|Home | About | Journals | Submit | Contact Us | Français|
Gliomas arise through genetic and epigenetic alterations of normal brain cells, though the exact cell of origin for each glioma subtype is unknown. The alteration-induced changes in gene expression and protein function allow uncontrolled cell division, tumor expansion and infiltration into surrounding normal brain parenchyma. The genetic and epigenetic alterations are tumor subtype and tumor-grade specific. Particular alterations predict tumor aggressiveness, tumor response to therapy and patient survival. Genetic alterations include deletion, gain, amplification, mutation and translocation which result in oncogene activation and tumor suppressor gene inactivation, or in some instances the alterations may simply be a consequence of tumorigenesis. Epigenetic alterations in brain tumors include CpG island hypermethylation associated with tumor suppressor gene silencing, gene-specific hypomethylation associated with aberrant gene activation, and genome-wide hypomethylation potentially leading to loss of imprinting, chromosomal instability and cellular hyperproliferation. Other epigenetic alterations, such as changes in the position of histone variants and changes in histone modifications are also likely important in the molecular pathology of brain tumors. Given that histone deacetylases are targets for drugs that are already in clinical trial, surprisingly little is known about histone acetylation in primary brain tumors. While a majority of epigenetic alterations are independent of genetic alterations, there is interaction on specific genes, signaling pathways and within chromosomal domains. Next-generation sequencing technology is now the method of choice for genomic and epigenome profiling, allowing more comprehensive understanding of genetic and epigenetic contributions to tumorigenesis in the brain.
Epigenetics is defined as mitotically heritable changes in gene expression that are not due to changes in the primary DNA sequence. Epigenetic mechanisms include enzymatic modification of DNA and associated histone proteins that regulate and maintain gene expression states, and have important roles in chromosome structure and stability. The discovery of altered epigenetic profiles in human neoplasia has led to a new paradigm in which both genetic and epigenetic mechanisms contribute significantly to cancer and perhaps many other common human diseases. Because of their reversible nature, epigenetic alterations are being targeted therapeutically in cancer clinical trials.
Covalent modifications of DNA and amino acids on histones are two major mechanisms of epigenetic gene regulation. DNA methylation involves the addition of a methyl group to cytosine to create 5-methylcytosine. In mammals methylation occurs primarily at 5′-CpG-3′ dinucleotides, and occasionally at CpNpGs as well.1 DNA methylation is controlled by DNA methyltransferases (DNMT) that create (DNMT3A, DNMT3B) or maintain (DNMT1) patterns of methylation.2, 3 DNA methylation is required for maintaining gene silencing on the inactive X chromosome,4-6 parental allele-specific expression of imprinted loci,7 and tissue and cell-type-specific gene expression. Methylation is also required for silencing transposable elements and maintaining genome stability8, 9 and is a critical regulator of pluripotency genes.10-12
A second type epigenetic mechanism is the post-translational modification of N-terminal tails of histone proteins by acetylation, methylation, phosphorylation, ubiquitylation, sumoylation, ADP ribosylation, biotinylation and potentially other modifications.13 Like DNA methylation, specific enzymes also catalyze post-translational modifications of histones, and include acetyltransferases and deacetylases, methyltransferases and demethylases, among others. In contrast to DNA methylation, histone methylation can be mono-, di-, or tri-methylated on a single, specific lysine (i.e., at H3K4). Multiple types of modifications are present on a single histone molecule, increasing the combinatorial complexity, referred to as a “histone code.” In addition to DNA methylation and histone modifications, there are other interrelated, potentially epigenetic mechanisms including specific deposition of histone variants, non-coding RNAs, chromatin remodeling, and nuclear organization of DNA. These mechanisms add additional layers of regulation and maintenance of gene expression states in both normal and diseased tissues.
Epigenetic mechanisms are critical to the development and function of the mammalian central nervous system (CNS). Global DNA methylation levels change during brain development,14 and methylation of specific genes can vary among different brain regions, cell types and potentially even between the same cell type from different brain regions.15 The relationship between brain-region-specific gene expression16 and brain-region-specific DNA methylation has not been fully explored however. Distinct CNS cell types may be differentially marked by DNA methylation. In murine astrogliogenesis for example, DNA demethylation of the Gfap promoter including a critical STAT3 DNA binding site is associated with activation of Gfap transcription, a known marker of the astroglial lineage.17 One particular CpG site within the Gfap promoter is methylated in neural precursors and postmitotic neurons but unmethylated in astrocytes.17-19
DNA methyltransferases are important in CNS development and function. The maintenance methyltransferase DNMT1 is highly expressed in the mammalian brain20-22 including in postmitotic neurons, despite its proposed primary role in copying methylation during DNA replication. DNA methylation changes in response to neural activity may be one function of DNMT1 in postmitotic neurons.23 Conditional Dnmt1 deletion in murine postmitotic neurons does not affect overall DNA methylation levels or cell survival.24 On the other hand, conditional Dnmt1 deletion in embryonic day 12 neuroblasts resulted in DNA hypomethylation and lethality immediately after birth due to CNS-associated respiratory failure, indicating a requirement for DNA methylation in these cells. Mosaic mice with ~30% Dnmt1-/- cells survived into adulthood, but mutant cells were rapidly eliminated from the brain within three weeks of birth, further supporting the necessity of DNMT1 for CNS cell survival. Dnmt3b is expressed in the murine CNS for a short time during neurogenesis, whereas Dnmt3a is expressed in both prenatal and postnatal CNS.25 Dnmt3b-/- mice exhibit prenatal lethality and neural tube defects, demonstrating a critical role for DNMT3b in neurodevelopment.26 Mice with conditional deletion of Dnmt3a in the nervous system are born apparently healthy but die prematurely, displaying hypoactivity, abnormal walking, and poor performance on tests of neuromuscular function and motor coordination.27 Thus, the effects of Dnmt deficiency on brain functions are significant, but the specific effects on neurons and glia require further investigation.
Recent evidence indicates that neuronal differentiation is regulated in part by DNA demethylation and polycomb-mediated histone H3 K27 trimethylation (H3K27me3).28 DNA methylation contributes to repression of pluripotency in lineage-committed neural progenitors. Also, promoters marked by H3K27me3 in neural stem cells often gain DNA methylation during differentiation.29 Thus, context-dependent interactions between different epigenetic mechanisms guide neural differentiation, and underscore the general importance of epigenetics in normal CNS development and maintenance of cellular identity.
Several human neurodevelopmental disorders are caused by mutations in genes encoding proteins involved in epigenetic mechanisms. Rett syndrome, for example, is a severe neurodevelopmental disorder caused by mutations in MECP2, which encodes a protein that can bind to methylated DNA and regulate gene expression.30 ICF Syndrome, which includes mental retardation, is caused by mutations in the de novo DNA methyltransferase DNMT3B.31-33 Furthermore, the dependence of the CNS on epigenetic regulation extends beyond DNA methylation and DNA methyltransferases: mutations in genes encoding other epigenetic regulatory proteins can cause neurodevelopmental disorders. One example is JARID1/SMCX, encoding a JmjC-domain-containing histone demethylase, which, when mutated, causes a form of X-linked mental retardation.34-36 These examples illustrate the importance of epigenetic control of gene expression in the development and function of the CNS. This knowledge of the critical role of epigenetic mechanisms and marks in the CNS provides a foundation for understanding the role of epigenetic regulation in tumors arising from CNS cells.
Genome-wide or “global” hypomethylation occurs at a high frequency (~80%) in primary glioblastoma (GBM).37-39 The level of hypomethylation varies between tumors, ranging from near normal brain levels to approximately 50% of normal in approximately 20% of the cases, reflecting the massive demethylation of approximately 10 million CpG sites per tumor cell. Both single-copy loci and repetitive sequences can be hypomethylated in GBM. The most severe globally hypomethylated primary GBMs are also the most proliferative and are associated with demethylation and transcriptional activation of the putative oncogene MAGEA1.37, 40 MAGEA1, a member of the MAGE family of genes, is one of a group of germline-specific genes that become transcriptionally activated in multiple types of human cancers, and this activation is correlated with genome-wide hypomethylation and increased cellular proliferation.37, 41 The expression of the protein products of these “cancer-testis antigen” genes in tumors results in the recognition of tumor-specific antigens on the cell surface by cytolytic T lymphocytes and also inhibits p53 function and response to chemotherapy in cancer cell lines.42, 43 Interleukin-13Ra2 is another classic cancer-testis antigen that is epigenetically activated in gliomas.44, 45 IL13Ra2 expression is cancer cell-specific, and antibodies directed to it have been used to deliver toxic agents such as Pseudomonas exotoxin to glioma cells in patients. Although the full consequences of genomic hypomethylation are unknown, it is sufficient to initiate tumorigenesis in mice, and can modulate tumor incidence in mouse models of cancer driven by genetic alterations.8, 46-48 The molecular consequences of global hypomethylation in gliomas have not been studied in-depth.
Locus-specific hypermethylation, mostly at CpG island (CGI) promoters, is frequent in gliomas. CGIs are regions of ~ 500 bp – 1kb where CpG nucleotides are ~5 times more abundant compared to the rest of the genome.49 In gliomas, CGI promoter hypermethylation occurs at genes with diverse functions related to tumorigenesis and tumor progression, including cell-cycle regulation, DNA repair, apoptosis, angiogenesis, invasion, and drug resistance. For example, the retinoblastoma (RB), PI3K, and p53 pathways are affected by CGI promoter hypermethylation (including promoters of the genes CDKN2/p16, RB, PTEN, TP53 and p14ARF).50-55 CGI hypermethylation in gliomas may also occur at genes that are not expressed in the brain, suggesting not all CGI methylation events are functionally important for tumorigenesis.
Screens for promoter hypermethylation in GBM have identified new tumor suppressor candidates as well as bona fide tumor suppressors. One example is at chromosome 19q, where deletions in gliomas and neuroblastomas have suggested the presence of a tumor suppressor gene. Expression array analysis of glioma cells treated with a DNA demethylating agent discovered a candidate tumor suppressor, epithelial membrane protein 3 (EMP3), a myelin-related gene involved in cell proliferation and cell-cell interactions.56 EMP3 is silenced by hypermethylation in primary gliomas and reintroduction of EMP3 in neuroblastoma cell lines with EMP3 silencing resulted in reduced colony formation in vitro and decreased xenograft growth in mice, suggesting tumor suppressor function.56
Promoter hypermethylation regulates the oncogenic and proliferation-promoting transforming growth factor (TGF)-beta signaling pathway in aggressive, highly proliferative GBMs. High levels of TGF-beta signaling are normally associated with poor prognosis. TGF-beta signaling promotes proliferation through the induction of platelet-derived growth factor (PDGF)-B. However, epigenetic silencing of PDGF-B can override the increased proliferative effects of TGF-beta signaling. Specifically, PDGF-B promoter hypermethylation prevents PDGF-B transcriptional activation by TGF-beta-induced Smad proteins.57 The oncogenic affect of the TGF-beta pathway is therefore blocked by epigenetic alteration of one of its targets.
Genes involved in invasion and metastasis can also be affected by promoter hypermethylation in gliomas. A high frequency of astrocytomas (88%), GBMs (87%), and glioma cell lines (100%) exhibit CGI promoter hypermethylation of the protocadherin-gamma subfamily A11 (PCDH-gamma-A11) gene, which is thought to be important in invasion of cancer cells into normal brain parenchyma.58 However, the use of the sensitive methylation-sensitive PCR (MSP) method for detecting methylation may overestimate the percentage of tumors in which methylation reaches a biologically meaningful level.
Promoter hypermethylation can modulate sensitivity to drugs and radiotherapy in GBM. The best known example is MGMT promoter methylation and response to DNA alkylating agents (see below), but there are other cases and this research area is likely to expand in the future.59, 60 This suggests that epigenetic profiling might be one way to categorize GBMs and to rationally apply patient-specific therapy.
For other sets of genes, promoter DNA hypermethylation might not be causal or may not be required for gene silencing in cancer. For example, the enhancer of zeste 2 (EZH2) histone methyltransferase silences approximately 5% of gene promoters in prostate cancer cells by histone H3 lysine trimethylation independent of DNA methylation.61 The order and hierarchy of epigenetic modifications in the process of gene silencing is an area of extensive investigation and will likely yield clues about potential primary causes of epigenetic silencing in GBM and other cancers.
One particularly important example of epigenetic silencing of a drug-resistance gene is promoter hypermethylation of O6-methylguanine-DNA methyltransferase (MGMT) in GBM. MGMT encodes a DNA repair protein that removes alkyl adducts at the O6 position of guanine and (less frequently) at the O4 position of thymine.62 Expression of MGMT protects normal cells from carcinogens; however, it can also protect cancer cells from chemotherapeutic alkylating agents. MGMT promoter methylation is negatively correlated with expression and is associated with decreased transcription factor binding in GBM cell lines.63, 64 An initial report suggested a paradoxical positive correlation between MGMT methylation and expression in cancer cell lines.65 However, updates in the genomic structure of the MGMT gene showed that the paradoxically higher level of methylation in MGMT-expressing cells was located 50kb downstream of the MGMT promoter, well into the gene body. Hypermethylation of the MGMT promoter, leading to transcriptional silencing, occurs in multiple types of human cancers, including glioma, lymphoma, breast cancer, prostate cancer, and retinoblastoma. 66 In human colorectal cancer, MGMT promoter methylation is associated with increased G to A transition mutations in TP53 and K-RAS.67, 68 In GBM, MGMT promoter methylation correlates with mismatch repair deficiency and a putative hypermutator phenotype.69
MGMT hypermethylation is associated with significantly longer survival in patients with GBMs and low-grade gliomas treated with radiation and alkylating agents, including temozolomide,70, 71 although it is unclear if this is directly due to reduced MGMT expression. Furthermore, MGMT hypermethylation is a predictor of the incidence and outcome of pseudoprogression,72 the progressive and enhancing lesions observed by MRI immediately after the end of treatment that are an effect of the radiotherapy and temozolomide treatment rather than tumor progression.72 MGMT hypermethylation is a late event in progressive oligodendrogliomas, suggesting that MGMT methylation may have a different prognostic value for oligodendrogliomas than for GBMs and low-grade gliomas.73 The predictive value of MGMT methylation may be increased when used in conjunction with gene expression data.74 However, further technical refinement is needed for the commonly used MGMT methylation assay: particularly when applied to fixed tissue, the MSP method is prone to false positives.75
Different glioma subtypes and tumor grades display distinct aberrant DNA methylation profiles.76 Secondary GBMs have a higher overall frequency of promoter methylation compared to primary GBMs at least for the promoters of p14ARF, p16INK4a, RB1, MGMT and TIMP-3.77 Low-grade gliomas and secondary GBMs show PTEN promoter methylation and activation of the PI3K pathway as measured by protein kinase B (PKB/AKT) phosphorylation, whereas PTEN promoter methylation is a rare occurrence in primary GBMs.78 On the other hand, the same epigenetic alterations can be shared by different types of brain tumors, even in different species.79 CNS cancers other than gliomas also display distinct methylation profiles: for example, hMLH1, TIMP3, MGMT, p73, and THBS1 are frequently hypermethylated in schwannomas,80 and NF2 is genetically and epigenetically altered in meningioma.81
The progression of glioma over time is associated with distinct epigenetic patterns. Malignant progression and shorter survival in astrocytoma are associated with p14ARF but not MGMT hypermethylation, suggesting that these are two distinct pathways in astrocytoma with different clinical consequences.82 Recurrence is also characterized by specific epigenetic marks: hypermethylation of the pro-apoptotic gene caspase-8 is frequently associated with relapsed GBM.83 Increased recurrence-free survival is associated with CITED4 hypermethylation at 1p43.2 and 1p and 19q losses.84 Murine glioma models also provide evidence of a role for epigenetic mechanisms in modifying tumor progression. In a mouse model of astrocytoma with mutant Trp53 and Nf1, a genetic modifier controlling susceptibility to progression is itself epigenetically modified.85
DNA hypermethylation can alter the differentiation properties of the fraction of glioma cells thought to be putative cancer stem cells, also called tumor-initiating cells (TICs). Normal astroglial differentiation is regulated by the bone morphogenetic protein (BMP) - and ciliary neurotrophic factor (CNTF) - mediated Jak/STAT pathway. In a subset of GBM TICs, this differentiation pathway is inhibited by epigenetic silencing of the BMP receptor 1B (BMPR1B) gene.86 This silencing is dependent on the histone methyltransferase enhancer of zeste 2 (EZH2) and DNA methylation, and could be reversed by treatment with a DNA demethylating agent. These data demonstrate that a subset of human GBM cells are blocked from differentiation though a defect in the BMP signaling pathway. Furthermore, forced expression of BMPR1B could restore the differentiation potential of these cells and can decrease tumorigenicity. These findings reinforce the idea that future epigenetic therapies might be applied rationally, guided by patient-specific epigenetic profiling.
The gene encoding the CD133 cell-surface marker used to identify and enrich GBM TICs is itself subject to epigenetic regulation. In GBM cell lines, CD133 promoter CGI DNA methylation was negatively correlated with CD133 expression.87 In pure CD133+ GBM lines, this methylation was absent. Furthermore, CD133 methylation was not observed in normal brain but only in cultured and primary tumors, suggesting that it could be a tumor-specific epigenetic marker. In contrast, another study found CD133 promoter CGI methylation in brain and hypomethylation in GBMs with high expression of aberrant of CD133 transcripts.88 Follow-up studies are needed to more thoroughly appreciate the functional importance of these results.
There are multiple strategies for discovering novel epigenetic alterations in brain tumors. One approach is to examine candidate loci that are affected by deletion or mutation on one allele, based on the idea that DNA hypermethylation might inactivate the other allele, leading to complete silencing or loss of normal function. This candidate gene strategy is more likely to discover hypermethylation of tumor suppressors rather than hypomethylation events. A more unbiased approach is to scan the entire genome for both hyper- and hypomethylation events. For gliomas, this was approached first through the use of methyl-sensitive restriction enzymes using restriction landmark genomic scanning (RLGS).89-94 Newer methods such as reduced representation bisulfite sequencing95 and methylated DNA immunoprecipitation (MeDIP) combined with DNA microarrays96 or massively parallel sequencing97 more completely interrogate the epigenome and are poised to dramatically augment our understanding of epigenetic dysregulation in gliomas and other cancers.
An additional approach to uncovering epigenetically altered genes is to treat cells with the DNMT1 inhibitor 5-aza-2′-deoxycytidine (5-aza) and/or the histone deacetylase inhibitor trichostatin A (TSA) followed by gene expression microarray analysis.98 This strategy identified brain expressed X-linked 1 (BEX1) and 2 (BEX2) as candidate tumor suppressors in glioma99 and SPINT2 in medulloblastoma.100 Another study identified >160 genes upregulated by combined 5-aza and TSA treatment in glioma cell lines and provided evidence that several of the epigenetically silenced loci are aberrantly methylated by the combined action of DNMT1 and DNMT3B, whereas others are regulated only by a single DNMT.101 5-aza treatment of short-term cultured primary GBMs identified two genes, runt-related transcription factor 3 (RUNX3) and testin (TES), with aberrant methylation.102 In this chemical approach, numerous genes may be activated indirectly. Thus, follow-up analyses of the epigenetic marks at the activated genes in primary tumors are essential.
In addition to DNA hypermethylation, cancer genes can be silenced or activated by aberrant patterns of histone modifications, either alone, or in combination with DNA methylation. Silenced CGI promoters, for example, are characterized by increased histone H3K9 methylation and loss of H3K9 acetylation. In embryonic stem cells, the dual presence of inactivating H3K27 methylation and activation-associated H3K4 methylation, called bivalent domains, is thought to create a “poised” chromatin state for developmentally regulated genes, allowing silencing in embryonic stem cells and subsequent transcriptional activation or repression in differentiated cells.103 Bivalent domains, along with additional repressive marks (dimethylated H3K9 and trimethylated H3K9), are found in embryonal carcinoma cells at genes that are frequently silenced by DNA hypermethylation in adult human cancer cells. These histone modifications are hypothesized to predispose tumor suppressor genes to DNA hypermethylation and heritable gene silencing.104 Trimethylation of H3K27 marks silenced genes in the absence of DNA hypermethylation, although this has not yet been shown in gliomas.
Genes encoding DNMTs are not frequently mutated in human cancers, with one report of DNMT1 mutations in colorectal cancer.105 However, overexpression of DNMT1, 3a, and 3b has been reported in multiple tumor types.106 Furthermore, there are many instances of genetic alterations and/or deregulated expression of genes encoding histone-modifying enzymes. In acute leukemias for example, translocations involving the mixed lineage leukemia (MLL) gene, encoding an H3K4 methyltransferase, are common.107 These translocations result in MLL fusion proteins that have lost H3K4 methyltransferase activity. The polycomb protein EZH2, an H3K27 methyltransferase, is overexpressed and amplified in multiple types of human cancers.108
The gene encoding BMI-1, a member of the polycomb group complex that regulates histone H3K27 methylation, is subject to frequent copy number alterations in both low- and high-grade gliomas, and BMI-1 deletions are associated with poor prognosis in patients.109 Furthermore, BMI-1 is also upregulated in a proportion of gliomas.110 An important target of BMI-1 transcriptional repression is the Ink4a/Arf locus, harboring two tumor suppressors in gliomas. Repression of this locus by BMI-1 promotes cell proliferation.111, 112 However, in an orthotopic transplantation model, BMI-1 is also required for tumorigenesis independent of Ink4a/Arf, suggesting additional targets of epigenetic regulation by BMI-1 may be important in glial tumorigenesis.113, 114
Mutations resulting in altered histone acetyltransferase (HAT) activity also occur in cancer-related diseases: CBP mutations, abolishing HAT activity, cause Rubenstein-Taybi syndrome, a developmental disorder that is associated with a higher risk of cancer.115 In GBM, there is preliminary evidence for deregulation of genes controlling histone modifications. Expression of some histone deacetylase (HDAC) proteins is reported to be altered in GBM. Class II and IV HDACs displayed decreased mRNA expression in GBMs compared to low-grade astrocytomas and normal brain, and overall histone H3 was more acetylated in GBMs.116 Large-scale sequencing of protein-coding genes in GBMs uncovered mutations in many genes involved in epigenetic regulation, including histone deacetylases HDAC2 and HDAC9, histone demethylases JMJD1A and JMJD1B, histone methyltransferases SET7, SETD7, MLL, MLL4 and methyl-CpG binding domain protein 1 (MBD1).117 These intriguing initial studies suggest that alterations in epigenetic mechanisms could be a major defect in GBM.
The cause(s) and consequences of epigenetic dysregulation in GBM, and cancer in general, remain an area of intense investigation, with some progress being made in understanding the consequences. The causes, however, remain mostly obscure. Multiple causes have been proposed for aberrant DNA methylation. Alteration in DNA methyltransferase enzymes is a possibility, and aberrant expression of DNMTs has been observed. For example, DNMT3b splice variants and aberrant transcripts are overexpressed or ectopically expressed in some cancers and cancer cell lines. The splice variant DNMT3b4, which lacks the conserved methyltransferase motifs IX and X, is overexpressed in hepatocellular carcinoma.118 Furthermore, its forced overexpression caused pericentromeric hypomethylation, suggesting that DNMT3b4 might act in a dominant negative manner, blocking methylation of this region of repetitive DNA. Another study of primary acute leukemia cells and cancer cell lines (including GBM cell lines) identified expression of over 20 aberrant DNMT3b transcripts with altered 5′ splicing, coding for truncated proteins that lack the C-terminal catalytic domain.119 Although aberrant DNMT3b transcripts account for only a small fraction of total DNMT3b mRNA in this study, at least some of these ar translated to proteins, and their overexpression was linked to changes in DNA methylation level and mRNA abundance of other genes. Of these transcripts, DNMT3b4 is the only one shown to produce a dominant negative-acting protein in cancer cells, leading to hypomethylation. Increased DNMT1 and decreased DNMT3a expression have also been reported in GBM cell lines, and the decrease in DNMT3a is hypothesized to contribute to Sat2 hypomethylation in the tumor stem cell subpopulation.39
In theory, either an initial genetic or epigenetic abnormality could result in tumorigenesis and cause further downstream genetic and epigenetic changes. In glioma, epigenetic silencing of the tumor suppressor gene WNK lysine deficient protein kinase 2 (WNK2) was significantly associated with combined deletion of chromosomes 1p and 19q.120 WNK2 indirectly inhibits MEK1, enhancing growth promoting signals through EGFR, which is often itself overexpressed and constitutively active due to genomic amplification and rearrangement, respectively.121, 122 Thus, it is possible that the epigenetic silencing of WNK2 interacts on a functional level with genetic alteration of EGFR signaling, a common abnormality in GBM. A GBM with a point mutation in WNK2 was reported recently,117 while commonly occurring cancers exhibit relatively frequent point mutations in all four WNK genes. In meningioma, grade-specific epigenetic alterations in WNK2 are the dominant mechanism of gene inactivation, occurring in 83% and 71% of Grade II and III tumors.123
Epigenetic-based therapy such as the DNMT inhibitor decitabine (5-aza-2′-deoxycytidine) and the HDAC inhibitor (HDACi) suberoylanilide hydroxamic acid (SAHA; vorinostat) are currently being tested in multiple cancers, although only HDACis are in trials for GBM treatments. In contrast to genetic mutations, which are “hard-wired’ once mutated, epimutations such as promoter hypermethylation and histone acetylation status are theoretically reversible by drug treatment or possibly diets.
A major unresolved issue for epigenetic therapy of cancer is target specificity. First, some genes that require DNA methylation or histone deacetylation for silencing in normal cells could be unintentionally activated by agents that inhibit DNMTs or HDACs. Second, cancer genomes are characterized by both DNA hyper- and hypomethylation. Therefore, using drugs that reactivate silenced tumor suppressors may have the undesired effect of further activating oncogenes through hypomethylation. These problems should be addressed to gain a more complete understanding of the molecular events resulting from epigenetic-based therapy.
HDACs catalyze the deacetylation of lysine residues within the N-terminal tails of core histones and also within non-histone proteins. As a result, their effects are complex and involve histone and non-histone substrates, and the mechanism of specificity for each HDAC is not fully understood. In general, HDACs promote a closed chromatin structure that represses transcription. There are 18 known HDACs in humans, divided into 5 main classes, with different target specificities.124 HDACs of Class I (HDACs 1, 2, 3 and 8), Class IIA (HDACs 4, 5, 7 and 9), Class IIB (HDACs 6 and 10), and Class IV (HDAC 11) all contain zinc in their active sites and are inhibited by the HDACis TSA and SAHA. Class III HDACs (sirtuins) do not contain zinc and are not inhibited by TSA or SAHA.
The rationale for using HDACis in cancer therapy is two-fold. First, HDACis promote a more open chromatin conformation and might therefore permit better access of DNA damaging agents to the chromatin and increase the sensitivity to killing by these agents. Second, HDACis will reverse some of the aberrant epigenetic gene silencing in GBMs, presumably leading to enhanced cell-cycle arrest and apoptosis from DNA-damaging agents.125 While it is not clear how consistently HDACis activate specific pathways from one GBM to the next, HDACis do synergize with DNA damaging agents in arresting or killing glioma cells in vitro.
HDACis are comprised of several classes of compounds, including hydroxymates (SAHA, TSA), cyclic peptides (depsipeptide), aliphatic acids (valproic acid, butyrate), and benzamides. No single HDACi is effective against all HDACs. HDACis cause increased acetylation of histone and non-histone proteins and can reactivate p21, which contributes to cell-cycle arrest.126, 127 Non-cancerous cells are more resistant to the effects of HDACi, but the reasons for this selective sensitivity are unclear.128 HDACis alter the expression levels of only a subset of expressed genes in transformed cells (~2-10%), and both increases and decreases in transcript levels are observed.129-132
SAHA is currently being tested as a monotherapy or in combination therapy in 5 phase I or I-II clinical trials for gliomas. SAHA targets class I and II HDACs at micromolar concentrations, and preclinical studies found that it sensitizes glioma cells in vitro, ex vivo, and in vivo to chemotherapy and radiation.133-137 SAHA treatment increased p21 promoter histone H3 acetylation in the U87 glioma cell line, and inhibited the proliferation of GL26 glioma cells implanted in mice.126 Two of the current clinical trials test SAHA in combination with temozolomide, and one also includes radiotherapy. A third trial consists of SAHA plus isotretinoin and carboplatin, a fourth uses SAHA plus bortezomib, and a fifth is testing SAHA as a monotherapy.
Trials are also currently underway for two additional HDACis, valproic acid (Depakene; Depakote) and depsipeptide (Romedepsin; FK-228), and there are additional HDACi compounds that are not yet in clinical trials. Valproic acid is being tested against GBM in combination with temozolomide plus radiation, and in a broader second trial against neuronal tumors and brain metastases in combination with etoposide. Depsipeptide monotherapy is being tested against high-grade gliomas. Valproic acid is active against Class I and IIA HDACs at millimolar concentrations; depsipeptide is active against Class I HDACs at nanomolar concentrations. An additional HDACi not yet in clinical trials is pivaloyloxymethyl butyrate (AN-9), a derivative of butyrate. Butyrate is an aliphatic acid HDACi effective against Class I and IIA HDACs at millimolar concentrations. AN-9 shows efficacy in GBM cell culture and animal models.138, 139 Furthermore, AN-9 sensitized GBM mouse xenografts to radiation and showed decreased tumor growth and increased survival. There are several HDACis that have shown efficacy against cancer cells but have not yet been tested for gliomas including panobinostat (LBH589)140 and belinostat (PXD101).141 The discovery and development of new epigenetic enzyme-targeting compounds is an area of active research in the pharmaceutical industry.
Blood plasma in cancer patients contains DNA derived from tumor cells due to necrotic or apoptotic cancer cells releasing genomic DNA. This potentially provides a less invasive method for biomarker detection. Aberrantly hypermethylated cancer genes found in plasma could be one such type of biomarker. This type of analysis may provide clinically useful information relating to diagnosis, prognosis, and follow-up after therapy. Methylated tumor-specific biomarkers have, in fact, been found in plasma and serum for several types of cancer.142, 143 There is a significant amount of tumor DNA in the plasma of high-grade glioma patients, and in 60% of patients, the same methylated promoters (p16INK4a, MGMT, p73, and RARbeta were tested) could be detected in both tumor and plasma DNA.144 The frequency of detection of methylated promoters in plasma of GBM patients was similar to that in other cancers, suggesting that the blood-brain barrier does not completely block the transfer of tumor DNA to plasma. This method is theoretically highly sensitive, since small amounts of methylated DNA can be detected in a sample by MSP.
Epigenetic studies of glioma are poised to 1) make substantial contributions to the understanding of glioma biology, 2) identify new predictive biomarkers, and 3) discover novel targets for therapy. New models, such as patient GBM-derived tumor stem cells grown in neurosphere culture may be a valuable addition to epigenetic research into GBM, particularly if the epigenetic profiles of the corresponding primary tumors are retained, as has been shown for gene expression patterns and invasive growth patterns of these cells.145, 146 Epigenomic profiling of DNA methylation, histone modifications, and non-coding RNAs (such as microRNAs) in primary tumors, orthotopic xenografts, and tumor neurospheres are strategies that will likely uncover many additional epigenetic alterations in gliomas, and therefore targets for therapy.
There are still many questions remaining about the role of epigenetics in gliomas. The causes and consequences of epigenetic alterations are still mostly unknown, and the relative contributions of genetic and environmental factors in causing epigenetic alterations have not been quantified. Why are some genes or pathways more affected by epigenetic rather than genetic alterations or vice versa? It is clear that approaches simultaneously examining both genetic and epigenetic defects, complemented with functional studies, will be essential. It will also be important to understand the effects of HDACi on the entire cancer acetylome to elucidate the molecular consequences of this treatment strategy.
Combination therapy with both DNMT and HDACis might be one strategy against glioma. A dual treatment approach may have a synergistic effect on gene activation, and could allow lower doses of each drug to be used. Such a strategy is being tested in a clinical trial for myelodysplastic syndrome (MDS) and acute myelogenous leukemia (AML) using the DNMT inhibitor decitabine with or without valproic acid (clinicaltrials.gov ID NCT00414310).
An area that is mostly unexplored in glioma is the development and testing of drugs directed against histone modifications other than acetylation. H3K27 methylation at silenced tumor suppressor promoters could be targeted to reactivate these genes, for example using the S-adenosylhomocysteine hydrolase inhibitor 3-Deazaneplanocin A (DZNep).147 As epigenetic modifications are better understood and more types are discovered, additional epigenetic drug targets can be tested in GBMs and other cancers.
Consortiums using high-throughput genomic and epigenomic approaches are helping to unravel the genetic and epigenetic alterations that contribute to GBM. The Cancer Genome Atlas (TCGA; http://cancergenome.nih.gov/about/mission.asp) is examining genomic changes and DNA methylation in three cancers, including GBM.69 A complementary, privately funded project sequenced the exons of 20,661 protein coding genes in GBM, identifying recurrent mutations in the active site of isocitrate dehydrogenase 1 (IDH1) in 12% of GBM patients,117 and 70-80% of low-grade glioma patients. The International Cancer Genome Consortium (ICGC) is an additional large-scale effort, which will include genomic and epigenomic profiling. Considering there are 30,000 CGIs in the human genome, tiling microarrays or next-generation sequencing methods may be required to fully investigate aberrant CGI DNA methylation in CNS tumors. A full genome-scale map of DNA methylation and histone modifications in glioma will be a key reference for accelerating understanding of epigenetic mechanisms underlying this deadly disease.
JF Costello is supported in part by the National Brain Tumor Society, Accelerated Brain Cancer Cure, Goldhirsh Foundation and the Karen Osney Brownstein Endowed Chair. RP Nagarajan is supported by a postdoctoral T32 NRSA (NIH Institutional Research Service Award in Molecular and Cellular Mechanisms in Cancer).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.