Regulation of HDACi − Triggered Autophagy by the Tumor Suppressor Protein p53

Cancer is a complex genetic and epigenetic-based disease that has developed a multitude of mechanisms in evading cell death. Deregulation of apoptosis and autophagy are commonly encountered during the development of human tumors. Histone deacetylase inhibitors (HDACi) have been employed to reverse epigeneti-cally deregulated gene expression caused by aberrant post-translational protein modifications. These interfere with histone acetyltransferase- and deacetylase-mediated acetylation of histone and non-histone proteins, and thereby exert a wide array of HDACi-stimulated cytotoxic effects. Key determinants of HDACi lethality that interfere with cellular growth in a multitude of tumor cells are apoptosis and autophagy. Currently, the factors that determine the mode of HDACi-elicited cell death are mostly unclear however. Experimental evidence of the last decade convincingly reports that the frequently mutated tumor suppressor protein p53 can act either as an activator or as an inhibitor of autophagy depending on its subcellular localization, and linked to its mode of action. Consistently, we recently described p53 as a regulatory switch that governs if histone deacetylase inhibitor-adminis-tered uterine sarcoma cells undergo autophagy or apoptosis. By highlighting this novel finding, we summarize in this chapter the role of p53-mediated signaling during the activation of the autophagic pathway in tumor cells in response to HDACi.


Introduction
Evading cell death has been defined as a cornerstone of cancer development [1,2]. Exploring the pathogenetic mechanisms that determine different cell death modes therefore facilitated the avenue for increased specifically directed interference with these molecular pathways. Morphologically, apoptosis, autophagy, and necrosis could be distinguished as major categories of programmed cell death very early that either act mutually exclusive or in combination involving cross talk for the elimination of tumor cells. While apoptosis and necrosis lead to inevitable cellular demise, autophagy can have either a cytotoxic or a cytoprotective function [3]. Basal autophagy in normal eukaryotic cells provide a possibility to save energy and reuse damaged or aged macromolecules or organelles by redirecting them toward lysosomal degradation via so-called autophagosomes and can be triggered by nutrient-starving conditions [4,5]. Thus, in early phases of tumorigenesis, autophagy obviously presumes a cytoprotective or pro-survival role by suppressing necrosis and inflammation with autophagy but found as a general regulator of lysosomes and endosome formation during membrane trafficking [30]. Further maturation of autophagosomes involves the interaction of the PI3P-binding proteins WIPI 1/2 with ATG12-ATG5-ATG16L and LC3-phosphatidylethanolamine (LC3-PE) complexes, both representing ubiquitin-like conjugation systems [4,18,31]. LC3 (microtubule-associated protein 1A/1B-light chain 3) and p62/Sequestosome-1 are two markers that are regularly employed for documenting autophagic flux as they are involved in the maturation of autophagosomes [32]. While LC3-I is processed to LC3-II, the scaffold protein p62 interacts with LC3 via its LC3-binding motif and seems to have a role in selectively guiding ubiquitinated proteins toward the autophagosome via its ubiquitin-binding domain [33]; thus, p62 levels decease during the induction of autophagy and have been found moreover to regulate protein deacetylation and is associated with tumorigenesis [34][35][36]. Autophagosome-lysosome fusion that requires the transmembrane protein LAMP2 and small Rab GTPases, finally permits hydrolase and cathepsin-mediated processing of the autophagosome content [37,38].

Positive regulation of p53-mediated autophagy
The tumor suppressor protein and transcription factor p53, which represents a "guardian" of the cell, has a fundamental role in the regulation of cell integrity Nuclear p53-mediated transcription-dependent autophagy, apoptosis and cell cycle arrest in response to stress conditions. By upregulation of tuberous sclerosis complex 2 (TSC2) or phosphatase and tensin homolog (PTEN; not shown), or AMP-activated protein kinase (AMPK) or its activators sestrins (not shown) p53 prevailingly attenuates mammalian target of rapamycin (mTOR) and the unc-51 like autophagy activating kinase 1 (ULK1) complex (consisting of autophagy-related gene 13 (ATG13) and the focal adhesion kinase interacting protein of 200 kD (FIP200)) as the autophagic canonical pathway. ULK-1 then interacts with Beclin-1 (BECN1) to initiate autophagosome formation. A shortcut for activation of autophagy involves damageregulated autophagy modulator (DRAM), death associated protein kinase (DAPK), or autophagy-related gene 5 (ATG5) upregulation by the p53-family members, p63 and p73, or disruption of BCL2-family-or alternate reading frame protein product of the CDKN2A locus (p14ARF)-mediated release of BECN1 inhibition. In addition to autophagy, DRAM and p63/p73 are able to activate apoptosis. Arrowlines, upregulation or activation by indicated proteins; double arrow, major pathway activity. p53-mediated upregulation of the cyclin-dependent kinase inhibitor 1 (p21) enforces cell-cycle arrest. This figure is used under the terms and conditions of the creative commons attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/) derived from Mrakovcic and Fröhlich [57]. and homeostasis and consequently in tumor defense. It coordinates cellular responses such as cell cycle arrest, apoptosis, senescence, metabolism, differentiation, angiogenesis, and even modulates autophagy. Among a multitude of other post-translational modifications, acetylation assists the master regulator to sense and integrate a variety of endogenous and exogenous cellular stress signals such as DNA damage, epigenetic alterations due to DNA methylation, genotoxicity, hypoxia, oxidative stress, or oncogene activation [39,40]. In response, p53, as a central transcription factor translocates to the nucleus by detaching from the E3 ubiquitin ligase, mouse double minute 2 homolog (MDM2), and modulates the expression of multiple downstream target genes that regulate processes such as cell cycle progression and cell death [41,42]. In appropriate conditions, p53 induces apoptosis by transactivating, i.e., transcriptional activation of pro-apoptotic genes or in the cytoplasm by direct interaction with anti-apoptotic proteins located in the mitochondrial membrane [43].
Several signaling pathways involving the transactivational activity of p53-in normal as well as cancer cells regulate autophagy in the classical canonical mTOR pathway as specified in the previous chapter (Figure 1) [44]. As pro-autophagic factors emanating from p53, these pathways involve on the one hand the tumor suppressor proteins tuberous sclerosis complex 2 (TSC2) and phosphatase and tensin homolog (PTEN), and on the other hand the nutrient energy sensor AMPactivated protein kinase (AMPK) or its activators sestrins 1 and 2 [45][46][47]. A further path that bypasses mTOR and can directly modulate p53 stress-activated signal transduction, is damage-regulated autophagy modulator (DRAM) that can activate the autophagic as well as apoptotic program [48]. As a protein located in the lysosome, it can intervene at different steps of autophagosome formation [49]. Furthermore, by either upregulating pro-apoptotic protein expression (BAX, BAD, BNIP3, or PUMA) or downregulating anti-apoptotic protein expression (BCL-2, BCL-xL and MCL-1) of the B-cell Lymphoma-2 (BCL-2) family, p53 can enforce dual activation of autophagy and apoptosis [50,51]. In the inactivated state these proteins directly interact with the BH3 domain of BECLIN-1 and block the direct activation of BECLIN-1-dependent autophagy [52,53]. Direct interaction of the nuclear full-length form of p53-modulated tumor suppressor protein p14ARF (an alternate reading frame protein of the CDKN2A locus) with the BCL-xL protein is a further similar mechanism promoting the induction of autophagy although the predominant role of p14ARF seems to stabilize p53 to protect the cell against hyperproliferative growth and associated activation of oncogenes [54,55]. Additionally, p53-elicited upregulation of DAPK has been reported to result in autophagic activation either by DAPK-mediated phosphorylation of Beclin-1 that blocks its degradation by BCL-2/BCL-xL, or by impeding the anti-autophagic LC3-interacting MAP1B protein [30,56].

Negative regulation of p53-mediated autophagy
Beyond the nuclear-based transactivating pro-autophagic effects mediated by p53, additional inhibitory anti-autophagic responses related to cytoplasm-localized p53 protein have been uncovered by Tasdemir et al. in the past 10 years (Figure 2) [58]. While the transactivation-dependent nuclear autophagic response of p53 is stimulated by stress induction, the cytoplasmic blockage of autophagic induction is a steady-state function that is present under physiological conditions and seems to engage direct protein interaction. This cytoplasm-mediated inhibition of autophagy was also characterized to activate the canonical p53-AMPK-mTOR signaling cascade. In contrast to transcription-dependent pathway, however, the positive autophagic regulator AMP-dependent kinase is inhibited by p53 which in turn activates mTOR [59]. Accordingly, either pharmacological interference, depletion of basal p53 levels, or p53 variants that possess a genetically modified nuclear export domain rendered cells more resistant toward metabolic stress through elevated autophagy. The underlying obscure mechanism that is found not only in mammalians but also in nematodes could involve direct binding of p53 to FIP200 (ATG17) as experimentally evidenced [60]. Negative regulation of autophagy by cytoplasmic p53 has been also linked to its target gene TIGAR (TP53induced glycolysis and apoptosis regulator) that suppresses glycolysis and ROS generation when the cell is exerted to stress [61]. Nevertheless, although TIGAR mediates ROS-mediated induction of autophagy, it is not a likely candidate for the above described mechanism but rather represents an alternative path since it does not inhibit mTOR obviously. A similar but also unclear anti-autophagic mechanism could be verified in embryonal carcinoma cells, where p53-Beclin-1 interaction facilitated Beclin-1 ubiquitination and subsequent degradation, which could be de-activated by depletion of p53 [62]. Another report identified Beclin-1 as a regulator of de-ubiquitination of p53 which was mediated by USP10 and USP13 ubiquitin-specific peptidases [63]. This mechanism might therefore also relate to the previously mentioned Beclin-1-induced autophagy which enables bidirectional dual activation of apoptosis and autophagy [64].

Regulation of autophagy in p53-inactivated cells
p53 is one of the most frequently inactivated tumor suppressor genes in human tumors [65]. Particularly, single point mutations that provoke a loss of p53 function  [57].
were documented that in several cases apply a dominant-negative effect to the remaining non-mutant allele thereby enhancing their oncogenic effect [41,66,67]. Often, such p53 variants result in increased genomic instability, attenuated chemotherapeutic success and a poor prognosis for patients [68]. One of the underlying reasons therefore could be that many tumor-derived p53 variants also inactivate cytoprotective or cytotoxic autophagy [69][70][71]. Nevertheless, although nucleus-based transcription-dependent autophagy might be shut down in these cases, cytoplasm-induced activation of the autophagic program might still be available, due to p53-deficiency or functional inactivation. Interestingly, the studies of Morselli et al. demonstrated that several tumor-derived mutants of p53 that reside in the cytoplasm are still able to block autophagic induction, presumably by direct protein interaction [70]. Such experiments underline the significance why it is meaningful to discriminate p53 mutant variants with regard to their potential effects. This finding also highlights the role of context-dependent autophagy during tumorigenesis as disabled autophagy by mutant p53 was found to prolong tumor cell survival while it inactivated its tumor suppressor function. Thus, increased proliferation of pancreas and breast cancer cells could be uncovered in a report that confirmed inhibition of autophagy by mutant gain-of-function p53 proteins. This counteractivity was evidenced by stimulation of AMPK-mTOR genes with concomitant downregulation of Beclin-1, DRAM, ATG12, and sestrin genes [72].
Various investigations found also a counter-acting surveillance mechanism between autophagy and since as mutant p53 blocks autophagic induction in one way but autophagy can stimulate the sequestering of mutant p53 in order to suppress tumorigenesis in the other way. Mechanistically, this mutual crosstalk is translated by the regulatory actions of the two suppressor genes Beclin-1 and p53 on autophagy as specified in the previous chapter [63]. While p53 exerts control on Beclin-1 via the canonical autophagic pathway, Beclin-1 also directly regulates p53 via controlling its deubiquitination activity which explains the mirrored effect on the phenotype of p53-and Beclin-1 ablated mice [73]. Additionally, with respect to Beclin-1 mediated autophagy, further reports documented the possibility of autophagic activation via the tumor suppressor protein p14ARF in p53-silenced or -inhibited cells [74]. Studies using doxorubicin-treated p53-wildtype or -deficient (p53−/−) mouse embryonic fibroblasts furthermore verified that the p53 family members p63 and p73 can substitute the loss of p53 (Figure 2) [75]; this mechanism involved nucleus translocation of p63/p73 and the increased expression of an extensive network of ATG proteins, such as ATG4a, ATG4c, ULK1, ULK2, UVRAG, and ATG5. This finding might explain resistance in doxorubicin-mediated chemotherapeutic treatment of cancer tissues.
Depending on function or structure, four classes (class I-IV) have been allocated that comprise 18 members of the HDAC family [80]. The "classical HDACs" contain classes I and II and are functionally dependent on zinc as co-factor, while class III HDACs include the sirtuin proteins (Sirt1-7; homology to yeast Sir2) and require NAD+ [81]. Nevertheless, HDACs also differ in subcellular localization, and expression pattern [78]. While class I HDACs are expressed ubiquitously as they are located in the cell nucleus providing them with superior enzymatic activity, class II HDACs possess restricted tissue-specific expression pattern. Thus, they have been subdivided into class IIa HDACs (HDAC4, 5, 7 and 9) which shuttle between nucleus and cytoplasm as well as class IIb HDACs (HDAC 6 and 10) that are located mostly in the cytoplasm [82]. SIRTs exhibit specific subcellular presence in the nucleus (Sirt1, 6 and 7), in the cytoplasm (Sirt2), or in mitochondria (Sirt3, 4 and 5) which is not interchangeable. HDAC11, the single less-well explored member of class IV HDACs, has narrowed tissue expression [83].
HDAC inhibitors (HDACi) have been explored as a new category of anticancer drugs that reverses epigenetic changes established by the deregulated activities of HDACs in hematological as well as solid cancers [84]. HDACi treatment induces transcriptional de-repression of genes that are eminent regulators of tumor cell activities such as cell cycle arrest, differentiation, and programmed cell death and even the expression and stability of oncoproteins [85]. HDACi categories encompass hydroxamic acids (hydroxamates), cyclic tetrapeptides, benzamides, electrophilic ketones, and aliphatic acids that include natural but also synthetic derivatives that exhibit different structures [84]. Favored representatives of the hydroxamates are SAHA (suberoylanilide hydroxamic acid, vorinostat, and Zolinza) which is a preferred derivative of naturally occurring trichostatin A (TSA) as well as the CBHA (m-carboxycinnamic acid bishydroxamate)-derived tubacin, LAQ-824 (dacinostat), LBH-589 (panobinostat), or PXD-101 (belinostat) [86][87][88][89]. The class I-selective FK-228 (romidepsin, FR901228, istodax) belongs to the group of cyclic tetrapeptides [90]. MS-275 (entinostat) and MGCD0103 (mocetinostat) exhibiting enhanced HDAC class I selectivity are members of benzamide-based HDACi [91,92]. The minor effective class I-and IIa-specific HDACi, VPA (valproic acid), PBA (phenylbutyrate), NaB (sodium butyrate), or AN-9 (pivaloyloxymethyl butyrate) belong to the category of aliphatic acids [93,94]. This classification mainly depends on the chemical structure of their zinc-binding group but in addition HDACi can also be subdivided into zinc-dependent, pan-or broad-spectrum inhibitors that inhibit all class I, II and IV HDACs in contrast to primarily class I-specific HDACi [95]. Representatives of pan-inhibitors are TSA, SAHA, LBH589, and PXD-101 while valproic acid and butyrate inhibit exclusively class I HDACs. MS-275 and depsipeptide inhibits only a few members of class I HDACs, respectively. To date, the HDAC6-specific inhibitor tubacin is the only representative of an isoform-specific HDACi [86]. With the exception of nicotinamide, no clinical useful SIRT inhibitors have been uncovered yet [96].
Clinical trials of single or combined treatments of several HDACi with diverse results have been or are in the progress of being tested in hematological and solid cancers (www.clinicaltrials.gov) [97,98]. Up to now, exclusively the evaluation of pan-inhibitors have succeeded in the admittance of four licensed HDACi, namely, SAHA, panobinostat (LBH589), belinostat (PXD-101), and romidepsin (FK228) for the treatment of cutaneous T cell lymphoma, multiple myeloma, or peripheral T cell lymphoma, respectively [99][100][101][102][103]. Although preclinical studies using single treatment regimen of many HDACi were encouraging, almost all entities of solid tumors (e.g., ovarian, breast, renal, prostate, and head and neck cancer) lacked positive effects in phase II clinical trials [104,105]. In addition, patients suffered from trivial (e.g., dehydration, anorexia, diarrhea) to toxic (e.g., cardiotoxicity, thrombocytopenia myelosuppression) non-selective side effects [85,94,106]. The reasons for these drawbacks are presently non-elucidated and were assumed to be due to a combination of failing blood vessel supply, endogenous molecular heterogeneity owing to epigenetic modifications, and the development of treatment resistance. In response, selective HDAC-specific inhibitors are being developed, that target only one or two isozymes [107]. The design of novel or improved specific inhibitors will allow the full exploration of individual functions of distinct HDAC activity and may furthermore provide improved therapeutic efficacy together with less toxicity.

Mechanisms of histone deacetylase inhibitor-induced cell death
Owing to the various posttranslational histone and non-histone protein acetylation targets, HDACi exert a multitude of anti-tumor effects that concern interference with growth, differentiation, migration, senescence, and death [108]. Although there may be tumor cell-type and HDACi-specific effects which are unclear presently, common mechanisms are shaping for different HDACi [85,108]. Induction of apoptosis is by far the prevailing avenue of HDACi triggered cell death in transformed cells which is prepared by re-induction of cell cycle arrest and induction cell differentiation, e.g., by the downregulation of positive cell growth regulators [109][110][111][112]. G1 or G2 phase induced cell cycle arrest in the G1 or G2 phase can occur in a p53-dependent or -independent manner by stimulating the upregulation of (p21waf-1/cip1) expression which is a cyclin-dependent kinase (CDKN) inhibitor of cyclins D1/D2 [113][114][115]. As an underlying mechanism it is assumed that the inability to exit the cell cycle from unfinished mitosis might sensitize the activation of apoptosis due to compiled DNA damage such as double-strand breaks [116,117]. HDACistimulated activation of the intrinsic (mitochondrial) pathway of apoptosis involves either down-regulation of anti-apoptotic genes (e.g., BCL-2, BCL-XL, XIAP, MCL-1, and survivin) or overexpression of pro-apoptotic genes (e.g., BAX, BAK) belonging to the B-cell Lymphoma-2 (BCL-2) family [113,114,118,119]. In the extrinsic (death-receptor) pathway, HDACi predominantly re-establish the expression of death receptors such as DR4 and DR5, or their corresponding ligands (e.g., TRAIL, FAS, FAS-L, and TNF-alpha) [120][121][122]. Furthermore, also the induction of reactive oxygen species (ROS) by HDACi is a second important anticancer-mechanism that is also responsible for cell death induction and associated with DNA damage; presumably, ROS is scavenged in normal, but not in malignant cells due to the compiled expression of thioredoxin (TXN) which represents an endogenous cellular antioxidant [123]. HDACs have moreover been detected to control histone deacetylation at damaged DNA sites undergoing repair that involves DNA damage-related response proteins [95,[124][125][126][127][128][129][130][131][132]. Thus, only in tumor cells upregulated expression of a marker for DNA double strand breaks, H2AX, was detected, when these were treated with SAHA [117]. In this context also the induction of autophagy as a means to maintain genomic was noticed, for instance, following MSH2-regulated DNA mismatch repair deregulation upon HDAC6 inhibition. Cell signaling pathways that were shut down in cancer cells can be furthermore re-established by immediately modifying acetylation of non-histone proteins such as transcription factors (e.g., NF-κB, p53, and STATs) [79]. As a prominent example, half-life and stability of p53 was influenced by MDM2 E3-ligase in HDACi-treated H1299 carcinoma cells [42]. In this way also chaperone protein function and the regulation of stress response pathways in the endoplasmic reticulum can be achieved, which affects the removal of misfolded proteins but also interference with stability and expression of oncoproteins [111,133]. Additional mechanisms of HDACi-regulated lethality in tumor cells were evidenced in the interference with migration-and invasion capability due to re-established expression of metastasis-related genes and in the disruption of angiogenesis by altering pro-and anti-angiogenic gene expression [134][135][136]. In recent years, autophagy as a form of programmed cell death was added to the list of further determinants of HDACi-mediated effects that impedes cellular growth in a range of tumor cells [137][138][139][140].

Mechanisms of HDACi-induced autophagic cell death
An impressive diversity of mechanisms have been uncovered in cancer cells that promote HDACi-elicited autophagy which include mostly attenuation of mTOR signaling that can occur in combination combined with increased expression of LC3, Beclin-1, or ATG and can be provoked by endoplasmic reticulum stress (reviewed in [57] and Table 1 [141]). mTOR is a well-known regulator of the canonical pathway of autophagy involving the regulation of the ULK1 complex and Beclin-1. The pivotal role of mTOR attenuated by SAHA-treatment which reestablishes ULK1 function could be initially verified by our own studies and those of Gammoh et al. using endometrial sarcoma cells, and were subsequently reiterated in many studies [71,142]. It should be noticed that HDACi-induced autophagy is frequently accompanied by the additional induction of apoptosis.
Further predominant mechanisms of autophagic induction involve ROS accumulation, p21 upregulation, NF-κB hyperacetylation, and sirtuin-mediated acetylation of p53 [143][144][145][146]. In addition to mTOR downregulation, substantial intracellular ROS production interfering with mitochondrial function and energy metabolism has been demonstrated to facilitate SAHA-induced autophagy in tumor cells. ROS-induced autophagy can go along with additional increased expression of cathepsin D, a lysosomal protease, or decreased expression of TRX, representing its substrate and/or activation of the mitogen activated protein kinases ERK1/2 and JNK [143,147]. Generally, enzymes related to energy metabolism, anti-oxidative stress and cellular redox control have been entangled by a proteomic study involving SAHA-administered Jurkat T-leukemia cells [147]. Cell cycle arrest, differentiation, and autophagy due to upregulated p21 expression were caused by treatment of PC-3 M and HL-60 cells with HDACi SAHA and H40 [148]. The same mechanism could be elicited by adding the novel HDACi, MRJF4, to prostate cancer cells autophagy [144]. As a further cause of SAHA/MS-275-induced autophagy in PC3 prostate cancer cells, re-activation of NF-κB associated target genes due to hyperacetylation of NF-κB RELA/p65, or downsizing of pERK/NF-κB signaling together with upregulated p21 expression, were described; however, the exact mechanism remains obscure [146].
Individual studies also noted nuclear translocation of the apoptosis inducing factor (AIF), apoptosome inactivation, FoxO1-stimulated expression, upregulation of DAPK or Nrf2, and p53-deficiency as regulatory mechanisms in HDACi-induced autophagy [71,146,[149][150][151][152][153]. Thus, apoptosis, necrosis or autophagy were triggered in malignant rhabdoid tumor cells in response to FK228 (depsipeptide) treatment and upon silencing of the apoptosis inducing factor (AIF) that translocates into the nucleus for caspase-induced death, autophagy was suppressed as supported by transmission electron microscopy and LC3 measurements [149]. Following the blockage of the mitochondrial pathway of apoptosis by deleting caspase-9 or Apaf-1 ATG gene upregulation *

Independent of p53 acetylation
Tenovin-6 CLL [155] or in Eμ-lymphomas the autophagic pathway was activated by the HDACi LAQ824 and LBH589 as evidenced morphologically and biochemically [152]. SAHA and TSA-induced autophagic cell survival via the transcription factor FoxO1 in HepG2 and HCT116 cells was furthermore mediated by sestrin 3 (SESN3)-induced mTOR inhibition and increased ATG protein expression [150]. Protein interaction or phosphorylation of the MAPK-interacting calcium-or calmodulin-regulated DAPK at serine 308 in HCT116 colon cancer cells, rather than its enzymatic function, moreover stimulated LBH589-induced autophagy [151]. Recently, even microRNAmediated regulation of mTOR involving the transcription factor Nrf2 (nuclear factor erythroid 2 like-2) was implicated in HDACi-induced autophagy [153]. HDACi-induced Nrf2 mRNA and protein expression thereby promoted augmented transcription of miR-129-3p which facilitated mTOR attenuation. Nonetheless, even HDACi-mediated suppression of autophagy could be documented in two studies. Negative regulation of HDACi-mediated autophagy but upregulation of autophagic flux could be induced in myeloid-leukemic cells treated with valproic acid, SAHA, TSA, panobinostat, or JQ2 by acetylation and decreased expression of ATG7, a protein important for fusion of peroxisomal and vacuolar membranes [154]. Additionally, increased ATG expression following treatment with sirtuin inhibitor tenovin-6 provoked autophagic suppression in chronic lymphocytic leukemia (CLL) cells which was evident by upregulated genes of the autophagiclysosomal pathway and LC3-II/p62 [154,155].

HDACi-induced autophagy mediated by p53
p53 as the first described representative subjected non-histone protein acetylation can in response to stress positively as well as negatively regulate cell cycle arrest, senescence as well as apoptosis and autophagy [39,40]. Acetylated residues attached by distinct HAT-mediated acetylation can be detected for p53 at distinct sites that could not only affect DNA binding and thereby its transactivational ability but also coactivator recruitment and/or its stability via proteasomal degradation [166][167][168]. For example, HDAC1-specific inhibition allows p53 to stay in an accessible state associated with transcriptional activity [169]. Furthermore, by mutating a combination of C-terminal sites that undergo acetylation p53-dependent transcription of p21 can be eliminated [170]. Nevertheless, the exact modalities of these mechanisms still need clarification. As previously specified, HDACi-mediated apoptosis that is commonly escorted by p21-mediated cell cycle arrest and ROS generation, has been documented as the most frequently encountered form of HDACi-triggered cell death [112,119]. However, since transcription of proapoptotic genes, such as Bax, Noxa, and Puma, by p53 may be limited by posttranslational acetylation, the role of p53 in this relation is discussed. These assumptions are supported for example by the finding that p53-independent p21 induction and apoptosis upon HDACi administration and the anticancer effect of HDACi is not influenced by the mutational status of p53 in the tumor [109,113]. Other reports in contrast verified p53 acetylation and stabilization in several tumor models in response to HDACi administration that presented cell cycle arrest and apoptosis [111,171]. Conclusively, p53-dependent but also -independent signaling pathways may add to HDACi-mediated apoptotic processes and HDACi may induce p53, but do not unconditionally require p53 for providing anticancer effects. In recent time, the range of HDACi-exerted mechanisms resulting in cellular demise of cancer cells have been expanded by the induction of autophagic cell death which can alternatively or additionally to apoptosis activate autophagy (reviewed in [57,141,172] and Table 1). Also, involvement of posttranslational modification of the non-histone p53 has been linked to the control of HDACi-stimulated autophagy as evident from its key regulatory role in normal cells. This might of crucial advantage if tumor cells have developed resistance toward apoptotic cell death induction. Thus, experimental evidence from our studies of endometrial sarcoma (ESS) cells support a major regulatory function for p53 in directing cell death either toward HDACi-elicited apoptosis or autophagy [165]. In our model, the detection of HDAC2 overexpression in malignant endometrial stroma sarcoma cell lines led to the establishment of therapeutic SAHA treatment and the evaluation of its mechanism of action [173]. Significantly advanced cell death in MES-SA and ESS-1 cells was accompanied by previous p21induced cell cycle arrest at the G1/S transition and reduced expression of HDAC2 and 7 [156,173]. Either predominant caspase-dependent apoptotic (48%) or caspase-independent autophagic cell death (80%) was attested in SAHA-treated MES-SA and ESS cells after 24 h, respectively [156]. In line with the induction of the canonical pathway of autophagy, attenuated mTOR protein expression could be evidenced in ESS-1 cells in contrast to MES-SA cells [165]. Further screening of key regulatory molecules for apoptosis and for autophagy, upstream of mTOR, were performed to explain the differences in the modes of SAHA-induced cell death. This search uncovered entire absence of detectable p53 protein and lowered levels of PUMA protein in ESS-1 cells. Investigation of p53 gene and mRNA led to the detection of a novel nonsense mutation (p53 R213X ) in the transactivating domain of p53 of ESS-1 cells that obviously provoked a degradation of the entire p53 transcript and could not be documented in MES-SA cells. Consistent with this finding, restoration of ESS-1 cells with a wild-type p53 variant restored induction of caspase-dependent apoptosis as supported by PUMA and caspase-9 upregulation as well as activation of the effector caspases-3 and -7 and final PARP-1 cleavage. Increased mTOR levels demonstrated the re-induction of basal autophagic flux in addition to apoptosis induction as verified by LC3 staining. Generalization of this finding could be obtained by several other p53-deficient tumor cell lines (such as PANC-1, Jurkat, HL-60, and U937) that are known to induce autophagy in response to SAHA and were supplied with wild-type p53.
We concluded that the molecular switch between SAHA-induced apoptosis and autophagy was thus mediated by the occurrence of functional p53 protein.
Our experimental evidence thus underlines an overall major regulatory role for p53 not only in HDACi-mediated apoptosis but also in HDACi-stimulated autophagy (Figure 3) (reviewed in [57,141]). As a consequence, p53-deficiency could moreover explain apoptosis resistance as well predominant induction of HDACi-provoked autophagy in cancer cells. The presumptive negative regulation of autophagy by functional cytoplasmic p53 protein in SAHA-treated ESS-1 cells is  [57]. moreover very consistent with the above discussed role of p53 as a dual regulator of autophagy by Tasdemir et al. [59]. Their findings convincingly describe nuclear p53 protein as an activator of transcription-dependent autophagy, in contrast to the inhibitory autophagic control by cytoplasmic p53 protein. In addition, by our report we link SAHA-induced acetylation of p53 to the mTOR signaling pathway which has been less evaluated to date, nonetheless, future experiments are needed to directly address this question.
In line with our experiments, the class III Sirt1 and 2-specific HDACi sirtinol, that affects acetylation of p53, has also been documented to determine HDACiinduced cell fate in several reports. For instance, p53 was entangled in balancing sirtinol-mediated apoptosis and autophagy in MCF-7 breast cancer cells [145]. Sirtinol treatment on the one hand preferentially induced predominant autophagy as shown by LC3-II upregulation, while addition of the autophagic inhibitor 3-methyladenine augmented cell cycle arrest and cytochrome C-triggered apoptotic cell death caused by increased BAX and diminished BCL-2 protein expression. In a similar experiment, inhibition by sirtinol and the novel SIRT1, -2, and -3 protein inhibitor, MHY2256, a similar phenotype that included cell cycle arrest and both types of programmed cell death could be provoked [164]. As a mechanistic explanation, SIRT1 and 2-induced acetylation of p53 at lysine 382 were found to inhibit ubiquitination of p53 via MDM2 which stabilized and increased its functional activity. MDM2-mediated degradation of p53 was also documented in MHY2256-treated Ishikawa cells that are derived from endometrial cancer which elicited activation of apoptosis together with autophagy as supported by elevated levels of p21, BAX and BCL-2, cytochrome C release, and cleaved PARP-1 [174].

Conclusions and perspectives
In recent times, epigenetic studies gained increasing significance in reports investigating the development of cancer. For this purpose, aberrant epigenetic patterns such as DNA methylation including the misguided expression of HDACs activity has been defined to some extent in many tumors which explains their selection as targets for anticancer therapy. Posttranslational modifications of histones and non-histones in the form of acetylation and deacetylation particularly enable pharmacological interference by different kind of inhibitors such as HDACi. The ability to sensitize apoptosis-resistant tumor cells by the disruption of autophagy was considered a promising route for cancer therapy as this process heightens the pro-apoptotic effects of HDACi. In addition to restrain the extents of tumor necrosis and inflammation however, autophagy might be required for the cancer cell to deal with metabolic stress and cytotoxicity during chemotherapy. Furthermore, by expediting the autophagic pathway in advanced stages of the cancer cell, autophagy may promote cell death by mostly non-elucidated mechanisms. Consistently, it is of pivotal importance to define the factors and mechanisms that influence the balance between HDACi-elicited apoptosis, autophagy or even necrosis in the cancer cell. In this regard, considerable research efforts are in progress to investigate the molecular pathways regulating HDACi-mediated cell death in tumor cells. The expansion of the knowledge about p53 as a mediator of apoptotic and autophagic cell death may as thus help to achieve progress not only in unraveling pathogenetic insights but also in the development of novel therapeutic strategies of such disease conditions as cancer.