Regain control of p53: targeting leukemia stem cells by isoform-specific HDAC inhibition

Abstract

Leukemia stem cells (LSCs) are self-renewable leukemia-initiating populations that are often resistant to traditional chemotherapy and tyrosine kinase inhibitors (TKI) currently used for treatment of acute or chronic myeloid leukemia (AML or CML). The persistence and continued acquisition of mutations in resistant LSCs represent a major cause for refractory disease and/or relapse following remission. Understanding the mechanisms regulating LSC growth and survival is critical for devising effective therapies that will improve treatment response and outcome. Several recent studies now indicate that the p53 tumor suppressor pathway is often inactivated in de novo myeloid leukemia through oncogenic specific mechanisms, which converge on aberrant p53 protein deacetylation. Here, we summarize our current understanding of various mechanisms underlying deregulation of histone deacetylases (HDACs), which could be exploited to restore p53 activity and enhance targeting of LSCs in molecularly defined patient subsets.

Introduction

Leukemia stem cells (LSCs), characterized by unlimited self-renewal capacity, are shown to be central to the initiation, growth and relapse of acute and chronic myelogenous leukemia (AML and CML). Studies in recent years have led to the view that the persistence of these clonal LSC subpopulations could be a major driving mechanism contributing to treatment refractory and/or relapse following remission [1-3]. It has also recently been brought to light that after chemotherapy treatment, clonal evolution from preleukemic hematopoietic stem cells (HSCs) could occur and promote development of chemoresistant relapse [4-6]. The heterogeneity and the dynamic nature of malignant disease progression appear increasingly complex. Meanwhile, it is now clear that new therapies more effective in targeting quiescent and chemoresistant LSCs are needed to improve treatment outcome and cure.

The tumor suppressor protein p53 is arguably the most studied molecule due to its central role in coordinating regulatory circuits that sense and respond to a wide variety of stressors including DNA damage and oncogenic events and ultimately control fundamental cell fate decisions such as cell cycle progression, apoptosis, senescence, metabolism, and autophagy [7,8]. The important role of p53 in cancer is underscored by the fact that genetic mutations in TP53 have been detected in approximately half of all human cancers and disruption of other p53 pathway components is prevalent in the remainder [9]. In myeloid leukemias, however, TP53 mutations are relatively infrequent (less than 10%) and mostly associate with complex karyotype and therapy related neoplasms [10-13]. Nevertheless, TP53 mutation is recognized as an adverse risk factor for chemotherapy response and prognosis [14,15]. As a master coordinator of important cellular processes, p53 function is regulated by a wide spectrum of post-translational modifications including phosphorylation, ubiquitination, acetylation, methylation and sumoylation [7,16-19]. It has been suggested that inactivation of non-mutated p53 frequently occurs through binding to its principal regulator MDM2, a E3 ubiquitin ligase that mediates degradation of p53 [20-22]. Compounds that directly interfere with the binding of p53 and MDM2, including Nutlins and MI-series inhibitors, have been developed and evaluated for anti-leukemia efficacy [23-32]. Multiple mechanisms have been observed to influence the efficacy of MDM2 inhibitors, underscoring the need to further dissect the heterogeneity and oncogene-specific mechanisms inhibiting p53 response in various types of leukemia. In particular, LSCs pertinent to refractory disease and relapse could rely heavily on alternative p53-inactivating mechanisms for survival and continued evolution during and following chemotherapy. Understanding these mechanisms presents new opportunities to specifically reactivate p53 and elicit LSC-selective vulnerability.

Histone deacetylases (HDACs) are enzymes that catalyze the removal of acetyl moieties from ε-amino groups of lysine residues in a variety of proteins, including histones and nonhistone proteins [33]. Based on homology to the yeast HDACs and their enzymatic activities, HDAC proteins are categorized into four classes, including class I (HDAC1, 2, 3 and 8), class II (HDAC4, 5, 6, 7, 9 and 10), class III (SIRT1, 2, 3, 4, 5, 6 and 7) and class IV (HDAC11). HDACs are widely recognized as important epigenetic regulators of gene expression via histone modification and chromatin remodeling. Many broad spectrum HDAC inhibitors have potent anticancer activities and are in various stages of clinical trials [33-38]. However, these inhibitors are highly toxic and lack selectivity, which have greatly hampered their clinical application and efficacy. More selective inhibition of mechanistically defined HDAC targets is needed to effectively eliminate cancer cells and minimize toxicity. Several members of the class I (HDAC1, 2 and 3) [39-41] and class III HDACs [42,43] are known to deacetylate the p53 protein. Given that acetylation modification of the p53 protein is essential for stabilization, nuclear localization, and transcriptional activation [44,45], p53 activity can be specifically altered by deregulation of HDACs. Here we focus on recent advances in our understanding of divergent p53-inactivating mechanisms and how deregulation of specific HDAC proteins could be exploited to restore p53 activity and enhance targeting of LSCs.

BCR-ABL activates SIRT1 expression in CML

CML has served as a paradigm for neoplasia evolution and targeted molecular therapy [46]. CML usually presents in a chronic phase and progresses through an accelerated phase followed by a terminal acute leukemia-like blast crisis [47]. It is uniformly associated with a chromosomal translocation t(9;22)(q34; q11) which results in the generation of BCR-ABL fusion gene. A unique feature of CML is that a single genetic lesion encoding the BCR-ABL fusion protein is sufficient to initiate malignant transformation of hematopoietic stem cells (HSCs). The use of tyrosine kinase inhibitor (TKI) to target BCR-ABL signaling has revolutionized the standard of care and greatly improved patient outcome. Treatment with TKI such as imatinib (IM), nilotinib, and dasatinib has been effective in inducing complete cytogenetic remissions and prolonging survival of chronic phase CML patients, but less effective against advanced phases of disease [48]. Even though TKI treatment effectively inhibited BCR-ABL kinase activity and reduced proliferation of primitive CML LSCs, it has been unable to eliminate residue LSC populations that may be potential sources of relapse [49-52]. In addition, mutations in BCR-ABL that confer resistance to TKI are common [53-55].

Sirtuin 1 (SIRT1) is a member of the sirtuin family of nicotinamide adenosine dinucleotide (NAD)-dependent deacetylases that regulate numerous biological processes, including aging, DNA repair, cell cycle, metabolism, and cell survival [56,57]. SIRT1 is shown to play important roles in the maintenance and differentiation of HSCs, especially under conditions of stress [58,59]. In CML, Wang et al. showed that SIRT1 deacetylase promotes acquisition of TKI resistant BCR-ABL mutations [60]. Given that acetylation is indispensible for transcriptional activation of p53 protein [44,45], SIRT1 functions as a negative regulator of p53 by deacetylating several lysine sites [42,43,61,62]. SIRT1 expression can be upregulated by multiple mechanisms including epigenetic silencing of a negative regulator HIC1 [63] or altered miRNA regulation [64]. In a study by Yuan et al., it was shown that BCR-ABL activates SIRT1 through STAT5 signaling and SIRT1 act as a survival pathway, which promotes oncogenic transformation and leukemogenesis [65]. Meanwhile, Li et al. showed that SIRT1 is overexpressed in primitive CML stem and progenitor cells compared to their normal counterparts [66]. Genetic knock-down of SIRT1 or pharmacological inhibition by the small molecule inhibitor tenovin-6 (TV-6) [67] impaired proliferation and induced apoptosis of CML stem and progenitor cells. In addition, combination of TV-6 with IM TKI treatment significantly reduced CML LSC growth and prolonged survival in vivo. Inhibition of SIRT1 led to enhanced p53 acetylation, and p53 activation is required for observed growth inhibitory effects of CML stem/progenitor cells. Another recent study by Wang et al. further demonstrated that genetic loss of SIRT1 depleted maintenance of CML LSCs [68]. Collectively, these studies establish that inhibiting the SIRT1-dependent survival pathway effectively activates p53 response and enhances targeting of CML LSCs. Combination of SIRT1 inhibitors with TKI could be efficacious for treating advanced CML disease and/or eradicating minimal residual disease.

FLT3-ITD induces SIRT1-c-MYC network in AML

AML is a form of highly heterogeneous hematopoietic malignancy with diverse cytogenetic, genetic and molecular abnormalities [69]. Identification of cytogenetic and genetic lesions has revolutionized AML disease classification and prognosis stratification [70-73]. However, treatment outcome in the majority of patients remains poor, with frequent and fatal relapse. Seminal work by Lapidot et al. provided the first proof that the continued growth and propagation of AML depends on a rare population of leukemia-initiating LSCs [74]. With the advent of next generation sequencing technologies, the profound heterogeneity in genomic and epigenetic landscapes in AML is undoubtedly clear [75,76]. It has also allowed detection of stepwise acquisition of AML driving mutations and infer clonal architecture [4-6,77]. In addition, it has led to identification of preleukemic stem cells harboring one or few founding mutations and the ability to acquire additional mutations contributing to relapse. The dynamic clonal and subclonal evolution during or following treatment further contributes to the complexity and heterogeneity of therapy response and outcome in AML. In-depth understanding of molecular alterations and oncogenic mechanisms underlying diverse genetic lesions and LSC resistance is needed to devise effective targeted therapies.

Activating mutations in receptor tyrosine kinases and signaling components constitute one of the classical types of mutations associated with AML. FMS-like tyrosine kinase-3 (FLT3) internal tandem duplication (ITD) is observed in 25–30% of AML patients and predicts poor prognosis [78-84]. The ITD mutation disrupts the negative regulatory function of the juxtamembrane domain, rendering FLT3 receptor constitutively active [85-87]. FLT3-ITD mutation activates canonical receptor tyrosine kinase signaling, most prominently via STAT5, RAS/MAPK, and PI3K [86,88-91]. Expression of FLT3-ITD from the endogenous promoter results in loss of HSC quiescence and a myeloproliferation neoplasm, which is reversible by FLT3-TKI treatment [92]. There are several small molecule FLT3 TKIs including quizartinib (AC220) and sorafenib being evaluated in clinical trials; however, responses have been heterogeneous and transient [93-96]. These results suggest that the leukemia-initiating LSCs may be escaping FLT3 TKI-induced cytotoxicity [96-99].

In an effort to better understand drug resistance mechanisms, Li et al. showed that FLT3-ITD caused increased SIRT1 protein expression via enhanced expression of USP22 deubiquitinase induced by c-MYC [100,101], which is activated by PIM1 as well as SIRT1-c-MYC feed forward loop in FLT3-ITD AML cells [102,103]. Inhibition of SIRT1 by shRNA-mediated knock-down or pharmacological inhibitor TV-6 reciprocally increased c-Myc acetylation and reduced its stability. SIRT1 knock-down or inhibition by TV-6 resulted in enhanced p53 acetylation and p53 target gene expression. Combination of TV6 with AC220 reduced FLT3-ITD⁺ AML CD34⁺ cell growth and survival, and enhanced TKI-mediated targeting of AML LSCs in vivo [100]. Meanwhile, Sasca et al. demonstrated that tyrosine kinase signaling including STAT5 and RAS activation likely acts in concert to activate SIRT1 expression [104]. In addition, it is proposed that FLT3-ITD regulates p53 acetylation via the ATM-DBC1-SIRT1 axis, which could also be regulated by irradiation-induced genotoxic stress [104]. In murine AML models driven in combination with MLL-AF9 or RUNX1-ETO, the combination of TV-6 and TKI modestly enhanced inhibition of proliferation [104]. The impact of additional genetic and cytogenetic aberrations on the sensitivity to SIRT1 inhibition remains to be determined. It is noteworthy that this p53 activating effect elicited by SIRT1 inhibition was not seen in FLT3 non-mutated AML cells or normal cells, underscoring the importance of identifying oncogene-specific adaptive response in the face of chemotherapy and other targeted therapy. However, there appears to be some discrepancy regarding whether SIRT1 activation was selective for FLT3-ITD⁺ AML and not for AML with FLT3-TKD mutations. Further studies are needed to clarify the nature and spectrum of oncogenic stimuli rendering SIRT1 activation and sensitivity to SIRT1 inhibition.

HDAC8 mediates deacetylation of p53 in inv(16) AML

In AML, chromosomal abnormalities frequently result in transcription factor fusion proteins that contribute to the unique etiology and prognosis of distinct cytogenetic subsets [105]. As a master transcriptional regulator of hematopoiesis, the core-binding factor (CBF) complex is a common target of leukemia-associated mutations [106,107]. Among the most common cytogenetic aberrations found in AML patients is chromosome 16 inversion inv(16)(p13.1q22) or translocation t(16;16)(p13.1;q22) [108]. Inv(16) generates a fusion gene Cbfb-MYH11, leading to expression of a fusion protein CBFβ-SMMHC [109,110]. A series of studies revealed that CBFβ-SMMHC dominantly inhibits CBF function, impairs hematopoietic differentiation and predisposes for leukemia transformation [111-115]. Dominant inhibition of RUNX proteins, either through cytoplasmic sequestration [116,117] or constitutive repression [118,119], was considered the main leukemogenic mechanism of CBFβ-SMMHC chimeric protein. However, more recent studies indicate that functional RUNX proteins are in fact needed for CBFβ-SMMHC leukemogenesis and growth of CBF AML cells [120-124]. It was previously reported that p53 response was reduced by CBFβ-SMMHC [125], although the underlying mechanism was not clear. A recent study by Qi et al. revealed that CBFβ-SMMHC gains p53-inhibiting function via aberrant protein-protein interaction with HDAC8 and the p53 protein [126]. HDAC8 is a member of the zinc-dependent class I HDAC enzyme known to deacetylate lysine residues in a variety of proteins, including histones and transcription factors [127-129]. Qi et al. showed that like other members of class I HDAC, HDAC8 is capable of deacetylating the p53 protein. Thus, CBFβ-SMMHC promoted HDAC8-mediated deacetylation of p53 by recruiting HDAC8 and p53 into an aberrant protein complex. Consequently, p53 induction and target gene expression is largely inhibited in the presence of CBFβ-SMMHC. Although CBFβ-SMMHC binds p53 and HDAC8 independently via distinct protein domains, the p53-inhibiting activity is dependent on the presence of both p53 and HDAC8 proteins in the ternary complex. Depleting CBFβ-SMMHC or HDAC8 resulted in restoration of p53 acetylation and activation upon exposure to genotoxic stress such as irradiation. Genetic deletion of Hdac8 in a conditional CBFβ-SMMHC knock-in mouse model dramatically diminished LSC transformation, as evidenced by greatly reduced AML incidence and delayed onset. Qi et al. also found that HDAC8 expression was significantly higher in the primitive CD34⁺ population and that inv(16)⁺ AML CD34⁺ cells express 5-12 fold higher levels of HDAC8 compared to non-inv(16) AML or normal CD34⁺ cells. In line with the differential HDAC8 expression, pharmacologic inhibition of HDAC8 enzyme using HDAC8 isoform-selective inhibitors (HDAC8i) [130,131] resulted in enhanced p53 acetylation, p53 target gene activation, and p53-dependent apoptosis selectively in inv(16)⁺ AML CD34⁺ cells while sparing the normal CD34⁺ stem/progenitor population. This activity further translated into elimination of AML propagation and leukemia-initiating activity in both murine AML and human AML xenograft models in vivo. Importantly, HDAC8i treatment was capable of enhancing the chemosensitivity of inv(16)⁺ CD34⁺ cells. Despite having a relatively favorable prognosis, only approximately half of the patients with inv(16) AML eventually achieve long-term survival with the standard chemotherapy regimens [132,133]. These results highlight the potential efficacy of HDAC8i in overcoming chemotherapy resistance and relapse of inv(16)⁺ AML.

Conclusion

In recent years, several alternative p53 inactivation mechanisms specific to the underlying oncogenic lesions have been shown for CML, FLT3-ITD⁺ AML, and inv(16)⁺ AML. These divergent pathways converge on inhibiting p53 acetylation via deregulation of alternative protein deacetylases (Figure 1). These results partly explain the heterogenous response to other p53 activating agents such as MDM2 inhibitors. Given that TP53 is rarely mutated in de novo myeloid neoplasm, these findings present new opportunities to regain control of p53 activity and enhance response to chemotherapy or other targeted therapies. Importantly, LSC populations relevant to refractory disease and relapse are selectively sensitive to perturbation of the specific protein deacetylase defined by the specific oncogenic mechanism. Thus, selective inhibition of context-specific HDAC isoforms is a promising approach to eradicate residual drug resistant LSCs, prevent further acquisition of mutations and reduce relapse. This highlights the importance to dissect the genetic and molecular heterogeneity, particularly in AML. These studies also demonstrate that by attacking cancer-specific vulnerability, the normal HSC counterpart can largely be spared. Further development of isoform specific HDAC inhibitors is critical to translate these insights into the clinic.

Figure 1. Multiple leukemogenic pathways converge on p53-inactivation through enhanced p53 protein deacetylation.

SIRT1 deacetylase is stabilized and activated through multiple mechanisms downstream of BCR-ABL in CML (left) and FLT3-ITD (center) signaling in FLT3-ITD⁺ AML. In inv(16) AML (right), CBFβ-SMMHC fusion protein recruits HDAC8 and p53 in a protein complex, thereby promoting deacetylation of p53 by HDAC8. Inhibition of oncogenic context-specific deacetylase is a promising approach to specifically activate p53 pathway and enhance sensitivity of leukemia-initiating LSCs to TKI or chemotherapy.