Abstract
Introduction
H
The transfer of the acetyl group on specific lysines is catalyzed by a group of enzymes known as histone acetyltransferases (HATs) (23). The enzymes responsible for deacetylation are histone deacetylases (HDACs). The balance of protein acetylation and deacetylation plays a critical role in the regulation of gene expression as well as in several other cellular processes. Indeed, dysfunctions of the HATs/HDACs balance are associated with various diseases such as cancer (144), diabetes, asthma, cardiac hypertrophy, retroviral pathogenesis, and neurodegenerative disorders (145).
The goal of this article is to review current understanding and to highlight the most recent advances with regard to HDACs and their connection with cell metabolism, especially in cancer.
Among HDACs, the sirtuins family has long been recognized as a target and regulator of cell metabolism; in fact, several studies have extensively described the role of these proteins in controlling cellular metabolism (148); therefore, they will not be covered in this review. Excluding sirtuins, less information has been generated regarding non-sirtuins or classical HDACs and cell metabolism. This is extremely important, as in recent years, several HDACs inhibitors are becoming a promising class of anti-cancer drug targets, already approved by FDA or in evaluation in clinical trials (169). In this regard, the exploration of the role of these proteins in controlling cell metabolism as well as whether metabolism controls HDACs assume an important meaning, especially in cancer cells, as the recent revival of interest in cell metabolism alterations as a key hallmark of tumorigenesis (84). This finding, indeed, has propelled the idea that metabolism could control several aspects of tumor biology and has led several researchers to the identification and the use in clinical trials of anti-cancer drugs targeting metabolism (110). Therefore, the knowledge on the connections between deacetylation and cell metabolism is a challenge to improve our understanding of cancer mechanisms and to develop novel therapeutic protocols against this disease.
Classification, Expression, Regulation, and Function of HDACs
Mammalian genomes encode 11 classical HDACs (from here indicated only as HDACs) that are grouped into three classes according to their phylogenetic conservation, namely class I, II, and IV, characterized by a highly conserved deacetylase domain requiring a Zn2+ ion as cofactor (57) (Fig. 1).
Class I
This class is composed by HDAC1, 2, 3, and 8. Their expression is ubiquitous, and they show a different subcellular localization. HDAC1 and HDAC2, which have partially redundant roles, are localized exclusively in the nucleus (86, 121) (Fig. 1). HDAC3 and HDAC8 localize both in the cytoplasm and in the nucleus (158a, 171). HDACs do not bind DNA directly, but interact with DNA through multi-proteins complexes, including co-repressors and co-activators. HDAC1 and 2 are often associated together in multi-protein co-repressor complexes formed by suppressor of defective silencing 3 protein, nucleosome-remodeling deacetylase, and Co-repressor for element-1-silencing transcription factor (190). HDAC3 is able to form oligomers both in vitro and in vivo with other HDACs, itself, and nuclear receptor corepressor (N-CoR) protein (189). Less information is available for HDAC8, as recent reports demonstrated that recombinant purified HDAC8 catalyzes deacetylation and displays substrate selectivity also in the absence of additional protein cofactors (35, 183). Class I HDACs null mice are lethal. HDAC1 knockout (KO) die before embryonic day 10.5 for severe growth retardation (99, 121), and HDAC2 KO succumb during the perinatal period due to cardiac defects (121). HDAC3 KO die before E9.5 for defects in gastrulation, while the conditional deletion in the liver and heart disrupts lipid/cholesterol homeostasis and glycogen storage in the liver (91, 122). HDAC8 KO die within 4–6 h after birth due to deficiency of cranial neural crest cells (62) (Fig. 1).
Class I members possess a simple structure consisting of the conserved deacetylase domain and short N- and C-terminus. As shown in Figure 2, HDAC1 and 2 possess the dimerization domain, which is necessary for their association and a nuclear localization signal (NLS); HDAC3 contains both nuclear export signal (NES) and NLS; and HDAC8 contains only an NLS (57). Class I HDACs are regulated by post-translational modifications (Fig. 2). HADC1 activity is increased by Casein kinase 2 (CK2)-mediated phosphorylation (139) and SUMOylation (18), while it is decreased by ubiquitination (196), cysteine carbonylation (150), and acetylation (187). HDAC2 activity is increased by CK2-phosphorylation (139) and decreased by s-nitrosylation (153), tyrosine-nitrosylation (129), and ubiquitination that specifically leads to its degradation (196). HDAC3 activity is stimulated by CK2-phosphorylation and antagonized by protein phosphatase 4-dephosphorylation (197). HDAC8 activity is decreased on protein kinase A (PKA)-phosphorylation (103).
Class II
Class II members are further subdivided in class IIA and IIB (Fig. 1). HDACs 4, 5, 7, and 9, comprising class IIA, are able to shuttle between the cytoplasm and the nucleus; while HDACs 6 and 10 make up class IIB, mainly cytoplasmic.
Class IIA
HDACs 4, 5, and 7 interact with silencing mediator of retinoid and thyroid receptors/N-CoR, Bcl-6-interacting co-repressor, C-terminal-binding protein 1, and HDAC3 in the nucleus (43). Class II HDACs expression is tissue specific; they are implicated in the development and differentiation of cardiac and skeletal striated muscle. HDAC4 KO die within the second week of life owing to ectopic ossification, while HDAC5 deletion causes cardiac defects, but mice are viable and fertile (17). HDAC7 KO results in embryonic lethality, owing to a loss of endothelial integrity (63). HDAC9 KO are viable but show severe cardiac defects (17) (Fig. 1).
Class II members have the catalytic domain located at the C-terminus, NLS, NES (125) and a protein–protein interaction domain located at N-terminus, involved in the association with different molecular partners (Fig. 2). Class IIA localization is regulated by specific post-translational modifications. HDAC4 phosphorylation by calcium/calmodulin-dependent kinase (CaMK) and salt-inducible kinase 1 (SIK1) creates docking sites for nuclear export (59), while PKA triggers its proteolytic cleavage antagonizing the action of CaMK and generating a product that relocalizes into the nucleus (7). HDAC4 activity is reduced by SUMOylation (89). HDAC5 results in a strong nuclear accumulation after PKA-phosphorylation (16). HDAC7 may be phosphorylated by CaMK and protein kinase C (PKC) that impair nuclear import (68); the same effect is also yielded through HDAC7 proteolytic cleavage (149). Three different isoforms of HDAC9 generated by alternative splicing, with conserved structure but distinct cellular localization and cell specificity, have been found (136). HDAC9 SUMOylation reduces its activity (89), while phosphorylation by PKC kinase impairs the nuclear import (68).
Class IIB
HDAC6 regulates cell motility, adhesion (198), and chaperone function (95). HDAC6-deficient mice are viable and fertile with hyperacetylated tubulin in most tissues (199) (Fig. 1). HDAC10, present in two variants, is the most recently discovered member of the class II. It interacts with different other HDACs, suggesting that it might exert cellular functions independently from its deacetylase activity by recruiting other deacetylases (42) (Fig. 1). HDAC6 contains two catalytic domains arranged in tandem and NLS and NES both located at N-terminus (Fig. 2). HDAC6 activity is increased on phosphorylation by CK2 (177) and oncogenic Aurora A kinase (142). HDAC5 and 6 have been also identified to be poly-ubiquitinated, but this modification does not target them for proteasomal degradation (130). HDAC10 has a catalytic domain on its N-terminus, and an NES and a putative second catalytic domain (Leucine-rich domain) on the C-terminus (Fig. 2).
Class IV
Class IV comprises only HDAC11, which is localized in the nucleus and is not found to reside in any of the known HDAC complexes (49). It is structurally related to both class I and II and in particular to HDAC3 and HDAC8. HDAC11 contains a catalytic domain located at the N-terminus. Little information is available about its expression and function, but in recent years, some evidence of a role in the regulation of the immune system has been collected (167).
HDACs targets
Histone substrates
DNA is organized into chromatin. The basic subunit of chromatin is the nucleosome, which is composed of 147 bp of DNA, coiled around an octamer of histones, with two molecules each of histone H2A, H2B, H3, and H4. Histone H1 associates with chromatin outside the core octamer unit and regulates the higher order of the chromatin structure (195). Each histone in the nucleosome has lysine-rich flexible tails, and DNA accessibility is controlled by post-translational modifications of some of these lysines (Fig. 3). In particular, their deacetylation, resulting in an increase of the affinity between histones and DNA, makes gene transcription more difficult.
Non-histone targets
Histones are not the only acetylated/deacetylated proteins. To date, more than 2000 acetylated proteins with various functions and involved in a wide range of cellular processes, including nuclear transport, chromatin remodeling, cytoskeleton rearrangements, cell cycle, DNA repair, and cell metabolism, have been identified (119) (Fig. 3). In particular, proteomic approaches have permitted the identification of a large number of acetylated metabolic proteins (Fig. 3) (70, 186), whose regulation is dependent by deacetylating HDACs activity. Conversely, a growing number of studies also report a fine regulation of HDACs by metabolism (Fig. 3). In this scenario, metabolism and HDACs appear to have different levels of communication: direct control of metabolic enzymes activity by HDACs-mediated deacetylation, direct control of HDACs catalytic activity by metabolites, direct control of HDACs catalytic activity by metabolic responsive signaling, and lastly direct control of HDACs catalytic activity by level of free acetyl groups into the cells (179). In an attempt to collect the knowledge relative to these different levels of communications, next we will report some old and more recent findings about how metabolism regulates and is regulated by HDACs in physiological and pathological conditions.
HDACs and Metabolism: Metabolites, Diet, and Metabolic Responsive Kinases as Control Mechanisms of HDAC Activity
Metabolites and diet
Recent studies suggest that several metabolites, generated by different intracellular biochemical pathways, may directly regulate HDAC activity (168). In vitro and in vivo activity of recombinant HDAC1 and HDAC2 or immunoprecipitated HDAC1- and HDAC2-containing complexes increases on addition of NADPH and CoA-derivates. In particular, acetyl-CoA, acetoacetyl-CoA, succinyl-CoA, butyryl-CoA, isobutyryl-CoA, glutaryl-CoA, HMG-CoA, malonyl-CoA, methylmalonyl-CoA, crotonyl-CoA, and methylcrotonyl-CoA (molecules generated from degradation of glucose, or involved in the biosynthesis of fatty acids and sterols or in the metabolism of several amino acids and ketone bodies utilization) increase HDACs activity between 1.5 and 3-fold (Fig. 4). On the contrary, similar kinetic analyses indicate that the long chain fatty acid derivative palmitoyl-CoA as well as the free CoA inhibit HDAC1 and HDAC2 activity (Fig. 4). Importantly the metabolite concentration used in these HDAC activity assays are almost comparable with their physiological intracellular levels (168).
Further examples of metabolic regulation on HDACs activity are represented by sphingosine-1 phosphate, a bioactive lipid produced by nuclear sphingolipid metabolism, and
Interestingly, it has also been shown, in different cancer cells, that pyruvate and lactate may act as specific inhibitors of HDACs. In particular, pyruvate shows an inhibitory effect comparable to butyrate, known histone deacetylase inhibitor (HDACi), on HDAC1 and HDAC3 (164) and lactate deregulates genes resembling significantly those affected by known HDACi such as butyrate and trichostatin A (TSA) (101). Although the lactate IC50 values for HDACs appear higher as compared with known HDACi, it is likely that lactate also may act as HDACi at a concentration similar to the physiological one as that observed; for instance, in skeletal muscle during prolonged exercise or in tumors which prevalently use glycolysis (146, 170).
Recently, it has been shown that low intracellular pH favors HDACs activity and a significant decrease of global histones acetylation. This effect, leading to the release of free acetate anions, prevents further pH reduction, as they are transported along with the acid H+ protons to the extracellular space. Notably, internal pH is significantly influenced by a reduction of extracellular pH, as for example, on accumulation of lactic acid in cancer cells, suggesting an association between acidic cancer cell environment and low levels of histone acetylation. The mechanism by which pH influences the HDACs activity is still to be determined (114).
As mentioned, butyrate, a short chain fatty acid that is formed in the human colon by bacterial fermentation of carbohydrates, has a non-competitive inhibitory effect on HDACs (25). Recent data have shown that butyrate at a concentration of 0.5 mM inhibits the growth of HCT116 colon cancer cells, while it has a stimulatory effect on the growth of non-cancerous cells, which is, however, lost at higher concentrations (2–5 mM) (34). The authors demonstrate that this effect is caused by the differences between normal and cancer cells in terms of energy metabolism. Normal cells, in fact, being characterized by an oxidative metabolism, also use a low dose of butyrate through the tricarboxylic acid (TCA) cycle generating acetyl-CoA that ultimately favors HAT activity and proliferation. Conversely, cancer colonocytes being characterized by a glycolytic metabolism use a relatively small amount of butyrate through TCA cycle, enabling it to accumulate inside of the nucleus where it acts as an HDAC inhibitor. At higher concentrations, butyrate accumulates into the nucleus and also acts as an inhibitor in normal cells (34, 81).
β-Hydroxybutyrate (βOHB), the major source of energy for mammals, present in organisms at millimolar concentrations during prolonged exercise or starvation (141), has been also identified as a new physiological inhibitor of HDACs activity (154), in particular of HDAC1, HDAC3, and HDAC4 isoforms (154).
An association between HDACs activity and metabolism has also been observed at the organismal level. In fact, metabolic diseases such as diabetes and obesity involve epigenetic modifications influenced by diets (137). Foods such as high-fat and simple sugars affect HDAC activity in different animal models. Japanese macaques, fed with a high-fat diet during pregnancy, give birth to offspring with an increased histone H3 acetylation and decreased HDAC1 expression in the liver compared with macaques fed with a low-fat diet (1). The liver is a center of energy metabolism and the principal site of detoxification within the body; therefore, detrimental effects induced by an unhealthy diet can usually be observed in this organ for failure of correct tissue regeneration on partial hepatectomy (116). Mice and rats fed with a high-fat diet have shown impaired ability to regenerate liver on partial hepatectomy as compared with animals fed with a normal diet (29, 161). Notably, liver regenerative ability of high-fat diet animals or old mice hepatocyte proliferation are inversely correlated with hepatic HDAC1 protein levels (161, 173). Similarly, it has been shown that hepatic HDAC3 higher expression levels are associated to metabolic syndrome after a high-fat diet (157). Similar phenotypes are obtained when mice are fed with a simple sugar diet, and this effect could be linked to HDACs regulation (124).
Taken together, all these studies suggest that HDACs are regulated by intermediates of cell metabolism, are responsive to metabolic changes, and are directly involved in the epigenetic modifications induced by diet.
Metabolic responsive kinases
Several findings demonstrate that not only the activity of HDACs against their substrates are physiologically relevant but also their function as substrates for other modifying enzymes, mainly kinases, is important for the development and maintenance of cellular homeostasis and cellular metabolism.
In recent years, several studies have highlighted the role of metabolism in acetylation modifications through the regulation of HDACs activity. In this regard, some kinases, previously described for their ability to regulate HDACs activity through post-translational modifications, are known to be sensitive to exogenous and endogenous stimuli, leading to a change of intracellular metabolism (39, 200) (Fig. 5).
Adenosine monophosphate-activated protein kinase (AMPK) is a key metabolic kinase that is activated in response to a decline of cell energetic status and AMP/ATP ratio increase, which regulates both catabolic and anabolic processes. In primary hepatocytes and hepatoma cell lines, class IIa HDACs (HDAC4,5,7) are phosphorylated and excluded from the nucleus by AMPK family kinases (Fig. 5A), leading to the inhibition of gluconeogenic genes transcription (118). On the other hand, PKA kinase, which is involved in several cellular functions including regulation of glycogen, sugar, and lipid metabolism (78, 163), may inhibit the AMPK activity, permitting the accumulation of HDAC4,5,7 in the nucleus, where they act as scaffolds for HDAC3. The stimulation of HDAC3 leads to deacetylation and consequent activation of the transcription factor Forkhead box class O 1 (FoxO1) that induce the expression of gluconeogenic genes (118) (Fig. 5A).
The inhibition of HDAC4 and HDAC5 by CaMKII and SIK1, a kinase that belongs to the AMPK family (77), leads to the activation of the glucose transporter GLUT4 and of peroxisome proliferator-activated receptor-γ coactivator (PGC-1α) transcription (115, 127) and to the inhibition of glyconeogenesis (159), therefore controlling glucose metabolism and mitochondrial biogenesis (Fig. 5B).
In Hela cells subjected to hypoxia (Fig. 5C), HDAC1 and HDAC2 activity is stimulated by CK2-phosphorylation (139). HDACs activation leads to transcriptional repression of the E3-ubiquitin ligase von Hippel-Lindau (VHL) and, hence, to the stabilization of its main target, the transcriptional factor Hypoxia-inducible factor 1α (HIF1α) (87).
Whether other kinases as well as other stimuli are able to control cell metabolism through the regulation of HDACs activity remains to be investigated; however, the ability of metabolic responsive kinases to regulate HDACs activity is a newly discovered important connection linking different metabolic states and transcriptional programs.
HDACs and Cancer Metabolism
Protein hypoacetylation and deregulated metabolism are distinctive features of tumors
Due to their important role in the regulation of gene transcription and protein activity, HDACs are essential for proper growth and behavior of cells and tissues. Therefore, deregulated function of HDACs can promote several developmental defects, as described in Figure 1, and also cancer onset and progression (53). Several studies have observed that tumors are characterized by the unbalance between protein acetylation and deacetylation (119). Although some tumors are characterized by a relative increase of acetylation, as a consequence of functional mutations and overexpression of HAT proteins (24, 133), several evidence indicates mainly an unbalance favoring the activity of HDACs, as a result of their over-expression (182) and/or aberrant recruitment to target promoters (50, 185), that leads to histone hypoacetylation. In this regard, significant evidence of the involvement of HDACs in cancer progression has been well documented through selective knockdown of single HDACs or by using specific HDAC inhibitors (63, 144). Interestingly, such approaches have also permitted to disclose specific targets and molecular mechanisms that link HDACs to tumorigenesis.
As described earlier, HDACs have histone and non-histone targets and among them, several metabolic proteins or proteins involved in the regulation of the cellular metabolism have been identified. Indeed, recently, it has been shown that lysine acetylation is a prevalent modification in enzymes that catalyze intermediary metabolism (106). Importantly, such a modification, directed by an HAT and HDAC pair, appear to be synchronized according to growth conditions and regulate the metabolic enzyme activity and the metabolic fluxes (175). Therefore, deregulated HDACs could be involved in metabolic reprogramming that characterizes cancer cells as compared with normal ones (117).
In this regard we would like to briefly summarize the most relevant among these metabolic differences (Fig. 6).
Cancer cells are very often characterized by an intense glycolytic activity that leads to a large production of lactic acid even at normal oxygen availability, the so-called Warburg effect. Several reports have shown that this enhanced glycolysis is associated to an increase in the activity of the glycolytic enzymes such as hexokinase1 and 2 (HK1-2), pyruvate kinase M2 (PKM2), lactate dehydrogenase A (LDHA), and monocarboxylate transporter 4 or to the up-regulation of glucose transporters GLUT1-3 (20). Warburg effect has also been related to the alteration of different oncogenic pathways such as protein kinase B (Akt or PKB) and phosphoinositide 3-kinase, whose activation is sufficient to functionally drive glucose uptake and aerobic glycolysis (14, 31, 37). Several studies have also demonstrated that the enhanced glycolytic activity is a tumor cell metabolic strategy to survive in hypoxia by the induction of HIF1α transcription factor (13). Conversely, the switch to glycolysis of cancer cells is also promoted by the inactivation of tumor suppressor proteins. In fact, phosphatase and tensin homolog deleted in chromosome 10 (PTEN) and p53 inactivation, leading to chronic activation of Akt or to a reduction of mitochondrial respiration (9), respectively, favor the increase of glycolysis observed in cancer cells.
Another metabolic adaptation of cancer cells is given by their tendency to increase glutamine consumption (30, 181); in fact, they often show an increased expression of membrane glutamine transporters as well as enzymes involved in the anabolic use of glutamine backbone such as glutaminase, glutamate dehydrogenase, aspartate amino transferase, and glutathione sinthetase (181). In addition, cancer cells often present an increased lipogenesis as reflected in the over-expression and hyperactivity of lipogenic enzymes such as ATP-citrate lyase, acetyl-coenzyme A carboxylase (ACC), or fatty acid synthase (97). Further metabolic differences between normal and cancer cells have been found in the mitochondrial respiratory system. In fact, cancer cell proliferation as well as tumor aggressiveness correlates with a low-mitochondrial respiratory chain activity (4, 19), established by the fact that the enhancement of oxidative phosphorylation (OXPHOS) activity reduces tumor growth (72, 147).
Histones and non-histones deacetylation affects cellular metabolism through gene expression and protein function regulation
In a previous section, we already mentioned that HDACs might control metabolic genes through their effect on transcriptional machinery such as the activation of gluconeogenic and GLUT4 genes. Here, we would like to underline the regulatory activity of HDACs on the level of expression and/or activity of important proteins involved in cell metabolism control.
Peroxisome proliferator-activated receptor-γ coactivator
PGC-1α is a transcription co-activator that plays a key role in the regulation of cellular energy metabolism. PGC-1α, in fact, in order to increase respiratory capacity of mitochondria, may stimulate mitochondrial biogenesis and/or coordinate organelle remodeling through mammalian target of rapamycin (mTOR) pathway (6, 10). Moreover, it promotes gluconeogenesis, participates in the regulation of both carbohydrate and lipid metabolism, and promotes the remodeling of muscle tissue to a fiber-type composition that is metabolically more oxidative and less glycolytic (143). PGC-1α promoter is down-regulated in response to a change of cell energy demands by specific deacetylation of histone H3 HDAC5-mediated that leads to a reduced mitochondrial activity (27) (Fig. 6). We might suggest that the effects of HDACs deacetylation on PGC-1α could be considered an important mechanism to control cell bioenergetics that need further investigation.
Phosphatase and tensin homolog deleted in chromosome 10
PTEN is a phosphatase that determines the inhibition of the Akt signaling pathway. It is a tumor suppressor that is frequently mutated, deleted, or transcriptionally repressed in a large number of tumors (155). Several reports have indicated that an important mechanism for PTEN repression is represented by the HDACs-dependent deacetylation of its promoter (47, 48, 126, 132). In particular, such a repression is due to the recruitment of HDAC1 to its promoter by Sp1. This mechanism has been proved both in normal and in cancer cells and convincingly confirmed in experiments in which the treatment with TSA strongly increases PTEN expression (94, 128, 132). On the other hand, acetylation/deacetylation of PTEN protein has also been recognized as a mechanism to control its activity; in fact, it is positively regulated by HATs and by TSA treatment. However, the HDACs involved in the deacetylation of PTEN protein are not known, at least for the E1 A-associated protein p300 (p300)/CBP-associated factor (PCAF)-acetylated residues. Regardless of the mechanism, PTEN repression leads to the activation of the oncogenic Akt pathway and, thus, to a metabolic reprogramming (52) (Fig. 6). Unfortunately, no papers have yet directly addressed the role of the axis HDACs-PTEN in the occurrence of cancer cell metabolic alterations.
Hypoxia-inducible factor 1α
HIF1 is a transcriptional factor that regulates the expression of genes which are critical to cellular response and adaptation to hypoxia, among which there are several glycolytic enzymes, glucose transporters, and TCA cycle inhibitors such as pyruvate dehydrogenase kinase 1 (PDK1) (152), favoring glycolysis and glucose fermentation over mitochondrial OXPHOS. HIF1α is constantly synthesized but rapidly degraded under non-hypoxic conditions by the VHL/proteasome-mediated degradation (151).
However, HIF1α is also regulated by acetylation/deacetylation. HDACs of class II are important for stability of HIF1α. It has been shown that HDAC7 co-translocates to the nucleus with HIF1α under hypoxic conditions, increasing its transcriptional activity (85). Moreover it has been shown that HDAC6 stabilizes HIF1α by deacetylation of the chaperone Hsp90. In accordance, HDAC6 inhibition results in hyperacetylation of Hsp90 that leads to HIF1α proteasomal degradation (92). Conversely, HDAC4 deacetylates N-terminal lysine residues of HIF1α, increasing protein stability under hypoxia. In fact, HDAC4 knockdown increases HIF1α acetylation and attenuates cancer cell response to hypoxia, inhibiting the HIF1α-mediated transcriptional activation of specific genes, among which are the glycolytic genes (GLUT1 and LDHA), and significantly reducing the hypoxia-dependent lactic acid production (51). Therefore, HDACs have an essential role in the glycolytic phenotype induced by hypoxia especially in cancer cells, which are often under reduced oxygen availability (134). HIF1α is also expressed in tumors independently from the oxygen environment (20). In this regard, it has been observed that the Metastasis-associated protein 1, whose elevated expression is correlated with tumors aggressiveness (104), enhances the stability of HIF1α by recruiting HDAC1 and permitting the HIF1α deacetylation (193). Such a mechanism, in a context of HDACs over-expression characteristic of a large number of tumors (182), may also lead to HIF1α stabilization and relative metabolic effects in an oxygen-independent manner (Fig. 6).
p53
p53 is the most commonly mutated tumor suppressor in human cancer. It codifies for a transcription factor that activates a large number of target genes which are involved in different processes, among which is cellular metabolism. In fact, p53, through its controlling role in specific proteins, for example, TP53-induced glycolysis and apoptosis regulator (9), cytochrome c oxidase 2 (113), and malic enzyme (ME1–2) (79), may regulate processes such as glycolysis (9), OXPHOS (113), pentose phosphate pathway (80), lipogenesis (79), NADPH production (79), glutamine metabolism (79, 158), and reactive oxygen species (ROS) levels (158). Therefore, its inactivation leads to a strong metabolic reprogramming that favors tumorigenesis (Fig. 6). It has been demonstrated that p53 activity is increased by acetylation at specific lysine residues in its C-terminal domain (60). While acetylation is essential for p53 function, class I HDACs deacetylation is necessary to maintain the protein in a functionally inactive state (83, 108). Acetylation/deacetylation of p53 also regulates its stability. In fact, since the acetylated lysine residues in C-terminal are also the sites for the mouse double minute 2 homolog (MDM2)-mediated ubiquitination and degradation of p53, on acetylation such a process is prevented. Interestingly, MDM2, in order to induce p53 degradation, first recruits an HDAC1-containing complex that promotes p53 deacetylation (75). More recently, it has been observed that HDAC2 also modulates the p53 transcriptional activity, indeed its knockdown increases p53-DNA binding activity through the modulation of histone acetylation and chromatin composition (66). Altogether, these observations strongly support a role of HDACs in the p53 loss of function that favors the cancer metabolic switch (Fig. 6).
Cellular myelocytomatosis viral oncogene homolog
Cellular myelocytomatosis viral oncogene homolog (c-Myc) is an oncoprotein commonly up-regulated in cancer (112). c-Myc has an essential role in regulating cancer cell metabolism. In particular, c-Myc favors the oxidation of glucose to lactate, increasing the expression of glucose transporters and glycolytic enzymes, suppressing mitochondrial respiration, and inducing PDK1 (28). Moreover, c-Myc can stimulate glutaminolysis, conferring glutamine-addiction to the cells (180) (Fig. 6). c-Myc activity is strongly regulated through different mechanisms (12). Several reports have indicated the c-Myc-recruitment of different HATs or of HDAC1 and HDAC3 as the primary mechanism for the regulation of its ability to activate or repress several target genes (38, 44, 98, 166). c-Myc is also regulated by acetylation. In fact, it has been shown that its stability is reduced by acetylation (38). However, there are no direct studies addressing the role and the HDACs involved in the deacetylation of c-Myc. On the other hand, several evidence indicates that the HDACi may down-regulate c-Myc expression (71, 96). Since there is an important role of c-Myc in different aspects of cancer metabolism, the characterization of the role of HDACs in c-Myc activity could be a vast area of investigation, leading to new elements for understanding tumorigenesis and cancer metabolism.
Signal transducer and activator of transcription 3
Signal transducer and activator of transcription 3 (STAT3) is a transcriptional factor that is activated by cytokines, growth factors, and oncogenes. On stimulation, phosphorylated STAT3 dimerizes and translocates into the nucleus or mitochondria to activate target genes (160). STAT3 controls different cellular processes, among which is cell metabolism (32, 55, 178) (Fig. 6). In fact, when phosphorylated at tyrosine 705, it translocates into the nucleus, where, through the transcriptional induction of HIF1α, it promotes aerobic glycolysis. In addition, also acting as a transcriptional repressor for the nuclear-encoded genes involved in mitochondrial function, it strongly reduces mitochondrial respiration (32). On the other hand, on phosphorylation at serine 727, STAT3 migrates into the mitochondria where it is able to modulate OXPHOS (55, 178), an effect that in Ras-transformed cells has been shown to favor aerobic glycolysis associated with transformation (55). Importantly, it has been demonstrated that STAT3 is also regulated by (de)acetylation. In particular, in different solid cancer cells, acetylation of lysine 685 is critical for STAT3 dimerization, DNA binding, and transcriptional activity, as HDAC1-3 activation has negative effects on the protein function (194). On the contrary, in diffuse large B-cell lymphoma, the activity of HDAC3 is necessary for the nuclear function of STAT3. In fact, inhibition or knockdown of HDAC3 increases STAT3 lysine 685 acetylation, preventing tyrosine 705 phosphorylation and STAT3 nuclear translocation (61). Therefore, the role of acetylation in regulating STAT3 appears important for its function (Fig. 6). No studies have been addressed so far to clarify the role of HDACs in the dual metabolic function of STAT3 into the nucleus or mitochondria.
Yin Yang 1
Yin Yang 1 (YY1) is a sequence-specific DNA-binding transcription factor that is able to activate or repress gene expression depending on the receiving stimuli and the factor to which it is associated (54). Its transcriptional activator function depends on its association with p300, while the transcriptional repressor activity is mediated by HDAC1, 2, and 5 (156, 191). Growing evidence reports YY1 as a downstream effector of the mTOR pathway (74, 100). A genomic analysis reveals that mTOR balances the cell energy metabolism by modulating the ability of YY1 to interact with transcriptional cofactors, such as PGC-1α, that are involved in energy metabolism regulation (6). In fact, knockdown of YY1 causes a significant decrease in mitochondrial gene expression, a decrease of mitochondrial respiration, and the lost of the mTOR-dependent gene activation (10, 26) (Fig. 6). Moreover, it has been shown that YY1 is a negative regulator of p53 protein, and a positive regulator of HIF1α given that its inhibition reduces the accumulation and the activity of HIF1α under hypoxic condition (184). Such an observation suggests the involvement of HDAC-induced repressor activity of YY1 in cancer metabolic switch.
Forkhead box class Os
FoxO belongs to the Forkhead box family of transcription regulators of cell cycle, apoptosis, and response to oxidative stress.
FoxO subclass comprises four members in mammals: FoxO1, FoxO3a, FoxO4, and FoxO6. FoxOs promote tumor suppression (131) and have relevant roles in metabolic pathways regulating ROS detoxification (93), energy metabolism, glucose homeostasis (58), and expression of nuclear mitochondrial encoding genes (41).
FoxOs are acetylated. Acetylation results in a reduced DNA binding and in an alteration of their cellular localization and, in consequence, FoxOs are inactivated (165). Conversely, deacetylation, due to class I/IIA HDACs, results in a strong activation of the transcriptional activity of FoxOs proteins (118, 172).
Mihaylova et al. (118) directly correlate FoxO1 and 3a deacetylation/activation by HDACs to glucose homeostasis in mice. Furthermore, Shimazu et al. (154) show that βOHB acts as an endogenous HDAC inhibitor on HDAC1, increasing FoxO3a activity and leading to protection from oxidative stress (Fig. 6).
Adenosine monophosphate-activated protein kinase
As previously described, AMPK can regulate localization and activity of HDACs (115, 159). Recently, it has been shown that AMPK is regulated by p300 and HDAC1 (107). p300 acetylates and inhibits AMPK, while the deacetylation of the kinase by HDAC1 favors its interaction with the upstream liver kinase B1 that on phoshorylation favors the AMPK activation. Deacetylated activated AMPK can phosphorylate its downstream targets, among which is the enzyme ACC. In cancer cells, where the equilibrium between HATs and HDACs is often lost, the metabolic kinase AMPK could be, consequently, deregulated (Fig. 6). To date, the role of AMPK as tumor suppressor or oncogene is still controversial and not yet completely clarified, as it has been suggested that its role may change depending on the stage of the tumor (65). In fact, AMPK, in general, switches off the anabolic pathways (65) and has been recognized as a negative regulator of the Warburg effect (40) but, on the other hand, it appears, especially in established tumors, essential for cancer development because of its fundamental role in maintaining energy homeostasis and protecting from metabolic stress (65, 105). Therefore, it may be very interesting to further investigate its regulation by HDACs, also considering the increasing use of HDACi as well as AMPK activators (200) as anticancer drugs.
Pyruvate kinase M2
PK regulates the final step of glycolysis. Alternate splicing leads to the generation of PKM2, which is often over-expressed in cancer cells, where it plays a central role in metabolic reprogramming, in the regulation of gene expression, and in the subsequent cell cycle progression (188) (Fig. 6). Lv et al. (109) demonstrate that PKM2 is acetylated by PCAF under high glucose concentrations and that such a modification leads to its reduced activity and to its lysosomal-dependent degradation. This mechanism permits tumor cells to accumulate or redirect some glycolytic intermediates to other anabolic pathways, for example, serine synthesis (192), in order to sustain cell growth especially under a rich glucose condition. Moreover, the authors show that PKM2 acetylation is enhanced after treatment with the HDACi TSA suggesting an involvement of different HDACs families in PKM2 deacetylation. To date, no more information is reported on this point, so it is possible to speculate that PKM2 deacetylation and allosteric activation by fructose 1, 6-bisphosphate in cancer positively regulate PKM2 activity, and, as a consequence, favor fermentative glycolysis and, hence, the Warburg effect.
Drugs inhibiting HDACs activity induce large metabolic rearrangement in cancer cells
HDACs have become one of the emerging targets for cancer therapy, and HDACi show promising anticancer activities through their ability to induce differentiation and/or apoptosis in cancer cell lines (11). However, the overall impact of these inhibitors widely varies depending on the cell type, the dose, and the targeting of factors outside of transcription. In this regard, in recent years, significant attention has been focused on the ability of HDACi to alter cell metabolism along with their effect on histone acetylation. In particular, it has been shown that valproic acid and suberoylanilide hydroxamic acid (SAHA) treatments of multiple myeloma cells disturb acetyl-CoA intracellular equilibrium, as a consequence of a decreased fatty acid β-oxidation and a parallel reduction of glucose uptake and utilization. These effects are the results of the ability of the two inhibitors to down-regulate the expression of GLUT1 and to inhibit the enzymatic activity of HK1 (176). In addition, an increased oxidative metabolism of amino acids has been observed, indicating that the inhibitors lead to a large metabolic reprogramming in cancer cells. Of note, these effects have been also observed in NB4 acute promyelocytic leukemia cells, suggesting a more general mechanism at least in non-solid tumors. Alteration of glucose metabolism following SAHA and TSA treatments has also been observed in different colon cancer cells lines, where HDACi induce a transcriptional-dependent reduction of epidermal growth factor receptor (EGFR) expression. Since EGFR has been reported to be involved in the stabilization of the active Sodium-glucose transport protein 1, such a reduction is also associated to a significant reduction of glucose uptake and to an increase of the HDACi-induced cell death. It should be noted that cell death is almost rescued by increasing extracellular glucose availability (21), suggesting that glucose metabolism alterations participate in the cell death effect of HDACi. A more detailed bioenergetics analysis, performed after treatment with sodium butyrate or TSA in lung cancer cells, has further confirmed the role of HDACs in affecting cancer cell metabolism. In fact, both inhibitors increase mitochondrial respiration over glycolysis. Such a change in cell metabolism is associated with a decreased and increased expression of GLUT1 and GLUT3 respectively, an enhancement of mitochondrial HK, a stimulation of glucose-6-phosphate dehydrogenase activity, and an increase of oxygen consumption coupled to ATP production. Importantly, the authors suggest that the negative effect of HDACi on HIF1α expression could justify all the observations on these lung cancer cells (5). An association between the increase of histone acetylation, inhibition of cell cycle, and remodeling of cell metabolism has been also observed in colon (HT29) and prostate (PC3) cancer cells as well as in HT29 xenograft tumors on Belinostat treatment, an HDACi similar to SAHA. This inhibitor induces an increase in the phosphocoline and in branched-chain amino acids (comprising valine, leucine, and isoleucine) levels and a reduction of glucose flux to lactate in favor of a rise in alanine formation (8). Altogether, these reports strongly support the role of HDACs in the positive control of cancer cell metabolism either through transcriptional regulation or through non-histone protein deacetylation.
On the other hand, such a role has also been confirmed by using HDACi in a physiological condition. In fact, it has been shown that class I HDACi enhances whole-body energy expenditure, improves insulin sensitivity, and stimulates OXPHOS and mitochondrial function in mice skeletal muscle and adipose tissue. In contrast, class II HDACi do not exhibit these actions, suggesting that HDAC isoforms have specific energy metabolic targets in vivo (46, 176).
Cancer cell metabolic reprogramming correlates with HDACs increased activity
As illustrated earlier, several metabolic intermediates are able to regulate HDACs activity both positively and negatively. Interestingly, several of these metabolites as well as physical changes are well correlated with the metabolic alterations observed in cancer cells (Fig. 7). Among the physical changes, hypoxia, a typical feature of developing tumors that arise after the reduction of vascularization (33, 67), activates HDACs. Similarly, also the pH lowering, a common feature in tumors as consequence of an increased fermentative metabolism of glucose and accumulation of lactic acid in the space surrounding the tumor (20), induces HDACs activity. Likewise, several metabolites regulating HDACs activity are products of metabolic pathways that are deregulated in cancer. Tumors show increased levels of NADPH, an HDACs activator, in order to sustain several anabolic reactions; for example, nucleotides, glutathione, and, the most important, fatty acid synthesis (36, 79, 80). In addition, different intermediates of the mevalonate pathway (e.g., acetoacetyl-CoA, HMG-CoA), a metabolic pathway involved in cholesterol synthesis and protein prenylation and often highly activated in cancer (45), positively regulate HDACs. Equally, several intermediates of amino-acid degradation (acetyl-CoA, crotonyl-CoA, glutaryl-CoA, methylcrotonyl-CoA, methylmalonyl-CoA, and succinyl-CoA) that cancer cells use for fatty acid synthesis and other anabolic needs (15, 102, 135, 138, 174) activate HDACs. Interestingly, a correlation between HDACs activity and deregulated metabolism could be also observed for the negative regulators. Palmitolyl-CoA, for instance, being a substrate for the synthesis of sphingolipids, often increased in cancer, is generally rapidly used in tumors. Contrasting findings have been observed for sphingosine-1P. Nevertheless, this metabolic molecule has an important role in cancer proliferation and survival, and its balance is tightly regulated in tumors (123, 140).
Conclusion and Outlook
Here, we have discussed the cross-talk between classical HDACs and cell metabolism with particular emphasis regarding cancer cell metabolism. We have seen that metabolism affects HDACs activity through several intermediates of different metabolic pathways as well as through signaling pathways which are responsive to metabolic changes (Fig. 8). As a consequence, metabolism may control transcription and several other processes regulated by the HDACs activity on non-histone proteins. On the other hand, HDACs control cell metabolism by deacetylating metabolic enzymes (Fig. 8). Such a relationship is clearly shown through the use of novel and more sophisticated technologies that are permitting the isolation of a high number of acetylated metabolic enzymes whose activity is changed on this post-translational modification. Interestingly, the acetylation state of these enzymes has been suggested to change depending on extracellular nutrient availability. In addition, recent reports, at least in rodent models, indicate that glucose homeostasis is actively regulated by HDACs as well as HDACs activity and expression is involved in metabolic changes, observed in different organisms such as rat and monkey, triggered by the diet (Fig. 8).
Despite the mechanisms controlling the connection between acetylation/deacetylation and metabolism still remaining unclear, the identification of a physiological role of HDACs in controlling metabolism has further increased the interest for their role in pathological conditions, such as diabetes, obesity, and cancer, as they are considered possible therapeutic targets for these diseases. However, the use of pan-HDACs inhibitors in cell and mouse models as well as in phase I–III clinical trials for malignant diseases has clearly indicated two important points: First, the mechanism of action of HDACi is only partially related to altered gene expression, as all the inhibitors on average modulate 2% of total genes (56, 111), and they are more related to changes in non-histone proteins via regulation at the post-translational modification level. Second, HDACi have varying antitumor activity, as their effects are influenced by the cell type, the dose, and the type of tumors. Both points may suggest that the different repertoire of non-histone protein targets in a tumor may be the reason for the variability of their efficacy. In this scenario, the identification of HDACs substrates will be increasingly interesting. Recent data have indicated metabolic enzymes as an important class of targets. In fact, it has been observed that virtually each enzyme of numerous metabolic pathways is modified by acetylation, as also shown by the large metabolic alterations observed on using HDACi treatment in cell models, indicating an important role of HDACs in their function (Fig. 8).
Altogether, these findings have to be taken more into account while considering the revival of cancer metabolism as a key hallmark of tumorigenesis. In fact, the discovery that different metabolic intermediates, highly produced by some of the pathways deregulated in cancer cells, such as fatty acid synthesis and pentose phosphate pathways, or chemical changes characterizing the tumors, such as hypoxia and low pH, may positively regulate HDAC activity (Fig. 9A) suggests a tight association between the two processes and probably nutrient availability. On the other hand, the ability of HDACs, as largely discussed throughout the text, to modulate the same metabolic pathways through their effect on oncosuppressors and oncogene activity implies the existence of a “positive feedback” for cancer in which metabolism and HDACs cross-talk to drive tumorigenesis. However, as anticipated throughout the review, several gaps in our knowledge regarding this “feedback” need to be filled. For instance, defining the relationship between the different metabolic profiles of the tumors, for example, glycolytic or glutaminolytic, and the specific class of HDACs active in the same type of the tumors will help delineate the reciprocal influences along the tumorigenic process (Fig. 9B). Another challenge for the future will be also to discover the pathological roles of individual HDACs in cancer metabolic alterations by using specific KO mice or cell lines that will also permit to identify HDAC isoform-specific targets (Fig. 9B). Similarly, it is currently almost unclear whether really cancer-specific metabolic intermediates may participate in HDAC activation in tumors. Altogether, these aims may find answers in the novel field of metabolomics, as the analysis of metabolites represents a sensitive measure of biological status in health or disease and may give fingerprints that are sometimes unique to every individual or every disease (Fig. 9B). In this regard, a systems approach is often useful to better understand the execution and control of complex biological functions (90). The first step of this approach generally requires a clear definition of the gene partners involved in specific modules (for instance, in our case, the regulation of a metabolic pathway by a given HDAC) (3). The availability of an interactome map of the human HDAC family (82) may enable a definition of the molecular network that underlies the function under investigation. The construction of a concept map following lines reported in the literature [e.g., Alzheimer Disease map (120)] brings a step further toward the construction of a dynamic molecular model, which generally involves a restricted number of molecular players—the ones involved in a more critical feature of the function under observation. Simulation analysis and experimental validation are finally able to shed light on the design principles that are crucial for the establishment of the physiological behavior.
The replication of this procedure to several aspects of HDAC regulation of cancer cell metabolism may help identify more effective (possible combinatorial) drug targets.
Footnotes
Acknowledgments
The authors apologize to the many authors whose work they were not able to cite given space constraints. Research in Chiaradonna laboratory has been supported by grants to F.C. from the Italian Government (FAR). L.A. and F.C. have been partially supported by SysBioNet, an MIUR grant for the Italian Roadmap of ESFRI Infrastructures. R.P. and G.V. have been supported by fellowships of SysBioNet. C.C. has been supported by a grant from MIUR (FIRB, 2008). The authors wish to thank Neil Campbell for English editing.