Year: 2018
Type: article
Abstract: Mutations in the gene encoding the metabolic enzymes isocitrate dehydrogenase (IDH)-1 and -2 is a hallmark of low-grade gliomas (LGGs). IDH mutations are thought to occur early during tumor evolution and are retained in recurrent tumors, implying that IDH mutations are integral for tumor initiation and progression. Recent reports, however, have shown that IDH mutations are lost during recurrence and selected against in in vitro cultures, suggesting that IDH mutations may not be required in all cases of recurrence.1,2 To address this, Joseph Costello and colleagues examined longitudinally collected IDH mutant (IDH-MUT) gliomas and performed a comprehensive analysis of the genetic and epigenetic alterations in these tissues.3 The authors examined a cohort of 50 paired primary and recurrent IDH-MUT LGGs. While a majority of the recurrent tumors retained IDH mutations with similar allele frequency, deletion of the IDH mutant allele as well as gain of both wild-type and mutant allele was observed in six recurrent cases. The genetic alteration of the IDH allele was also reflected in changes at the mRNA and the protein level of IDH. Brain tumor-initiating cultures (BTICs) from some of these recurrent tumors exhibited impaired production of 2-hydroxyglutarate (2-HG), an oncometabolite that accumulates in IDH mutant gliomas, irrespective of gain or loss the IDH mutant allele. The BTICs were also highly proliferative in vitro, formed tumors in intracranial implants, and had deletions in chromosome 2 encompassing IDH, but this event was undetected in IDH wild-type BTICs. These results implied that chromosome 2 deletion conferred a growth advantage specifically to IDH-MUT tumors. Because 2-HG causes epigenetic reprogramming, the authors examined the global DNA methylation in these tumors. Interestingly, the recurrent tumors retained the glioma CpG island methylator phenotype, a hallmark of IDH-MUT tumors, but exhibited marked hypomethylation in non-CpG island regions. Taken together, these results from the first longitudinal study on IDH-MUT LGGs uncover a possible tumor suppressive role for IDH in some instances where loss of the mutant allele leads to increased proliferation and demonstrate that IDH mutation is always not associated with recurrence. More significantly, with IDH-MUT inhibitors being tested in the clinic, it is important to assess the IDH status of the patients before administration of these agents. Tumor heterogeneity at the cellular and molecular level is now widely recognized as a feature of glioblastomas (GBMs). Spatial differences in the tumor microenvironment can substantially contribute to the heterogeneity of the tumor. For example, tumor cells at the invading front are exposed to a different environment than the central necrotic regions that can cause activation of oncogenic pathways differentially. Stem-like, tumor initiating cells in GBM (termed glioma stem cells, GSCs) are enriched in micronenvironmental niches, but the behavior of these cells and the molecular pathways altered in these specific niches are unknown. In a recent study, Jin et al1 used in silico approaches as well as profiling of GSCs from multiple regions of the same tumors and examined their molecular signatures with a goal to examine the niche-dependent heterogeneous distribution of GSCs and exploit molecular pathways for therapeutic targeting. Interestingly, they found that tumor cells from areas of angiogenesis were enriched with GSCs of the proneural (PN) subtype. In contrast, necrotic regions were enriched in cells with mesenchymal (MES) signatures. The authors hypothesized that these transcriptional differences could be governed by chromatin regulation and remodeling. Using chromatin immunoprecipitation of histone H3 K27 trimethylation and histone H2A K119 mono-ubiquitination representative of polycomb repressive complex (PRCs) catalysts EZH2 and BMI1, respectively, the authors found chromatin marks for PRC1 (H2AK119Ub) or PRC2 (H3K27 me3) in CD15+ GSCs depending on the region in which they were localized. H3K27Me3 targets were predominantly neuronal, overlapped with EZH2 activity and protein levels as well as OLIG2, and had a mature vascular signature. On the other hand, H2AK119Ub marked MES signaling and higher BMI1 and CD44 proteins. EZH2 or BMI activity was independently associated with poor survival, but the combined activity of these proteins/signatures associated with even worse outcome, implying the cooperative oncogenic signaling of PRC1 and PRC2. Intriguingly, while the expression of EZH2 mRNA expression correlated with its activation signature, BMI1 mRNA did not correlate with the BMI1 signature. Instead, the authors found that BMI1 protein was different between the PN and MES subtypes of GSCs and that its stability was regulated by RNF144A, a known ubiquitin E3 ligase. Depletion of RNF144A caused reduced poly-ubiquitination and degradation of BMI1, and RNF144A associated negatively with tumor grade and malignancy. Although EZH2 and BMI1 were independently required for survival of PN and MES GSCs, respectively, BMI1 was required for cell survival under stress, whereas EZH2 was dispensable in these conditions. Cell mixing experiments demonstrated a direct role for BMI1 in GSC propagation in necrotic regions. The authors then used pharmacological agents that target EZH2 and BMI1 and showed specific growth inhibitory effects in vitro as well as brain penetration and efficacy in intracranial xenografts. Importantly, combinatorial inhibition of BMI1 and EZH2 was more effective that single agents, suggesting that growth inhibition of heterogeneous PN and MES subtype of GSCs that are differentially localized in the brain tumor microenvironment can be achieved by simultaneous targeting of these proteins. This study uncovers growth-promoting microenvironmental niches that play a role and can be exploited for therapeutic targeting of GBMs. Alterations in metabolism are increasingly being explored as important drivers of gliomagenesis and, thus, as possible therapeutic targets. Using brain tumor initiating cells (BTICs) Wang et al1 explored the complex interplay between metabolic pathways and MYC, which is specifically overexpressed in the majority of GBMs and BTICs. The authors elegantly demonstrated in vitro as well as in vivo with mouse models that BTICs overexpress mevalonate pathway genes and the promoters of mevalonate pathway genes have higher levels of activity in BTICs. The mevalonate metabolic pathway produces cholesterol and plays a role in the localization of proteins in the cell. The rate-limiting step in the mevalonate pathway is the enzyme 3-hydroxyl-methylglutaryl coenzyme A reductase (HMGCR). Expression of HMGCR was elevated in BTICs and, when the authors looked at TCGA data, increased HMGCR was associated with a worse prognosis. Blocking HMGCR with a statin and a non-statin blocker of mevalonate pathway led to a decrease in BTIC viability and prolonged survival in mice implanted with BTICs, which was most likely from increased apoptosis rather than autophagy or senescence. To explore the link between MYC and the mevalonate pathway, the authors blocked MYC with a lentiviral-transduced shRNA. The result was reduced expression of all six mevalonate pathway genes tested at the mRNA and protein levels. The mouse model also demonstrated a modest increase in activated ERK levels,which, ultimately, led the authors to conclude that the mevalonate pathway regulates MYC through a RAS–ERK–miR33b pathway by creating a feed-forward loop. Overall, despite the grain of salt needed when using BTICs as a model, these results raise interesting questions about the potential therapeutic role for well-tolerated drugs like statins in glioma. Along these lines there was even a trial combining atorvastin with chemoradiation in newly diagnosed GBM that has completed accrual (NCT02029573). Therapeutic targeting of metabolic alterations is becoming a very promising avenue, since metabolism is a fundamental requirement for cell survival, and thus warrants more detailed exploration in humans via prospective clinical trials. Cancer stem cells (CSC) are considered to give rise to all progeny in a tumor. In order to investigate the contribution of CSC fate to the intratumoral heterogeneity of glioblastoma, Lan et al1 traced the evolution of individual cell clones upon serial orthotopic transplantations in immune-compromised mice. To this end, freshly resected glioblastoma cells were DNA barcoded before transplantation into mice. This allows tracing of the clonal evolution of individual cells containing a unique DNA barcode that can be revealed and quantified by sequencing to estimate the number and size of clones. Upon sacrifice of the tumor-bearing mice, the authors isolated human cells and re-injected them. Part of the cells underwent barcode analysis for quantification of the number and relative size of surviving cell clones, which was complemented by characterization of mutations and the transcriptome. Separate analyses of the injected versus the contralateral side informed on the invasiveness of the individual clones. Barcode analysis for a given patient sample revealed over 1500 detectable clones (from <3% barcoded cells) on the injected side, of which one third was also detectable on the contralateral side. These clones also overlapped largely with the clones found again in subsequent passages, suggesting that invasiveness is an inherent feature of CSCs. The number of clones was reduced over the passages; however, some clones became detectable only at later passages, from which the authors concluded that only a fraction of the cells had tumor-sustaining CSC properties. The authors tested the hypotheses of inherent differential fitness, e.g. driven by acquired genetic and epigenetic alterations versus equipotency of clones. The broad distribution of clone size that fit the model of “negative binomial dependence” was a common observation across the patient samples, serial passages, and hemispheres, which pointed to a model of equipotency of clones. The authors proposed a single-parameter model of conservative proliferative hierarchy of CSC to explain the observed heterogeneity of clones. Treatments with temozolomide exerted similar effects on most clones, with some exceptions of clone enrichment that together with the observation of the emergence of new clones suggested preexisting resistance. The presence of larger clones, outside the distribution, was considered to fit the proposed model after adding acquired resistance to apoptosis as an additional parameter. Some larger outlier clones were also observed in untreated tumors. In vitro drug screening followed by barcode analysis and chromatin state evaluation indicated interesting drug combinations, including epigenetic drugs, as promising strategies. This is an interesting and comprehensive study that provides a fresh look at the role of CSCs in the evolution of tumor heterogeneity and that may impact future treatment strategies. However, the effect of the microenvironment on clone selection remains to be determined. The fact that one third of glioblastomas failed to grow in the in vitro and in vivo assays highlights the limitation of our understanding of tumor-host interaction. Recent research suggests that mechanisms of gene splicing are aberrant in cancer cells and could be targeted in novel anti-cancer strategies.1 Deep RNA sequencing has revealed that cells carry a pool of partially spliced transcripts with “detained introns” (DIs) that remain in the pre-mRNA after all other introns have been removed. These DIs retain the RNA in the nucleus and are only spliced in response to specific cellular cues,2 thus regulating productive gene expression. An important regulatory factor of RNA splicing is the protein PRMT5, which methylates arginine residues in proteins that form part of the splicing complex (“spliceosome”). PRMT5 has been previously identified as an important therapeutic target in glioblastomas,3 but its mechanisms have not been elucidated. Using a screen of shRNAs against epigenetic regulatory factors, Braun et al4 observed that glioblastoma cells expressing PRMT5 shRNA were depleted from the overall cell population under a variety of in vitro and in vivo conditions. Genetic and pharmacological inhibition of PRMT5 caused cell senescence and inhibition of tumor growth, underscoring the relevance of this enzyme for tumor cell maintenance. Interestingly, PRMT5 inhibition caused the accumulation of DI-containing pre-mRNAs in the cell nucleus, arresting the productive expression of those genes. This effect was likely caused by the inhibition of PRMT5 methylation of core spliceosome proteins, specifically affected DI splicing with minimal impact on the splicing of other introns. Importantly, the pool of pre-mRNAs arrested in the nucleus after PRMT5 inhibition contained a preponderance of genes that regulate cell proliferation. PRMT5 inhibition also downregulated other genes involved in cell proliferation even when they lacked DIs, suggesting that inhibition of this enzyme has ample negative effect on the genetic program(s) that control the cell cycle. Overall, the investigators suggest that the upregulation of PRMT5 in glioblastoma makes these cells highly dependent on this enzyme to control the splicing of the unique DI-containing mRNA pool. This dependence creates an opportunity for strategies focused on targeting RNA splicing to cause widespread, inescapable damage in tumor cells.
Source: Neuro-Oncology
Language: English
Subfield: Genetics
Field: Medicine
Domain: Health Sciences
Open Access status: bronze
OpenAlex ID: W4252432706