Histone deacetylases (HDACs) regulate the expression and activity of numerous proteins involved in both cancer initiation and cancer progression. By removal of acetyl groups from histones, HDACs create a non-permissive chromatin conformation that prevents the transcription of genes that encode proteins involved in tumorigenesis. In addition to histones, HDACs bind to and deacetylate a variety of other protein targets including transcription factors and other abundant cellular proteins implicated in control of cell growth, differentiation and apoptosis. This review provides a comprehensive examination of the transcriptional and post-translational mechanisms by which HDACs alter the expression and function of cancer-associated proteins and examines the general impact of HDAC activity in cancer.
During the normal lifecycle, a cell encounters numerous opportunities where it must decide whether to proliferate, differentiate or die. A defect in any of these processes may result in cancer, or the uncontrolled growth of the cell. It has been appreciated for decades that histone deacetylases (HDACs) play an important role in cancer development since the first observation that sodium butyrate (NaB), later identified as an HDAC inhibitor, can cause the morphological reversion of the transformed cell phenotype (Ginsburg et al., 1973; Altenburg et al., 1976; Boffa et al., 1978). Currently, HDAC inhibitors are in clinical trials to treat a variety of malignancies. HDAC inhibitors can block cell proliferation, promote differentiation and induce apoptosis, any of which is desirable to thwart the growth of rogue cells (Figure 1). Despite decades of research, the precise mechanism(s) by which these inhibitors work remains far from clear. In many cases, they may influence multiple targets. To understand how HDAC inhibitors function, we must first understand the functions of the enzymes they inhibit. In this review, we will address the molecular mechanisms by which HDACs act and how these actions relate to cancer.
Chromatin modifications and cancer
Histones comprise the protein backbone of chromatin. Histone acetyltransferases (HATs) acetylate the ɛ-amino group of lysine (K) residues on histones, thereby neutralizing their positive charge, which diminishes their ability to bind to negatively charged DNA. In addition, acetylation of histone H4-K16 specifically disrupts the formation of higher-order chromatin structures (Shogren-Knaak et al., 2006). An open chromatin configuration provides accessibility for specific transcription factors and the general transcription machinery. HDACs remove the acetyl groups, allowing compacted chromatin to reform (Figure 2). Among their many properties, HDACs are required for cell-cycle progression. Indeed, HDAC inhibitors, including NaB, trichostatin A (TSA) and trapoxin, cause mammalian cells to undergo cell-cycle arrest. Using a trapoxin affinity column, HDAC1 was purified biochemically, subsequently cloned and shown to have HDAC activity (Taunton et al., 1996). In an unrelated series of experiments, HDAC2 was cloned on the basis of its interaction with the transcriptional activator/repressor YY1 and was shown to have transcriptional repressor activity (Yang et al., 1996). As such, the first mammalian HDACs were cloned independently, using distinct methods. The same roles that HDACs play in influencing cell-cycle progression and repressing transcription also have important implications in their connection to cancer.
Since the initial cloning of HDAC1 and HDAC2, at least 16 additional human HDACs have been identified and are grouped into four classes: class I comprises HDAC1–3 and 8, which show similarity to yeast RPD3; class II comprises HDAC4–7, 9 and 10, which show similarity to yeast HDA1; class III the NAD+-dependent HDACs, comprises SIRT1–7, which show similarity to yeast SIR2; class IV comprises HDAC11, which has some features of both class I and II HDACs (Gregoretti et al., 2004). Class III HDACs are not affected by the cytostatic HDAC inhibitors, but are inhibited by nicotinamide.
Histone acetylation and deacetylation does not occur in a vacuum. Specific patterns of histone acetylation and deacetylation are typically influenced by other histone modifications. Together, these post-translational modifications potentially generate a ‘histone code’ (Jenuwein and Allis, 2001). For example, histone H3-K9 acetylation and H3-K4 methylation are associated with active transcription. In contrast, loss of histone H3-K9 acetylation and gain of H3-K9 and H3-K27 methylation is indicative of heterochromatin (Maison et al., 2002). In many cases, modifications of a single residue are mutually exclusive. Furthermore, the presence of one modification may set the stage for additional modifications at nearby amino acids, thus expanding the coding information. Adding another layer of complexity, promoters of genes de-repressed by HDAC inhibitors often contain hypermethylated DNA, indicating crosstalk between histone acetylation/methylation and DNA methylation. Each of these events can have profound implications for gene expression in normal and cancerous cells.
Epigenetic changes, including global DNA hypomethylation and promoter CpG island hypermethylation, have long been recognized as characteristics that distinguish cancer cells from their normal counterparts. Recently, very specific changes in post-translational modification of histones that characterize cancer cells have been identified as well. For example, the coupled loss of acetylation of histone H4-K16 and trimethylation of histone H4-K20 can be considered a common feature of cancer cells as compared to their normal counterparts (Fraga et al., 2005). These changes occur early and accumulate as the cancer progresses. Hypoacetylation of histone H4-K16 is caused at least in part by diminished recruitment of the MYST (MOZ, Ybf2/Sas3, Sas2 and Tip60) family of HATs (Fraga et al., 2005); however, an increase in HDAC recruitment and/or activity cannot be ruled out. Indeed, the class III HDAC SIRT2 shows preference for deacetylation of histone H4-K16. Although this specific deacetylation occurs at the G2/M transition of the cell cycle, it appears to influence the timing of S-phase entry (Vaquero et al., 2006). In addition to defining cancer cells, specific histone modifications can be used to predict cancer recurrence. Histone H3-K18 acetylation coupled with histone H3-K4 dimethylation confers lowered risk for prostate cancer recurrence (Seligson et al., 2005). While these findings presently apply only to prostate cancer, it is likely that similar modifications will be found in other cancer types. In addition, the observation that low acetylation levels correlate with negative outcomes may provide a basis for the clinical use of HDAC inhibitors.
Histone deacetylation: the classic examples
Upon the identification and cloning of HDACs, it was quickly ascertained that HDACs were recruited to promoters to repress transcription, thereby counteracting the activating effects of HATs. In the simplest view, recruitment of HDACs causes decreased histone acetylation that compacts chromatin and precludes access by the transcription machinery, resulting in transcriptional repression (Figure 2). HDACs act on multiprotein complexes and several distinct complexes were rapidly identified that clarified the mechanism of active repression of transcription. For example, the transcription factor Max is the heterodimeric binding partner of both the transcriptional activator Myc and the transcriptional repressor Mad. Both Myc–Max and Mad–Max heterodimers bind the same E box-containing DNA sequence, but Mad–Max complexes associate with the mSin3 scaffold protein that recruits HDAC1 and HDAC2. Transcriptional repression of Mad–Max target genes requires HDAC activity (Hassig et al., 1997; Laherty et al., 1997). In addition, DNA-bound unliganded nuclear receptors, such as the retinoic acid receptor, bind the co-repressors SMRT and N-CoR. These co-repressors bind directly to HDAC3, thereby recruiting HDAC3 to promoters and preventing transcription (Wen et al., 2000; Guenther et al., 2001). Furthermore, repression of E2F-mediated transcription by Rb requires HDAC1 and recruitment of HDAC1 coincides with decreased histone acetylation at E2F-regulated promoters (Brehm et al., 1998; Luo et al., 1998; Magnaghi-Jaulin et al., 1998). In each of these cases of ‘classic repression’, treatment with an HDAC inhibitor de-represses the promoter in question. Importantly, repressed genes often are tumor suppressors, cell-cycle inhibitors and differentiation factors or apoptosis inducers. Loss of expression of any combination of these classes of proteins would be advantageous to a cancerous cell. This offers a clue as to why HDAC inhibitors were initially shown to reverse the transformed cell phenotype and why they may be useful for effective cancer therapy.
Histone deacetylation and cancer initiation
Histone acetylation and deacetylation affect the expression of numerous genes implicated in oncogenesis (Figure 3). Countless experiments have been performed, in which cancer cells are treated with HDAC inhibitors and gene expression and histone acetylation status are examined. These experiments have provided a basic understanding of the mechanisms by which HDACs modulate the expression of these important genes.
Histone deacetylation and cellular proliferation
The cyclin-dependent kinase inhibitor p21 is one of the best studied targets of HDAC inhibitor-mediated de-repression. Treatment of many cancer cell types with any of several HDAC inhibitors causes the transcriptional upregulation of this antiproliferative gene. This is largely independent of p53. Expression of p21 coincides with hyperacetylation of histones H3 and H4 in its promoter region (Sambucetti et al., 1999; Richon et al., 2000). Multiple HDACs repress p21 expression in different cell types. In normal development, HDAC1 is the most important because targeted disruption of HDAC1 causes early embryonic lethality owing to a lack of proliferation caused by increased expression of p21. This indicates that other HDACs cannot compensate for the loss of HDAC1 in early embryonic development, despite a concomitant increase in HDAC2 and HDAC3 expression (Lagger et al., 2002).
Like many genes required for proper development, dysregulated function of HDACs can lead to cancer. As such, overexpression of HDACs is observed in many cancer types, with corresponding decreases in p21 expression. For example, prostate cancer cells overexpress HDAC1 (Halkidou et al., 2004). Gastric carcinomas, colorectal carcinomas, cervical dysplasias and endometrial stromal sarcomas all overexpress HDAC2 as compared to their normal counterparts (Huang et al., 2005; Song et al., 2005; Hrzenjak et al., 2006). Increased HDAC2 expression correlates with reduced p21 expression and correspondingly, HDAC2 knockdown increases p21 expression (Huang et al., 2005; Hrzenjak et al., 2006). Loss of function of the tumor suppressor adenomatosis polyposis coli (APC) is one mechanism that causes increased HDAC2 levels in intestinal mucosa (Zhu et al., 2004). HDAC inhibitors, such as valproic acid (VPA), reduce adenoma formation in APC mutant mice that express increased levels of HDAC2 (Zhu et al., 2004). Furthermore, reduced tumor formation upon HDAC inhibitor treatment of APC mutant mice correlates with increased histones H3 and H4 acetylation in the promoter regions of p21 as well as the proapoptotic Bax gene (Myzak et al., 2006). HDAC2 knockdown also increases apoptosis. The critical role of HDAC2 in determining the sensitivity of colon cancer cells to HDAC inhibitors is supported by sporadic colorectal carcinomas that carry a frameshift mutation encoding truncated, non-functional HDAC2. These cells express wild-type APC, yet resist the apoptotic effects of HDAC inhibitor treatment (Ropero et al., 2006). In addition, APC expression appears to be required for HDAC inhibitor-induced downregulation of survivin, which sensitizes cells to apoptosis (Huang and Guo, 2006). HDAC3 is also overexpressed in colon cancer and inhibits p21 expression. Similarly, silencing of HDAC3 increases p21 promoter activity and expression (Wilson et al., 2006). HDAC6, a class II HDAC, also represses expression. Runx2 recruits HDAC6 to the p21 promoter in osteosarcoma cells (Westendorf et al., 2002). The upregulation of HDACs that repress an important growth suppressive gene is an important mechanism to promote cancer cell proliferation. As such, HDAC inhibition can be used to stop the growth of cancer cells.
The cell-cycle activator cyclin D1 is frequently overexpressed in cancer. Overexpression of this proproliferative gene usually is caused by gene amplification, but can also occur by loss of the tumor suppressor SMAR1, a matrix-associated protein. SMAR1 resides on chromosome 16q24, a region where loss of heterozygosity is known to deregulate cyclin D1 expression in breast, prostate and other cancers. Decreased levels of SMAR1 correlate with increased cyclin D1 expression. SMAR1 normally binds the cyclin D1 promoter and recruits the HDAC1/mSin3 repression complex, causing reduced cyclin D1 expression through histone deacetylation. Correspondingly, SMAR1-expressing cells show reduced acetylation of histones H3-K9 and H4-K8 in the cyclin D1 promoter and surrounding region. In contrast, knockdown of SMAR1 causes an increase in histone acetylation (Rampalli et al., 2005). In this case, a tumor suppressor recruits an HDAC to repress expression of a proproliferative gene and restrain cell growth.
Restraint of inappropriate cell growth is critical to prevent cancer. In normal cells, transforming growth factor-β (TGF-β) inhibits cell growth. In contrast, cancer cells are refractory to the inhibitory effect of TGF-β, which often is due to the loss of TGF-β receptors (TGF-βR). TGF-βRII binds TGF-β, and TGF-βRI transduces the signal by activation of Smad family members. While Smad family members recently were shown to be targets of acetylation themselves, the role of Smad acetylation in cancer remains to be elucidated (Simonsson et al., 2006; Inoue et al., 2007). Lung cancer cells that do not express TGF-βRII resist endonuclease digestion in the area surrounding the transcription start site, indicating a closed chromatin conformation. Acetylation of histones H3 and H4 is severely reduced in these cell types. Treatment with TSA or NaB induces expression of TGF-βRII and also increases sensitivity to nuclease digestion, indicative of an open chromatin conformation (Osada et al., 2001). Furthermore, loss of TGF-βRI expression is frequently observed in many cancer types. Treatment of breast cancer cells with HDAC inhibitors increases the levels of acetylated histones H3 and H4 at the TGF-βRI promoter and correspondingly increases the expression of TGF-βRI. Likewise, overexpression of the HAT p300 increases TGF-βRI promoter activity. HDAC inhibitor treatment restores TGF-β sensitivity by impairing HDAC1 activity associated with Sp1/Sp3 at the TGF-βRI promoter (Ammanamanchi and Brattain, 2004). By repressing the expression of a receptor for a growth-restraining signaling molecule, HDACs render a cancer cell unresponsive to this signal and allow for unfettered cell growth.
HDAC repression of epithelial differentiation
Loss of expression of proliferation-restraining genes is one common feature of cancer cells. Inhibition of differentiation also causes inappropriate proliferation, which can lead to cancer. For example, loss of expression of the GATA family of differentiation factors is observed in several cancer types. The mechanism behind the loss of these critical transcription factors appears to be cell type dependent. Loss of GATA4 and GATA5 expression in gastric and colo-rectal carcinoma is exclusively due to promoter hypermethylation and not histone deacetylation (Akiyama et al., 2003). GATA4 and GATA6 are highly expressed in normal ovarian epithelial cells, but not in ovarian cancer cells. Indeed, loss of GATA6 expression strongly correlates with ovarian epithelial morphological transformation and de-differentiation. Hypoacetylation of histones H3 and H4, and loss of histone H3-K4 methylation within the GATA4 and GATA6 promoter regions coincide with lack of GATA expression. Treatment with TSA increases histones H3 and H4 acetylation at these promoters and increases the promoters’ sensitivity to DNaseI, indicating an open chromatin conformation. Importantly, TSA treatment restores GATA factor expression and the expression of GATA targets, including the tumor suppressor Dab2 (Caslini et al., 2006).
The mucin Muc2 plays a role in gastrointestinal cell differentiation. Loss of expression of this gene is implicated in pancreatic and colorectal cancer and Muc2-null mice show increased adenoma formation (Velcich et al., 2002). Increased acetylation of histones H3-K9 and H3-K27 in the promoter region of Muc2 correlates with expression of this tumor-suppressor gene. Cells expressing Muc2 also display histone H3-K4 methylation and decreased CpG island methylation in the promoter region. TSA treatment of pancreatic cancer cells that do not express Muc2 induces histone acetylation and Muc2 mRNA and protein expression (Yamada et al., 2006). Regulation of Muc2 by HDACs may be cell type dependent, because NaB treatment of undifferentiated adenocarcinoma cells inhibits Muc2 expression (Augenlicht et al., 2003). This underscores the requirement for empirical determination of the efficacy of HDAC inhibitor treatment in any given tumor type.
HDAC repression of hematopoietic differentiation
Proper expression of differentiation factors is especially important in step-wise differentiation programs such as hematopoiesis. The progression from pluripotent stem cell to mature hematopoietic cell requires a complex interplay of a variety of specific molecules. Blockade at any step along the differentiation pathway may result in the proliferation of leukemic cells. Chromosomal translocations are a common hallmark of leukemias and lymphomas. Such translocations often result in the aberrant recruitment of HDACs to promoters, preventing appropriate gene expression. The critical regulator of definitive hematopoiesis, Runx1 (AML1) is frequently disrupted in leukemia. Runx1-ETO, the product of the t(8;21) translocation is frequently observed in acute myelocytic leukemia, binds HDAC1–3 as well as the co-repressors mSin3a, SMRT and N-CoR (Amann et al., 2001). In addition, Runx1 binds HDAC5, HDAC6 and HDAC9 with varying affinities (Durst et al., 2003). ETO directly binds HDACs but has no specific DNA binding ability by itself. Since the fusion protein retains the DNA binding domain of Runx1, Runx1-ETO acts as a dominant negative inhibitor of wild-type Runx1 activity. Among the genes that are repressed by this chimeric protein is the tumor suppressor p14ARF, which may extend the lifespan of myeloid progenitor cells due to loss of senescence (Linggi et al., 2002). Another target of Runx1-ETO repression is c-fms, which encodes the colony-stimulating factor 1 receptor. Cells expressing Runx1-ETO display decreased amounts of acetylated histones H3-K9 and H3-K14 at the c-fms intronic regulatory element, and recruit HDAC1 to this region causing decreased expression of this important macrophage differentiation factor (Follows et al., 2003). In addition, Runx1 recruits the histone methyltransferase SUV39H1, which methylates histone H3-K9, leading to another mark of inactive chromatin (Reed-Inderbitzin et al., 2006). ETO also interacts with promyelocytic leukemia zinc finger (PLZF), a member of the BTB/POZ family of repressor proteins that recruit HDACs and co-repressors. ETO synergistically acts with PLZF to enhance transcriptional repression of genes required for differentiation in an HDAC-dependent manner (Melnick et al., 2000). Similarly, ETO binds another BTB/POZ member, BCL6. The BCL6 repressor is frequently expressed inappropriately in B-cell lymphomas and recruits both class I and II HDACs (Lemercier et al., 2002). ETO interacts with DNA-bound BCL6 and enhances repression in an HDAC-dependent manner (Chevallier et al., 2004). In each of these cases, the ability of chimeric proteins to inappropriately recruit HDACs to regulatory regions of genes involved in differentiation effectively prevents differentiation and allows for continued proliferation of undifferentiated progenitor cells, resulting in leukemia or lymphoma.
Histone deacetylation and cancer progression: angiogenesis and metastasis
In addition to regulation of genes involved in the genesis of cancer, histone acetylation and deacetylation modulate genes involved in cancer progression (Figure 3). This includes the regulation of angiogenesis that permits increased tumor growth as well as the regulation of adhesion, cell migration and invasion required for metastasis. Hypoxia, such as encountered in the center of a solid tumor, enhances angiogenesis and also induces HDAC expression and activity. Overexpression of HDAC1 represses the tumor suppressors p53 and von Hippel–Lindau (VHL) but induces the hypoxia-responsive genes hypoxia inducible factor-α (HIF-1α) and vascular endothelial growth factor (VEGF) and increases angiogenesis. Conversely, HDAC inhibitors de-repress the tumor suppressors p53 and VHL and repress HIF-1α and VEGF, and correspondingly, VEGF signaling. In vivo, hypoxic regions of tumors express increased levels of HDAC1 but treatment with TSA inhibits hypoxia-induced angiogenesis (Kim et al., 2001; Deroanne et al., 2002). Multiple HDAC inhibitors prevent new vessel formation in other models of angiogenesis as well, and this is coincident with global hyperacetylation of histone H4 (Kim et al., 2004; Michaelis et al., 2004).
Metastasis depends on loss of the ability of a cell to interact properly with its environment and its neighboring cells. The ability of class I HDACs to regulate extracellular matrix-related genes is highly conserved from Caenorhabditis elegans to humans. HDAC1 represses cystatin, a peptidase inhibitor that suppresses tumor invasion. Knockdown of HDAC1 or overexpression of cystatin reduces cellular invasion (Whetstine et al., 2005). Similarly, treatment with low doses of HDAC inhibitors reduces v-Fos-transformed fibroblast invasion. HDAC inhibition de-represses several genes in this model system, including RYBP, STAT6 and a protocadherin. Overexpression of any of these genes inhibits invasion, but it is unclear whether the role of HDAC activity is direct or indirect (McGarry et al., 2004).
Class I HDACs also directly regulate the expression of E-cadherin, a gene important for cell–cell adhesion. Loss of E-cadherin expression causes epithelial invasion, a necessary first step in metastasis. The repressor Snail recruits HDAC1 and HDAC2 and the co-repressor mSin3A to the E-cadherin promoter, repressing E-cadherin expression. Snail overexpression correlates with an increase in hypoacetylated histones H3 and H4 at the E-cadherin promoter, as well as increased methylated histone H3-K9 and decreased methylated histone H3-K4. Each of these modifications is consistent with repressed gene expression. TSA treatment abolishes Snail-mediated repression (Peinado et al., 2004). In addition, combined treatment of prostate cancer cells with a peroxisome proliferator-activated receptor-γ (PPARγ) agonist and the HDAC inhibitor VPA reduces the invasiveness of these cells coincident with increased expression of E-cadherin mRNA and protein. HDAC3 and PPARγ together bind to the E-cadherin promoter, and repression in this case is mediated by HDAC3. In the presence of the combination of a PPARγ agonist and an HDAC inhibitor, HDAC3 and PPARγ are no longer bound to the promoter, histone H4 is hyperacetylated at the promoter and E-cadherin is expressed (Annicotte et al., 2006). Since E-cadherin expression often is lost in prostate cancer, combination therapy may reduce metastatic potential of prostate cancer.
HDAC inhibitors downregulate expression of the chemokine receptor CXCR4 in acute lymphoblastic leukemia (ALL) cells. This reduces the migration that targets the ALL cells to the spleen, liver, lymph nodes and brain, all of which express high levels of the chemoattractant stromal cell derived-factor (Crazzolara et al., 2002). HDAC inhibitors also increase expression of the intercellular adhesion molecule ICAM1 on tumor-derived endothelial cells. This enhances the ability of lymphocytes to adhere to endothelial cells, allowing for better tumor infiltration by the lymphocytes. The ICAM1 promoter from tumor-derived endothelial cells contains hypoacetylated histone H3 and hypomethylated histone H3-K4. Treatment with HDAC inhibitors and methyltransferase inhibitors reverses these modifications (Hellebrekers et al., 2006).
Deacetylation of non-histone targets implicated in cancer
Over the past decade, it has become increasingly apparent that histones are not the only targets of acetylation and deacetylation. Indeed, a plethora of non-histone proteins are also regulated by the reversible actions of HATs and HDACs (Figure 4). Acetylation and deacetylation of non-histone proteins have pleiotrophic effects on protein function, including modulation of protein–protein interactions, protein stability and subcellular localization (for review see Glozak et al., 2005). As the list of non-histone proteins grows almost daily, it is becoming obvious that acetylation and deacetylation may rival phosphorylation and dephosphorylation as post-translational modifications that influence the activity of many proteins (Kouzarides, 2000). Importantly, many of the proteins modulated by acetylation play key roles in oncogenesis and cancer progression.
Deacetylation of transcription factors
Transcription factors comprise one of the largest families of proteins modulated by acetylation status. Acetylation and deacetylation affect many properties of transcription factors, including, but not limited to, their ability to bind DNA and activate transcription. Indeed, the tumor suppressor p53 is the founding member of the collection of non-histone targets of acetylation and deacetylation. Depending on the trigger, activation of p53 can prompt cell-cycle arrest or induce apoptosis. Acetylation of p53 by p300/CBP at K305, K370, K372, K373, K381 and K382 and by PCAF at K320 increases its DNA-binding ability and consequently increases its ability to activate transcription of its target genes (Gu and Roeder, 1997; Sakaguchi et al., 1998; Liu et al., 1999; Wang et al., 2003; Luo et al., 2004). Recently, the role of site-specific p53 acetylation in transcriptional regulation has become the target of intense scrutiny. Activation of p53 by different types of damage causes different acetylation events. For example, DNA damage caused by alkylating agents induces acetylation of K320, which activates genes with high-affinity p53 binding sites, including p21, and promotes cell-cycle arrest. In contrast, DNA damage caused by topoisomerase inhibition induces acetylation of K373, which activates genes with low-affinity p53 sites, such as Bax, and promotes apoptosis. These differential acetylation patterns affect subsequent p53 phosphorylation, which enhances the nuclear localization of p53. In addition, acetylation of K373 stabilizes the interaction of p53 with HDAC1 and SIRT1 (Knights et al., 2006). SIRT1 binds to and deacetylates p53 at K382, reducing its activator function (Luo et al., 2001; Vaziri et al., 2001). Abrogation of acetylation of the K320 homolog in mice (K317) enhances p53-mediated apoptosis after DNA damage by inducing expression of proapoptotic genes (Chao et al., 2006). Tip60 and hMOF, members of the MYST family of HATs, acetylate p53 following DNA damage (Sykes et al., 2006; Tang et al., 2006). Acetylation by MYST family members occurs specifically on K120 and is required for inducing apoptosis, but is dispensable for initiating cell-cycle arrest. Intriguingly, K120R mutations have been identified in human cancer. Since arginine cannot be acetylated and because acetylation is required for p53 to activate the proapoptotic genes Bax and PUMA, cells carrying this mutation resist apoptosis, a distinct advantage for a cancer cell.
Treatment of lung cancer cells with the HDAC inhibitor depsipeptide causes the specific acetylation of p53 at K373 and K382, which recruits p300 and increases expression of p21 (Zhao et al., 2006). On the other hand, treatment of prostate cancer cells with the HDAC inhibitors TSA or CG-1521 differentially stabilizes the acetylation of K382 or K373, respectively. Each of these distinct acetylation events recruit mutually exclusive co-activator complexes and only acetylation of K373 in this case is sufficient to assemble the basal transcriptional machinery on the p21 promoter (Roy and Tenniswood, 2007). Thus, treatment of different cancer cells with different HDAC inhibitors each can have unique outcomes.
Acetylation and deacetylation may also modulate p53's stability. Since many of the lysines that are acetylated are also sites of ubiquitination, acetylation can protect p53 from ubiquitination and subsequent degradation by the proteasome. The E3 ubiquitin ligase Mdm2 recruits HDAC1 to p53, and deacetylation promotes ubiquitination and degradation (Ito et al., 2002). The importance of this in vivo, however, is currently a matter of debate due to recent studies indicating that acetylation may not be required for p53 stability. Replacement of the C-terminal lysines with arginines in knock-in mice showed that p53 half-life is normal, although p53 activation following DNA damage differs (Feng et al., 2005; Krummel et al., 2005). Such findings underscore the importance of verifying in vitro results with in vivo experiments as results may vary based on the experimental system utilized.
In addition to p53, many other transcription factors are regulated by acetylation and deacetylation. Interestingly, many transcription factors that are acetylated are regulated transcriptionally by histone acetylation, indicating multiple levels where deacetylation can act. The functions of the Runx family of transcription factors are regulated by acetylation and deacetylation. Runx1, a critical regulator of definitive hematopoiesis may act as a transcriptional activator or repressor. In addition to its ability to inappropriately recruit HDACs to chromatin when expressed as the Runx1-ETO fusion as discussed above, Runx1 must be acetylated to function as a transcriptional activator. p300 acetylates Runx1 at K24 and K43, and mutation of these lysines impairs the transforming ability of Runx1 (Yamaguchi et al., 2004). Another Runx family member, Runx3, is required for T-cell development and acts as a gastric tumor suppressor. Upon TGF-β stimulation, p300 binds to and acetylates Runx3 at K148, K186 and K182. HDAC4 and HDAC5, and to a lesser extent, HDAC1 and HDAC2 reduce this acetylation. Acetylation enhances Runx3 stability by preventing ubiquitination by Smurf ubiquitin ligases (Jin et al., 2004). Paradoxically, Runx3 can act as an oncogene, based on its overexpression in basal cell carcinomas. The effect of acetylation in this setting, however, remains unknown (Salto-Tellez et al., 2006).
BTB/POZ repressor proteins that recruit HDACs and co-repressors to chromatin, including BCL6 and PLZF, are targets of protein acetylation. Acetylation by p300 at K376, K377 and K379 inhibits the ability of BCL6 to bind to HDAC2, thereby inhibiting its ability to repress transcription. Both TSA and nicotinamide enhance BCL6 acetylation levels, indicating both class I/II and III HDACs can deacetylate BCL6. Such treatment leads to cell-cycle arrest and apoptosis of B-cell lymphoma cells. In addition, mutation of BCL6 that mimics acetylation reduces the transforming ability of BCL6 (Bereshchenko et al., 2002). PLZF is acetylated by p300 at K562, K565, K647, K650 and K653 in zinc fingers six and nine. Acetylation has no effect on the ability of PLZF to recruit HDACs, but it does increase the ability of PLZF to bind DNA and therefore enhances transcriptional repression of growth promoting genes. Mutation of acetylation sites severely impairs the ability of PLZF to suppress cell growth (Guidez et al., 2005).
B-cell follicular lymphomas, characterized by the t(14;18) translocation, overexpress the antiapoptotic gene Bcl-2 as a consequence of the translocation. Treatment of lymphoma cells with HDAC inhibitors causes a significant decrease in Bcl-2 expression and induces apoptosis. The transcription factors Sp1 and C/EBPα, both of which can be acetylated, regulate Bcl-2 expression. TSA treatment increases the acetylation levels of Sp1 and C/EBPα; however, it decreases their binding to their cognate promoters, inhibiting transcription. TSA treatment also decreases the interaction of Sp1 and C/EBPα with HDAC2 and reduces the amount of HDAC2 bound to the Bcl-2 promoters. Overexpression of HDAC2, but not HDAC1 or HDAC3, increases Bcl-2 expression. Correspondingly, overexpression of HDAC2 decreases the amount of acetylated Sp1 and enhances DNA binding (Duan et al., 2005). In this case, HDAC2 is required to promote hypoacetylation of transcription factors, which allows for promoter binding and activation of an antiapoptotic gene. As such, HDAC inhibitors promote lymphoma cell death.
GATA factors are another family of transcription factors modulated by protein acetylation and deacetylation. Each of the hematopoietic GATA factors (GATA1–3) is acetylated. Acetylation of GATA1 by CBP/p300 within the zinc fingers is required for chromatin binding in vivo and is critical for its ability to activate transcription of genes required for terminal differentiation (Boyes et al., 1998; Hung et al., 1999; Lamonica et al., 2006). Deacetylation by HDAC5 represses GATA1-mediated transcription, but the association of GATA1 and HDAC5 decreases during differentiation (Watamoto et al., 2003). DNA-bound, acetylated GATA1 is targeted by MAPK-dependent phosphorylation for ubiquitination and subsequent degradation. In this way, a cell can attenuate its response to growth factors by removing a transcription factor once its job is complete (Hernandez-Hernandez et al., 2006). Acetylation of GATA2 is required for its DNA binding and transcription activity (Hayakawa et al., 2004). In addition, mutation of GATA3 that prevents acetylation prolongs the survival of T cells and impairs T-cell homing (Yamagata et al., 2000). Given the importance of GATA factors in hematopoietic differentiation, the dysregulation of their function by deacetylation may directly contribute to hematopoietic malignancy by maintenance of highly proliferative, undifferentiated precursor cells.
Nuclear factor-κB (NF-κB) plays an important role in the development of many human cancers (Basseres and Baldwin, 2006). The activity and duration of NF-κB activity is tightly regulated by acetylation and deacetylation at multiple levels. The ramifications of these modifications depend on the cell type and which lysine residue is modified. The transcription factor NF-κB comprises a heterodimer of RelA/p65 and p50 or p52 subunits. In resting cells, NF-κB resides in the cytoplasm, associated with its repressor, IκB. Upon stimulation, IκB kinase (IKK) phosphorylates IκB, causing its ubiquitination and degradation by the proteasome. Free NF-κB translocates to the nucleus, where the p65 subunit is acetylated by p300/CBP or PCAF. Acetylation of p65 at K218, K221 and K310 by p300 enhances the DNA binding and transcriptional activation ability of NF-κB and prevents its association with IκB, which prevents nuclear export (Chen et al., 2001, 2002). In this case, deacetylation by HDAC3 promotes NF-κB and IκB binding, facilitating nuclear export (Chen et al., 2001). SIRT1 also deacetylates p65 at K310, leading to decreased expression of NF-κB target genes (Yeung et al., 2004). The targets activated by NF-κB include antiapoptotic genes such as inhibitors of apoptosis (Yeung et al., 2004; Hoberg et al., 2006) and Bcl-xL (Chen et al., 2000). Therefore, repression by HDACs would be advantageous to downregulate these genes and promote apoptosis of cancerous cells. Indeed, treatment of osteosarcoma cells with topoisomerase inhibitors such as doxorubicin induces p65 to associate with HDAC1–3 leading to repression of antiapoptotic genes, including Bcl-xL, and correspondingly, decreased histone H3 acetylation at the Bcl-xL promoter (Campbell et al., 2004). In addition, doxorubicin treatment of breast cancer cells inhibits the p65 acetylation, reducing its affinity for binding DNA (Ho et al., 2005).
HDACs are responsible for the basal repression mediated by NF-κB p50 or p52 homodimers, which recruit SMRT/N-CoR–HDAC3 complexes. De-repression, caused by the exchange of co-repressor complexes for co-activator complexes, requires phosphorylation of SMRT. This releases HDAC3, allowing recruitment of p300 to acetylate p65 at K310, thus activating transcription (Hoberg et al., 2006). In contrast, acetylation of p65 at K122 and K123 by p300 or CBP reduces the DNA-binding affinity of NF-κB allowing binding to IκB and removal from the nucleus (Kiernan et al., 2003).
Upstream of NF-κB activation is the signal transducers and activators of transcription (STAT) family of signaling molecules, which play critical roles in oncogenesis. Acetylation regulates both STAT1 and STAT3. Dimerization of STAT3 depends on acetylation of K685 by CBP/p300, which enhances both DNA binding and transactivation. Deacetylation, primarily by HDAC3, inhibits transcription of STAT3 target genes including growth-promoting genes such as cyclin D1 and antiapoptotic genes such as Bcl-xL (Wang et al., 2005; Yuan et al., 2005). In addition, activation of STAT3 by acetylation enhances the proteolytic cleavage of NF-κB p100 to p52. In prostate cancer cells, increased amounts of p52 allows for cell survival following chemotherapy (Nadiminty et al., 2006). In the above scenarios, upregulation of HDAC activity would be advantageous to treat cancer. On the other hand, in other tumor types, acetylated STAT1 preferentially binds p65. This enhances the nuclear export of p65 and prevents expression of antiapoptotic NF-κB target genes. CBP acetylates STAT1 at K410 and K413, making melanoma cells susceptible to apoptosis (Kramer et al., 2006). In a situation such as this, inhibition of HDAC activity can be advantageous in cancer treatment. This exemplifies the need for empirical testing to determine the role of deacetylation in any given process.
Deacetylation of other cellular proteins with roles in cancer
In addition to transcription factors, acetylation and deacetylation regulate other classes of cellular proteins that have roles in cancer development and progression. Ku70 is a multifunctional protein regulated by acetylation. Ku70 binds the proapoptotic protein Bax, sequestering Bax in the cytoplasm. CBP and PCAF acetylate Ku70 at K539 and K542. This releases Bax, allowing it to translocate to the mitochondria and initiate apoptosis. Both class I/II and III HDACs deacetylate Ku70, which serves to keep Ku70–Bax complexes in the cytosol (Cohen et al., 2004b). As such, treatment with HDAC inhibitors promotes Bax-dependent apoptosis in several cell types, including neuroblastoma (Cohen et al., 2004a; Subramanian et al., 2005). Ku70 and Ku80 play a critical role in DNA repair as the DNA-binding subunits of DNA-dependent protein kinase. Although the mechanism is unclear, HDAC inhibitors decrease the overall levels of Ku70 and Ku80 in melanoma cells, which diminishes DNA repair, thus enhancing the radiation sensitivity of the melanoma cells (Munshi et al., 2005).
Acetylation modulates the activity of the molecular chaperone HSP90. Multiple client proteins must interact with the HSP90 complex to achieve their mature conformation and localization. Acetylation of HSP90 at K294 inhibits its ability to form these complexes with clients and co-chaperones (Scroggins et al., 2007). This prevents activation or enhances degradation by ubiquitination of client proteins. HDAC6 binds to and deacetylates HSP90, allowing for maturation or stabilization of its client(s). As such, treatment with HDAC inhibitors can promote the degradation of oncogenic client proteins such as BCR-ABL or Runx1-ETO (Bali et al., 2005; Kovacs et al., 2005; Yang et al., 2007). In addition, coupling HDAC inhibitor treatment with HSP90 antagonists may prove even more beneficial for cancer treatment.
Cellular migration is an important determinant of cancer metastasis. HDAC6 plays a critical role in cell motility as a tubulin deacetylase. In contrast to stable microtubules that are highly acetylated, dynamic microtubules are largely hypoacetylated. HDAC6 directly deacetylates α-tubulin at K40, allowing tubulin depolymerization that is necessary for cell migration. Overexpression of HDAC6 causes α-tubulin hypoacetylation and depolymerization, promoting cell movement (Hubbert et al., 2002; Matsuyama et al., 2002; Zhang et al., 2003). The class III HDAC SIRT2 also deacetylates α-tubulin at K40 (North et al., 2003). HDAC6 is unique, in that it has two HDAC domains and both of which are required for deacetylation because mutation in the catalytic core of either HDAC domain destroys HDAC activity (Zhang et al., 2006). Estrogen induces HDAC6 in breast cancer cells leading to increased migration. Treatment with antiestrogens inhibits migration. In addition, treatment with tubacin, an HDAC6 inhibitor, inhibits the estrogen-induced migration (Saji et al., 2005). HDAC6 also enhances lymphocyte migration. Interestingly, while HDAC6 protein is required, HDAC activity is not. Instead, HDAC6 may act as a scaffold to assemble the complexes required for rapid lymphocyte movement (Cabrero et al., 2006). Thus, HDAC inhibition is an important means of preventing cell migration and consequently, metastasis.
Conclusions and future directions
Acetylation and deacetylation are important mechanisms to regulate the activity of histones and a plethora of other proteins. Early studies envisaged that HDACs had roles in the development and progression of human cancer, and recent studies overwhelmingly support these predictions. The potential use of HDAC inhibitors in the clinic to treat cancer and other disorders is exciting, but should be approached with caution. Considering the pleiotropic cellular effects of acetylation and deacetylation and because new targets of acetylation and deacetylation are being identified at a remarkable rate, it becomes even more critical to understand the ramifications of these modifications on protein function to foresee potential off-target effects.
We apologize to all investigators whose works were not cited in this article due to space limitations. Work in our laboratory is supported by grants from the National Institutes of Health and an endowment from the Kaul Foundation.
About this article
Epigenetic suppression of E-cadherin expression by Snail2 during the metastasis of colorectal cancer
Clinical Epigenetics (2018)