machinery also involves important miRNAs (either oncomiRs or tumor-suppressor miRNAs). Herein we describe a combined in silico and experimental approach for the detection of edited miRNAs and the identification and validation of their target genes potentially involved in cancer progression or invasion.

ADAR1 is a new target of METTL3 and plays a pro-oncogenic role in glioblastoma by an editing-independent mechanism.

BACKGROUND: N6-methyladenosine (m6A) and adenosine-to-inosine (A-to-I) RNA editing are two of the most abundant RNA modification events affecting adenosines in mammals. Both these RNA modifications determine mRNA fate and play a pivotal role in tumor development and progression. RESULTS: Here, we show that METTL3, upregulated in glioblastoma, methylates ADAR1 mRNA and increases its protein level leading to a pro-tumorigenic mechanism connecting METTL3, YTHDF1, and ADAR1. We show that ADAR1 plays a cancer-promoting role independently of its deaminase activity by binding CDK2 mRNA, underlining the importance of ADARs as essential RNA-binding proteins for cell homeostasis as well as cancer progression. Additionally, we show that ADAR1 knockdown is sufficient to strongly inhibit glioblastoma growth in vivo. CONCLUSIONS: Hence, our findings underscore METTL3/ADAR1 axis as a novel crucial pathway in cancer progression that connects m6A and A-to-I editing post-transcriptional events.

ADAR-mediated RNA editing of DNA:RNA hybrids is required for DNA double strand break repair.

The maintenance of genomic stability requires the coordination of multiple cellular tasks upon the appearance of DNA lesions. RNA editing, the post-transcriptional sequence alteration of RNA, has a profound effect on cell homeostasis, but its implication in the response to DNA damage was not previously explored. Here we show that, in response to DNA breaks, an overall change of the Adenosine-to-Inosine RNA editing is observed, a phenomenon we call the RNA Editing DAmage Response (REDAR). REDAR relies on the checkpoint kinase ATR and the recombination factor CtIP. Moreover, depletion of the RNA editing enzyme ADAR2 renders cells hypersensitive to genotoxic agents, increases genomic instability and hampers homologous recombination by impairing DNA resection. Such a role of ADAR2 in DNA repair goes beyond the recoding of specific transcripts, but depends on ADAR2 editing DNA:RNA hybrids to ease their dissolution.

De Novo A-to-I RNA Editing Discovery in lncRNA.

BACKGROUND: Adenosine to inosine (A-to-I) RNA editing is the most frequent editing event in humans. It converts adenosine to inosine in double-stranded RNA regions (in coding and non-coding RNAs) through the action of the adenosine deaminase acting on RNA (ADAR) enzymes. Long non-coding RNAs, particularly abundant in the brain, account for a large fraction of the human transcriptome, and their important regulatory role is becoming progressively evident in both normal and transformed cells. RESULTS: Herein, we present a bioinformatic analysis to generate a comprehensive inosinome picture in long non-coding RNAs (lncRNAs), using an ad hoc index and searching for de novo editing events in the normal brain cortex as well as in glioblastoma, a highly aggressive human brain cancer. We discovered >10,000 new sites and 335 novel lncRNAs that undergo editing, never reported before. We found a generalized downregulation of editing at multiple lncRNA sites in glioblastoma samples when compared to the normal brain cortex. CONCLUSION: Overall, our study discloses a novel layer of complexity that controls lncRNAs in the brain and brain cancer.

Quantifying RNA Editing in Deep Transcriptome Datasets.

Massive transcriptome sequencing through the RNAseq technology has enabled quantitative transcriptome-wide investigation of co-/post-transcriptional mechanisms such as alternative splicing and RNA editing. The latter is abundant in human transcriptomes in which million adenosines are deaminated into inosines by the ADAR enzymes. RNA editing modulates the innate immune response and its deregulation has been associated with different human diseases including autoimmune and inflammatory pathologies, neurodegenerative and psychiatric disorders, and tumors. Accurate profiling of RNA editing using deep transcriptome data is still a challenge, and the results depend strongly on processing and alignment steps taken. Accurate calling of the inosinome repertoire, however, is required to reliably quantify RNA editing and, in turn, investigate its biological and functional role across multiple samples. Using real RNAseq data, we demonstrate the impact of different bioinformatics steps on RNA editing detection and describe the main metrics to quantify its level of activity.

The Interplay between CD27dull and CD27bright B Cells Ensures the Flexibility, Stability, and Resilience of Human B Cell Memory.

Memory B cells (MBCs) epitomize the adaptation of the immune system to the environment. We identify two MBC subsets in peripheral blood, CD27dull and CD27bright MBCs, whose frequency changes with age. Heavy chain variable region (VH) usage, somatic mutation frequency replacement-to-silent ratio, and CDR3 property changes, reflecting consecutive selection of highly antigen-specific, low cross-reactive antibody variants, all demonstrate that CD27dull and CD27bright MBCs represent sequential MBC developmental stages, and stringent antigen-driven pressure selects CD27dull into the CD27bright MBC pool. Dynamics of human MBCs are exploited in pregnancy, when 50% of maternal MBCs are lost and CD27dull MBCs transit to the more differentiated CD27bright stage. In the postpartum period, the maternal MBC pool is replenished by the expansion of persistent CD27dull clones. Thus, the stability and flexibility of human B cell memory is ensured by CD27dull MBCs that expand and differentiate in response to change.

The adaptive potential of RNA editing-mediated miRNA-retargeting in cancer.

A-to-I RNA editing is a post-transcriptional mechanism that converts the genomically coded Adenosine (A) into Inosine (I) at the RNA level. This type of RNA editing is the most frequent in humans and is mediated by the ADAR enzymes. RNA editing can alter the genetic code of mRNAs, but also affect the functions of noncoding RNAs such as miRNAs. Recent studies have identified thousands of microRNA editing events in different cancer types. However, the important role played by miRNA-editing in cancer has been reported for just a few microRNAs. Herein, we recapitulate the current studies on cancer-related microRNA editing and discuss their importance in tumor growth and progression. This article is part of a Special Issue entitled: mRNA modifications in gene expression control edited by Dr. Soller Matthias and Dr. Fray Rupert.

Dynamic inosinome profiles reveal novel patient stratification and gender-specific differences in glioblastoma.

BACKGROUND: Adenosine-to-inosine (A-to-I) RNA editing is an essential post-transcriptional mechanism mediated by ADAR enzymes that have been recently associated with cancer. RESULTS: Here, we characterize the inosinome signature in normal brain and de novo glioblastoma (GBM) using new metrics that re-stratify GBM patients according to their editing profiles and indicate this post-transcriptional event as a possible molecular mechanism for sexual dimorphism in GBM. We find that over 85% of de novo GBMs carry a deletion involving the genomic locus of ADAR3, which is specifically expressed in the brain. By analyzing RNA editing and patient outcomes, an intriguing gender-dependent link appears, with high editing of Alus shown to be beneficial only in male patients. We propose an inosinome-based molecular stratification of GBM patients that identifies two different GBM subgroups, INO-1 and INO-2, which can identify novel high-risk gender-specific patient groups for which more aggressive treatments may be necessary. CONCLUSIONS: Our data provide a detailed picture of RNA editing landscape in normal brain and GBM, exploring A-to-I RNA editing regulation, disclosing unexpected editing implications for GBM patient stratification and identification of gender-dependent high-risk patients, and suggesting COG3 I/V as an eligible site for future personalized targeted gene therapy.