|Home | About | Journals | Submit | Contact Us | Français|
Normal transcription displays a high degree of flexibility over the choice, timing and magnitude of mRNA expression levels that tend to oscillate and cycle. These processes allow for combinatorial actions, feedback control and fine-tuning. A central role has emerged for the transcriptional co-repressor proteins such as NCOR1, NCOR2/SMRT, CoREST and CTBPs, to control the actions of many transcriptional factors, in large part, by recruitment and activation of a range of chromatin remodeling enzymes. Thus, co-repressors and chromatin remodeling factors are recruited to transcription factors at specific promoter/enhancer regions and execute changes in the chromatin structure. The specificity of this recruitment is controlled in a spatial-temporal manner. By playing a central role in transcriptional control, as they move and target transcription factors, co-repressors act as a key driver in the epigenetic economy of the nucleus. Co-repressor functions are selectively distorted in malignancy, by both loss and gain of function and contribute to the generation of transcriptional rigidity. Features of transcriptional rigidity apparent in cancer cells include the distorted signaling of nuclear receptors and the WNTs/β-catenin axis. Understanding and predicting the consequences of altered co-repressor expression patterns in cancer cells has diagnostic and prognostic significance, and also have the capacity to be targeted through selective epigenetic therapies.
A high level of organized complexity underpins homeostasis in humans, central to which are the dexterous and integrated actions of transcriptional networks. Frequently these networks result in transcriptional outputs that oscillate between on and off states. Alternating between these two states can be considered to be a cyclic behavior. As yet the master regulators of these alternating and cycling events remain incompletely understood. Developmental programs display such phasic spatial-temporal regulation, for example, in embryonic vertebrate formation, and combine with extrinsic inputs, to generate the circadian rhythm and metabolic control. This oscillating, or cycling behavior also occurs in response to environmental signals, such as in the regulation of inflammatory genes. The nuclear receptors (NRs) also entrain transcriptional cycling in a wide variety of target genes in response to both intrinsic hormonal and extrinsic dietary derived signals.
Transcriptional cycling arises from the actions of distinct activating and repressing complexes that together allow for fine control of mRNA regulation. The choreography of these actions is tightly regulated and allows these regulatory complexes to recognize, interpret and sustain histone modifications on target gene promoters. The number of histone modifications and their combinatorial nature generates a very diverse and malleable interface platform between DNA and transcription factors (reviewed in1) and is a further driver of transcriptional plasticity. The regulation of histone modifications acts as a type of biological ratchet in which sequential complexes feed forward to initiate and sustain transcription2, 3. It should be borne in mind that very few, if any, transcriptional targets are governed by a single factor, but rather the patterns of transcriptional regulation reflect the integration of multiple factors acting in concert. Collectively, the endpoint of these interactions is the spatial-temporal distribution of boundaries between transcriptionally rich euchromatin and transcriptionally restricted heterochromatin. The boundaries between these chromatin states are cemented further by regulation of CpG island methylation; thus histone modifications and CpG methylation are dynamically intertwined4.
Co-activators and co-repressors significantly regulate the control of transcriptional active and inactive states events respectively. Co-repressors repress transcription by recognizing, initiating, and sustaining repressive chromatin environments and contribute the points of minima and limit the maxima on the waves of transcription cycling and thereby significantly regulate the timing of transcription. Given their role to regulate epigenetic events that underlie transcriptional inactivation, the structurally diverse yet phenotypically related co-repressor proteins have emerged as key players in cancer etiology.
Post-genomic analyses of major transcription factor families, in both malignant and non-malignant cell types, have resulted in a revision of understanding. Transcription factor actions in cancer cell systems appear to exhibit a restricted repertoire of the dexterity and plasticity displayed by normal cell systems. The evolution of a restricted malignant transcriptome is seen clearly in the NR superfamily, but is also apparent in the MYC and AP-1 networks (reviewed in5).
NRs form one of the largest and most well understood superfamily of transcription factors. Generally, unliganded NRs repress transcription of targets whereas agonist binding promotes transcription; binding of NRs to either co-repressor or co-activator complexes respectively is essential to mediate these effects (reviewed in6). NR complex composition varies resulting in diverse patterns of transcriptional cycling, in terms of amplitude and magnitude. That is, the constituents of the NR complex change during the transcriptional cycle at different response elements. The implication of this is that the individual complexes both recognize and impart separate patterns of histone modifications and local chromatin structure to allow, for example, for chromatin looping7, 8.
The loss of transcriptional plasticity and the evolution of a restricted transcriptome are displayed by NRs in malignancy. A role for the androgen receptor (AR) in driving proliferation of epithelial progenitor cells within the prostate is well established, for example, through co-operation with WNT and mTOR pathways. However, the AR also controls cell cycle progression and differentiation through direct regulation of gene targets such as CDKN1A and NKX3.19. Genome wide studies have revealed that during cancer progression the AR transcriptome becomes altered and evolves towards the targeting of different promoters of genes that drive proliferation10, 11. For example, the TMPRSS2/ETS oncogenic fusion protein is a common critical event in prostate cancer, precisely because the TMPRSS2 promoter is sustained in an AR responsive state11 whereas the AR responsiveness of other targets such as NKX3.1 becomes silenced12. Equally in a range of solid tumors and myeloid leukemia, NRs that normally exert mitotic restraint, such as the VDR, RARs and PPARs, become skewed, with selective silencing of anti-proliferative target genes13-18. Combined, oncogenic transcriptional rigidity reflects the simultaneous distorted regulation of target loci such that proliferative and survival signals are enhanced and anti-mitotic inputs are either limited or lost. Co-repressor proteins contribute significantly to disruption of these processes.
NCOR-1 and NCOR2/SMRT are proto-typical co-repressors, cloned in parallel in 1995 by the groups of Rosenfeld19 and Evans20 respectively. These large regulatory proteins of approximately 270 kDa were isolated using NRs as bait to capture interacting proteins. Both proteins display a number of functional motifs including the NCOR2/SMRT and NCOR1 conserved (SNC), and SWI3, ADA2, NCOR1 and TFIIIB (SANT) domains. While the first SANT domain (SANT1) is dispensable, SANT2 is essential for interaction with HDAC3, HDAC1 and CoREST functions. NCOR1 also interacts allosterically with HDAC3 via the deacetylase interaction domain (DAD). There are three different Interaction Domains (IDs) in NCOR1 (N1-3) and NCOR2/SMRT (S1-3), which contain similar motifs to the co-activators' NR box, and are named “CoRNR box”.
Generally, in the absence of ligand, NR conformation facilitates interaction in large co-repressor complexes (~2.0 MDa)21. However more recently mechanisms have emerged whereby activated transcription factors can recruit NCOR1 and NCOR2/SMRT leading to active gene silencing. For example, ligand-dependent sumoylation of the PPAR-γ ligand-binding domain recruits the NCOR1 and HDAC3 complex to drive transrepression22. In turn, this prevents NCOR1 recruitment to the ubiquitylation/19S proteasome and promoter clearing.
The list of NCOR1 and NCOR2/SMRT targeted transcription factors is diverse and includes not only NRs, but also MAD/MXI, MYOD, ETO, CBF, TFIIB, AP-1 members and NF-κB factors. Reflecting this large number of transcription factor interactions, murine knockouts of Ncor1 and Ncor2/Smrt are embryonically lethal. More recently, stem cell components from Ncor1−/− and Ncor2/Smrt−/− mice and conditional approaches have revealed a network of interactions that includes both unique and shared targets23. Similarly, a knock-in approach circumvented the embryonic lethality and allowed the specific disruption of the interaction of Ncor2/Smrt with NRs. These models revealed dramatically enhanced differentiation rates, notably of adipocytes mediated by PPARγ24. NCOR2/SMRT also regulates FOXP1 that in turn governs a cohort of genes necessary for proper myocardial development. In wild type hearts, Ncor2/Smrt and Foxp1 are detected on the promoter of Cckn1a (encodes p21(waf1/cip1)) together with H3K9me2 enrichment, which represses p21(waf1/cip1) expression. Consequently, p21(waf1/cip1) levels are basally elevated in either Ncor2/Smrt or FoxP knockout animals, leading to a block of proliferation and thinned myocardium25.
Well established oncogenic roles for NCOR1 and NCOR2/SMRT have been elucidated in acute promyelocytic leukaemia that results from a fusion between the NR, RARα, and either the promyelocytic leukaemia (PML) or promyelocytic leukaemia zinc finger (PLZF) genes16. Both chimeric proteins sustain NCOR1 interactions and consequently RARα-mediated cell differentiation is blocked, in part, as a result of maintaining a condensed chromatin structure around the promoters of RARα target genes that govern normal hematopoietic differentiation. In the PML-RAR fusion, this can be overcome by pharmacological dosing with retinoic acid. The PLZF-RAR fusion is resistant to retinoic acid alone and treatment with a combination of retinoic acid and HDAC inhibitors has shown promising results. Similarly, in acute myeloid leukemia (AML), the AML1/ETO fusion protein promotes leukemogenesis by recruiting NCOR1 and again impeding transcriptional regulation26. The importance of NCOR1 binding in the treatment of these disease states exemplifies the relevance of the co-repressors in firstly driving critical oncogenic events, but secondly providing a rational targeted strategy towards HDACs.
Expression profiling in solid tumors has revealed altered NCOR1 and NCOR2/SMRT expression and localization, for example in breast, bladder, and prostate cancers13, 17, 27-30. However, to date, uncertainty remain over their precise role in solid tumors, especially in the case of breast and prostate cancers where the etiology of disease is intimately driven by the actions of steroid hormone NRs. Indeed the ability of the ligand free NR conformation to bind NCOR1 and NCOR2/SMRT is central to therapeutic exploitation with receptor antagonists such as Tamoxifen in the case of breast cancer. Therefore ambiguity exists over the extent and timing of NCOR1 and NCOR2/SMRT expression changes, as they relate to initiation and progression of disease. Secondly, it remains unclear how changes in NCOR1 and NCOR2/SMRT expression relate to different NRs and other transcription factors that exert either pro- or anti-mitotic and survival effects. Resolving these ambiguities has significant therapeutic implications in terms of targeting co-repressors as either epigenetic mono-therapies using HDAC inhibitors or in combinations with transcription factor targeting.
In prostate cancer cells elevated levels of NCOR2/SMRT have been detected and suppress VDR responsiveness13. Similarly, PPAR actions are disrupted and can be targeted selectively by using HDAC inhibitor co-treatments31, 32. More specifically elevated NCOR1, and to a lesser extent NCOR2/SMRT correlated with, and functionally drove, the selective insensitivity of PPARα/γ receptors towards dietary derived and therapeutic ligands32 most clearly in androgen-independent disease. Similar roles for NCOR1 and NCOR2/SMRT appear in the development of breast cancer. NCOR1 and NCOR2/SMRT levels appear unchanged in estrogen-dependent disease that responds to Tamoxifen, whereas they are down-regulated in murine models and patients samples of emergent Tamoxifen resistance27. ERα activation initiates a feed-forward effect on transcription by initiating NCOR1 proteosomal degradation. Possibly reflecting the loss of this negative control loop, elevated levels of NCOR1 occur in ERα negative disease and in turn attenuate anti-mitotic actions of VDR. Again, this molecular lesion can be targeted in ERα negative breast cancer cell lines with co-treatments of VDR ligand (e.g. 1α,25(OH)2D3) plus HDAC inhibitors resulting in selective re-expression of VDR target genes, notably VDUP1 and GADD45A17. Combined the studies in breast and prostate cancer suggest NR specificity of interactions with co-repressors. NCOR1 appears to be involved in the regulation of receptors such as the VDR and PPARs and NCOR2/SMRT with steroid hormone receptors and reflect the emergent specificities of NR interactions in the murine knockout models.
Given the list of targeted transcription factors it is unlikely that deregulation of NR actions is the only oncogenic impact of altered NCOR1 and NCOR2/SMRT expression. These questions are perhaps more clearly addressed in cancers where the biology is not so exquisitely associated with NRs, such as colon cancer. Considering post-translational modifications that regulate NCOR1 and NCOR2/SMRT activity reveals other de-regulated aspects of these co-repressors. Phosphorylated NCOR133 and NCOR2/SMRT translocate out of the nucleus. NCOR1 and NCOR2/SMRT impede β-catenin binding to LEF/TCF target genes such as CCND1, therefore mediating suppression of TCF4 transcriptome. Consequently, there appears to be increased co-repressor cytoplasmic export and location in colon cancer34-36. Similarly, roles have emerged for interleukin-1b to drive NCOR1 export from the nucleus37 allowing transactivation mediated in part by the TIP60 co-activator complex of a specific subset of nuclear factor-kappa-B (NF-κB)-regulated genes. Furthermore, enhanced NF-κB signaling also leads to NCOR1 and NCOR2/SMRT phosphorylation that enhances further cytoplasmic export. Also, repression domains (RD1-3) at the N-terminal region facilitate interaction with TBL1 and TBLR138 leading to ubiquitination and therefore governs co-repressor half-life of the complex. To-date it remains unclear if the capacity of TBL1/TBLR1 to initiate degradation of NCORs is distorted in malignancy.
Together these findings considering different transcription factors suggests that both enhanced and impeded NCOR1 and NCOR2/SMRT activity occur in cancer and will require comprehensive analyses to identify the critical transcription factor interactions. Similar contextual aspects to function have emerged for other co-repressors as outlined below. Generation of this knowledge concerning the timing and specificity of interactions with transcription factors and the epigenetic machinery may pave the way for more targetted therapeutic exploitation.
The C-terminal domain of viral E1A proteins were used as bait to identify interacting proteins including CTBP139 and CTBP240 that in turn modulate multiple aspects of transformation. For example CTBP1 and CTBP2 co-localize with polycomb group complexes thereby leading to gene silencing41. CTBP1 also displays dehydrogenaze activity and appears to serve as a cellular redox sensor with repression being acutely sensitive to NAD+/NADH levels42. CTBPs interact with transcription factors through a PXDLS motif to generates a large repressive complex, again, similar to the NCORs, in the order of 1.3 to 1.5 MDa, including components such as REST/CoREST, HDAC1 and HDAC2. More recently, to underline the co-operative role between co-repressor complexes, Rosenfeld and co-workers have established that CTBPs and NCOR2/SMRT complexes together mediate a combined transcriptional repression checkpoint that must be overcome to initiate transcription43.
Like the NCORs, the oncogenic actions of CTBPs also appear to be context-dependent. CTBP1 is implicated in the transrepression events mediated by ERα44, 45 and, again, its deregulation suggests a role that distorts and restricts the normal flexibility of receptor transcription. For example, CTBP1 was found to interact in-vitro and in-vivo with the zinc-finger protein ZNF366 and mediate selective estrogen-dependent and HDAC-mediated repression of ERα target genes in breast cancer cells46. Similarly, in hepatocellular carcinoma, CTBP1 binds to the INK4 family member, p19INK4D, suppressing its expression and leading to increased invasiveness47. Other workers have focused on the oncogenic role of WNT signaling in colon tumorigenesis, and demonstrated that upregulation of CTBP1 is a critical step downstream of loss of APC either prior to, or independently of, β-catenin nuclear localization further supporting a role for establishing transcriptional rigidity in the oncogenic process48. By contrast, CTBP1 is lost in melanoma cells allowing increased expression of a number of LEF/TCF target genes resulting in increased invasiveness49. Thus the specificity and timing of transcription factor interaction is critical to evaluate the CTBPs oncogenic properties.
REST/CoREST is present in a wide variety of cell types and complexes with RE1 silencing transcription factor/neural-restrictive silencing factor (REST/NRSF) to block expression of key target genes50 and, in turn, is part of a complex with HDAC1, -2, BHC80, BRAF35 and the histone demethylase KDM1/LSD1. The interaction with this KDM illustrates further aspects to co-repressor distortion.
KDM1/LSD1 recognizes methylated histone lysine residues, notably H3K4 and K9. However the demethylation of these residues leads to opposite phenotypic events; demethylation of H3K4 is associated with gene repression51, whereas demethylation of H3K9 leads to gene activation52. Therefore the specificity of complex sequestration for this important enzyme is pivotal. Specificity is in part contributed by the action of site-specific tri-demethylases whose actions govern the availability of the respective di-methylated substrate for KDM1/LSD1. Family members of the JARID proteins target H3K4me353 whereas JMJD2 members target H3K9me354, and these proteins themselves have emerging roles in cancer etiology. KDM1/LSD1 sequestration, for example into the REST/CoREST complex that promotes demethylation of H3K4me2 at active loci and subsequent recruitment of MeCP2 to facilitate CpG island methylation and the generation of repressive chromatin structure51, 55 (Figure 1).
Other CoREST complex functions include association with the chromatin-remodeling complex, SWI/SNF to lead to enhanced tumorigenicity56. CoREST can also form a larger complex with ZNF217, which is a Kuppel-like transcription factor, and CTBP1/2. Specifically, ZNF217 is a strong candidate oncogene found in breast cancer at the 20q13.2 amplicon and there appears to be a link between over-expression of ZNF217 and the loss of TGF-β responsiveness in breast cancer leading to the transcriptional suppression of targets such as p15ink4,57.
RUNX transcription factors were originally identified in embryonal carcinoma cells after they underwent RAR induced differentiation. The α-subunits (RUNX1/AML1/PEBP2αB, RUNX2/PEBP2αA or RUNX3/PEBP2αC) heterodimerize with the β subunit of PEBP2/CBF to increase DNA affinity. Although RUNX2 represents the best-understood family member, its actions are also highly context-dependent. RUNX2 haploinsufficiency is associated with cleidocranial dysplasia - a disorder characterized by a range of skeletal abnormalities. This phenotype is in part mediated by the loss of interaction of RUNX2 with the CDKN1A promoter activity, that is mediated by members of the transducin-like enhancer of split (TLE)/Groucho (Grg) family of proteins58, 59. Its carboxy terminal has also been shown to be a potent repressor, for example, by recruitment of HDAC6 from the cytoplasm to chromatin59.
The direct oncogenic potential of RUNX2 has been demonstrated by Vaillant et al creating a transgenic mouse with Runx2 under the control of the Cd2 promoter, that perturbed thymocyte development and lead to spontaneous lymphoma60. Increased expression of RUNX2 has also been observed in breast cancer cell lines and is thought to be involved in bone metastasis; a possible mechanism of action that reflects the significant bone biology of RUNX2. RUNX2 expression appears to be downstream of WNT signaling61 and forms a RUNX2 enriched network, for example in prostate cancer62, and therefore this co-repressor has been proposed as a critical novel drug target63.
Several key co-repressors have been identified in breast cancer. For example the MTA gene family has been examined in detail in breast cancer and linked to ERα signalling64 For more detail the reader is referred to an excellent review on MTA1 and MTA365. Another co-repressor identified in breast cancer is BRMS1, which was identified as part of strategy to reveal metastasis suppressors66, 67, by examining frequently deleted genomic regions in metastatic breast carcinoma. Critically, studies on BRMS1 illustrate the importance of loss of function of co-repressors in the oncogenic process. The protein is approximately 28.5 kDa with several phosphorylation sites, two functional NLS and interacts, for example, with Retinoblastoma Protein-1, the mSin3-HDAC complex and Hsp90. BRMS1 complexes modte ula NF-κB actions by blocking IκBα (inhibitor of NFκB68) phosphorylation and degradation and in turn suppressing NF-κB activity. BRMS1 participates in the direct repression of the RelA/p65 subunit via HDAC1-dependent deacetylation. Consequently, BRMS1 knockdown allows recruitment of acetylated-RelA/p65 to NF-κB-dependent anti-apoptotic genes69, 70. More recently it has emerged that other key targets regulated by BRMS1, that contribute to metastasis suppression, are microRNAs including miR-14671.
Another example of co-repressor loss of function leading to malignancy is illustrated by Scaffold/Matrix-Associated Region-1 (SMAR1)72. SMAR1 acts with Cux, either synergistically or independently, as transcriptional repressor for the Eβ enhancer. Studies have demonstrated that SMAR1 also forms heterocomplexes with HDAC1/SIN3 and pocket Rbs (p107, p130). Thus, depletion of SMAR1 generates an open chromatin structure as demonstrated by the acetylated histones at the CCND1 promoter73 indicating an alternative mechanism of cell cycle deregulation in cancer cells. Reflecting this, SMAR1 is down-regulated in malignant breast cancer tissues leading to a more active TGFβ signaling and promoting invasion with prognostic significance74. By contrast, over-expression of SMAR1 significantly delayed tumor growth in mice through direct interaction with p53 and downstream activation of p21(waf1/cip1),75.
ZBRK1/ZNF350 undergoes homo-oligomerization and a C-terminal domain represses transcription in a BRCA1-dependent, HDAC-dependent and promoter specific manner, for example of GADD45A76. Similar to other co-repressors, this complex interacts with downstream target genes, such as the HP1 family of heterochromatin-associated proteins thereby leading to stable silencing (Figure1). This repression is present in the absence of DNA damage whereas upon UV-methyl methanesulfonate treatment the ubiquitin-proteasome pathway mediates ZBRK1/ZNF350 BRCA1-independent degradation77. This damage-dependent de-repression of cell cycle gatekeepers exerts a tight control over cell proliferation. ZBRK1/ZNF350 also forms a repressor complex on the Angiotensin-1 (ANG-1) promoter and loss of expression is consistent with Brca1-deficient mice models where Ang1 is over-expressed and accelerates tumor growth rate78.
Altered expression of ZBRK1/ZNF350, both over-expression and under-expression, has been found on primary breast carcinoma, and not surprisingly mRNA levels are not prognostic. However, a significant correlation exists between two ZBRK1/ZNF350 polymorphisms and mRNA levels79 and expression in colon cancer is reduced80.
NRIP1/RIP140 was first identified as an ERα interacting protein that binds in a ligand-dependent manner to a wide range of NRs to limit activation. At higher ligand concentrations, NRIP/RIP140 binding is associated with ERα trans-repression functions. To achieve these functions it interacts with HDAC1 and HDAC3 as well as HDAC2, -5 and -6, possibly in a redundant manner
The primary role of co-repressors complexes is to re-model chromatin and attenuate the kinetics of transcriptional initiation. In malignancy elevated co-repressor expression leads to suppressed gene expression, most likely by suppressing the ability to initiate the positive aspects of mRNA cycling, as illustrated by NCORs. It has also emerged that loss of co-repressor expression, such as BRMS1, leads to the opposite, enhanced transcriptional activity. The specific details and consequences of co-repressor interactions are often more nuanced, depending on the interacting transcription factor and gene (or miRNA) being targeted. However the consequence of de-regulated co-repressor function is the same; normal plastic transcriptional cycling is lost and a state of rigidity emerges. Reflecting the biological importance of this control, the number of co-repressor proteins continues to grow, for example NRIP181, 82, COPS2/TRIP15/Alien83, TCERG1/CA15084, SLIRP85 and LCOR86 are well-characterized biologically but their roles in cancer remain to be clarified.
The transcriptional distortion and rigidity in cancer induced by changes of co-repressor expression, localization and interactions is mediated by the actions of a diverse number of histone modifying enzymes. It is likely that a greater understanding of how histone modifications combine1 in a so-called histone code87 will help to reveal further the specificities of target gene regulation. Of the varied histone modifications, deacetylation and demethylation are intimately linked and are significant mediators of transcriptional repression. In the absense of DNA methylation, these inter-relationships are highly dynamic, with target gene promoters often poised to be subsequently pushed towards a fully active, or a more stably repressed state. For example, non-expressed genes retain low level RNA POLII association, simultaneous HDAC and HAT activity, and modest transcriptional initiation88. It seems that the presence of H3K4 methylation perhaps holds these promoters in this poised state.
Elevated co-repressor expression can act to suppress this poised state. Sustained repressive histone modifications also have a significant role to restrict transcriptional plasticity further as they attract DNA methylating events associated with stable gene silencing. For example, NCOR1 and NCOR2/SMRT complexes deacetylate H3K9ac and thereby facilitate H3K9 methylation by several mechanisms. The loss of acetylation makes K9 accessible to methylation for example by SUV39H1. H3K9me2 status is subsequently targeted by HP1 and leads to further stable gene silencing through recruitment of DNA methyltransferase 1 (DNMT1) and DNA methylation of adjacent CpG islands. Similarly NCOR1-mediated gene silencing involves interaction with Kaiso, a methylation dependent transcriptional repressor that allows interaction with the methylated CpGs present in the promoter of target genes, silencing their transcription89. Recently, histone activating marks and CpG methylation have emerged to be even more dynamically intertwined than previously thought, with lower density CpG regions transiently regulated, for example during the transcriptional cycle of NRs90 adding further potential control to the plasticity of transcription. One possible consequence of increased CpG island hypermethylation is the generation of stably maintained transcriptional rigidity displayed in cancer cell systems. Thus, it is reasonable to suggest that elevated expression of co-repressors favors histone modifications that in turn are targets for the DNA methylation machinery and lead selectively to regions of the genome acquiring heterochromatin like characteristics. These repressive states are fixed in a more stable and heritable manner and thereby limit the choice of promoters (Figure 1). The loss of co-repressor expression and/or functions theoretically will induce the opposite chromatin states, but this remains to be established formally.
Importantly these epigenetic lesions are individually highly targetable with clinically available small molecular weight inhibitors targeted to specific histone deacetylation events and more recently this has been extended to include histone methylation events91, coupled with agents that target CpG methylation (reviewed in92). Thus comprehensive understanding of the key co-repressors in malignancy, delineating the key transcription factors interactions and the critical targets that are thereby dis-regulated may have considerable prognostic utility, specifically through the capacity to stratify patients for specific tailored epigenetic therapies.
MJC acknowledges the support of NucSys, a European Community FP6- Marie Curie Research Training Network, the Biotechnology and Biological Sciences Research Council, and in part, support from National Institute of Health grants R01 CA095367-06 and 2R01-CA-095045-06, and the NCI Cancer Center Support Grant to the Roswell Park Cancer Institute [CA016056].
The authors declare that there is no conflict of interest that would prejudice the impartiality.