TIPS-1113; No. of Pages 10
Review
Exploiting epigenetic vulnerabilities for cancer therapeutics Barbara Mair, Stefan Kubicek, and Sebastian M.B. Nijman CeMM Research Center for Molecular Medicine of the Austrian Academy of Sciences, Vienna, Austria
Epigenetic deregulation is a hallmark of cancer, and there has been increasing interest in therapeutics that target chromatin-modifying enzymes and other epigenetic regulators. The rationale for applying epigenetic drugs to treat cancer is twofold. First, epigenetic changes are reversible, and drugs could therefore be used to restore the normal (healthy) epigenetic landscape. However, it is unclear whether drugs can faithfully restore the precancerous epigenetic state. Second, chromatin regulators are often mutated in cancer, making them attractive drug targets. However, in most instances it is unknown whether cancer cells are addicted to these mutated chromatin proteins, or whether their mutation merely results in epigenetic instability conducive to the selection of secondary aberrations. An alternative incentive for targeting chromatin regulators is the exploitation of cancerspecific vulnerabilities, including synthetic lethality, caused by epigenetic deregulation. We review evidence for the hypothesis that mechanisms other than oncogene addiction are a basis for the application of epigenetic drugs, and propose future research directions. Anticancer therapy and epigenetics The success of an anticancer drug critically depends on its ability to kill cancer cells without causing severe side effects in healthy tissues. Hence, most therapeutics in the clinic today exploit inherent differences between cancer cells and normal cells such as defective DNA repair pathways and checkpoints, and the presence of oncogenic kinases. Another important feature that discriminates tumor from normal cells involves epigenetic deregulation. In normal cells, gene expression is tightly controlled by epigenetic mechanisms that regulate chromatin accessibility for the transcriptional machinery through post-translational modifications of histones (e.g., methylation, acetylation, and phosphorylation) and DNA (methylation and hydroxymethylation). Modified chromatin in turn recruits ‘reader’ proteins to carry out further regulatory functions. However, many cancers display global changes in histone modifications, aberrant Corresponding authors: Kubicek, S. ([email protected]); Nijman, S.M.B. ([email protected]). Keywords: chromatin-modifying enzymes; drugs; therapeutics; synthetic lethality; non-oncogene addiction. 0165-6147/$ – see front matter ß 2014 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.tips.2014.01.001
methylation patterns, and mutations in or altered expression levels of key chromatin-modifying proteins, and some tumors even harbor mutations in histone variants. Here we discuss how drugs that inhibit epigenetic modifiers can specifically target cancer cells by exploiting this deregulated chromatin state, including the targeting of synthetic lethal interactions (also known as non-oncogene addiction or induced essentiality; Box 1). We only briefly touch on epigenetics and cancer (Box 2) and refer readers to recent reviews on these topics [1–8]. Network rewiring in cancer generates new synthetic lethal targets During tumorigenesis, normal cells acquire a set of hallmark characteristics that allow survival and proliferation even in the presence of counteracting signals [9–11]. This dramatic transition is reflected by extensive rewiring of cellular signaling networks, and is driven by mutations and the deregulated expression of oncogenes and tumor suppressor genes [12,13]. Thus, the cellular networks of cancer cells are distinct from those of normal cells, and most of the mechanisms of action of anticancer drugs can be explained by understanding the rewiring involved. Such network changes can lead to a critical reliance on an oncogenic driver event (oncogene addiction, Box 1 and Figure 1A–C), which underlies most of the targeted agents that inhibit overactive oncogenes such as BCR-ABL1 (breakpoint cluster region c–abl oncogene 1 fusion protein), BRAF (v-raf murine sarcoma viral oncogene homolog B), EGFR (epidermal growth factor receptor), and HER2 (ERBB2; v-erb-b2 avian erythroblastic leukemia viral oncogene homolog 2) inhibitors. Network rewiring can also result in strict dependence on the activity of gene products that are not essential in normal cells. Thus, inactivation of such a de novo essential gene through mutation would lead to cell death or a decrease in fitness. This phenomenon has traditionally been studied in model organisms such as the budding yeast Saccharomyces cerevisiae and the fruit fly Drosophila melanogaster, and has been termed synthetic lethality or synthetic sickness (Box 1) [14–17]. The concept of synthetic lethality (in a broader sense also termed non-oncogene addiction or induced essentiality) has been proposed as a therapeutic avenue to specifically target cancer cells and leave normal cells largely unaffected: introduction of a perturbation (i.e., a drug) would represent a ‘single mutant’ and hence would be non-toxic to normal cells, but the transformed cancer cells would be ‘double mutant’ and thus synthetically lethal/sick [6,11,18–21]. Theoretically, Trends in Pharmacological Sciences xx (2014) 1–10
1
TIPS-1113; No. of Pages 10
Review Box 1. Genetic interactions, synthetic lethality, and oncogene addiction Background and nomenclature In complex cellular systems, genes do not act independently from each other but function in a concerted manner, constituting a genetic interaction network. The topology of genetic interaction networks has evolved to confer robustness such that an organism can cope with various types of internal and external perturbation while maintaining the capacity to evolve over generations. This has led to the identification of various mechanisms, and corresponding terms, that underlie but also undermine this robustness. Buffering: Robust genetic interaction networks buffer genetic variation, as illustrated by the observation that under standard laboratory culture conditions, the great majority of yeast single mutants are viable. Synthetic lethality and synthetic sickness: On occasion, single mutants become highly sensitive to specific second mutations, resulting in cell death or a decrease in fitness. These phenomena have been termed synthetic lethality and synthetic sickness, respectively (from the Greek syn-tithenai for to put together). Uncovering synthetic interactions is not trivial because they are rare and not always directly conserved from model organisms. Non-oncogene addiction and induced essentiality: Synthetic lethality has been revisited as a promising strategy for specific targeting of cancer, because cancer cells harbor numerous genetic aberrations that distinguish them from normal cells. To discern these applications from the precise genetic meaning, they have also been dubbed non-oncogene addiction or induced essentiality. Oncogene addiction: Distinct from synthetic lethality, oncogene addiction describes the rewired state of the signaling network of a cancer cell that has become critically dependent on the expression of a driver oncogene by shutting off parallel signaling pathways. A typical example of oncogene addiction with clinical relevance is the dependence of chronic myeloid leukemia (CML) cells on the BCR-ABL fusion gene, inhibition of which causes cell death.
exploitation of such synthetic interactions for therapeutic intervention by definition has less toxic side effects on normal tissue, thereby increasing the therapeutic index. Importantly, synthetic lethality would allow targeting of not only gain-of-function oncogenic events but also of lossof-function mutations in tumor suppressor genes, as well as oncogenes that are not directly druggable. Epigenetic deregulation as a cancer vulnerability Interestingly, several studies have highlighted that chromatin modifiers play a special role in buffering genetic interactions. One such study involved a screen for genetic interactions in the worm Caenorhabditis elegans [22]. By systematically screening 65 000 pairs of genes using RNA interference, the authors found that chromatin modifiers were strongly enriched among the gene pairs displaying synthetic lethality. This included components of several conserved chromatin modifier complexes such as the histone acetyl transferase NuA4–Tip60 complex, the transcriptional co-activator Mediator complex, and the Mi2–NuRD (nucleosome remodeling deacetylase) complex. Conceptually, it appears logical that the fine tuning of transcriptional processes is well positioned to buffer perturbations in other regulatory layers and is reminiscent of the role of the protein folding chaperone Hsp90 in buffering genetic variation [23]. Whereas heat shock proteins adjust protein folding and stability, epigenetic buffering directly targets transcription and may thus compensate for deleterious passenger mutations via transcriptional activation of the second allele, 2
Trends in Pharmacological Sciences xxx xxxx, Vol. xxx, No. x
Box 2. Epigenetics and cancer What evidence links chromatin and cancer? The earliest hints of a connection between epigenetics and cancer came from DNA methylation studies more than three decades ago, but were merely correlative in nature. With the advent of highthroughput sequencing technologies and the analysis of large sample numbers in recent years, it is now evident that epigenetic changes play a widespread role in the establishment of cancer via various mechanisms. For instance, many tumor suppressor genes are epigenetically inactivated by hypermethylation of their promoters. A number of mouse models have shown that epigenetic regulators are required for tumorigenesis, and most importantly, mutated chromatin-modifying enzymes have frequently been found in various types of cancer. What consequences arise from a deregulated epigenome? It has been proposed that aberrant epigenetic mechanisms allow for oncogenic reprogramming reminiscent of the differentiation process. In another model presented by Timp and Feinberg [46], it is hypothesized that deregulation of the epigenome occurs in the very early stages of tumorigenesis and causes flattening of the epigenetic landscape, leading to a stochastic departure from the normal epigenetic signature. This in turn allows for establishment of heterogeneity, acquisition of cancer hallmarks, and alteration of mutation rates. Changes in DNA methylation also affect the cell mutation rate, for instance, through higher rates of spontaneous deamination at methylated compared to unmethylated cytosines. Moreover, in the absence of DNMTs, cells suffer microsatellite instability, destabilization of repeats, and alterations in telomere length (and lengthening, ALT) and recombination. Not only DNA methylation but also histone methylation, especially H3K9me3, and other indicators of heterochromatin have been correlated with elevated mutation rates. Furthermore, it has been shown that overexpression of EZH2 contributes to genomic instability.
repression of the mutant allele, or regulation of alternative pathways. Enrichment of chromatin modifiers was also observed in a large screen in S. cerevisiae [24]. A genome-scale genetic interaction screen consisting of 5.4 million pairwise interactions showed that genes involved in chromatin function were among the most highly connected genes in the network. Furthermore, chromatin modifiers are often found bridging different cellular functions, further supporting a buffering function in the cell. Although similar high-throughput studies in human (cancer) cells are lacking, experiments using histone deacetylase (HDAC) inhibitors suggest a conserved function. There are reports that HDAC inhibitors frequently synergize with other compounds such as DNA methyltransferase (DNMT) inhibitors, proteasome inhibitors, or inhibitors of pro-survival proteins, resulting in greater anticancer effects than their additive individual effects [25]. Moreover, they seem to be enriched in drug combination screens aimed at killing cancer cells (S.K., unpublished observation). Together, these studies suggest that chromatin modifiers are important in maintaining homeostasis on genetic perturbations and that this function is conserved. Below we discuss some of the implications of this notion in the context of specific epigenetic drugs. Do approved HDAC and DNMT inhibitors work via synthetic lethal mechanisms? Chromatin emerged as a drug target long before epigenetics was considered a druggable area of biology or even
TIPS-1113; No. of Pages 10
Review
Trends in Pharmacological Sciences xxx xxxx, Vol. xxx, No. x
Direct targeng / oncogene addicon Acvang mutaons/translocaons (A)
(B)
(C) Y641F
GOF
HAT
EZH2
HATi Me Me
Ac Ac
Me
Ac
EZH2i
Me Me
Ac
Me Me
PHDi
PHF23
NUP98
JARID1A NUP98
Indirect epigenec vulnerability (D)
(E)
Recruitment by fusion proteins HDACi PML
PLZF
Chroman catastrophe and buffering exhauson DNMT3A/B DNMT1 mutaon
DNMTi
RARa
RARa
HDAC
DNMT
RUNX1 ETO
Normal CBF
5-azacydine HDACi
mC
MYH11
IDH mutaon TET2 mutaon
Cell death
mC hmC
DNMTi HMTi
Tumor
(F)
Chroman mark homeostasis Me Me Me
mut
(G)
mut
JMJD3
X
BRD4
UTX
BRD4
EZH2
NUT
NUT
BRD4 NUT Ac Ac Ac Ac
H3K27
(H)
Chroman mark dissipaon HDACi
HDAC
Ac Ac
BRD4
NUT
Ac
Ac Ac
BRD4 NUT
Ac
Ac
EZH2i
Repression of oncogenes
(I)
Chroman mark replacement MLL
X
BRD4
AF4/9
H3K4 Me
H3K79 Me
c-MYC
DOT1L
BETi DOT1Li
TRENDS in Pharmacological Sciences
Figure 1. Rationale and opportunities for targeting chromatin in cancer through potential synthetic lethality (and related) mechanisms. (A–C) Direct targeting or oncogene addiction, (D–I) various types of indirect epigenetic vulnerability. (A) Cancers with gain-of-function mutations/amplifications of histone acetyltransferases (HATs) may be susceptible to HAT inhibition. (B) Cancers harboring NUP98-JARID1A/PHF23 translocations that result in PHD-finger-containing fusion proteins may be sensitive to PHD inhibition. (C) In diffuse large B cell lymphoma (DLBCL) with GOF of enhancer of zeste homolog 2 (EZH2), treatment with EZH2 inhibitors leads to reduced proliferation and apoptosis. (D) Various fusion proteins recruit DNA methyltransferases (DNMTs) and histone deacetylases (HDACs), which can be exploited by treatment with corresponding inhibitors. (E) Mutations in DNMTs, TET2, and IDH1 result in altered levels of DNA methylation (methyl-C, mC) and hydroxymethylation (hydroxymethyl-C, hmC), which creates potential sensitivities to 5-azacytidine/HDAC inhibition or DNMT/HMT inhibition, respectively, through chromatin catastrophe and/or general exhaustion of buffering mechanisms. (F) In cancers with loss-of-function mutations in H3K27 demethylases JMJD3 or UTX, treatment with EZH2 inhibitors can restore normal methylation levels. (G) In nuclear protein in testis (NUT) midline carcinoma, the BRD4-NUT fusion protein accumulates at particular hyperacetylated sites. Treatment with HDAC inhibitors causes dissipation of chromatin acetylation marks and redistribution of BDR4-NUT. (H) It has been shown that treatment of leukemias transcriptionally represses c-MYC, a mechanism that could extend to other oncogenes. (I) Mixed-lineage leukemia (MLL)-rearranged leukemias recruit DOT1L, which leads to replacement of H3K4 by H3K27 methylation and dependence on DOT1L.
3
TIPS-1113; No. of Pages 10
Review established as a scientific field. For instance, DNAdamaging and -intercalating agents cause not only a DNA damage response and cell cycle phenotype but also global chromatin changes, and these may contribute to their specific effects on cancer cells [1,26]. Furthermore, the discovery and development of the first HDAC inhibitor were sparked by a serendipitous finding that the commonly used chemical solvent dimethyl sulfoxide (DMSO) causes differentiation of leukemia cells [27]. This observation inspired a 20-year quest that resulted in approval of the HDAC inhibitor SAHA/vorinostat for cutaneous T cell lymphoma (CTCL) in 2006 (Figure 2). A second, structurally unrelated natural product with HDAC inhibitor activity (romidepsin) was approved for the same disease in 2009. HDACs are best known for their ability to remove acetyl groups from histones, thereby increasing DNA binding and reducing accessibility of regulatory regions to transcription factors. In addition, both histone and nonhistone acetylation can exert their biological function via recruitment of specific acetyl-binding proteins typically containing a bromo domain, providing another layer of regulation by acetylation that might be druggable. Currently, more than 20 additional HDAC inhibitors are being evaluated in clinical trials across various malignancies, including solid tumors. However, despite their clinical success and promise, the mechanism of action of HDAC inhibitors that explains their cancer cell selectivity remains elusive [28]. HDACs themselves are rarely mutated, and it appears unlikely that increased HDAC activity per se contributes to tumorigenesis. Furthermore, HDAC expression or activity has not been linked to the response to HDAC inhibitors, and other predictive biomarkers are lacking. Thus, the response to HDAC inhibition of a variety of cancer cell lines cannot be easily explained by an oncogene addiction mode of action. In some instances the sensitivity to HDAC inhibitors may be due to an oncogenic driver that requires HDAC activity, resulting in induced essentiality. Examples illustrating this concept are oncogenic fusion proteins such as PML-RARa (promyelocytic leukemia–retinoic acid receptor alpha), PLZF-RARa (promyelocytic leukaemia zinc finger–RARa), CBF-MYH11 (core binding factor–myosin heavy chain 11), and RUNX1-ETO (runt-related transcription factor 1–RUNX1 translocated to 1), which recruit HDACs and induce aberrant gene silencing (Figure 1D). This may explain a specific sensitivity to HDAC inhibitors, because they can reverse some of the abnormal gene repression correlating with growth arrest, differentiation, and apoptosis in cancer cells [4,29–31]. At therapeutically relevant doses, HDAC inhibitors typically change the expression of thousands of genes [32,33], supporting the notion that they interfere with histone acetylation and thereby impinge on gene transcription. However, it remains unclear how these global effects would result in specific triggering of apoptosis in cancer cells. In addition, HDACs increase the acetylation of non-histone proteins, and this may also contribute to their cytotoxicity.
Trends in Pharmacological Sciences xxx xxxx, Vol. xxx, No. x
A second class of anticancer compounds that are often referred to as epigenetic drugs are the DNA methyltransferase (DNMT) inhibitors 5-azacytidine (vidaza) and 5-aza20 deoxy-cytidine (decitabine), both approved for myelodysplastic syndrome (MDS, a collection of hematological conditions characterized by ineffective production of myeloid cells) [34]. DNMTs catalyze the addition of methyl groups to cytosines in double-stranded DNA and thus contribute to the epigenetic silencing of tumor suppressor genes with CpG promoters that become methylated during tumorigenesis. However, DNA methylation, particularly in heterochromatic regions, is considered to play important roles in genomic stability. In line with this dual role of DNA methylation, cancer cells can often be characterized by global DNA hypomethylation and local DNA hypermethylation. Both DNMT inhibitors act via incorporation into the DNA, resulting in subsequent covalent crosslinking of DNMTs to the DNA. As a consequence, the drugs deplete the available pool of DNMTs and thereby reduce CpG methylation globally, resulting in re-expression of epigenetically silenced genes [35–37]. In addition, they can cause a DNA damage response, because the DNA–protein crosslinks need to be removed and repaired before cells can divide. This dual mechanism of action of DNMT inhibitors makes it difficult to determine what drives the sensitivity of cancer cells to these compounds. Inactivating mutations in the catalytic domain of DNMT3a and DNMT1 have been found in MDS and acute myeloid leukemia (AML), yet the functional consequences of these are poorly understood, because aberrant methylation patterns do not always correlate with mutations [38,39]. Moreover, from a mechanistic perspective it is unclear how inactivating mutations in DNMT3a would render cells sensitive to DNMT inhibitors, and these mutations do not explain the more general sensitivity of cancer cells to the drugs [40]. By contrast, DNMT1 and DNMT3B seem to be overexpressed in various other solid tumors. In analogy to HDACs, the PML-RARa fusion protein, as well as EZH2 (enhancer of zeste homolog 2), recruits DNMTs to promoters of tumor suppressor genes (Figure 1D), which may provide a selective sensitivity to DNMT inhibitors in specific cases [41,42]. However, as for HDAC inhibitors, these observations suggest that drug sensitivity is not solely a consequence of deregulated DNMT activity per se. Both HDAC and DNMT inhibitors cause global genomewide chromatin changes that result in dramatic gene expression changes in both cancer and normal cells. At least part of the mechanism of action of these drugs may be explained by synthetic genetic interactions and buffering principles. In light of epigenetic deregulation as a ubiquitous phenomenon in cancer, it is not surprising that further perturbation of this process by inhibition of DNMTs and HDACs is more detrimental to malignant cells. Furthermore, cancer cells harbor many mutations and we postulate that HDAC and DNMT inhibitors are specific for target cells because they perturb the buffering capacity of the cell (Figure 1E).
Abbreviations: BRD4, bromodomain containing 4; c-MYC, v-myc avian myelocytomatosis viral oncogene homolog; DOT1L, DOT1-like histone H3K79 methyltransferase; GOF/amp, gain-of-function/amplification; H3K4, Histone 3 lysine 4; H3K27, Histone 3 lysine 27; JMJD3, jumonji domain containing 3 (also known as KDM6B, lysine (K)-specific demethylase 6B); NUP98-JARID1A/PHF23, nucleoporin 98kDa-jumonji, AT rich interactive domain 1A (also known as KDM5A, lysine (K)-specific demethylase 5A)/PHD finger protein 23; PHD, Plant Homeo Domain; UTX, ubiquitously transcribed tetratricopeptide repeat, X chromosome (also known as KDM6A, lysine (K)-specific demethylase 6A).
4
TIPS-1113; No. of Pages 10
Review
Trends in Pharmacological Sciences xxx xxxx, Vol. xxx, No. x
Histone deacetylase inhibitors O
H N
N H
O
OH
Merck
Cutaneous T-cell lymphoma
HDAC
Approved
Romidepsin
Celgene
Cutaneous T-cell lymphoma
HDAC
Approved
Azacidine
Celgene
Myelodysplasic syndrome
DNMT
Approved
Decitabine
Astex
Myelodysplasic syndrome
DNMT
Approved
EPZ-7438
Epizyme
Non-hodgkin lymphoma
EZH2
Phase I/II
EPZ-5676
Epizyme
MLL-rearranged leukemia
DOT1L
Phase I
Oryzon
Acute myeloid leukemia
LSD1
Phase I
O
HN S O S HN NH
O
Vorinostat
HN
O
O O
DNA methyltransferase inhibitors O O HO
N N
HO
NH2 N
OH O O
N N
HO
NH2 N
HO
Histone methyltransferase inhibitors O N N
O
O HN
O
NH2
HN
N N O
N OH
N OH
H N N
Histone demethylase inhibitors ORY-1001
Undisclosed
Tranylcypromine
H2N
Anergic depression
MAO A/B Approved LSD1 off-target
Bromo domain inhibitors O O
OH
N NH O
RVX208
Resverlogix
Cardiovascular disease
BET
Phase IIb
OTX015
Oncoethix
Hematological malignancies
BET
Phase I
GSK
NUT midline carcinoma
BET
Phase I/II
O N N
N
S
NH
N O
OH
Cl
N N
N
NH
N
O
O
GSK525762
Cl
TRENDS in Pharmacological Sciences
Figure 2. Epigenetic drugs and chromatin-targeting compounds in clinical development. These include two approved histone deacetylase (HDAC) inhibitors (not shown are the more than 20 additional HDAC inhibitors in clinical development) and two approved DNA methyltransferase inhibitors. Compounds targeting histone methyltransferases, demethylases, and bromodomains are in early-stage development, and clinical trials for each class have recently been initiated.
Both HDAC and DNMT inhibitors are currently being evaluated in many different clinical trials across cancer types and in combination with other compounds. If our hypothesis is correct, cancers that are most dependent on epigenetic buffering should respond best to these compounds. To identify biomarkers for patient selection, a better understanding of the oncogenic pathways that confer synthetic lethality to DNMT and HDAC inhibitors will be necessary. Similarly, it is unclear if the effect requires polypharmacology of the compounds inhibiting multiple enzymes or if a single enzyme is the most important target.
Following the trend for targeted anticancer drugs, recent development efforts are aimed at increasing the specificity of epigenetic drugs. For HDAC inhibitors, this entails making them more selective for individual HDAC enzymes, whereas currently available compounds all simultaneously target HDAC 1, 2, and 3, as well as HDAC6 and 8 to various degrees [43]. For DNMT inhibitors, competitive inhibitors are desirable to allow the separation of effects caused by DNA damage from loss of CpG methylation. However, given that it is presently unclear which effects are most central to the antitumor activities of the 5
TIPS-1113; No. of Pages 10
Review existing agents, it remains to be determined whether increases in selectivity will coincide with improved therapeutic selectivity. Most emerging epigenetic drugs target oncogene addiction The recent cancer genome resequencing efforts have highlighted many recurrent mutations in chromatin regulators [44]. For reasons that are at present unclear, mutations in chromatin regulators appear particularly frequent in hematological cancers. Perhaps the need for especially stringent epigenetic control of cell fate renders these lineages vulnerable to malignant transformation on epigenetic perturbation. However, mutations in chromatin modifiers are also found in solid tumors [45], where they mainly affect rare aggressive adult cancers, pediatric neoplasms, and cancers that relapse or are resistant to therapy [4,46]. Together, these studies have yielded an increasing appreciation of epigenetics as a druggable target class (Box 2) [4,31,47]. In a few cases, overactive chromatin modifiers have inspired the development of potent novel compounds that are highly selective for enzyme classes and subclasses. The targets of these compounds were selected based on strong genetic evidence of an oncogene addiction mode. The first compounds are currently entering early-stage clinical trials in selected populations with corresponding mutations and include EZH2, DOT1L, and BRD4 inhibitors. Importantly, for these drugs and other epigenetic drug targets, non-oncogene addiction applications can also be envisioned. This is of particular relevance because most mutations in chromatin modifiers and other driver mutations do not constitute readily accessible drug targets. Thus, a broader evaluation of epigenetic drugs in cancer may provide additional anticancer therapeutic applications. Some of these mostly speculative indications are discussed below and summarized in Table 1. Potential non-oncogene addiction applications for chromatin-targeting compounds Given that chromatin modifiers frequently display synthetic lethality, we postulate that tumors with mutations in this class of genes might be particularly suitable for therapeutic strategies that exploit this phenomenon. In particular, we would expect that targeting of additional chromatin complexes in these tumors would be especially cytotoxic through a variety of different mechanisms (see below and Table 1). For instance, we could imagine targeting of chromatin regulators that exist in pairs of enzymes with counteracting activities, such as HDACs– HATs (histone acetyltransferases), HMTs–HDMs (histone methyltransferases and demethylases), DNMTs– TETs (tet methylcytosine dioxygenases), and PRMTs– PADIs (protein arginine methyltransferases and deiminases). Inhibition of the counteractor, which would reestablish homeostatic levels for histone modifications, might compensate loss-of-function mutations in one of the partners or further deregulate homeostasis, resulting in lethality. One concrete example includes EZH2 inhibitors. Oncogenic EZH2 gain-of-function mutations cause 6
Trends in Pharmacological Sciences xxx xxxx, Vol. xxx, No. x
increased levels of H3K27me3 and occur in up to 20% of diffuse large B cell lymphoma (DLBCL) cells of germinal center origin [48]. DLBCL cells with EZH2 mutations are addicted to EZH2 activity for proliferation and survival [49–51]. However, EZH2 inhibitors may also be applied in EZH2 wild type neoplasms. Besides EZH2, mutations of other members of the polycomb repressive complex 2 (PRC2) have been found in cancer and may be candidates for therapeutic intervention through targeting of EZH2 [5,31,47]. Furthermore, the loss of EZH2 opposing demethylases UTX/KDM6A and JMJD3/KDM6B would be expected to result in similar sensitivity to EZH2 inhibitors. Indeed, mutations in these genes have been found in a wide range of tumor types, and it has been suggested that the H3K27 demethylases are tumor suppressors [52,53]. The histone mark homeostasis model predicts that UTX mutant cells will respond to EZH2 inhibitors, and this has been observed for one cell line, at least anecdotally (Figure 1F) [50]. Similar reciprocal targeting might apply for other methyl-transferase/demethylase pairs, such as for cancers harboring mutations in MLL (mixed-lineage leukemia) 2 and 3, which have been detected in various malignancies. BET domain inhibitors Other ways of applying the new highly potent chromatintargeting compounds are emerging and BET bromodomain inhibitors are attracting special interest. Bromodomains are histone acetylation binders that translate the presence of this activating mark to a transcriptional response. In NUT midline carcinoma, translocations occur between the NUT (nuclear protein in testis) protein and the bromodomain of BRD4 [54]. Interestingly, HDAC inhibitors can cause differentiation of NUT midline carcinoma cells, possibly mediated by the global increase in acetylation, thereby titrating away the BRD4 protein from critical sites [55]. Thus, HDAC inhibition in cancers with translocations of acetylation binders represents another example of how epigenetic drugs can act through non-oncogene addiction (Figure 1G). Potentially exploiting oncogene addiction, direct BRD4 inhibitors have recently been developed by targeting the BET bromodomain [56,57]. Two compounds have now entered the clinic and are first being evaluated in NUT midline carcinomas. They might also be suitable for targeting melanoma with overexpression of BRD4 [58]. However, cancer cells without mutations in BRD4 can also display profound sensitivity to this compound class. In particular, it has been shown that they downregulate transcription of the driver oncogene c-MYC in AML and multiple myeloma cells (Figure 1H) [59,60]. The selective effect observed is probably not solely attributable to cMYC downregulation, and the detailed mode of action of BET inhibitors is incompletely understood [61,62]. However, BRD4 inhibition seems to affect only a particular set of genes with exceptionally large enhancers that are marked by co-occupation of BRD4 and Mediator (superenhancers). Many of these genes are oncogenes themselves, suggesting that BRD4 inhibition could target a variety of cancer cells [63].
TIPS-1113; No. of Pages 10
Review
Trends in Pharmacological Sciences xxx xxxx, Vol. xxx, No. x
Table 1. Rationale and opportunities for targeting chromatin in cancer through potential synthetic lethality (and related) mechanisms Gene/protein
Function
Aberration Cancer type (examples)
Direct targeting Oncogene addiction HAT CBP
Mut, amp
p300
Mut
MYST3/MOZ/KAT
Mut, TL
ELP4 AIB
Amp Amp
EZH2
HMT
Suggested mechanism
Potential therapeutic Outlook targeting
HAT inhibition
Mut (GOF)
Ovarian/breast/lung/ oncogene addiction bladder cancer, lymphoma, ALL pancreatic/bladder cancer, HNSCC, lymphoma leukemia, medulloblastoma, breast cancer Breast cancer Breast/gastric/esophageal cancer, lymphoma Prostate/breast/bladder oncogene addiction cancer DLBCL oncogene addiction
sensitivity to BET inhibitors PHD inhibition
OE
EZH2 inhibition
BRD4-NUT
BET
TL
NUT midline carcinoma
oncogene addiction
JARID1A/KDM5A PHF23-NUP98
PHD finger PHD finger
TL, OE TL, mut
AML, gastric cancer AML
sensitivity to mutations impairing binding of the PHD finger to H3K4me3
Lymphoma, renal/bladder cancer, medulloblastoma Various (rare)
Altered H3K27me3 levels
EZH2 inhibition to restore H3K27me3
AML, lymphoma, colorectal cancer AML, MDS Pancreatic/liver/bladder/ breast cancer Breast cancer MPN, MDS, CMML
Vulnerability due to DNA methylation changes
Hypomethylating agents or HDAC inhibition
Decrease in 5-hmC and increase in 5-mC Glioma, chondrosarcoma, a) 2-HG accumulation AML, astrocytoma inhibits TET2, mimics its LOF b) 2-HG inhibits a-KGdependent HDMs NUT midline carcinoma Fusion causes sequestration of acetylated histones, hypoacetylation, differentiation block
DNMT inhibition
Indirect epigenetic vulnerability Chromatin alterations Mut HDM UTX/KDM6A Mut
JMJD3/KDM6B DNMT1
DNMT
Mut Mut OE
DNMT3A DNMT1
IDH1/2
OE DNA Mut (LOF) hydroxylase TCA cycle Mut (LOF)
BRD4-NUT
BET
DNMT3B TET2
Induced essentiality HMT MLL1, 2, 3
TL
TL, mut
Induction of transcriptional changes TF TL PML-RARa PLZF-RARa
TF
TL
RUNX1-ETO
TF
TL
CBF-MYH11
TF
TL
PML-RARa
TF
TL
EZH2
HMT
OE, mut
HDAC inhibition (epigenetic differentiation therapy)
DOT1L essential in MLL fusion cancers MLL fusion addicted to menin
AML, PML, leukemias AML, PML, leukemias AML, PML, leukemias AML, PML, leukemias AML, PML, leukemias
Recruitment of HDACs and HDAC inhibition induction of aberrant gene silencing
various
Similar reciprocal targeting for other methyltransferase/ demethylase pairs
a) DNMT inhibition b) HMT inhibition
MLL/ALL/AML, medulloblastoma, lymphoma, HNSCC, bladder/breast/renal/lung/ gastric/prostate cancer
various
targeting other PRC2 components (directly or through EZH2 inhibition)
DOT1L inhibition Menin inhibition
Similar dependence suggested with PRMT1, BRD4, LSD1, PRC2, IKK, and MEK inhibition
various various various
Recruitment of DNMTs to promoters of tumor suppressor genes
DNMT inhibition
7
TIPS-1113; No. of Pages 10
Review
Trends in Pharmacological Sciences xxx xxxx, Vol. xxx, No. x
Table 1 (Continued ) Gene/protein
BRD4
Function
BET
Aberration Cancer type (examples)
OE
MDS, MPN, myeloid malignancies, T-ALL, HNSCC, prostate, breast, bladder cancer, DLBCL Melanoma, various leukemias
Suggested mechanism
Potential therapeutic Outlook targeting
BRD4 loss downregulates c-MYC transcription
BRD4 inhibition
Affects genes with super-enhancers (many oncogenes)
HAT, histone acetyltransferase; HMT, histone methyltransferase; HDM, histone demethylase; DNMT, DNA methyltransferase; HDAC, histone deacetylase; BET, BET bromodomain; TF, transcription factor; mut, mutated; amp, amplified; TL, translocation; LOF, loss of function; GOF, gain of function; LOH, loss of heterozygosity; OE, overexpression; ALL, acute lymphoblastic leukemia; HNSCC, head and neck squamous cell carcinoma; DLBCL, diffuse large B cell lymphoma; AML, acute myeloid leukemia; MPN, myeloproliferative neoplasms; MDS, myelodysplastic syndrome; CMML, chronic myelomonocytic leukemia; PML, promyelocytic leukemia; MSI, microsatellite instability; ALT, alternative lengthening of telomeres. If not indicated otherwise or in the main text, information on genes, mutations, cancer types, and frequencies is from [4,5,30,40,41,56] and references therein.
MLL translocations result in induced essentiality of DOT1L and menin MLL translocations are the most common lesion in childhood AMLs, and they also occur in adult leukemias [64]. Normally, the MLL protein contain a SET domain conferring HMT activity that generates histone H3 lysine 4 methylation. In leukemic translocations, the catalytic activity of MLL is lost and replaced with other protein domains. More than 70 different fusion partners have been identified, and interestingly they very often contain other chromatin-modifying activities. For example, acetyltransferase and arginine methyltransferase domains have been identified in fusion partners. The most common fusion partners are AF4 and AF9, which do not harbor chromatin-modifying activities themselves. However, they recruit DOT1L, a methyltransferase that generates histone H3 lysine 79 methylation. Therefore, at the chromatin level, translocation results in the replacement of H3K4 with H3K79 methylation at MLL target sites (Figure 1I) [65– 68]. Based on these protein-protein interaction data, the effects of knockdown and knockout of DOT1L in MLLtranslocated leukemias have been evaluated in cell culture and animal models [69,70]. Pronounced non-oncogene dependence has been detected, and this result endorsed the development of a highly potent DOT1L inhibitor, EPZ5676. This compound competes with the cofactor Sadenosyl-methionine that is used by all histone methyltransferases, yet it is highly selective for DOT1L and inhibits this enzyme at picomolar concentrations. In xenograft models with MLL-rearranged leukemia lines, EPZ5676 leads to tumor regression and sustained responses, even after compound withdrawal [71]. This compound is currently being evaluated in clinical trials (NCT01684150). In addition, it has been shown that MLL-translocated malignancies are addicted to menin, an interaction that was recently successfully targeted by a small-molecule inhibitor [72,73]. Similar dependence of MLL-fusion leukemias has been suggested with PRMT1, BRD4, LSD1, PRC2, IKK, and MEK inhibition [74–76]. Concluding remarks The recognition that aberrant epigenetic regulators are causally implicated in cancer, as well as the fact that many 8
of them are druggable, has fueled the development of small-molecule inhibitors for various epigenetic enzyme classes. The field of epigenetics is now firmly established as an important area of drug development. However, most efforts focus on a few selected proteins with a strong genetic basis such as amplifications, translocations, and activating mutations to exploit oncogene addiction. As illustrated by several examples in this review, the inhibition of many epigenetic regulators can result in specific effects in cancer cells through mechanisms akin to synthetic lethality. Thus, many other chromatin proteins might be attractive targets for inhibitor development based on non-oncogene addiction through a variety of mechanisms. Therefore, we envision that systematic investigations of epigenetic regulators as drug targets will be a rewarding approach towards the identification of novel anticancer drugs. Along the same lines, epigenetic drugs are attractive candidates for combination with other drugs. Supporting this notion, numerous synergies of HDAC and DNMT inhibitors with other molecules have been found in vitro. These synergies and their potent biological activities have kept the interest in HDAC and DNMT inhibitors high, and more than 400 clinical trials are currently being performed with these compound classes. Ultimately, to prioritize relevant targets from the more than 400 chromatin-modifiers, systematic evaluation of the involvement of these proteins in synthetic lethal interactions will be essential. Recent advances in genome engineering and isogenic mutant collections will provide technical avenues to achieve this goal. Acknowledgements We thank Christoph Bock and Helen Pickersgill for critical reading and discussions. B.M. is supported by a DOC fellowship from the Austrian Academy of Sciences.
References 1 Baylin, S.B. and Jones, P.A. (2011) A decade of exploring the cancer epigenome – biological and translational implications. Nat. Rev. Cancer 11, 726–734 2 Brough, R. et al. (2011) Searching for synthetic lethality in cancer. Curr. Opin. Genet. Dev. 21, 34–41 3 Chan, D.A. and Giaccia, A.J. (2011) Harnessing synthetic lethal interactions in anticancer drug discovery. Nat. Rev. Drug Discov. 10, 351–364
TIPS-1113; No. of Pages 10
Review 4 Dawson, M.A. and Kouzarides, T. (2012) Cancer epigenetics: from mechanism to therapy. Cell 150, 12–27 5 Helin, K. and Dhanak, D. (2013) Chromatin proteins and modifications as drug targets. Nature 502, 480–488 6 Kaelin, W.G., Jr (2005) The concept of synthetic lethality in the context of anticancer therapy. Nat. Rev. Cancer 5, 689–698 7 Rodriguez-Paredes, M. and Esteller, M. (2011) Cancer epigenetics reaches mainstream oncology. Nat. Med. 17, 330–339 8 Skene, P.J. and Henikoff, S. (2013) Histone variants in pluripotency and disease. Development 140, 2513–2524 9 Hanahan, D. and Weinberg, R.A. (2000) The hallmarks of cancer. Cell 100, 57–70 10 Hanahan, D. and Weinberg, R.A. (2011) Hallmarks of cancer: the next generation. Cell 144, 646–674 11 Luo, J. et al. (2009) Principles of cancer therapy: oncogene and nononcogene addiction. Cell 136, 823–837 12 Gonzalez-Angulo, A.M. et al. (2010) Future of personalized medicine in oncology: a systems biology approach. J. Clin. Oncol. 28, 2777–2783 13 Vidal, M. et al. (2011) Interactome networks and human disease. Cell 144, 986–998 14 Bridges, C.B. (1922) The origin of variation. Am. Nat. 56, 51–63 15 Dobzhansky, T. (1946) Genetics of natural populations. XIII. Recombination and variability in populations of Drosphila pseudoobscura. Genetics 31, 269–290 16 Lucchesi, J.C. (1968) Synthetic lethality and semi-lethality among functionally related mutants of Drosophila melanogaster. Genetics 59, 37–44 17 Sturtevant, A.H. (1956) A highly specific complementary lethal system in Drosophila melanogaster. Genetics 41, 118–123 18 Hartwell, L.H. et al. (1997) Integrating genetic approaches into the discovery of anticancer drugs. Science 278, 1064–1068 19 Kaelin, W.G., Jr (1999) Choosing anticancer drug targets in the postgenomic era. J. Clin. Invest. 104, 1503–1506 20 Kaelin, W.G., Jr (2009) Synthetic lethality: a framework for the development of wiser cancer therapeutics. Genome Med. 1, 99 21 Reinhardt, H.C. et al. (2009) Exploiting synthetic lethal interactions for targeted cancer therapy. Cell Cycle 8, 3112–3119 22 Lehner, B. et al. (2006) Systematic mapping of genetic interactions in Caenorhabditis elegans identifies common modifiers of diverse signaling pathways. Nat. Genet. 38, 896–903 23 Queitsch, C. et al. (2002) Hsp90 as a capacitor of phenotypic variation. Nature 417, 618–624 24 Costanzo, M. et al. (2010) The genetic landscape of a cell. Science 327, 425–431 25 Bots, M. and Johnstone, R.W. (2009) Rational combinations using HDAC inhibitors. Clin. Cancer Res. 15, 3970–3977 26 Soria, G. et al. (2012) Prime, repair, restore: the active role of chromatin in the DNA damage response. Mol. Cell 46, 722–734 27 Marks, P.A. and Breslow, R. (2007) Dimethyl sulfoxide to vorinostat: development of this histone deacetylase inhibitor as an anticancer drug. Nat. Biotechnol. 25, 84–90 28 Khan, O. and La Thangue, N.B. (2012) HDAC inhibitors in cancer biology: emerging mechanisms and clinical applications. Immunol. Cell Biol. 90, 85–94 29 Petruccelli, L.A. et al. (2013) Expression of leukemia-associated fusion proteins increases sensitivity to histone deacetylase inhibitor-induced DNA damage and apoptosis. Mol. Cancer Ther. 12, 1591–1604 30 Ropero, S. and Esteller, M. (2007) The role of histone deacetylases (HDACs) in human cancer. Mol. Oncol. 1, 19–25 31 Shen, H. and Laird, P.W. (2013) Interplay between the cancer genome and epigenome. Cell 153, 38–55 32 Wang, Z. et al. (2009) Genome-wide mapping of HATs and HDACs reveals distinct functions in active and inactive genes. Cell 138, 1019– 1031 33 Kubicek, S. et al. (2012) Chromatin-targeting small molecules cause class-specific transcriptional changes in pancreatic endocrine cells. Proc. Natl. Acad. Sci. U.S.A. 109, 5364–5369 34 Yang, X. et al. (2010) Targeting DNA, methylation for epigenetic therapy. Trends Pharmacol. Sci. 31, 536–546 35 Christman, J.K. (2002) 5-Azacytidine and 5-aza-20 -deoxycytidine as inhibitors of DNA methylation: mechanistic studies and their implications for cancer therapy. Oncogene 21, 5483–5495
Trends in Pharmacological Sciences xxx xxxx, Vol. xxx, No. x
36 Juttermann, R. et al. (1994) Toxicity of 5-aza-20 -deoxycytidine to mammalian cells is mediated primarily by covalent trapping of DNA methyltransferase rather than DNA demethylation. Proc. Natl. Acad. Sci. U.S.A. 91, 11797–11801 37 Orta, M.L. et al. (2013) 5-Aza-20 -deoxycytidine causes replication lesions that require Fanconi anemia-dependent homologous recombination for repair. Nucleic Acids Res. 41, 5827–5836 38 Ley, T.J. et al. (2010) DNMT3A mutations in acute myeloid leukemia. N. Engl. J. Med. 363, 2424–2433 39 Walter, M.J. et al. (2011) Recurrent DNMT3A mutations in patients with myelodysplastic syndromes. Leukemia 25, 1153–1158 40 Metzeler, K.H. et al. (2012) DNMT3A mutations and response to the hypomethylating agent decitabine in acute myeloid leukemia. Leukemia 26, 1106–1107 41 Di Croce, L. et al. (2002) Methyltransferase recruitment and DNA hypermethylation of target promoters by an oncogenic transcription factor. Science 295, 1079–1082 42 Vire, E. et al. (2006) The polycomb group protein EZH2 directly controls DNA methylation. Nature 439, 871–874 43 Bradner, J.E. et al. (2010) Chemical phylogenetics of histone deacetylases. Nat. Chem. Biol. 6, 238–243 44 Garraway, L.A. and Lander, E.S. (2013) Lessons from the cancer genome. Cell 153, 17–37 45 Kandoth, C. et al. (2013) Mutational landscape and significance across 12 major cancer types. Nature 502, 333–339 46 Timp, W. and Feinberg, A.P. (2013) Cancer as a dysregulated epigenome allowing cellular growth advantage at the expense of the host. Nat. Rev. Cancer 13, 497–510 47 Geutjes, E.J. et al. (2012) Targeting the epigenome for treatment of cancer. Oncogene 31, 3827–3844 48 Morin, R.D. et al. (2010) Somatic mutations altering EZH2 (Tyr641) in follicular and diffuse large B-cell lymphomas of germinal-center origin. Nat. Genet. 42, 181–185 49 Knutson, S.K. et al. (2012) A selective inhibitor of EZH2 blocks H3K27 methylation and kills mutant lymphoma cells. Nat. Chem. Biol. 8, 890– 896 50 McCabe, M.T. et al. (2012) EZH2 inhibition as a therapeutic strategy for lymphoma with EZH2-activating mutations. Nature 492, 108–112 51 Qi, W. et al. (2012) Selective inhibition of Ezh2 by a small molecule inhibitor blocks tumor cells proliferation. Proc. Natl. Acad. Sci. U.S.A. 109, 21360–21365 52 Ene, C.I. et al. (2012) Histone demethylase Jumonji D3 (JMJD3) as a tumor suppressor by regulating p53 protein nuclear stabilization. PLoS ONE 7, e51407 53 van Haaften, G. et al. (2009) Somatic mutations of the histone H3K27 demethylase gene UTX in human cancer. Nat. Genet. 41, 521–523 54 French, C.A. et al. (2004) Midline carcinoma of children and young adults with NUT rearrangement. J. Clin. Oncol. 22, 4135–4139 55 Schwartz, B.E. et al. (2011) Differentiation of NUT midline carcinoma by epigenomic reprogramming. Cancer Res. 71, 2686–2696 56 Filippakopoulos, P. et al. (2010) Selective inhibition of BET bromodomains. Nature 468, 1067–1073 57 Nicodeme, E. et al. (2010) Suppression of inflammation by a synthetic histone mimic. Nature 468, 1119–1123 58 Segura, M.F. et al. (2013) BRD4 sustains melanoma proliferation and represents a new target for epigenetic therapy. Cancer Res. 73, 6264–6276 59 Delmore, J.E. et al. (2011) BET bromodomain inhibition as a therapeutic strategy to target c-Myc. Cell 146, 904–917 60 Zuber, J. et al. (2011) RNAi screen identifies Brd4 as a therapeutic target in acute myeloid leukaemia. Nature 478, 524–528 61 Dawson, M.A. et al. (2012) Targeting epigenetic readers in cancer. N. Engl. J. Med. 367, 647–657 62 Popovic, R. and Licht, J.D. (2012) Emerging epigenetic targets and therapies in cancer medicine. Cancer Discov. 2, 405–413 63 Loven, J. et al. (2013) Selective inhibition of tumor oncogenes by disruption of super-enhancers. Cell 153, 320–334 64 Muntean, A.G. and Hess, J.L. (2012) The pathogenesis of mixedlineage leukemia. Annu. Rev. Pathol. 7, 283–301 65 Bernt, K.M. et al. (2011) MLL-rearranged leukemia is dependent on aberrant H3K79 methylation by DOT1L. Cancer Cell 20, 66–78 66 Chang, M.J. et al. (2010) Histone H3 lysine 79 methyltransferase Dot1 is required for immortalization by MLL oncogenes. Cancer Res. 70, 10234–10242 9
TIPS-1113; No. of Pages 10
Review 67 Jo, S.Y. et al. (2011) Requirement for Dot1l in murine postnatal hematopoiesis and leukemogenesis by MLL translocation. Blood 117, 4759–4768 68 Nguyen, A.T. et al. (2011) DOT1L, the H3K79 methyltransferase, is required for MLL-AF9-mediated leukemogenesis. Blood 117, 6912– 6922 69 Daigle, S.R. et al. (2011) Selective killing of mixed lineage leukemia cells by a potent small-molecule DOT1L inhibitor. Cancer Cell 20, 53– 65 70 Neff, T. and Armstrong, S.A. (2013) Recent progress toward epigenetic therapies: the example of mixed lineage leukemia. Blood 121, 4847–4853 71 Daigle, S.R. et al. (2013) Potent inhibition of DOT1L as treatment of MLL-fusion leukemia. Blood 122, 1017–1025
10
Trends in Pharmacological Sciences xxx xxxx, Vol. xxx, No. x
72 Grembecka, J. et al. (2012) Menin-MLL inhibitors reverse oncogenic activity of MLL fusion proteins in leukemia. Nat. Chem. Biol. 8, 277– 284 73 Yokoyama, A. et al. (2005) The menin tumor suppressor protein is an essential oncogenic cofactor for MLL-associated leukemogenesis. Cell 123, 207–218 74 Kuo, H.P. et al. (2013) Epigenetic roles of MLL oncoproteins are dependent on NF-kappaB. Cancer Cell 24, 423–437 75 Miller, P.G. et al. (2013) In vivo RNAi screening identifies a leukemiaspecific dependence on integrin beta 3 signaling. Cancer Cell 24, 45–58 76 Pemovska, T. et al. (2013) Individualized systems medicine (ISM) strategy to tailor treatments for patients with chemorefractory acute myeloid leukemia. Cancer Discov. 3, 1416–1429
Review
Exploiting epigenetic vulnerabilities for cancer therapeutics Barbara Mair, Stefan Kubicek, and Sebastian M.B. Nijman CeMM Research Center for Molecular Medicine of the Austrian Academy of Sciences, Vienna, Austria
Epigenetic deregulation is a hallmark of cancer, and there has been increasing interest in therapeutics that target chromatin-modifying enzymes and other epigenetic regulators. The rationale for applying epigenetic drugs to treat cancer is twofold. First, epigenetic changes are reversible, and drugs could therefore be used to restore the normal (healthy) epigenetic landscape. However, it is unclear whether drugs can faithfully restore the precancerous epigenetic state. Second, chromatin regulators are often mutated in cancer, making them attractive drug targets. However, in most instances it is unknown whether cancer cells are addicted to these mutated chromatin proteins, or whether their mutation merely results in epigenetic instability conducive to the selection of secondary aberrations. An alternative incentive for targeting chromatin regulators is the exploitation of cancerspecific vulnerabilities, including synthetic lethality, caused by epigenetic deregulation. We review evidence for the hypothesis that mechanisms other than oncogene addiction are a basis for the application of epigenetic drugs, and propose future research directions. Anticancer therapy and epigenetics The success of an anticancer drug critically depends on its ability to kill cancer cells without causing severe side effects in healthy tissues. Hence, most therapeutics in the clinic today exploit inherent differences between cancer cells and normal cells such as defective DNA repair pathways and checkpoints, and the presence of oncogenic kinases. Another important feature that discriminates tumor from normal cells involves epigenetic deregulation. In normal cells, gene expression is tightly controlled by epigenetic mechanisms that regulate chromatin accessibility for the transcriptional machinery through post-translational modifications of histones (e.g., methylation, acetylation, and phosphorylation) and DNA (methylation and hydroxymethylation). Modified chromatin in turn recruits ‘reader’ proteins to carry out further regulatory functions. However, many cancers display global changes in histone modifications, aberrant Corresponding authors: Kubicek, S. ([email protected]); Nijman, S.M.B. ([email protected]). Keywords: chromatin-modifying enzymes; drugs; therapeutics; synthetic lethality; non-oncogene addiction. 0165-6147/$ – see front matter ß 2014 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.tips.2014.01.001
methylation patterns, and mutations in or altered expression levels of key chromatin-modifying proteins, and some tumors even harbor mutations in histone variants. Here we discuss how drugs that inhibit epigenetic modifiers can specifically target cancer cells by exploiting this deregulated chromatin state, including the targeting of synthetic lethal interactions (also known as non-oncogene addiction or induced essentiality; Box 1). We only briefly touch on epigenetics and cancer (Box 2) and refer readers to recent reviews on these topics [1–8]. Network rewiring in cancer generates new synthetic lethal targets During tumorigenesis, normal cells acquire a set of hallmark characteristics that allow survival and proliferation even in the presence of counteracting signals [9–11]. This dramatic transition is reflected by extensive rewiring of cellular signaling networks, and is driven by mutations and the deregulated expression of oncogenes and tumor suppressor genes [12,13]. Thus, the cellular networks of cancer cells are distinct from those of normal cells, and most of the mechanisms of action of anticancer drugs can be explained by understanding the rewiring involved. Such network changes can lead to a critical reliance on an oncogenic driver event (oncogene addiction, Box 1 and Figure 1A–C), which underlies most of the targeted agents that inhibit overactive oncogenes such as BCR-ABL1 (breakpoint cluster region c–abl oncogene 1 fusion protein), BRAF (v-raf murine sarcoma viral oncogene homolog B), EGFR (epidermal growth factor receptor), and HER2 (ERBB2; v-erb-b2 avian erythroblastic leukemia viral oncogene homolog 2) inhibitors. Network rewiring can also result in strict dependence on the activity of gene products that are not essential in normal cells. Thus, inactivation of such a de novo essential gene through mutation would lead to cell death or a decrease in fitness. This phenomenon has traditionally been studied in model organisms such as the budding yeast Saccharomyces cerevisiae and the fruit fly Drosophila melanogaster, and has been termed synthetic lethality or synthetic sickness (Box 1) [14–17]. The concept of synthetic lethality (in a broader sense also termed non-oncogene addiction or induced essentiality) has been proposed as a therapeutic avenue to specifically target cancer cells and leave normal cells largely unaffected: introduction of a perturbation (i.e., a drug) would represent a ‘single mutant’ and hence would be non-toxic to normal cells, but the transformed cancer cells would be ‘double mutant’ and thus synthetically lethal/sick [6,11,18–21]. Theoretically, Trends in Pharmacological Sciences xx (2014) 1–10
1
TIPS-1113; No. of Pages 10
Review Box 1. Genetic interactions, synthetic lethality, and oncogene addiction Background and nomenclature In complex cellular systems, genes do not act independently from each other but function in a concerted manner, constituting a genetic interaction network. The topology of genetic interaction networks has evolved to confer robustness such that an organism can cope with various types of internal and external perturbation while maintaining the capacity to evolve over generations. This has led to the identification of various mechanisms, and corresponding terms, that underlie but also undermine this robustness. Buffering: Robust genetic interaction networks buffer genetic variation, as illustrated by the observation that under standard laboratory culture conditions, the great majority of yeast single mutants are viable. Synthetic lethality and synthetic sickness: On occasion, single mutants become highly sensitive to specific second mutations, resulting in cell death or a decrease in fitness. These phenomena have been termed synthetic lethality and synthetic sickness, respectively (from the Greek syn-tithenai for to put together). Uncovering synthetic interactions is not trivial because they are rare and not always directly conserved from model organisms. Non-oncogene addiction and induced essentiality: Synthetic lethality has been revisited as a promising strategy for specific targeting of cancer, because cancer cells harbor numerous genetic aberrations that distinguish them from normal cells. To discern these applications from the precise genetic meaning, they have also been dubbed non-oncogene addiction or induced essentiality. Oncogene addiction: Distinct from synthetic lethality, oncogene addiction describes the rewired state of the signaling network of a cancer cell that has become critically dependent on the expression of a driver oncogene by shutting off parallel signaling pathways. A typical example of oncogene addiction with clinical relevance is the dependence of chronic myeloid leukemia (CML) cells on the BCR-ABL fusion gene, inhibition of which causes cell death.
exploitation of such synthetic interactions for therapeutic intervention by definition has less toxic side effects on normal tissue, thereby increasing the therapeutic index. Importantly, synthetic lethality would allow targeting of not only gain-of-function oncogenic events but also of lossof-function mutations in tumor suppressor genes, as well as oncogenes that are not directly druggable. Epigenetic deregulation as a cancer vulnerability Interestingly, several studies have highlighted that chromatin modifiers play a special role in buffering genetic interactions. One such study involved a screen for genetic interactions in the worm Caenorhabditis elegans [22]. By systematically screening 65 000 pairs of genes using RNA interference, the authors found that chromatin modifiers were strongly enriched among the gene pairs displaying synthetic lethality. This included components of several conserved chromatin modifier complexes such as the histone acetyl transferase NuA4–Tip60 complex, the transcriptional co-activator Mediator complex, and the Mi2–NuRD (nucleosome remodeling deacetylase) complex. Conceptually, it appears logical that the fine tuning of transcriptional processes is well positioned to buffer perturbations in other regulatory layers and is reminiscent of the role of the protein folding chaperone Hsp90 in buffering genetic variation [23]. Whereas heat shock proteins adjust protein folding and stability, epigenetic buffering directly targets transcription and may thus compensate for deleterious passenger mutations via transcriptional activation of the second allele, 2
Trends in Pharmacological Sciences xxx xxxx, Vol. xxx, No. x
Box 2. Epigenetics and cancer What evidence links chromatin and cancer? The earliest hints of a connection between epigenetics and cancer came from DNA methylation studies more than three decades ago, but were merely correlative in nature. With the advent of highthroughput sequencing technologies and the analysis of large sample numbers in recent years, it is now evident that epigenetic changes play a widespread role in the establishment of cancer via various mechanisms. For instance, many tumor suppressor genes are epigenetically inactivated by hypermethylation of their promoters. A number of mouse models have shown that epigenetic regulators are required for tumorigenesis, and most importantly, mutated chromatin-modifying enzymes have frequently been found in various types of cancer. What consequences arise from a deregulated epigenome? It has been proposed that aberrant epigenetic mechanisms allow for oncogenic reprogramming reminiscent of the differentiation process. In another model presented by Timp and Feinberg [46], it is hypothesized that deregulation of the epigenome occurs in the very early stages of tumorigenesis and causes flattening of the epigenetic landscape, leading to a stochastic departure from the normal epigenetic signature. This in turn allows for establishment of heterogeneity, acquisition of cancer hallmarks, and alteration of mutation rates. Changes in DNA methylation also affect the cell mutation rate, for instance, through higher rates of spontaneous deamination at methylated compared to unmethylated cytosines. Moreover, in the absence of DNMTs, cells suffer microsatellite instability, destabilization of repeats, and alterations in telomere length (and lengthening, ALT) and recombination. Not only DNA methylation but also histone methylation, especially H3K9me3, and other indicators of heterochromatin have been correlated with elevated mutation rates. Furthermore, it has been shown that overexpression of EZH2 contributes to genomic instability.
repression of the mutant allele, or regulation of alternative pathways. Enrichment of chromatin modifiers was also observed in a large screen in S. cerevisiae [24]. A genome-scale genetic interaction screen consisting of 5.4 million pairwise interactions showed that genes involved in chromatin function were among the most highly connected genes in the network. Furthermore, chromatin modifiers are often found bridging different cellular functions, further supporting a buffering function in the cell. Although similar high-throughput studies in human (cancer) cells are lacking, experiments using histone deacetylase (HDAC) inhibitors suggest a conserved function. There are reports that HDAC inhibitors frequently synergize with other compounds such as DNA methyltransferase (DNMT) inhibitors, proteasome inhibitors, or inhibitors of pro-survival proteins, resulting in greater anticancer effects than their additive individual effects [25]. Moreover, they seem to be enriched in drug combination screens aimed at killing cancer cells (S.K., unpublished observation). Together, these studies suggest that chromatin modifiers are important in maintaining homeostasis on genetic perturbations and that this function is conserved. Below we discuss some of the implications of this notion in the context of specific epigenetic drugs. Do approved HDAC and DNMT inhibitors work via synthetic lethal mechanisms? Chromatin emerged as a drug target long before epigenetics was considered a druggable area of biology or even
TIPS-1113; No. of Pages 10
Review
Trends in Pharmacological Sciences xxx xxxx, Vol. xxx, No. x
Direct targeng / oncogene addicon Acvang mutaons/translocaons (A)
(B)
(C) Y641F
GOF
HAT
EZH2
HATi Me Me
Ac Ac
Me
Ac
EZH2i
Me Me
Ac
Me Me
PHDi
PHF23
NUP98
JARID1A NUP98
Indirect epigenec vulnerability (D)
(E)
Recruitment by fusion proteins HDACi PML
PLZF
Chroman catastrophe and buffering exhauson DNMT3A/B DNMT1 mutaon
DNMTi
RARa
RARa
HDAC
DNMT
RUNX1 ETO
Normal CBF
5-azacydine HDACi
mC
MYH11
IDH mutaon TET2 mutaon
Cell death
mC hmC
DNMTi HMTi
Tumor
(F)
Chroman mark homeostasis Me Me Me
mut
(G)
mut
JMJD3
X
BRD4
UTX
BRD4
EZH2
NUT
NUT
BRD4 NUT Ac Ac Ac Ac
H3K27
(H)
Chroman mark dissipaon HDACi
HDAC
Ac Ac
BRD4
NUT
Ac
Ac Ac
BRD4 NUT
Ac
Ac
EZH2i
Repression of oncogenes
(I)
Chroman mark replacement MLL
X
BRD4
AF4/9
H3K4 Me
H3K79 Me
c-MYC
DOT1L
BETi DOT1Li
TRENDS in Pharmacological Sciences
Figure 1. Rationale and opportunities for targeting chromatin in cancer through potential synthetic lethality (and related) mechanisms. (A–C) Direct targeting or oncogene addiction, (D–I) various types of indirect epigenetic vulnerability. (A) Cancers with gain-of-function mutations/amplifications of histone acetyltransferases (HATs) may be susceptible to HAT inhibition. (B) Cancers harboring NUP98-JARID1A/PHF23 translocations that result in PHD-finger-containing fusion proteins may be sensitive to PHD inhibition. (C) In diffuse large B cell lymphoma (DLBCL) with GOF of enhancer of zeste homolog 2 (EZH2), treatment with EZH2 inhibitors leads to reduced proliferation and apoptosis. (D) Various fusion proteins recruit DNA methyltransferases (DNMTs) and histone deacetylases (HDACs), which can be exploited by treatment with corresponding inhibitors. (E) Mutations in DNMTs, TET2, and IDH1 result in altered levels of DNA methylation (methyl-C, mC) and hydroxymethylation (hydroxymethyl-C, hmC), which creates potential sensitivities to 5-azacytidine/HDAC inhibition or DNMT/HMT inhibition, respectively, through chromatin catastrophe and/or general exhaustion of buffering mechanisms. (F) In cancers with loss-of-function mutations in H3K27 demethylases JMJD3 or UTX, treatment with EZH2 inhibitors can restore normal methylation levels. (G) In nuclear protein in testis (NUT) midline carcinoma, the BRD4-NUT fusion protein accumulates at particular hyperacetylated sites. Treatment with HDAC inhibitors causes dissipation of chromatin acetylation marks and redistribution of BDR4-NUT. (H) It has been shown that treatment of leukemias transcriptionally represses c-MYC, a mechanism that could extend to other oncogenes. (I) Mixed-lineage leukemia (MLL)-rearranged leukemias recruit DOT1L, which leads to replacement of H3K4 by H3K27 methylation and dependence on DOT1L.
3
TIPS-1113; No. of Pages 10
Review established as a scientific field. For instance, DNAdamaging and -intercalating agents cause not only a DNA damage response and cell cycle phenotype but also global chromatin changes, and these may contribute to their specific effects on cancer cells [1,26]. Furthermore, the discovery and development of the first HDAC inhibitor were sparked by a serendipitous finding that the commonly used chemical solvent dimethyl sulfoxide (DMSO) causes differentiation of leukemia cells [27]. This observation inspired a 20-year quest that resulted in approval of the HDAC inhibitor SAHA/vorinostat for cutaneous T cell lymphoma (CTCL) in 2006 (Figure 2). A second, structurally unrelated natural product with HDAC inhibitor activity (romidepsin) was approved for the same disease in 2009. HDACs are best known for their ability to remove acetyl groups from histones, thereby increasing DNA binding and reducing accessibility of regulatory regions to transcription factors. In addition, both histone and nonhistone acetylation can exert their biological function via recruitment of specific acetyl-binding proteins typically containing a bromo domain, providing another layer of regulation by acetylation that might be druggable. Currently, more than 20 additional HDAC inhibitors are being evaluated in clinical trials across various malignancies, including solid tumors. However, despite their clinical success and promise, the mechanism of action of HDAC inhibitors that explains their cancer cell selectivity remains elusive [28]. HDACs themselves are rarely mutated, and it appears unlikely that increased HDAC activity per se contributes to tumorigenesis. Furthermore, HDAC expression or activity has not been linked to the response to HDAC inhibitors, and other predictive biomarkers are lacking. Thus, the response to HDAC inhibition of a variety of cancer cell lines cannot be easily explained by an oncogene addiction mode of action. In some instances the sensitivity to HDAC inhibitors may be due to an oncogenic driver that requires HDAC activity, resulting in induced essentiality. Examples illustrating this concept are oncogenic fusion proteins such as PML-RARa (promyelocytic leukemia–retinoic acid receptor alpha), PLZF-RARa (promyelocytic leukaemia zinc finger–RARa), CBF-MYH11 (core binding factor–myosin heavy chain 11), and RUNX1-ETO (runt-related transcription factor 1–RUNX1 translocated to 1), which recruit HDACs and induce aberrant gene silencing (Figure 1D). This may explain a specific sensitivity to HDAC inhibitors, because they can reverse some of the abnormal gene repression correlating with growth arrest, differentiation, and apoptosis in cancer cells [4,29–31]. At therapeutically relevant doses, HDAC inhibitors typically change the expression of thousands of genes [32,33], supporting the notion that they interfere with histone acetylation and thereby impinge on gene transcription. However, it remains unclear how these global effects would result in specific triggering of apoptosis in cancer cells. In addition, HDACs increase the acetylation of non-histone proteins, and this may also contribute to their cytotoxicity.
Trends in Pharmacological Sciences xxx xxxx, Vol. xxx, No. x
A second class of anticancer compounds that are often referred to as epigenetic drugs are the DNA methyltransferase (DNMT) inhibitors 5-azacytidine (vidaza) and 5-aza20 deoxy-cytidine (decitabine), both approved for myelodysplastic syndrome (MDS, a collection of hematological conditions characterized by ineffective production of myeloid cells) [34]. DNMTs catalyze the addition of methyl groups to cytosines in double-stranded DNA and thus contribute to the epigenetic silencing of tumor suppressor genes with CpG promoters that become methylated during tumorigenesis. However, DNA methylation, particularly in heterochromatic regions, is considered to play important roles in genomic stability. In line with this dual role of DNA methylation, cancer cells can often be characterized by global DNA hypomethylation and local DNA hypermethylation. Both DNMT inhibitors act via incorporation into the DNA, resulting in subsequent covalent crosslinking of DNMTs to the DNA. As a consequence, the drugs deplete the available pool of DNMTs and thereby reduce CpG methylation globally, resulting in re-expression of epigenetically silenced genes [35–37]. In addition, they can cause a DNA damage response, because the DNA–protein crosslinks need to be removed and repaired before cells can divide. This dual mechanism of action of DNMT inhibitors makes it difficult to determine what drives the sensitivity of cancer cells to these compounds. Inactivating mutations in the catalytic domain of DNMT3a and DNMT1 have been found in MDS and acute myeloid leukemia (AML), yet the functional consequences of these are poorly understood, because aberrant methylation patterns do not always correlate with mutations [38,39]. Moreover, from a mechanistic perspective it is unclear how inactivating mutations in DNMT3a would render cells sensitive to DNMT inhibitors, and these mutations do not explain the more general sensitivity of cancer cells to the drugs [40]. By contrast, DNMT1 and DNMT3B seem to be overexpressed in various other solid tumors. In analogy to HDACs, the PML-RARa fusion protein, as well as EZH2 (enhancer of zeste homolog 2), recruits DNMTs to promoters of tumor suppressor genes (Figure 1D), which may provide a selective sensitivity to DNMT inhibitors in specific cases [41,42]. However, as for HDAC inhibitors, these observations suggest that drug sensitivity is not solely a consequence of deregulated DNMT activity per se. Both HDAC and DNMT inhibitors cause global genomewide chromatin changes that result in dramatic gene expression changes in both cancer and normal cells. At least part of the mechanism of action of these drugs may be explained by synthetic genetic interactions and buffering principles. In light of epigenetic deregulation as a ubiquitous phenomenon in cancer, it is not surprising that further perturbation of this process by inhibition of DNMTs and HDACs is more detrimental to malignant cells. Furthermore, cancer cells harbor many mutations and we postulate that HDAC and DNMT inhibitors are specific for target cells because they perturb the buffering capacity of the cell (Figure 1E).
Abbreviations: BRD4, bromodomain containing 4; c-MYC, v-myc avian myelocytomatosis viral oncogene homolog; DOT1L, DOT1-like histone H3K79 methyltransferase; GOF/amp, gain-of-function/amplification; H3K4, Histone 3 lysine 4; H3K27, Histone 3 lysine 27; JMJD3, jumonji domain containing 3 (also known as KDM6B, lysine (K)-specific demethylase 6B); NUP98-JARID1A/PHF23, nucleoporin 98kDa-jumonji, AT rich interactive domain 1A (also known as KDM5A, lysine (K)-specific demethylase 5A)/PHD finger protein 23; PHD, Plant Homeo Domain; UTX, ubiquitously transcribed tetratricopeptide repeat, X chromosome (also known as KDM6A, lysine (K)-specific demethylase 6A).
4
TIPS-1113; No. of Pages 10
Review
Trends in Pharmacological Sciences xxx xxxx, Vol. xxx, No. x
Histone deacetylase inhibitors O
H N
N H
O
OH
Merck
Cutaneous T-cell lymphoma
HDAC
Approved
Romidepsin
Celgene
Cutaneous T-cell lymphoma
HDAC
Approved
Azacidine
Celgene
Myelodysplasic syndrome
DNMT
Approved
Decitabine
Astex
Myelodysplasic syndrome
DNMT
Approved
EPZ-7438
Epizyme
Non-hodgkin lymphoma
EZH2
Phase I/II
EPZ-5676
Epizyme
MLL-rearranged leukemia
DOT1L
Phase I
Oryzon
Acute myeloid leukemia
LSD1
Phase I
O
HN S O S HN NH
O
Vorinostat
HN
O
O O
DNA methyltransferase inhibitors O O HO
N N
HO
NH2 N
OH O O
N N
HO
NH2 N
HO
Histone methyltransferase inhibitors O N N
O
O HN
O
NH2
HN
N N O
N OH
N OH
H N N
Histone demethylase inhibitors ORY-1001
Undisclosed
Tranylcypromine
H2N
Anergic depression
MAO A/B Approved LSD1 off-target
Bromo domain inhibitors O O
OH
N NH O
RVX208
Resverlogix
Cardiovascular disease
BET
Phase IIb
OTX015
Oncoethix
Hematological malignancies
BET
Phase I
GSK
NUT midline carcinoma
BET
Phase I/II
O N N
N
S
NH
N O
OH
Cl
N N
N
NH
N
O
O
GSK525762
Cl
TRENDS in Pharmacological Sciences
Figure 2. Epigenetic drugs and chromatin-targeting compounds in clinical development. These include two approved histone deacetylase (HDAC) inhibitors (not shown are the more than 20 additional HDAC inhibitors in clinical development) and two approved DNA methyltransferase inhibitors. Compounds targeting histone methyltransferases, demethylases, and bromodomains are in early-stage development, and clinical trials for each class have recently been initiated.
Both HDAC and DNMT inhibitors are currently being evaluated in many different clinical trials across cancer types and in combination with other compounds. If our hypothesis is correct, cancers that are most dependent on epigenetic buffering should respond best to these compounds. To identify biomarkers for patient selection, a better understanding of the oncogenic pathways that confer synthetic lethality to DNMT and HDAC inhibitors will be necessary. Similarly, it is unclear if the effect requires polypharmacology of the compounds inhibiting multiple enzymes or if a single enzyme is the most important target.
Following the trend for targeted anticancer drugs, recent development efforts are aimed at increasing the specificity of epigenetic drugs. For HDAC inhibitors, this entails making them more selective for individual HDAC enzymes, whereas currently available compounds all simultaneously target HDAC 1, 2, and 3, as well as HDAC6 and 8 to various degrees [43]. For DNMT inhibitors, competitive inhibitors are desirable to allow the separation of effects caused by DNA damage from loss of CpG methylation. However, given that it is presently unclear which effects are most central to the antitumor activities of the 5
TIPS-1113; No. of Pages 10
Review existing agents, it remains to be determined whether increases in selectivity will coincide with improved therapeutic selectivity. Most emerging epigenetic drugs target oncogene addiction The recent cancer genome resequencing efforts have highlighted many recurrent mutations in chromatin regulators [44]. For reasons that are at present unclear, mutations in chromatin regulators appear particularly frequent in hematological cancers. Perhaps the need for especially stringent epigenetic control of cell fate renders these lineages vulnerable to malignant transformation on epigenetic perturbation. However, mutations in chromatin modifiers are also found in solid tumors [45], where they mainly affect rare aggressive adult cancers, pediatric neoplasms, and cancers that relapse or are resistant to therapy [4,46]. Together, these studies have yielded an increasing appreciation of epigenetics as a druggable target class (Box 2) [4,31,47]. In a few cases, overactive chromatin modifiers have inspired the development of potent novel compounds that are highly selective for enzyme classes and subclasses. The targets of these compounds were selected based on strong genetic evidence of an oncogene addiction mode. The first compounds are currently entering early-stage clinical trials in selected populations with corresponding mutations and include EZH2, DOT1L, and BRD4 inhibitors. Importantly, for these drugs and other epigenetic drug targets, non-oncogene addiction applications can also be envisioned. This is of particular relevance because most mutations in chromatin modifiers and other driver mutations do not constitute readily accessible drug targets. Thus, a broader evaluation of epigenetic drugs in cancer may provide additional anticancer therapeutic applications. Some of these mostly speculative indications are discussed below and summarized in Table 1. Potential non-oncogene addiction applications for chromatin-targeting compounds Given that chromatin modifiers frequently display synthetic lethality, we postulate that tumors with mutations in this class of genes might be particularly suitable for therapeutic strategies that exploit this phenomenon. In particular, we would expect that targeting of additional chromatin complexes in these tumors would be especially cytotoxic through a variety of different mechanisms (see below and Table 1). For instance, we could imagine targeting of chromatin regulators that exist in pairs of enzymes with counteracting activities, such as HDACs– HATs (histone acetyltransferases), HMTs–HDMs (histone methyltransferases and demethylases), DNMTs– TETs (tet methylcytosine dioxygenases), and PRMTs– PADIs (protein arginine methyltransferases and deiminases). Inhibition of the counteractor, which would reestablish homeostatic levels for histone modifications, might compensate loss-of-function mutations in one of the partners or further deregulate homeostasis, resulting in lethality. One concrete example includes EZH2 inhibitors. Oncogenic EZH2 gain-of-function mutations cause 6
Trends in Pharmacological Sciences xxx xxxx, Vol. xxx, No. x
increased levels of H3K27me3 and occur in up to 20% of diffuse large B cell lymphoma (DLBCL) cells of germinal center origin [48]. DLBCL cells with EZH2 mutations are addicted to EZH2 activity for proliferation and survival [49–51]. However, EZH2 inhibitors may also be applied in EZH2 wild type neoplasms. Besides EZH2, mutations of other members of the polycomb repressive complex 2 (PRC2) have been found in cancer and may be candidates for therapeutic intervention through targeting of EZH2 [5,31,47]. Furthermore, the loss of EZH2 opposing demethylases UTX/KDM6A and JMJD3/KDM6B would be expected to result in similar sensitivity to EZH2 inhibitors. Indeed, mutations in these genes have been found in a wide range of tumor types, and it has been suggested that the H3K27 demethylases are tumor suppressors [52,53]. The histone mark homeostasis model predicts that UTX mutant cells will respond to EZH2 inhibitors, and this has been observed for one cell line, at least anecdotally (Figure 1F) [50]. Similar reciprocal targeting might apply for other methyl-transferase/demethylase pairs, such as for cancers harboring mutations in MLL (mixed-lineage leukemia) 2 and 3, which have been detected in various malignancies. BET domain inhibitors Other ways of applying the new highly potent chromatintargeting compounds are emerging and BET bromodomain inhibitors are attracting special interest. Bromodomains are histone acetylation binders that translate the presence of this activating mark to a transcriptional response. In NUT midline carcinoma, translocations occur between the NUT (nuclear protein in testis) protein and the bromodomain of BRD4 [54]. Interestingly, HDAC inhibitors can cause differentiation of NUT midline carcinoma cells, possibly mediated by the global increase in acetylation, thereby titrating away the BRD4 protein from critical sites [55]. Thus, HDAC inhibition in cancers with translocations of acetylation binders represents another example of how epigenetic drugs can act through non-oncogene addiction (Figure 1G). Potentially exploiting oncogene addiction, direct BRD4 inhibitors have recently been developed by targeting the BET bromodomain [56,57]. Two compounds have now entered the clinic and are first being evaluated in NUT midline carcinomas. They might also be suitable for targeting melanoma with overexpression of BRD4 [58]. However, cancer cells without mutations in BRD4 can also display profound sensitivity to this compound class. In particular, it has been shown that they downregulate transcription of the driver oncogene c-MYC in AML and multiple myeloma cells (Figure 1H) [59,60]. The selective effect observed is probably not solely attributable to cMYC downregulation, and the detailed mode of action of BET inhibitors is incompletely understood [61,62]. However, BRD4 inhibition seems to affect only a particular set of genes with exceptionally large enhancers that are marked by co-occupation of BRD4 and Mediator (superenhancers). Many of these genes are oncogenes themselves, suggesting that BRD4 inhibition could target a variety of cancer cells [63].
TIPS-1113; No. of Pages 10
Review
Trends in Pharmacological Sciences xxx xxxx, Vol. xxx, No. x
Table 1. Rationale and opportunities for targeting chromatin in cancer through potential synthetic lethality (and related) mechanisms Gene/protein
Function
Aberration Cancer type (examples)
Direct targeting Oncogene addiction HAT CBP
Mut, amp
p300
Mut
MYST3/MOZ/KAT
Mut, TL
ELP4 AIB
Amp Amp
EZH2
HMT
Suggested mechanism
Potential therapeutic Outlook targeting
HAT inhibition
Mut (GOF)
Ovarian/breast/lung/ oncogene addiction bladder cancer, lymphoma, ALL pancreatic/bladder cancer, HNSCC, lymphoma leukemia, medulloblastoma, breast cancer Breast cancer Breast/gastric/esophageal cancer, lymphoma Prostate/breast/bladder oncogene addiction cancer DLBCL oncogene addiction
sensitivity to BET inhibitors PHD inhibition
OE
EZH2 inhibition
BRD4-NUT
BET
TL
NUT midline carcinoma
oncogene addiction
JARID1A/KDM5A PHF23-NUP98
PHD finger PHD finger
TL, OE TL, mut
AML, gastric cancer AML
sensitivity to mutations impairing binding of the PHD finger to H3K4me3
Lymphoma, renal/bladder cancer, medulloblastoma Various (rare)
Altered H3K27me3 levels
EZH2 inhibition to restore H3K27me3
AML, lymphoma, colorectal cancer AML, MDS Pancreatic/liver/bladder/ breast cancer Breast cancer MPN, MDS, CMML
Vulnerability due to DNA methylation changes
Hypomethylating agents or HDAC inhibition
Decrease in 5-hmC and increase in 5-mC Glioma, chondrosarcoma, a) 2-HG accumulation AML, astrocytoma inhibits TET2, mimics its LOF b) 2-HG inhibits a-KGdependent HDMs NUT midline carcinoma Fusion causes sequestration of acetylated histones, hypoacetylation, differentiation block
DNMT inhibition
Indirect epigenetic vulnerability Chromatin alterations Mut HDM UTX/KDM6A Mut
JMJD3/KDM6B DNMT1
DNMT
Mut Mut OE
DNMT3A DNMT1
IDH1/2
OE DNA Mut (LOF) hydroxylase TCA cycle Mut (LOF)
BRD4-NUT
BET
DNMT3B TET2
Induced essentiality HMT MLL1, 2, 3
TL
TL, mut
Induction of transcriptional changes TF TL PML-RARa PLZF-RARa
TF
TL
RUNX1-ETO
TF
TL
CBF-MYH11
TF
TL
PML-RARa
TF
TL
EZH2
HMT
OE, mut
HDAC inhibition (epigenetic differentiation therapy)
DOT1L essential in MLL fusion cancers MLL fusion addicted to menin
AML, PML, leukemias AML, PML, leukemias AML, PML, leukemias AML, PML, leukemias AML, PML, leukemias
Recruitment of HDACs and HDAC inhibition induction of aberrant gene silencing
various
Similar reciprocal targeting for other methyltransferase/ demethylase pairs
a) DNMT inhibition b) HMT inhibition
MLL/ALL/AML, medulloblastoma, lymphoma, HNSCC, bladder/breast/renal/lung/ gastric/prostate cancer
various
targeting other PRC2 components (directly or through EZH2 inhibition)
DOT1L inhibition Menin inhibition
Similar dependence suggested with PRMT1, BRD4, LSD1, PRC2, IKK, and MEK inhibition
various various various
Recruitment of DNMTs to promoters of tumor suppressor genes
DNMT inhibition
7
TIPS-1113; No. of Pages 10
Review
Trends in Pharmacological Sciences xxx xxxx, Vol. xxx, No. x
Table 1 (Continued ) Gene/protein
BRD4
Function
BET
Aberration Cancer type (examples)
OE
MDS, MPN, myeloid malignancies, T-ALL, HNSCC, prostate, breast, bladder cancer, DLBCL Melanoma, various leukemias
Suggested mechanism
Potential therapeutic Outlook targeting
BRD4 loss downregulates c-MYC transcription
BRD4 inhibition
Affects genes with super-enhancers (many oncogenes)
HAT, histone acetyltransferase; HMT, histone methyltransferase; HDM, histone demethylase; DNMT, DNA methyltransferase; HDAC, histone deacetylase; BET, BET bromodomain; TF, transcription factor; mut, mutated; amp, amplified; TL, translocation; LOF, loss of function; GOF, gain of function; LOH, loss of heterozygosity; OE, overexpression; ALL, acute lymphoblastic leukemia; HNSCC, head and neck squamous cell carcinoma; DLBCL, diffuse large B cell lymphoma; AML, acute myeloid leukemia; MPN, myeloproliferative neoplasms; MDS, myelodysplastic syndrome; CMML, chronic myelomonocytic leukemia; PML, promyelocytic leukemia; MSI, microsatellite instability; ALT, alternative lengthening of telomeres. If not indicated otherwise or in the main text, information on genes, mutations, cancer types, and frequencies is from [4,5,30,40,41,56] and references therein.
MLL translocations result in induced essentiality of DOT1L and menin MLL translocations are the most common lesion in childhood AMLs, and they also occur in adult leukemias [64]. Normally, the MLL protein contain a SET domain conferring HMT activity that generates histone H3 lysine 4 methylation. In leukemic translocations, the catalytic activity of MLL is lost and replaced with other protein domains. More than 70 different fusion partners have been identified, and interestingly they very often contain other chromatin-modifying activities. For example, acetyltransferase and arginine methyltransferase domains have been identified in fusion partners. The most common fusion partners are AF4 and AF9, which do not harbor chromatin-modifying activities themselves. However, they recruit DOT1L, a methyltransferase that generates histone H3 lysine 79 methylation. Therefore, at the chromatin level, translocation results in the replacement of H3K4 with H3K79 methylation at MLL target sites (Figure 1I) [65– 68]. Based on these protein-protein interaction data, the effects of knockdown and knockout of DOT1L in MLLtranslocated leukemias have been evaluated in cell culture and animal models [69,70]. Pronounced non-oncogene dependence has been detected, and this result endorsed the development of a highly potent DOT1L inhibitor, EPZ5676. This compound competes with the cofactor Sadenosyl-methionine that is used by all histone methyltransferases, yet it is highly selective for DOT1L and inhibits this enzyme at picomolar concentrations. In xenograft models with MLL-rearranged leukemia lines, EPZ5676 leads to tumor regression and sustained responses, even after compound withdrawal [71]. This compound is currently being evaluated in clinical trials (NCT01684150). In addition, it has been shown that MLL-translocated malignancies are addicted to menin, an interaction that was recently successfully targeted by a small-molecule inhibitor [72,73]. Similar dependence of MLL-fusion leukemias has been suggested with PRMT1, BRD4, LSD1, PRC2, IKK, and MEK inhibition [74–76]. Concluding remarks The recognition that aberrant epigenetic regulators are causally implicated in cancer, as well as the fact that many 8
of them are druggable, has fueled the development of small-molecule inhibitors for various epigenetic enzyme classes. The field of epigenetics is now firmly established as an important area of drug development. However, most efforts focus on a few selected proteins with a strong genetic basis such as amplifications, translocations, and activating mutations to exploit oncogene addiction. As illustrated by several examples in this review, the inhibition of many epigenetic regulators can result in specific effects in cancer cells through mechanisms akin to synthetic lethality. Thus, many other chromatin proteins might be attractive targets for inhibitor development based on non-oncogene addiction through a variety of mechanisms. Therefore, we envision that systematic investigations of epigenetic regulators as drug targets will be a rewarding approach towards the identification of novel anticancer drugs. Along the same lines, epigenetic drugs are attractive candidates for combination with other drugs. Supporting this notion, numerous synergies of HDAC and DNMT inhibitors with other molecules have been found in vitro. These synergies and their potent biological activities have kept the interest in HDAC and DNMT inhibitors high, and more than 400 clinical trials are currently being performed with these compound classes. Ultimately, to prioritize relevant targets from the more than 400 chromatin-modifiers, systematic evaluation of the involvement of these proteins in synthetic lethal interactions will be essential. Recent advances in genome engineering and isogenic mutant collections will provide technical avenues to achieve this goal. Acknowledgements We thank Christoph Bock and Helen Pickersgill for critical reading and discussions. B.M. is supported by a DOC fellowship from the Austrian Academy of Sciences.
References 1 Baylin, S.B. and Jones, P.A. (2011) A decade of exploring the cancer epigenome – biological and translational implications. Nat. Rev. Cancer 11, 726–734 2 Brough, R. et al. (2011) Searching for synthetic lethality in cancer. Curr. Opin. Genet. Dev. 21, 34–41 3 Chan, D.A. and Giaccia, A.J. (2011) Harnessing synthetic lethal interactions in anticancer drug discovery. Nat. Rev. Drug Discov. 10, 351–364
TIPS-1113; No. of Pages 10
Review 4 Dawson, M.A. and Kouzarides, T. (2012) Cancer epigenetics: from mechanism to therapy. Cell 150, 12–27 5 Helin, K. and Dhanak, D. (2013) Chromatin proteins and modifications as drug targets. Nature 502, 480–488 6 Kaelin, W.G., Jr (2005) The concept of synthetic lethality in the context of anticancer therapy. Nat. Rev. Cancer 5, 689–698 7 Rodriguez-Paredes, M. and Esteller, M. (2011) Cancer epigenetics reaches mainstream oncology. Nat. Med. 17, 330–339 8 Skene, P.J. and Henikoff, S. (2013) Histone variants in pluripotency and disease. Development 140, 2513–2524 9 Hanahan, D. and Weinberg, R.A. (2000) The hallmarks of cancer. Cell 100, 57–70 10 Hanahan, D. and Weinberg, R.A. (2011) Hallmarks of cancer: the next generation. Cell 144, 646–674 11 Luo, J. et al. (2009) Principles of cancer therapy: oncogene and nononcogene addiction. Cell 136, 823–837 12 Gonzalez-Angulo, A.M. et al. (2010) Future of personalized medicine in oncology: a systems biology approach. J. Clin. Oncol. 28, 2777–2783 13 Vidal, M. et al. (2011) Interactome networks and human disease. Cell 144, 986–998 14 Bridges, C.B. (1922) The origin of variation. Am. Nat. 56, 51–63 15 Dobzhansky, T. (1946) Genetics of natural populations. XIII. Recombination and variability in populations of Drosphila pseudoobscura. Genetics 31, 269–290 16 Lucchesi, J.C. (1968) Synthetic lethality and semi-lethality among functionally related mutants of Drosophila melanogaster. Genetics 59, 37–44 17 Sturtevant, A.H. (1956) A highly specific complementary lethal system in Drosophila melanogaster. Genetics 41, 118–123 18 Hartwell, L.H. et al. (1997) Integrating genetic approaches into the discovery of anticancer drugs. Science 278, 1064–1068 19 Kaelin, W.G., Jr (1999) Choosing anticancer drug targets in the postgenomic era. J. Clin. Invest. 104, 1503–1506 20 Kaelin, W.G., Jr (2009) Synthetic lethality: a framework for the development of wiser cancer therapeutics. Genome Med. 1, 99 21 Reinhardt, H.C. et al. (2009) Exploiting synthetic lethal interactions for targeted cancer therapy. Cell Cycle 8, 3112–3119 22 Lehner, B. et al. (2006) Systematic mapping of genetic interactions in Caenorhabditis elegans identifies common modifiers of diverse signaling pathways. Nat. Genet. 38, 896–903 23 Queitsch, C. et al. (2002) Hsp90 as a capacitor of phenotypic variation. Nature 417, 618–624 24 Costanzo, M. et al. (2010) The genetic landscape of a cell. Science 327, 425–431 25 Bots, M. and Johnstone, R.W. (2009) Rational combinations using HDAC inhibitors. Clin. Cancer Res. 15, 3970–3977 26 Soria, G. et al. (2012) Prime, repair, restore: the active role of chromatin in the DNA damage response. Mol. Cell 46, 722–734 27 Marks, P.A. and Breslow, R. (2007) Dimethyl sulfoxide to vorinostat: development of this histone deacetylase inhibitor as an anticancer drug. Nat. Biotechnol. 25, 84–90 28 Khan, O. and La Thangue, N.B. (2012) HDAC inhibitors in cancer biology: emerging mechanisms and clinical applications. Immunol. Cell Biol. 90, 85–94 29 Petruccelli, L.A. et al. (2013) Expression of leukemia-associated fusion proteins increases sensitivity to histone deacetylase inhibitor-induced DNA damage and apoptosis. Mol. Cancer Ther. 12, 1591–1604 30 Ropero, S. and Esteller, M. (2007) The role of histone deacetylases (HDACs) in human cancer. Mol. Oncol. 1, 19–25 31 Shen, H. and Laird, P.W. (2013) Interplay between the cancer genome and epigenome. Cell 153, 38–55 32 Wang, Z. et al. (2009) Genome-wide mapping of HATs and HDACs reveals distinct functions in active and inactive genes. Cell 138, 1019– 1031 33 Kubicek, S. et al. (2012) Chromatin-targeting small molecules cause class-specific transcriptional changes in pancreatic endocrine cells. Proc. Natl. Acad. Sci. U.S.A. 109, 5364–5369 34 Yang, X. et al. (2010) Targeting DNA, methylation for epigenetic therapy. Trends Pharmacol. Sci. 31, 536–546 35 Christman, J.K. (2002) 5-Azacytidine and 5-aza-20 -deoxycytidine as inhibitors of DNA methylation: mechanistic studies and their implications for cancer therapy. Oncogene 21, 5483–5495
Trends in Pharmacological Sciences xxx xxxx, Vol. xxx, No. x
36 Juttermann, R. et al. (1994) Toxicity of 5-aza-20 -deoxycytidine to mammalian cells is mediated primarily by covalent trapping of DNA methyltransferase rather than DNA demethylation. Proc. Natl. Acad. Sci. U.S.A. 91, 11797–11801 37 Orta, M.L. et al. (2013) 5-Aza-20 -deoxycytidine causes replication lesions that require Fanconi anemia-dependent homologous recombination for repair. Nucleic Acids Res. 41, 5827–5836 38 Ley, T.J. et al. (2010) DNMT3A mutations in acute myeloid leukemia. N. Engl. J. Med. 363, 2424–2433 39 Walter, M.J. et al. (2011) Recurrent DNMT3A mutations in patients with myelodysplastic syndromes. Leukemia 25, 1153–1158 40 Metzeler, K.H. et al. (2012) DNMT3A mutations and response to the hypomethylating agent decitabine in acute myeloid leukemia. Leukemia 26, 1106–1107 41 Di Croce, L. et al. (2002) Methyltransferase recruitment and DNA hypermethylation of target promoters by an oncogenic transcription factor. Science 295, 1079–1082 42 Vire, E. et al. (2006) The polycomb group protein EZH2 directly controls DNA methylation. Nature 439, 871–874 43 Bradner, J.E. et al. (2010) Chemical phylogenetics of histone deacetylases. Nat. Chem. Biol. 6, 238–243 44 Garraway, L.A. and Lander, E.S. (2013) Lessons from the cancer genome. Cell 153, 17–37 45 Kandoth, C. et al. (2013) Mutational landscape and significance across 12 major cancer types. Nature 502, 333–339 46 Timp, W. and Feinberg, A.P. (2013) Cancer as a dysregulated epigenome allowing cellular growth advantage at the expense of the host. Nat. Rev. Cancer 13, 497–510 47 Geutjes, E.J. et al. (2012) Targeting the epigenome for treatment of cancer. Oncogene 31, 3827–3844 48 Morin, R.D. et al. (2010) Somatic mutations altering EZH2 (Tyr641) in follicular and diffuse large B-cell lymphomas of germinal-center origin. Nat. Genet. 42, 181–185 49 Knutson, S.K. et al. (2012) A selective inhibitor of EZH2 blocks H3K27 methylation and kills mutant lymphoma cells. Nat. Chem. Biol. 8, 890– 896 50 McCabe, M.T. et al. (2012) EZH2 inhibition as a therapeutic strategy for lymphoma with EZH2-activating mutations. Nature 492, 108–112 51 Qi, W. et al. (2012) Selective inhibition of Ezh2 by a small molecule inhibitor blocks tumor cells proliferation. Proc. Natl. Acad. Sci. U.S.A. 109, 21360–21365 52 Ene, C.I. et al. (2012) Histone demethylase Jumonji D3 (JMJD3) as a tumor suppressor by regulating p53 protein nuclear stabilization. PLoS ONE 7, e51407 53 van Haaften, G. et al. (2009) Somatic mutations of the histone H3K27 demethylase gene UTX in human cancer. Nat. Genet. 41, 521–523 54 French, C.A. et al. (2004) Midline carcinoma of children and young adults with NUT rearrangement. J. Clin. Oncol. 22, 4135–4139 55 Schwartz, B.E. et al. (2011) Differentiation of NUT midline carcinoma by epigenomic reprogramming. Cancer Res. 71, 2686–2696 56 Filippakopoulos, P. et al. (2010) Selective inhibition of BET bromodomains. Nature 468, 1067–1073 57 Nicodeme, E. et al. (2010) Suppression of inflammation by a synthetic histone mimic. Nature 468, 1119–1123 58 Segura, M.F. et al. (2013) BRD4 sustains melanoma proliferation and represents a new target for epigenetic therapy. Cancer Res. 73, 6264–6276 59 Delmore, J.E. et al. (2011) BET bromodomain inhibition as a therapeutic strategy to target c-Myc. Cell 146, 904–917 60 Zuber, J. et al. (2011) RNAi screen identifies Brd4 as a therapeutic target in acute myeloid leukaemia. Nature 478, 524–528 61 Dawson, M.A. et al. (2012) Targeting epigenetic readers in cancer. N. Engl. J. Med. 367, 647–657 62 Popovic, R. and Licht, J.D. (2012) Emerging epigenetic targets and therapies in cancer medicine. Cancer Discov. 2, 405–413 63 Loven, J. et al. (2013) Selective inhibition of tumor oncogenes by disruption of super-enhancers. Cell 153, 320–334 64 Muntean, A.G. and Hess, J.L. (2012) The pathogenesis of mixedlineage leukemia. Annu. Rev. Pathol. 7, 283–301 65 Bernt, K.M. et al. (2011) MLL-rearranged leukemia is dependent on aberrant H3K79 methylation by DOT1L. Cancer Cell 20, 66–78 66 Chang, M.J. et al. (2010) Histone H3 lysine 79 methyltransferase Dot1 is required for immortalization by MLL oncogenes. Cancer Res. 70, 10234–10242 9
TIPS-1113; No. of Pages 10
Review 67 Jo, S.Y. et al. (2011) Requirement for Dot1l in murine postnatal hematopoiesis and leukemogenesis by MLL translocation. Blood 117, 4759–4768 68 Nguyen, A.T. et al. (2011) DOT1L, the H3K79 methyltransferase, is required for MLL-AF9-mediated leukemogenesis. Blood 117, 6912– 6922 69 Daigle, S.R. et al. (2011) Selective killing of mixed lineage leukemia cells by a potent small-molecule DOT1L inhibitor. Cancer Cell 20, 53– 65 70 Neff, T. and Armstrong, S.A. (2013) Recent progress toward epigenetic therapies: the example of mixed lineage leukemia. Blood 121, 4847–4853 71 Daigle, S.R. et al. (2013) Potent inhibition of DOT1L as treatment of MLL-fusion leukemia. Blood 122, 1017–1025
10
Trends in Pharmacological Sciences xxx xxxx, Vol. xxx, No. x
72 Grembecka, J. et al. (2012) Menin-MLL inhibitors reverse oncogenic activity of MLL fusion proteins in leukemia. Nat. Chem. Biol. 8, 277– 284 73 Yokoyama, A. et al. (2005) The menin tumor suppressor protein is an essential oncogenic cofactor for MLL-associated leukemogenesis. Cell 123, 207–218 74 Kuo, H.P. et al. (2013) Epigenetic roles of MLL oncoproteins are dependent on NF-kappaB. Cancer Cell 24, 423–437 75 Miller, P.G. et al. (2013) In vivo RNAi screening identifies a leukemiaspecific dependence on integrin beta 3 signaling. Cancer Cell 24, 45–58 76 Pemovska, T. et al. (2013) Individualized systems medicine (ISM) strategy to tailor treatments for patients with chemorefractory acute myeloid leukemia. Cancer Discov. 3, 1416–1429
