Small Molecules Targeting HATs, HDACs, and BRDs in Cancer Therapy

Evidence for research over the past decade shows that epigenetic regulation mechanisms run through the development and prognosis of tumors. Therefore, small molecular compounds targeting epigenetic regulation have become a research hotspot in the development of cancer therapeutic drugs. According to the obvious abnormality of histone acetylation when tumors occur, it suggests that histone acetylation modification plays an important role in the process of tumorigenesis. Currently, as a new potential anti-cancer therapeutic drugs, many active small molecules that target histone acetylation regulatory enzymes or proteins such as histone deacetylases (HDACs), histone acetyltransferase (HATs) and bromodomains (BRDs) have been developed to restore abnormal histone acetylation levels to normal. In this review, we will focus on summarizing the changes of histone acetylation levels during tumorigenesis, as well as the possible pharmacological mechanisms of small molecules that target histone acetylation in cancer treatment.


INTRODUCTION
Histone post-translational modifications (PTMs) directly impact gene transcription by regulating the chromatin architecture (1). Histone acetylation is one of the most well-studied and important PTMs, which mainly affects the status of local chromatin relaxation through changing the distribution of histone acetylation marks in the local chromatin region, thereby regulating gene transcription activation (2). In more detail, the acetylation of histones occurs in the lysine residues on the N-terminal tail of the nucleosome histones composed of H2A, H2B, H3, and H4, and the histone deacetylases (HDACs) and the histone acetyltransferases (HATs) are responsible for adding or removing acetyl groups from the N-terminal tail of the nucleosome histones (3). A large amount of research data demonstrated that histone acetylation widespread in cells is involved in various cellular activities, including genome maintenance, biological processes, DNA damage repair, cell cycle, and apoptosis (4). Once the dynamic balance between acetylation/deacetylation in cells is disrupted, it will cause various diseases, such as Parkinson's disease, leukemia, and even cancer (5-7). The following will specifically explain the changes in histone acetylation levels during cancer development, and how small molecules as cancer therapeutic drugs target and regulate intracellular acetylation levels.

IMBALANCED HISTONE ACETYLATION LEVELS IN TUMORIGENESIS
Based on the role of histone acetylation in the activation of gene expression, researchers speculated the mechanisms by which histone acetylation participated in and regulated progression of tumorigenesis (8). Multiple histone N-terminal acetylation sites have been identified (Figure 1). And many lysine sites on histones are obviously abnormally modified by acetylation in cancer cells and tumor tissues, suggesting that changes in their acetylation levels are closely related to the occurrence of cancer. Consistent with this argument, it has been confirmed that some HATs or HDACs are abnormally expressed when cancer occurs, resulting in alteration of local chromatin structure by changing the distribution of histone acetylation, ultimately affecting the expression of genes related to tumorigenesis.
It has been reported that the level of acetyl-modification on some histone lysine sites in cancer cells or tissues is obviously abnormal, and the increase or decrease of the modification level varies according to the type of cancer. Regarding H2A, Hat1 knockdown-or Tip60 abrogation-mediated downregulation of HeLa cell H2A lysine 5 acetylation (H2AK5ac) decreases HeLa cell colony size, suggesting that this acetylation can regulate cell proliferation (9). Furthermore, Ras-ERK1/2 pathway activationinduced osteosarcoma proliferation and migration co-occurs with downregulated H2BK12ac, a phenotype rescued by HDAC1 knockdown-mediated H2BK12ac restoration (10). Relative to other types of histone acetylation, the H2BK20ac modification preferentially accumulates at promoters of cell type-specific genes, indicating a role in regulating cell-specific functions (11).
Previous data indicate that the acetylation of specific histone lysine sites is associated with the occurrence of certain cancers. Recent research reported that histone H3 acetylation level is correlated with the pathological stage of colorectal cancer, especially with the depth of tumor invasion (12). For instance, downregulation of H3K4ac and H3K9ac has been observed in oral squamous cell carcinoma and ovarian tumors, and the status of acetylation level is tightly correlated with tumor stage, perineural invasion and tumor prognosis (13-15). Part of the reason for the above results may be related to its distribution region on chromatin. Because subsequent studies found that H3K4ac is enriched in the promoter regions of genes which associated with cancer-related phenotypic features, such as the estrogen response and the epithelial-mesenchymal transition (EMT) pathway (16,17). In head and neck squamous cell carcinoma (HNSCC) cells, H3K4ac modulated by HDAC3 is enriched around the transcription start site of EMT related genes such like GLI1 and SMO, cooverexpression of which promotes HNSCC cell invasion and migration ability (18). In addition to H3K4ac and H3K9ac, highlevel of H3K23ac, which is correlated with TRIM24, has been observed in patients with HER2-positive breast cancer, and this correlates with a shorter survival interval (19). Moreover, H3K27 represents a site vulnerable to multiple modification types, including methylation and acetylation, and upregulated H3K27ac in colon cancer and glioma cells is correlated with tumor invasive capability (20, 21). In esophageal squamous cell carcinoma (ESCC), H3K27ac activates long non coding RNA colon cancer associated transcript-1 (CCAT1), thereby promotes ESCC cells proliferation and migration (22). It is worth noting that some lysine-sites acetylation on histone H3 have been used as biomarkers. For example, H3K18ac and H3K4me2 has been used as biomarker in prostate, pancreatic, lung, and kidney cancers (23, 24). Taken together, unbalanced acetylation level of histone H3 in various cancer tissues or cells suggests that H3 acetylation may be involved in the transcriptional regulation of cancer-related genes.
In summary, biological mechanisms employing acetylated histones are much more diverse than chromatin structure regulation alone. The numerous N-terminal tail lysine residue acetylation sites of H2A, H2B, H3, and H4 allow them to participate in various signaling pathways, and facilitate their multi-faceted roles in cancer cell biology. Indeed, various cancers exhibit a globally dysregulated histone acetylation pattern, correlating with progression, pathological stage, and prognosis. As such, acetylation patterns may have potential as valuable prognostic markers (24).

HATS, HDACS AND BRDS
ACT AS "WRITERS", "ERASERS," AND "READERS" RESPECTIVELY Biological mechanisms employing acetylated histones are much more diverse than chromatin structure regulation alone. The numerous N-terminal tail lysine residue acetylation sites of H2A, H2B, H3, and H4 allow them to participate in various signaling pathways, and facilitate their multi-faceted roles in cancer cell biology. As mentioned, various cancers exhibit a globally dysregulated histone acetylation pattern, correlating with progression, pathological stage, and prognosis. As such, acetylation patterns may have potential as valuable prognostic markers (24). Noting that the dynamic change and reversible process of the acetyl-group at the N-terminal lysine site of histones can be controlled by certain proteins just like writers, erasers and readers. Cancer-associated abnormal histone acetylation profiles are due to corresponding aberrant expression or catalytic activities of these enzymes. HATs function as "writers", transferring the acetyl group (-COCH 3 ) from acetyl-CoA (Ac-CoA) to a target histone, whereas HDACs function as "erasers", removing the acetyl group of a target histone (36, 37). However, whether it is to remove the acetyl group or recruit proteins to a specific acetylmodified lysine site, the proteins usually have to recognize the acetyl group on a specific protein just like a reader.
Histone-mark readers often recognize marks through the functional domain contained in itself. Based on published literatures, the readers that can recognize histone acetylation are roughly divided into three categories including bromodomaincontaining protein (BRD), PHD finger and YEATS domains. Among them, PHD finger and YEATS domain proteins have a wide range of functions. In addition to acetyl-group, they can also recognize methyl-group or other proteins. For example, PHD finger proteins can able to acquaint acetylated or unacetylated and methylated histones. However, BRD is the only protein group featuring a domain that is able to recognize and bind acetylated histone lysine residues. BRD-containing proteins are widely present in most tissues. According to the sequence or structure similarity, BRDs are divided into eight families exhibiting various activities, including histone modification and chromatin remodeling ( Figure  2) (38, 39). For example, one of the most well-known BRD family members, BRD4, accumulates in highly acetylated and transcriptionally prone chromatin regions (including promoters and enhancers) and promotes RNA polymerase II (RNA Pol II) activity, thereby stimulating transcription initiation and transcript elongation. BRD4 is involved in HCC cell growth and invasiveness in vitro, and it is significantly upregulated in HCC tissue (a feature also associated with HCC progression) (40). Such functions are largely dependent on the ability of BRD4 to recognize acetylated proteins (41).
Considering the above description, the addition, removal and recognition of acetyl groups on histones is an indispensable dynamic balance. In other words, acetylation profiles regulated by HATs, HDACs, and BRDs, ultimately impact an abundance of target genes involved in tumorigenesis, thus regulating numerous cellular processes. For example, downregulation of TIP60 in 61% of primary gastric cancer patients is correlated with invasiveness and metastasis (42). Later research data supports this result. Currently, it is generally believed that alteration of HATs or HDACs level is involved in the occurrence and progression of cancer. From the published literature, the decrease of HATs and its enzymatic activity or the excessively high activity of HDACs can directly or indirectly affect the global acetylation level in cells. HAT MOF expression is downregulated in numerous cancers, including RCC, ovarian cancer, gastric cancer, and CRC (33). For additional detail, accumulating data reveals mutation residues on HATs in certain cancer, such as TIP60 in CRC (8). On the contrary, higher level of HDACs such as SIRT1, SIRT2, and SIRT7 were detected in cancer cells (43-45). Given this close relationship, an increasing number of small molecules targeting histone acetylation-regulating proteins are being investigated for their anti-cancer therapeutic potential.

SMALL MOLECULES TARGETING HATS, HDACS, AND BRDS IN CANCER THERAPY HDAC Inhibitors (HDACis)
HDACs are enzymes that remove acetyl group on Lys residues of histone proteins, the following four classes of HDACs are recognized: I (HDAC1, 2, 3, and 8), II (A: HDAC4, 5, 7, and 9; B: HDAC6 and 10), III (SIRT1-7), and IV (HDAC11) (Figure 2) (46). Given that the HDACs frequently show higher expression levels in cancer cells, small molecules targeting HADCs were first investigated. At present, many small molecules have been developed as HDAC inhibitors (HDACis). These HDACis may target different stages of cancer or different signaling pathways, and ultimately achieve the purpose of inhibiting or treating cancer.
So far, five HDACis Vorinostat (SAHA), Belinostat (PXD-101), Panobinostat (LBH589), and chidamide (CS055, HBI-8000) and Romidepsin (FK228) have been approved by the U.S. FDA (Food and Drug Administration) as medicines for treatment of skin Tcell lymphoma (TCL) and peripheral TCL (47, 48). The former three HDACis inhibit class I, II, and IV HDACs, while Romidepsin selectively targets class I (47). As one of the beststudied and pan-HDACi SAHA induces autophagy of chronic lymphocytic leukemia, breast cancer as well as colon cancer cell lines, and the induced autophagy modulates mutant p53 degradation, further affects cancer cell survival (49, 50). In addition to use alone, SAHA induces radio treatment pancreatic cancer cell cycle arrest and apoptosis by targeting RAD51, clarifying the function of SAHA in enhance radiosensitivity (51). In combination with other anti-cancer drugs, such as oxaliplatin (Eloxatin) and ruxolitiniband, SAHA optimally inhibits cancer cell proliferation (52, 53). In addition, an isotretinoin-SAHA combination for the treatment of neuroblastoma is currently undergoing phase I clinical trials (54).
In addition to SAHA, there are already more than 20 kinds of HDACis are in different stages of clinical research, indicating that the research and development of HDACis is very popular and has broad development prospects. Most of the HDACis studied extensively are aimed at the proliferation of tumor cells by targeting cell cycle and apoptosis, growth, and migration capability (55). CG200745, is a pan HDACi, targets HDACs and modulates acetylation, thereby regulates down-stream genes including p53, myeloid cell leukemia-1 (Mcl-1) and B-cell lymphoma-extra large (Bcl-xL) (56, 57). In detail, CG200745 inhibits NSCLC cell growth by modulating the profile of H4K16ac at the transcription start site of cell proliferation related genes (58). Moreover, CG200745 (59, 60) enhances the expression of p53 target genes by regulating p53 acetylation, thereby inducing clonogenic cell death (56). (61) In pancreatic cancer, CG200745 elevates the H3 acetylation level and induces the expression of apoptotic proteins, furthermore, CG200745 works better in combination with gemcitabine or erlotinib in suppressing cancer cell proliferation (62). The ability of CG200745 to sensitize tumor cells to existing chemotherapeutic drugs (such as 5-fluorouracil (5-FU), cisplatin, and oxaliplatin) has also been demonstrated (57, 62-64). These data recommend the pan-HDACi CG200745 as a candidate antitumor drug or chemotherapy adjuvant, and is currently undergoing the phase I/II clinical trials for pancreatic cancer (62, 65-67).
Sirtuins (SIRT1-7) are human homologs of the yeast Sir2 (silent information regulator-2) protein and are divided into four main classes: SIRT1-3 are class I, SIRT4 is class II, SIRT5 is class III and SIRT6-7 are class IV (92). SIRT proteins belong NADdependent deacetylases that act as intracellular regulators and are thought to have ADP-ribosyltransferase activity (93). It has been reported that (94-97) SIRT1 and SIRT2 as deacetylases modulate the acetylation of p53, thereby regulating p53 target genes and cancer cell progression (81,98). JQ-101, which inhibits SIRT1mediated H4K16 and p53 acetylation, thereby inducing A549 cell senescence and inhibiting tumor growth and invasiveness, similar phenomenon and mechanism has been detected in SIRT1 specific inhibitor EX527 treated glioma cells (82,99). Moreover, AEM1 and AEM2 also can facilitate p53 acetylation by targeting SIRT2 and further regulating the expression of p53 target genes (e.g., cell cycle regulator p21), thereby sensitizing NSCLC cells to genotoxic stress (100). However, tenovin-6 modulates the mRNA and protein level of p21 in cancer cell lines but through a p53-independent mechanism (101)(102)(103)(104).
Recently, with the development of HDAC inhibitors, many newly synthesized, derived derivatives or modified compounds have come out, and pre-clinical experiments have begun. For instance, a novel HDACi (OH-VPA) was developed by modifying a traditional HDACi (VPA), representing a new approach to novel HDACi development. The derivative HDACi is more effective in inhibiting HeLa cell proliferation than its parent molecule (105

Small Molecules Targeting HATs
This review limits its scope to discussing only HAT inhibitors which have been approved for cancer therapy or commercialization, since the specific mechanisms of HAT modulation-mediated anti-cancer effects are complex and ambiguous (177, 178). It appears that HAT influence during carcinogenesis is contextspecific because HATs are able to act as both oncogenes and tumor suppressors (179). The possible reason is that different tumors show mutations in different HAT members, which directly or indirectly affects any steps in the continuous process of tumor progression from tumorigenesis to carcinogenesis and metastasis (180). Based on sequence homology and shared structural features, HATs can be divided into two different classes. One is the GCN5-related Nacetyltransferases (GNATs) family, including GCN5 and p300/ CBP-associating factor (PCAF), that can acetylate lysine residues on histones and non-histone proteins (181). In lung cancer cells, p300 may promote Snail-dependent EMT (epithelialmesenchymal transition) by acetylating Snail at K187 site  (185). Similarly, the molecule PU141, a selective CBP/ P300 inhibitor, suppresses murine SK-N-SH neuroblastoma xenograft survival (186). Another HAT inhibitor C646 suppresses gastric cancer cell survival and invasive capability through competitively disrupting the interaction between Ac-CoA and CBP/P300 (187,188). Recently discovered compounds CCT077791 and CCT077792 were also found to target P300 and PCAF, and resulting in the reduction of global acetylation level in colon tumor cell acetylation levels and inhibiting tumor cell growth (189).
Another HAT family is the MYST superfamily, exhibiting a conserved catalytic MYST domain, and large group membership, including MOZ, Ybf2, Sas2, TIP60, and hMOF (181). The role of MYST family in tumorigenesis is beyond doubt. Based on laboratory research data, Tip60 can harbor substrates including histones and non-histone proteins like p53 and ATM kinase, through which TIP60 plays critical roles in regulating cancer progression such as cell cycle, invasiveness and metastasis in gastric cancer and breast cancer cells (190,191). Importantly, changes of downregulation of TIP60 is correlated with overall survival of breast cancer patients (42,192). In addition, by regulating PI3K/AKT pathway, Tip60 suppresses the proliferation and migration of cholangiocarcinoma (111,193) Given the critical role of TIP60 (a HAT which forms part of the TIP60/NuA4 complex) in DNA damage repair, several TIP60 inhibitors have been investigated for their anti-cancer therapeutic potential, including TH1834, NU9056, and 6alkylsalicylates. Indeed, TH1834 (which blocks the binding site of TIP60) disrupts DNA damage repair to induce breast cancer cell apoptosis (194), and NU9056 both inhibits prostate cancer cell growth and induces apoptosis (195). Similarly, frequent downregulation of MOF has been detected in numerous cancers, including RCC, ovarian cancer, gastric cancer, and CRC (33). Developed MOF inhibitor DC-M01-7 downregulates H4K16ac, inhibiting proliferation of human colon cancer (HCT116) cells (196). Furthermore, through the role of HATs in DNA damage repair, several novel HAT inhibitors sensitize cancer cells to the cytotoxic effects of radiation therapy, suggesting their potential as adjuvants in this context (197,198). However, there are few reports on selective inhibitors targeting members of this family.

BRD Inhibitors
It is common for both histone acetylation and BRDs to become dysregulated in cancer. Current BRD inhibitors (e.g., isoxazoles, purines, quinolinones, tetrahydroquinolines, naphthyridines, and acetylated lysine analogs) exhibit high affinity and specificity for the BET bromine domain (199). Both I-BET 151 and I-BET 762 down-regulate c-Myc transcription, result in inhibition of myeloma cell proliferation (177). Moreover, I-BET 762 suppresses pancreatic cancer cell proliferation (178), and I-BET 762 inhibits breast and lung cancer cell proliferation through cell growth arrest and immune modulation (200).
Whereas another BRD inhibitor JO1, by competing with histone acetylated residues, releases BRD4 from chromatin, thereby modulating RNA-Pol II activity to regulate the transcription of key cancer-associated genes (201). In addition, JQ1 decreases the acetylation level and activity of mutant p53, inducing cell growth arrest and subsequent senescence in HNSCC (202). OTX015 (MK-8628, birabresib), one of BRD and extra-terminal domain inhibitors, exhibits antitumor activity in medulloblastoma, B-cell lymphoma, and lung cancer (179,184,203). In addition, BET inhibitors such like PLX51107 and NHWD-870 have been identified the activity of tumor proliferation suppression (204,205). By targeting the interaction of BRDs and acetylated lysine residues on histone, BRD inhibitors modulate chromosome structure and cancerassociated gene expression including c-Myc.

CONCLUSIONS AND PERSPECTIVES
Altered histone acetylation-one of the earliest-identified and best-studied epigenetic modifications-is associated with tumorigenesis and tumor progression. Aberrant acetylation profiles are present across various cancer cells, tissues, and types. Given that dynamic histone acetylation/deacetylation is regulated by HDACs, HATs, and BRDs, many small molecules and novel synthesized compounds targeting enzyme catalytic activity or BRD/histone interaction are under investigation for their anti-cancer therapeutic potential (Figure 3). While several agents are already FDA-approved for clinical use, many more are undergoing clinical trials, and additional novel agents are being developed and tested. Indeed, the full clinical therapeutic scope and commercial value of such agents in the field of oncology is only just emerging.

AUTHOR CONTRIBUTIONS
ZQ and DL designed the review. DW, YQ, ZQ, YJ, and DL contributed to manuscript preparation. DW and YQ contributed equally. All authors contributed to the article and approved the submitted version.

ACKNOWLEDGMENTS
We would like to thank Editage (www.editage.cn) for English language editing; Miss Mo Bai for figures editing.