A simplified integrated molecular and immunohistochemistry-based algorithm allows high accuracy prediction of glioblastoma transcriptional subtypes.

Glioblastomas (GBM) can be classified into three major transcriptional subgroups (proneural, mesenchymal, classical), underlying different molecular alterations, prognosis, and response to therapy. However, transcriptional analysis is not routinely feasible and assessment of a simplified method for glioblastoma subclassification is required. We propose an integrated molecular and immunohistochemical approach aimed at identifying GBM subtypes in routine paraffin-embedded material. RNA-sequencing analysis was performed on representative samples (n = 51) by means of a "glioblastoma transcriptional subtypes (GliTS) redux" custom gene signature including a restricted number (n = 90) of upregulated genes validated on the TCGA dataset. With this dataset, immunohistochemical profiles, based on expression of a restricted panel of gene classifiers, were integrated by a machine-learning approach to generate a GliTS based on protein quantification that allowed an efficient GliTS assignment when applied to an extended cohort (n = 197). GliTS redux maintained high levels of correspondence with the original GliTS classification using the TCGA dataset. The machine-learning approach designed an immunohistochemical (IHC)-based classification, whose concordance was 79.5% with the transcriptional- based classification, and reached 90% for the mesenchymal subgroup. Distribution and survival of GliTS were in line with reported data, with the mesenchymal subgroup given the worst prognosis. Notably, the algorithm allowed the identification of cases with comparable probability to be assigned to different GliTS, thus falling within overlapping regions and reflecting an extreme heterogeneous phenotype that mirrors the underlying genetic and biological tumor heterogeneity. Indeed, while mesenchymal and classical subgroups were well segregated, the proneural types frequently showed a mixed proneural/classical phenotype, predicted as proneural by the algorithm, but with comparable probability of being assigned to the classical subtype. These cases, characterized by concomitant high expression of EGFR and proneural biomarkers, showed lower survival. Collectively, these data indicate that a restricted panel of highly sensitive immunohistochemical markers can efficiently predict GliTS with high accuracy and significant association with different clinical outcomes.

Genetic Evolution of Glioblastoma Stem-Like Cells From Primary to Recurrent Tumor.

Glioblastoma (GBM) is a lethal tumor that displays remarkable genetic heterogeneity. It is also known that GBM contains a cell hierarchy driven by GBM stem-like cells (GSCs), responsible for tumor generation, therapeutic resistance, and relapse. An important and still open issue is whether phylogenetically related GSCs can be found in matched primary and recurrent GBMs, and reflect tumor genetic evolution under therapeutic pressure. To address this, we analyzed the mutational profile of GSCs isolated from either human primary GBMs (primary GSCs) or their matched tumors recurring after surgery and chemoradiotherapy (recurrent GSCs). We found that recurrent GSCs can accumulate temozolomide-related mutations over primary GSCs, following both linear and branched patterns. In the latter case, primary and recurrent GSCs share a common set of lesions, but also harbor distinctive mutations indicating that primary and recurrent GSCs derive from a putative common ancestor GSC by divergent genetic evolution. Interestingly, TP53 mutations distinctive of recurrent GSCs were detectable at low frequency in the corresponding primary tumors and likely marked pre-existent subclones that evolved under therapeutic pressure and expanded in the relapsing tumor. Consistently, recurrent GSCs displayed in vitro greater therapeutic resistance than primary GSCs. Overall, these data indicate that (a) phylogenetically related GSCs are found in matched primary and recurrent GBMs and (b) recurrent GSCs likely pre-exist in the untreated primary tumor and are both mutagenized and positively selected by chemoradiotherapy. Stem Cells 2017;35:2218-2228.

Protocol for in vitro establishment of heterogeneous stem-like cultures derived from whole human glioblastoma tumors.

Cultures enriched in glioblastoma stem-like cells (GSCs) are prominent in vitro models to investigate molecular determinants and therapeutic targets of glioblastoma; however, conventional GSC derivation protocols fail to preserve GSC heterogeneity. Here, we present a protocol for the propagation of heterogeneous GSC cultures starting from cell resuspensions containing the entire tumor mass. We describe steps for isolation of GSCs and their maintenance and expansion in culture. We then detail procedures for preliminary analysis to be performed on freshly isolated material. For complete details on the use and execution of this protocol, please refer to De Bacco et al.1.

Liquid Biopsy of Cerebrospinal Fluid Enables Selective Profiling of Glioma Molecular Subtypes at First Clinical Presentation.

PURPOSE: Current glioma diagnostic guidelines call for molecular profiling to stratify patients into prognostic and treatment subgroups. In case the tumor tissue is inaccessible, cerebrospinal fluid (CSF) has been proposed as a reliable tumor DNA source for liquid biopsy. We prospectively investigated the use of CSF for molecular characterization of newly diagnosed gliomas. EXPERIMENTAL DESIGN: We recruited two cohorts of newly diagnosed patients with glioma, one (n = 45) providing CSF collected in proximity of the tumor, the other (n = 39) CSF collected by lumbar puncture (LP). Both cohorts provided tumor tissues by surgery concomitant with CSF sampling. DNA samples retrieved from CSF and matched tumors were systematically characterized and compared by comprehensive (NGS, next-generation sequencing) or targeted (ddPCR, droplet digital PCR) methodologies. Conventional and molecular diagnosis outcomes were compared. RESULTS: We report that tumor DNA is abundant in CSF close to the tumor, but scanty and mostly below NGS sensitivity threshold in CSF from LP. Indeed, tumor DNA is mostly released by cells invading liquoral spaces, generating a gradient that attenuates by departing from the tumor. Nevertheless, in >60% of LP CSF samples, tumor DNA is sufficient to assess a selected panel of genetic alterations (IDH and TERT promoter mutations, EGFR amplification, CDKN2A/B deletion: ITEC protocol) and MGMT methylation that, combined with imaging, enable tissue-agnostic identification of main glioma molecular subtypes. CONCLUSIONS: This study shows potentialities and limitations of CSF liquid biopsy in achieving molecular characterization of gliomas at first clinical presentation and proposes a protocol to maximize diagnostic information retrievable from CSF DNA.

Coexisting cancer stem cells with heterogeneous gene amplifications, transcriptional profiles, and malignancy are isolated from single glioblastomas.

Glioblastoma (GBM) is known as an intractable, highly heterogeneous tumor encompassing multiple subclones, each supported by a distinct glioblastoma stem cell (GSC). The contribution of GSC genetic and transcriptional heterogeneity to tumor subclonal properties is debated. In this study, we describe the systematic derivation, propagation, and characterization of multiple distinct GSCs from single, treatment-naive GBMs (GSC families). The tumorigenic potential of each GSC better correlates with its transcriptional profile than its genetic make-up, with classical GSCs being inherently more aggressive and mesenchymal more dependent on exogenous growth factors across multiple GBMs. These GSCs can segregate and recapitulate different histopathological aspects of the same GBM, as shown in a paradigmatic tumor with two histopathologically distinct components, including a conventional GBM and a more aggressive primitive neuronal component. This study provides a resource for investigating how GSCs with distinct genetic and/or phenotypic features contribute to individual GBM heterogeneity and malignant escalation.

ERBB3 as a therapeutic target in glioblastoma: overexpression can make the difference.

By exploiting an integrated experimental platform based on patient-derived cancer stem cells, we identified a glioblastoma subset characterized by inheritable Erb-B2 Receptor Tyrosine Kinase 3 (ERBB3) overexpression, metabolic dependency on ERBB3 signaling, and liability to ERBB3 targeting. We provide insights on why some glioblastomas may rely on ERBB3 and how to recognize them.

ERBB3 overexpression due to miR-205 inactivation confers sensitivity to FGF, metabolic activation, and liability to ERBB3 targeting in glioblastoma.

In glioblastoma (GBM), the most frequent and lethal brain tumor, therapies suppressing recurrently altered signaling pathways failed to extend survival. However, in patient subsets, specific genetic lesions can confer sensitivity to targeted agents. By exploiting an integrated model based on patient-derived stem-like cells, faithfully recapitulating the original GBMs in vitro and in vivo, here, we identify a human GBM subset (∼9% of all GBMs) characterized by ERBB3 overexpression and nuclear accumulation. ERBB3 overexpression is driven by inheritable promoter methylation or post-transcriptional silencing of the oncosuppressor miR-205 and sustains the malignant phenotype. Overexpressed ERBB3 behaves as a specific signaling platform for fibroblast growth factor receptor (FGFR), driving PI3K/AKT/mTOR pathway hyperactivation, and overall metabolic upregulation. As a result, ERBB3 inhibition by specific antibodies is lethal for GBM stem-like cells and xenotransplants. These findings highlight a subset of patients eligible for ERBB3-targeted therapy.

A Molecularly Annotated Model of Patient-Derived Colon Cancer Stem-Like Cells to Assess Genetic and Nongenetic Mechanisms of Resistance to Anti-EGFR Therapy.

Purpose: Patient-derived xenografts ("xenopatients") of colorectal cancer metastases have been essential to identify genetic determinants of resistance to the anti-EGFR antibody cetuximab and to explore new therapeutic strategies. From xenopatients, a genetically annotated collection of stem-like cultures ("xenospheres") was generated and characterized for response to targeted therapies.Experimental Design: Xenospheres underwent exome-sequencing analysis, gene expression profile, and in vitro targeted treatments to assess genetic, biological, and pharmacologic correspondence with xenopatients, and to investigate nongenetic biomarkers of therapeutic resistance. The outcome of EGFR family inhibition was tested in an NRG1-expressing in vivo model.Results: Xenospheres faithfully retained the genetic make-up of their matched xenopatients over in vitro and in vivo passages. Frequent and rare genetic lesions triggering primary resistance to cetuximab through constitutive activation of the RAS signaling pathway were conserved, as well as the vulnerability to their respective targeted treatments. Xenospheres lacking such alterations (RASwt) were highly sensitive to cetuximab, but were protected by ligands activating the EGFR family, mostly NRG1. Upon reconstitution of NRG1 expression, xenospheres displayed increased tumorigenic potential in vivo and generated tumors completely resistant to cetuximab, and sensitive only to comprehensive EGFR family inhibition.Conclusions: Xenospheres are a reliable model to identify both genetic and nongenetic mechanisms of response and resistance to targeted therapies in colorectal cancer. In the absence of RAS pathway mutations, NRG1 and other EGFR ligands can play a major role in conferring primary cetuximab resistance, indicating that comprehensive inhibition of the EGFR family is required to achieve a significant therapeutic response. Clin Cancer Res; 24(4); 807-20. ©2017 AACRSee related commentary by Napolitano and Ciardiello, p. 727.

MET inhibition overcomes radiation resistance of glioblastoma stem-like cells.

Glioblastoma (GBM) contains stem-like cells (GSCs) known to be resistant to ionizing radiation and thus responsible for therapeutic failure and rapidly lethal tumor recurrence. It is known that GSC radioresistance relies on efficient activation of the DNA damage response, but the mechanisms linking this response with the stem status are still unclear. Here, we show that the MET receptor kinase, a functional marker of GSCs, is specifically expressed in a subset of radioresistant GSCs and overexpressed in human GBM recurring after radiotherapy. We elucidate that MET promotes GSC radioresistance through a novel mechanism, relying on AKT activity and leading to (i) sustained activation of Aurora kinase A, ATM kinase, and the downstream effectors of DNA repair, and (ii) phosphorylation and cytoplasmic retention of p21, which is associated with anti-apoptotic functions. We show that MET pharmacological inhibition causes DNA damage accumulation in irradiated GSCs and their depletion in vitro and in GBMs generated by GSC xenotransplantation. Preclinical evidence is thus provided that MET inhibitors can radiosensitize tumors and convert GSC-positive selection, induced by radiotherapy, into GSC eradication.

TNF-α promotes invasive growth through the MET signaling pathway.

The inflammatory cytokine Tumor Necrosis Factor Alpha (TNF-α) is known to trigger invasive growth, a physiological property for tissue healing, turning into a hallmark of progression in cancer. However, the invasive response to TNF-α relies on poorly understood molecular mechanisms. We thus investigated whether it involves the MET oncogene, which regulates the invasive growth program by encoding the tyrosine kinase receptor for Hepatocyte Growth Factor (HGF). Here we show that the TNF-α pro-invasive activity requires MET function, as it is fully inhibited by MET-specific inhibitors (small-molecules, antibodies, and siRNAs). Mechanistically, we show that TNF-α induces MET transcription via NF-κB, and exploits MET to sustain MEK/ERK activation and Snail accumulation, leading to E-cadherin downregulation. We then show that TNF-α not only induces MET expression in cancer cells, but also HGF secretion by fibroblasts. Consistently, we found that, in human colorectal cancer tissues, high levels of TNF-α correlates with increased expression of both MET and HGF. These findings suggest that TNF-α fosters a HGF/MET pro-invasive paracrine loop in tumors. Targeting this ligand/receptor pair would contribute to prevent cancer progression associated with inflammation.

MET signaling in colon cancer stem-like cells blunts the therapeutic response to EGFR inhibitors.

Metastatic colorectal cancer remains largely incurable, although in a subset of patients, survival is prolonged by new targeting agents such as anti-EGF receptor (anti-EGFR) antibodies. This disease is believed to be supported by a subpopulation of stem-like cells termed colon cancer-initiating cell (CCIC), which may also confer therapeutic resistance. However, how CCICs respond to EGFR inhibition has not been fully characterized. To explore this question, we systematically generated CCICs through spheroid cultures of patient-derived xenografts of metastatic colorectal cancer. These cultures, termed "xenospheres," were capable of long-term self-propagation in vitro and phenocopied the original patient tumors in vivo, thus operationally defining CCICs. Xenosphere CCICs retained the genetic determinants for EGFR therapeutic response in vitro and in xenografts; like the original tumors, xenospheres harboring a mutated KRAS gene were resistant to EGFR therapy, whereas those harboring wild-type RAS pathway genes (RAS(wt)) were sensitive. Notably, the effects of EGFR inhibition in sensitive CCICs could be counteracted by cytokines secreted by cancer-associated fibroblasts. In particular, we found that the MET receptor ligand hepatocyte growth factor (HGF) was especially active in supporting in vitro CCIC proliferation and resistance to EGFR inhibition. Ectopic production of human HGF in CCIC xenografts rendered the xenografts susceptible to MET inhibition, which sensitized the response to EGFR therapy. By showing that RAS(wt) CCICs rely on both EGFR and MET signaling, our results offer a strong preclinical proof-of-concept for concurrent targeting of these two pathways in the clinical setting.

The MET oncogene is a functional marker of a glioblastoma stem cell subtype.

The existence of treatment-resistant cancer stem cells contributes to the aggressive phenotype of glioblastoma. However, the molecular alterations that drive stem cell proliferation in these tumors remain unknown. In this study, we found that expression of the MET oncogene was associated with neurospheres expressing the gene signature of mesenchymal and proneural subtypes of glioblastoma. Met expression was almost absent from neurospheres expressing the signature of the classical subtype and was mutually exclusive with amplification and expression of the EGF receptor (EGFR) gene. Met-positive and Met-negative neurospheres displayed distinct growth factor requirements, differentiated along divergent pathways, and generated tumors with distinctive features. The Met(high) subpopulation within Met-pos neurospheres displayed clonogenic potential and long-term self-renewal ability in vitro and enhanced growth kinetics in vivo. In Met(high) cells, the Met ligand HGF further sustained proliferation, clonogenicity, expression of self-renewal markers, migration, and invasion in vitro. Together, our findings suggest that Met is a functional marker of glioblastoma stem cells and a candidate target for identification and therapy of a subset of glioblastomas.

Induction of MET by ionizing radiation and its role in radioresistance and invasive growth of cancer.

BACKGROUND: Ionizing radiation (IR) is effectively used in cancer therapy. However, in subsets of patients, a few radioresistant cancer cells survive and cause disease relapse with metastatic progression. The MET oncogene encodes the hepatocyte growth factor (HGF) receptor and is known to drive "invasive growth", a regenerative and prosurvival program unduly activated in metastasis. METHODS: Human tumor cell lines (MDA-MB-231, MDA-MB-435S, U251) were subjected to therapeutic doses of IR. MET mRNA, and protein expression and signal transduction were compared in treated and untreated cells, and the involvement of the DNA-damage sensor ataxia telangiectasia mutated (ATM) and the transcription factor nuclear factor kappa B (NF-κB) in activating MET transcription were analyzed by immunoblotting, chromatin immunoprecipitation, and use of NF-κB silencing RNA (siRNA). Cell invasiveness was measured in wound healing and transwell assays, and cell survival was measured in viability and clonogenic assays. MET was inhibited by siRNA or small-molecule kinase inhibitors (PHA665752 or JNJ-38877605). Combinations of MET-targeted therapy and radiotherapy were assessed in MDA-MB-231 and U251 xenografts (n = 5-6 mice per group). All P values were from two-sided tests. RESULTS: After irradiation, MET expression in cell lines was increased up to fivefold via activation of ATM and NF-κB. MET overexpression increased ligand-independent MET phosphorylation and signal transduction, and rendered cells more sensitive to HGF. Irradiated cells became more invasive via a MET-dependent mechanism that was further enhanced in the presence of HGF. MET silencing by siRNA or inhibition of its kinase activity by treatment with PHA665752 or JNJ-38877605 counteracted radiation-induced invasiveness, promoted apoptosis, and prevented cells from resuming proliferation after irradiation in vitro. Treatment with MET inhibitors enhanced the efficacy of IR to stop the growth of or to induce the regression of xenografts (eg, at day 13, U251 xenografts, mean volume increase relative to mean tumor volume at day 0: vehicle = 438%, 5 Gy IR = 151%, 5 Gy IR + JNJ-38877605 = 76%; difference, IR vs JNJ-38877604 + IR = 75%, 95% CI = 59% to 91%, P = .01). CONCLUSION: IR induces overexpression and activity of the MET oncogene through the ATM-NF-κB signaling pathway; MET, in turn, promotes cell invasion and protects cells from apoptosis, thus supporting radioresistance. Drugs targeting MET increase tumor cell radiosensitivity and prevent radiation-induced invasiveness.