Hybrid inhibitors of DNA and HDACs remarkably enhance cytotoxicity in leukaemia cells

Abstract Chlorambucil is a nitrogen mustard-based DNA alkylating drug, which is widely used as a front-line treatment of chronic lymphocytic leukaemia (CLL). Despite its widespread application and success for the initial treatment of leukaemia, a majority of patients eventually develop acquired resistance to chlorambucil. In this regard, we have designed and synthesised a novel hybrid molecule, chloram-HDi that simultaneously impairs DNA and HDAC enzymes. Chloram-HDi efficiently inhibits the proliferation of HL-60 and U937 leukaemia cells with GI50 values of 1.24 µM and 1.75 µM, whereas chlorambucil exhibits GI50 values of 21.1 µM and 37.7 µM against HL-60 and U937 leukaemia cells, respectively. The mechanism behind its remarkably enhanced cytotoxicity is that chloram-HDi not only causes a significant DNA damage of leukaemia cells but also downregulates DNA repair protein, Rad52, resulting in the escalation of its DNA-damaging effect. Furthermore, chloram-HDi inhibits HDAC enzymes to induce the acetylation of α-tubulin and histone H3.


Introduction
Genotoxic drugs represent an important class of chemotherapy and constitute a major treatment modality of human cancers. Genotoxic drugs such as cisplatin, busulfan, cyclophosphamide, chlorambucil, and temozolomide, cause various types of DNA damage, exerting their anticancer activity ( Figure 1). Among many different DNA lesions resulting from genotoxic drugs, doublestrand breaks (DSBs) are the most deleterious and a single irreparable DSB is sufficient to induce apoptosis 1 . Despite such potent lethality of genotoxic drugs, cancer cells often mitigate DNA damage by their intrinsic DNA repair capability, render drug resistance and ultimately lead to treatment failure. DNA DSBs are repaired by two distinct pathways, non-homologous end joining (NHEJ) and homologous recombination (HR). NHEJ allows for direct ligation of the DSB ends in the G 0 /G 1 phase of the cell cycle, involving KU70/80, DNAPK and DNA ligase IV, while HR utilises homologous DNA sequences as a template accurately to restore the genomic sequence in the S/G 2 phase of the cell cycle, mediated by BRCA1, BRCA2 and Rad52 2,3 . In eukaryotic cells, DNA damage triggers a sophisticated network of DNA damage response (DDR), which includes a sensing of DNA damage, the assembly of DNA repair factors, cell cycle transit arrest, and the activation of DNA repair for maintaining genomic integrity 4,5 . Conversely, dysregulation of DDR causes genomic instability, leading to activation of cell death pathway. The physiological importance of DDR is clearly illustrated by the severe cancer susceptibility of inherited defects in DDR factors, such as BRCA1 and BRCA2 [6][7][8] . Consequently, drugs that interfere DNA damage response (DDR) network, such as PARP1, ATM, and DNAPK, have been intensely investigated, resulting in the first clinical approval of PARP1 inhibitor, olaparib (tradename; Lynparza) as an anticancer drug in 2014 9,10 .
Chlorambucil is a bifunctional DNA alkylating drug, belonging to a member of the nitrogen mustard. First synthesised in 1953, chlorambucil still remains one of the front-line treatment of chronic lymphocytic leukaemia (CLL) and malignant lymphomas [11][12][13] . The nitrogen mustards share a common mechanism of action stemming from the presence of N,N-bis(2-chloroehtyl)amine moiety. Owing to the high electrophilic reactivity of the nitrogen mustard moiety, chlorambucil readily reacts with separate DNA bases, forms DNA interstrand crosslinks and causes DSBs 14 . In this regard, exposure of chlorambucil promotes apoptotic cell death of human cancers via the accumulation of persistent DNA damage.
Although approximately 60-80% of patients respond to chlormabucil for years, eventually all patients become resistant to this drug 11,15 . The mechanism of chlorambucil resistance is poorly understood. However, there are several reports suggesting that DNA repair system including homologous recombination (HR) are the major culprit to deactivate its therapeutic potency [15][16][17] . Accordingly, the occurrence of drug resistance is a serious impediment to the successful treatment of chronic lymphocytic leukaemia (CLL) and malignant lymphomas with chlorambucil.
HDAC inhibitors are grouped structurally into four classes, hydroxamic acids, cyclic tetrapeptides, benzamides, and shortchain fatty acids 20,21 . Most classes of HDAC inhibitors contain the common pharmacophore, which is composed of three distinct domains, a zinc-binding domain, a linker domain, and a cap domain 37,38 . The zinc-binding domain chelates the catalytic Zn 2þ ion in the active site, which is critical for catalytic function of HDACs, the cap domain is a surface recognition group that interacts with the entrance of the active site pocket, and the linker domain connects the cap domain to the zinc-binding domain.

Drug design and synthesis
Recently, it has been reported that the inhibition of HDAC enzymes suppress DNA repair machinery and sensitises cancers to genotoxic drugs 39,40 . Based on this, we determined to develop small molecule inhibitors, which could simultaneously attack DNA and HDACs. To this end, a novel hybrid, chloram-HDi (3) was designed by coupling chlorambucil with hydroxamic acid as a zinc-binding group, anticipating that chloram-HDi (3) could inhibit HDACs function by chelating the catalytic Zn 2þ ion in the active site of HDACs without losing the inherent DNA damaging capability of chlrambucil. We hypothesised that chloram-HDi (3) impaired DNA HR repair system via inactivation of HDACs, while nitrogen mustard moiety of 3 caused DSBs by alkylating DNA, and accumulated DNA damage. Besides, the fact that DNA and HDACs mostly exist in a same nuclear compartment, makes them more prone to dual inhibition by chloram-HDi (3), circumventing difficult pharmacokinetics of dual targeting. During our investigation of novel hybrid DNA-HDACs inhibitors, Yuan's group reported a DNA-HDAC inhibitor by attaching a benzamide zinc-binding group to chlorambucil. However, their inhibitor did not show good HDACs inhibitory activities as well as potent anti-proliferative activities against cancer cell lines 41 .
We first pursued the synthesis of chloram-HDi (3) from chlorambucil (1), which was equipped with hydroxamic acid moiety as zinc-binding group (ZBG), as shown in Scheme 1. To do so, amide coupling reaction of chlorambucil (1) with NH 2 OTHP was carried out in the presence of EDC, HOBt, and DIPEA in DMF to afford compound 2. Subsequent cleavage of THP group with 1 N HCl successfully provided compound 3 in 34% yield for two steps.
We next embarked on synthesis of compound 7 and 8a-b, which did not have electrophilic chloro group unlike chloram-HDi (Scheme 2). The synthesis of 7 and 8a-b commenced with the esterification of carboxylic acid 4 in the presence of sulphuric acid in methanol to give ester 5 in 86% yield. Methyl Ester 5 was then alkylated with ethyl iodide or propyl iodide to afford 6a-b in 75-76% yield. Finally, methyl ester 5 and 6a-b were treated with hydroxylamine in the presence of potassium hydroxide in methanol to afford 7 and 8a-b in 40-46% yield.

Biological evaluations of compounds
Upon completion of synthesis, compound 1, 3, 7, and 8a-b were evaluated for their anti-proliferative activity against various cancer cell lines, including human acute myeloid leukaemia (AML) cell lines (U-937 and HL-60), human breast cancer cell lines (MDA-MB-231 and MCF-7), and a human ovarian cancer cell line (A2780). Each cancer cell line was treated with the indicated compound for 3 days and its anti-proliferative effect on cancer cell lines was measure using MTS colorimetric assay. As shown in Table 1, chloram-HDi (3) displayed greater anti-proliferative activities than its parent drug, chlorambucil (1). Interestingly, chloram-HDi exerted the most potent anti-proliferative activity against AML cell lines, U-937 and HL-60 with GI 50 values of 1.75 mM and 1.24 mM respectively, which were 17-21 fold lower than GI 50 values of chlorambucil. In contrast, chloram-HDi showed relatively poor anti-proliferative activity against human breast cancer cell lines, MDA-MB-231 and MCF-7, in that its GI 50 values against MDA-MB-231 and MCF-7 were 95.9 mM and 244.9 mM, respectively. Compound 7 and 8a-b, which are lack of nitrogen mustard moiety furnished either poor or mediocre anti-proliferative activities against tested cancer cell lines, indicating that nitrogen mustard moiety is an essential warhead in this anticancer drug design.
We next determined the effect of chlorambucil and chloram-HDi on the growth inhibition of various cancer cell lines (HL60, U937, acute myeloid leukaemia; A549, non-small cell lung carcinoma; H1975, gefitinib-resistant non-small cell lung carcinoma (NSCLC); MCF-7/ADR, multidrug-resistant breast adenocarcinoma; MDA-MD-231, mammary carcinoma; U266, multiple myeloma; A2780, ovarian carcinoma). As shown in Figure 3, chloram-HDi displayed significantly enhanced anti-proliferative activities against various cancer cell lines, compared to its parent drug, chlorambucil. Especially, the GI 50 values (concentration in 50% growth inhibition) of chloram-HDi (1.24 mM and 1.75 mM) against HL-60 and U937 cancer cell lines were obviously much lower than those of chlorambucil (21.1 mM and 37.7 mM), indicating that chloram-HDi was superior to the clinical drug, chlorambucil in cellular potency ( Table 2).
The comet assay is a sensitive and rapid technique for determining the amount of DNA damage in a single cell 42 . This assay directly allows the microscopic observation of the "comet tail" that is associated with the damaged DNA content. Therefore, we investigated the impact of chloram-HDi and chlorambucil on the integrity of DNA in HL-60 cells using the comet assay ( Figure 4). We first treated HL-60 cells with chloram-HDi or chlorambucil for 24 h and then measured DNA damage using the comet assay. The assay indicated that chloram-HDi (10 mM) caused more noticeable DNA damage than chlorambucil (10 mM), in that % DNA content  We next investigated the precise mechanism of action behind the increased DNA-damaging effect of chloram-HDi compared with its parent drug, chlorambucil. Hence, we first treated HL-60 cells with the indicated concentrations of chlorambucil and chloram-HDi for 24 h and measured the expression levels of Rad52, H2AX, c-H2AX, and b-actin. c-H2AX is a well-established hallmark of DNA double-strand breaks and Rad52 is an important protein in the homologous recombination repair (HRR) of DNA double-strand breaks [43][44][45] . As shown in Figure 5(A), the treatment of cells with chloram-HDi more significantly increased the expression of c-H2AX than chlorambucil in a dose-dependent manner, indicating that chloram-HDi effectively induced more DNA damage than chlorambucil, which was consistent with the result that we observed in the comet assay. Furthermore, chlrom-HDi downregulated the protein level of Rad52 in a dose-dependent manner. It has been reported that the inhibition of HDACs leads to the reduction of Rad52 protein level. Nevertheless, the reduction of the DNA repair protein, Rad52 contributed to the increased DNA-damaging effect of chloram-HDi, offering a distinct advantage over chlorambucil.
To assess the comparative effect of chlorambucil and chloram-HDi on the acetylation status of a-tubulin and histone H3, we also measured the expression levels of a-tubulin, Ac-a-tubulin, histone H3, and Ac-histone H3 ( Figure 5(B)). a-Tubulin and histone H3 are well documented substrates of HDAC6 and HDAC1 46,47 . Accordingly, the inhibition of HDAC6 and HDAC1 enzymes epigenetically induced the acetylation of a-tubulin and histone H3, respectively. The treatment of HL-60 cells with chloram-HDi dosedependently increased the acetylation of a-tubulin and histone H3, while total amount of histone H3 remained unchanged. It is  interesting to note that chloram-HDi decreased the expression level of a-tubulin in a dose-dependent manner. In contrast, the treatment of HL-60 cells with chlorambucil (5 and 10 mM) failed to induce the acetylation of a-tubulin and histone H3, whereas 30 mM of chlorambucil resulted in little acetylation of a-tubulin and histone H3. Taken together, the result undoubtedly demonstrated that chloram-HDi efficiently inhibited the function of HDAC6 and HDAC1 enzymes and promoted the cellular acetylation status of a-tubulin and histone H3. The activation of apoptotic signals can suppress the proliferation of cancer cells, as it is regulated by the avoidance of the   apoptotic mechanism. To determine if the anti-proliferative effect of chloram-HDi on HL-60 cells is associated with the induction of apoptosis, we therefore measured the cleavage of apoptotic biomarkers, including PARP, caspase 3, and caspase 8. As shown in Figure 5(C), the exposure of HL-60 cells with chloram-HDi significantly promoted the cleavage of PARP, caspase 3, and caspase 8 in a dose-dependent manner. Conversely, when HL-60 cells were treated with chlorambucil, the administration of 30 mM concentration was required to induce the cleavage of PARP, caspase 3, and caspase 8, clearly indicating that chloram-HDi more potently led to the apoptotic cancer cells death than chlorambucil.
To determine whether chloram-HDi affected the cell cycle, we evaluated the effect of chloram-HDi on the cell cycle distribution of HL-60 cells using flow cytometry. After exposed to the indicated concentrations of chlorambucil and chloram-HDi for 24 h, the cell cycle distribution of HL-60 cells was measured. As shown in Figure 6, the exposure of HL-60 cells with chloram-HDi  remarkably promoted the cell cycle arrest in the G 2 /M phases, compared with the DMSO control and the reference drug chlorambucil in that the prominence of G 2 /M cell cycle arrest is a reported feature of the nitrogen mustards 48,49 . The result indicated that chloram-HDi caused the DNA double-strand breaks of HL-60 cells, leading to the activation of the homologous recombination repair (HRR) pathway through G 2 /M cell cycle arrest. Conversely, another DNA DSB repair pathway, Non-homologous end joining (NHEJ) is reported to primarily occur in G 0 /G 1 phase 2 . The treatment of HL-60 cells with 0.1 mM of chloram-HDi increased G 2 /M fraction from 21.8% to 30.8%, while the treatment of HL-60 cells with 1 mM of chlorambucil led to 32.2% G 2 /M arrest of HL-60 cells, which indicated that chloram-HDi was more efficacious than chlorambucil in the cell cycle arrest. When the cells were treated with the increased concentration of chloram-HDi, the G 2 /M arrest was more remarkable (52% G 2 /M fraction at 0.5 mM and 62.4% G 2 / M fraction at 1 mM). Conversely, the treatment of cells with chloram-HDi resulted in a reduction of the cell population at the G 0 /G 1 phases in a dose-dependent manner. It is also interesting to note that chloram-HDi increased the cell population at sub G 0 /G 1 phases in a dose-dependent manner. Taken together, the results suggested that chloram-HDi effectively induced the cell cycle arrest at the G 2 /M phases in HL-60 cells in a dose-dependent manner.
We next investigated the inhibitory activity of chloram-HDi against HDAC1, 3, 6, and 7 isoforms, in that the clinically approved HDAC inhibitor, SAHA was used as a reference drug (Table 3). Although chloram-HDi displayed less efficacious inhibitory activities than the reference drug SAHA against all tested HDACs enzymes, chloram-HDi still displayed modest inhibitory activities against HDACs enzymes. Interestingly, chloram-HDi more selectively inhibited HDAC6 enzyme with an IC 50 value of 694 nM than other HDAC enzymes (IC 50 of HDAC1¼4.497 mM, IC 50 of HDAC3¼6.201 mM, and IC 50 of HDAC7¼74.568 mM).
To determine whether chloram-HDi was toxic to the normal cells, we isolated primary human peripheral blood mononuclear cells (PBMCs) and examined the cytotoxicity of chloram-HDi on normal human PBMCs. Cells were first treated with various concentrations of chloram-HDi for 3 days and cell viability was measured using the colorimetric MTS assay. As shown in Figure 7, the result indicated that chloram-HDi was not toxic to normal human PBMCs up to 10 mM concentration, suggesting that chloram-HDi selectively targets the malignant leukaemia cell lines HL-60 (IC 50 ¼ 1.24 mM) and U937 (IC 50 ¼ 1.75 mM) over the normal PBMCs (IC 50 > 10 mM).
Cytochrome P450 (P450) enzymes have emerged as an important determinant in the occurrence of drug interactions that can lead to adverse drug reactions. Therefore, we investigated the effect of chloram-HDi (3) on the catalytic activities of clinically significant human P450s such as 1A2, 2C9, 2C19, 2D6, and 3A in human liver microsomes 50,51 . To do so, the inhibitory potency of chloram-HDi (3) was determined with cytochrome P450 assays in the absence and the presence of chloram-HDi (3) up to 50 mM final concentration using pooled human liver microsomes ( Table 4). The assay indicated that chloram-HDi (3) showed weak inhibitory effect (>50 mM) against five P450 isoforms. These findings suggest that clinical interactions between chloram-HDi (3) and substrate drugs of five P450 isoforms would not be expected.

Molecular modelling of chloram-HDi (3)
To investigate the binding mode of chloram-HDi (3), in silico docking studies were performed with HDAC6 enzymes (Figure 8). Modelling chloram-HDi (3) in the substrate-binding pocket of HDAC6 (PBD code: 5EF7) indicated that chloram-HDi (3) effectively bound into a deep substrate-binding pocket of HDAC6 (Figure 8(A)). The middle phenylpropyl group of chloram-HDi (3) well occupied the lipophilic channel and formed favourable p-p stacking interactions with F643 and F583 residues (Figure 8(B)). The hydroxamate C¼O and OH groups of chloram-HDi (3) chelated the active site Zn 2þ ion in a bidentate manner and formed additional hydrogen bond interactions with Y745, D612, H573, and G743 residues. In contrast, the nitrogen mustard moiety of chloram-HDi (3) was located in the rim of the substrate-binding pocket, participating in Van der Waals interactions with the hydrophobic patches of the rim, composed of F643 and F583 residues.

General methods and materials
Unless otherwise noted, all reactions were performed under argon atmosphere in oven-dried glassware. All purchased reagents and solvents were used without further purification. Thin layer chromatography (TLC) was carried out using Merck silica gel 60 F 254 plates. TLC plates were visualised using a combination of UV light and iodine staining. Column chromatography was conducted under medium pressure on silica (Merck Silica Gel 40-63 mm) or performed by MPLC (Biotage Isolera One instrument) on a silica column (Biotage SNAP HP-Sil) or C18 column (Biotage SNAP Ultra C18). NMR analyses were carried out using a JNM-ECZ500R (500 MHz) manufactured by Jeol resonance. 1 H and 13 C NMR chemical shifts are reported in parts per million (ppm). The deuterium lock signal of the sample solvent was used as a reference, and coupling constants (J) are given in hertz (Hz). The splitting pattern abbreviations are as follows: s, singlet; d, doublet; t, triplet; q, quartette; dd, doublet of doublets; m, multiplet. The purity of all tested compounds was confirmed to be higher than 95% by HPLC analysis performed with a dual pump Shimadzu LC-6AD system equipped with VP-ODS C18 column (4.6 mm Â 250 mm, 5 mm, Shimadzu).

4-(4-Aminophenyl)-N-hydroxybutanamide (7)
Hydroxylamine hydrochloride (0.95 g, 13.7 mmol) in methanol (4 ml) was added to a solution of potassium hydroxide (0.78 g, 13.7 mmol) in methanol (4 ml) at 0 C. The mixture was stirred for 15 min at 0 C and the precipitated potassium chloride was removed and the filtrate was used as such; To a solution of the compound 5 (0.13 g, 0.34 mmol) in tetrahydrofuran (4 ml) was added to freshly prepared hydroxylamine at 0 C and stirred at the same temperature for 2 h. The mixture was neutralised with 3 N HCl to pH 7 and extracted with ethyl acetate. The organic layer was dried over Na 2 SO 4, concentrated under reduced pressure and purified by HPLC to afford compound 7 in 40% yield. R f ¼ 0.27 (9:1 ethyl acetate: methanol). 1

4-(4-(Diethylamino)phenyl)-N-hydroxybutanamide (8a)
Hydroxylamine hydrochloride (4.11 g, 59.19 mmol) in methanol (5 ml) was added to a solution of potassium hydroxide (3.32 g, 59.19 mmol) in methanol (5 ml) at 0 C. The mixture was stirred for 15 min at 0 C and the precipitated potassium chloride was removed and the filtrate was used as such; To a solution of the compound 6a (0.33 g, 1.32 mmol) in tetrahydrofuran (10 ml) was added to freshly prepared hydroxylamine at 0 C and stirred at the same temperature for 2 h. The mixture was neutralised with acetic acid to pH 7 and extracted with ethyl acetate.

4-(4-(Dipropylamino)phenyl)-N-hydroxybutanamide (8 b)
Hydroxylamine hydrochloride (2.25 g, 32.44 mmol) in methanol (5 ml) was added to a solution of potassium hydroxide (1.82 g, 32.44 mmol) in methanol (5 ml) at 0 C. The mixture was stirred for 15 min at 0 C and the precipitated potassium chloride was removed and the filtrate was used as such; To a solution of the compound 6b (0.20 g, 0.72 mmol) in tetrahydrofuran (10 ml) was added to freshly prepared hydroxylamine at 0 C and stirred at the same temperature for 2 h. The mixture was neutralised with acetic acid to pH 7 and extracted with ethyl acetate.

Cell proliferation assay
Cells were seeded at a clear 96-well plate, the medium volume was brought to 100 mL, and cells were allowed to attach overnight. Various concentrations of compounds (1, 3, 7, 8a, 8b or DMSO) were added to the wells. Cells were then incubated at 37 C for 3 days. Cell viability was determined using the Promega Cell Titre 96 Aqueous One solution cell proliferation assay.
Absorbance at 490 nm and 690 nm as reference wavelength was read on Tecan Infinite F200 Pro plate reader, and values were expressed as percent of absorbance from cells incubated in DMSO alone. was added to the well, followed by various concentrations of compounds (5 mL) or SAHA (5 mL) as a positive control, and then the resulting mixture was incubated at 37 C for 30 min. After the incubation, 50 mL of undiluted 2Â HDAC developer was added to each well. After the mixture was incubated at rt for 15 min, fluorescence intensity was measured using a microplate reader at 360 nm excitation and 460 nm emission wavelengths.

Molecular docking
In silico docking of chloram-HDi with the 3 D coordinates of the X-ray crystal structure of HDAC6 (PDB code: 5EF7) was accomplished using the AutoDock 4.2 programme downloaded from the Molecular Graphics Laboratory of the Scripps Research Institute. In the docking experiments carried out, water was removed from the 3 D X-ray coordinates while Gasteiger charges were placed on the X-ray structures of HDAC6 along with chloram-HDi using tools from the AutoDock suite. A grid box centred on the substrate binding pocket of HDAC6 enzyme with definitions of 60 Â 60 Â 60 points and 0.375 Å spacing was chosen for ligand docking experiments. The docking parameters consisted of setting the population size to 150, the number of generations to 27,000, and the number of evaluations to 2,500,000, while the number of docking runs was set to 100 with a cut-off of 1 Å for the root-meansquare tolerance for the grouping of each docking run. The docking model of HDAC6 with chloram-HDi was depicted in Figure 8 and rendering of the picture was generated using PyMol software (DeLanoScientific).

Conclusion
Although chlorambucil remains one of the front-line treatment of CLL and malignant lymphomas, the occurrence of drug resistance is still a major hurdle to the successful cancer treatment. Among the mechanisms of drug resistance, it has been suggested that the activation of DNA repair machinery such as homologous recombination (HR) plays a critical role in the drug resistance of chlorambucil 17,52 .
In this regard, we designed a novel hybrid molecule, chloram-HDi that simultaneously impaired DNA and HDAC enzymes ( Figure  9). Chloram-HDi caused a remarkable DNA damage of HL-60 cells, in that % DNA content in the tail was xx% in comet assay and the DNA damage marker, c-H2AX was significantly upregulated in western blot analysis. Chloram-HDi also caused the reduction of the DNA repair protein, Rad52, magnifying its DNA-damaging effect by disrupting Rad52-mediated homologous recombination (HR). Furthermore, chloram-HDi inhibited HDAC enzymes to induce the acetylation of a-tubulin and histone H3. It has been reported that the inhibition of HDACs induces the acetylation of Hsp90, which impairs the chaperone activity of Hsp90 and consequently causes the depletion of its client protein, Rad52 53 . As a result, chloram-HDi efficiently promoted the cell cycle arrest at the G 2 /M phases and resulted in apoptotic cell death, evidenced by the cleavage of PARP, caspase 3, and caspase 8. Overall, these findings firmly supported that chloram-HDi could serve as a potential drug candidate for the treatment of leukaemia, warranting further studies on in vivo evaluation as well as ADMET profiles. These studies will be reported in due course.

Disclosure statement
No potential conflict of interest was reported by the author(s).