Key structural requirements of benzamide derivatives for histone deacetylase inhibition: design, synthesis and biological evaluation

Abstract

Background: Histone deacetylase inhibitors (HDACIs) are important as anticancer agents. Objective: This study aimed to investigate some key structural features of HDACIs via the design, synthesis and biological evaluation of novel benzamide-based derivatives. Methods: Novel structures, designed using a molecular modification approach, were synthesized and biologically evaluated. Results: The results indicated that a subset of molecules with CH₃/NH₂ at R₂ position possess selective antiproliferative activity. However, only those with an NH₂ group showed HDACI activity. Importantly, the shorter the molecule length, the stronger HDACI. Among all, 7j was the most potent HDAC1-3 inhibitor and antiproliferative compound. Conclusion: The results of the present investigation could provide valuable structural knowledge applicable for the development of the HDACIs and benzamide-based antiproliferative agents in the future.

Plain language summary

Summary points.

• Novel benzamide-based structures were designed by modifying the length of molecules and varying substitutions on the terminal benzene rings of the core scaffold.

• The designed compounds were synthesized.

• Class I histone deacetylase inhibitor (HDACI) potencies of designed compounds were evaluated. According to the results, the compounds with an NH₂ group in the R₂ position and a shorter molecular length are potent HDACIs.

• The antiproliferative activity of the designed compounds was evaluated on MCF-7 and T₄7D cells. The results showed that molecules with either CH₃ or NH₂ at the R₂ position possessed selective antiproliferative activity against cancer cells.

• Compound 7j was the most potent HDACI and antiproliferative in the synthesized compounds of this study.

• Two new potent antiproliferative molecules were proposed using a quantitative structure–activity relationship study.

• Docking of synthesized compounds on HDAC2 revealed that 7j establishes more interactions with the enzyme, which is almost identical to those shown experimentally for 2-aminobenzamide.

• Analyses of the root-mean-square deviation and potential energy plots during molecular dynamics simulation for the complexes of HDAC2 with those ligands showed stability for the proposed docked models.

• Drug likeness features of 7g and 7j compounds fit with Lipinski rules.

Among the most prevalent types of cancer, breast cancer stands as the second leading cause of cancer-related mortality of women in the USA. Cancer is caused by the uncontrolled division of cells and their rapid proliferation due to both genetic and epigenetic changes [1,2]. Cancer treatment strategies include chemotherapy, hormone therapy, radiation therapy and surgery [3]. However, there are still problems such as relapse, metastasis of breast cancer and resistance to available anticancer drugs that lead researchers to find novel potent chemotherapeutics. Epigenetic post-translational modification of histone protein is important in different biological processes for the expression or suppression of target genes. One of the most important histone modifications is histone acetylation and deacetylation [4]. HAT and HDAC balance the acetylation and deacetylation levels of histone protein [5]. The activities of HDAC enzymes lead to more tightly wrapped DNA around histones, which reduces gene transcription [6,7]. Members of the HDAC family of enzymes are categorized into four classes, namely class I, II and IV, which are Zn²⁺-dependent enzymes, and class III, which is a NAD⁺-dependent enzyme. Class I includes HDAC1–3 and 8; class II is divided into two groups consisting of class IIa (HDAC4, 5, 7 and 9); and class IIb (HDAC6 and 10), and class IV has only one member, called HDAC11. Class III includes sirtuins (SIRTs), including SIRT1–7 [8,9].

HDAC inhibitors (HDACIs) induce cancer cell death by various mechanisms, including activation of caspase-mediated apoptosis, autophagy and upregulation of p53 and p21 protein levels [10]. HDACIs comprise three main segments including a cap group, linker and zinc-binding group (ZBG) to interact with rim, tunnel of binding pocket (wall) and ZBD (zinc binding domain), respectively. The first generation of HDACIs belongs to compounds such as vorinostat, belinostat and panobinostat, which have hydroxamic acid in their ZBG. These inhibitors have received approval from the US FDA for the treatment of (cutaneous) peripheral T cells, but are associated with low potency and more side effects. For example, vorinostat is a pan inhibitor and has low potency in solid tumors and poor pharmacokinetics (rapid elimination) [7,11]. The two main strategies implemented to address these issues include the development of isoform-selective HDACIs and multitarget inhibitors.

To design isoform-selective inhibitors, it is essential to identify the precise structure of each isoform or class of enzymes. For instance, there is a subpocket at the end of the active site channel of HDAC1-3 enzymes called the foot-pocket. Therefore, designing HDACIs containing a bulky group such as an aromatic group in ZBG segments of the molecules may lead to HDAC1–3-selective inhibitors. In addition, the discovery of inhibitors with alternative ZBGs such as a benzamide group or ZBG-less HDACIs could be useful structural modifications for eliminating/reducing the off-target effects of first-generation inhibitors [12,13]. The second approach is designing multitargeted inhibitors that include two or three different structural segments to interact with more than one therapeutic target. These types of HDACIs are achieved by replacing or combining the cap group of an inhibitor with the structural moieties of other biologically active compounds capable of affecting related targets [14].

In addition, numerus studies are finding novel HDACIs by modifying the ZBG, linker and cap group regions of existing structures to improve their potency and pharmacokinetic profiles [15,16]. Benzamide-based HDACIs such as entinostat (MS-275) are class I selective inhibitors that have relatively better pharmacokinetics and high efficacy in solid tumors compared with pan inhibitors. In breast cancer, HDAC1-3 isoforms are overexpressed so that their selective inhibitions can be regarded as a rational therapeutic strategy. Entinostat, an orally administered inhibitor, is in a phase III clinical study (ENCORE 301) in combination with exemestane for the treatment of breast cancer [17]. Owing to the promising results of class I selective HDACIs, the development of new inhibitors could lead to novel antibreast cancer agents with low side effects.

Drug discovery is one of the most complicated, time- and cost-consuming processes in pharmaceutical research [18]. The first stage in drug discovery is finding leads and optimizing them to develop clinically useful drug candidates. Intuition, cognition and the creativity of medicinal chemists play important roles in drug design and discovery. As previously reported [19], benzamide-based compounds 7b and 7e, which were designed using an in silico structure/ligand-based design strategy, have shown encouraging HDAC1–3 inhibition. In the present study, some more benzamide-based derivatives of 7b were designed using a molecular modification approach. We have mainly focused on the length of molecules and R₁ and R₂ substitutions on terminal benzene rings A and C, respectively (Figure 1). The designed compounds were synthesized and their inhibitory potency on class I HDAC enzymes were evaluated. Moreover, the antiproliferative activities of the newly synthesized compounds on MCF-7 and T47D cancer cells were assessed, and their modes of interaction with the HDAC enzyme were investigated using molecular docking calculations.

Figure 1.

Structures of entinostat, 7b, 7e and the designed molecules.

ZBG: Zinc binding group.

Materials & methods

Chemistry

Melting points were determined using an Electrothermal 9200 melting point apparatus. IR spectra (KBr discs) were recorded with an FT-IR 8400S from Shimadzu (CA, USA). Nuclear magnetic resonance (NMR) spectra were recorded in DMSD-d6 using Bruker 400 MHz and 500 MHz spectrometers (Bruker Bioscience, MA, USA) with tetramethylsilane as an internal standard. Elemental (CHN) analyses were performed with the Costech Elemental Combustion System CHNS-O (ECS 4010; Costech Analytical, CA, USA). Mass spectra were recorded using the 5973 Network Mass Spectrometer System (Agilent Technologies, Inc., CA, USA). All commercially available chemicals and reagents were purchased and used without further purification.

General procedure for the preparation of benzoic acid derivatives (4a–g)

To 1 eq of corresponding carboxylic acid 1a–1g was added an excess amount of thionyl chloride (2–3 eq) and the reaction mixture was refluxed overnight. Excess thionyl chloride was removed under reduced pressure to generate the corresponding acid chloride 2a–2g quantitatively. Then, the produced acid chloride 2a–2g (1.2 mmol) was added dropwise to a stirred solution of 4-(aminomethyl) benzoic acid 3 (1 mmol) in a 1M solution of potassium hydroxide. Stirring was continued at room temperature for 5 h. After completion of the reaction, the pH of the reaction mixture was adjusted to 3–4 using hydrochloric acid (37%). The precipitate was filtered, washed with water and dried to obtain the key intermediates 4a–4g.

General procedure for preparation of benzamide derivatives (7a–l)

To the suspension of benzoic acid derivatives 4a–4g (1 mmol) in dry dichloromethane (20 ml) was added thionyl chloride (1.2 mmol) and a catalytic amount of dry dimethylformamide (DMF). The reaction mixture was stirred at reflux temperature for 5h. After the production of benzoyl chlorides 5a–5g, excess thionyl chloride and DMF were evaporated using a rotary evaporator under reduced pressure at room temperature. Finally, an appropriate amount of benzoyl chloride 5a–5g was added dropwise to the solution of ortho-phenylene diamine 6 (1.2 mmol) and triethylamine (1 mmol) in dry dichloromethane as solvent (20 ml) in an ice bath. Then, the temperature was increased to room temperature and the reaction mixture was stirred at reflux temperature overnight. The progress of the reaction was monitored by TLC. The reaction mixture was washed once with water, and then the organic phase was dried with anhydrous sodium sulfate. The organic solvent was evaporated and the obtained solid was recrystallized from ethyl acetate and ethanol to generate products 7a–7l.

Enzymatic in vitro HDACI assay

Recombinant human HDAC1, HDAC2 and HDAC3/NCOR1 were purchased from ENZO Life Sciences AG (Lausen, Switzerland). Recombinant human HDAC8 was produced as described in [20]. In vitro testing on recombinant HDAC1–3 was performed with a fluorogenic peptide derived from p53 (Ac-RHKK[acetyl]-AMC) [21]. Measurements were performed in assay buffer (50 mM Hepes, 150 mM NaCl, 5 mM MgCl₂, 1 mM TCEP and 0.2 mg/ml BSA, pH 7.4 adjusted with NaOH) at 37°C. Inhibitors at different concentrations were incubated with 10 nM HDAC1, 3 nM HDAC2 or 3 nM HDAC3 (final concentration) for at least 5 min. The reaction was started with the addition of the fluorogenic substrate (20 μM final concentration) and incubated for 30 min for HDAC2 and HDAC3 and 90 min for HDAC1. The reaction was stopped with a solution of 1 mg/ml trypsin and 20 μM SAHA in 1 mM HCl and incubated for 1 h at 37°C. The fluorescence intensity was recorded with an Envision 2104 Multilabel Plate Reader (PerkinElmer, MA, USA) with an excitation wavelength of 380 ± 8 nm and an emission wavelength of 430 ± 8 nm. The received fluorescence intensities were normalized with uninhibited reaction as 100% and the reaction without enzyme as 0%. A nonlinear regression analysis was carried out to determine the half-maximal inhibitory concentration (IC₅₀) value. The enzyme inhibition of HDAC8 was determined using a reported homogenous fluorescence assay [22]. The enzymes were incubated for 90 min at 37°C, with the fluorogenic substrate ZMAL (Z [Ac]Lys-AMC) at a concentration of 10.5 μM and at increasing concentrations of inhibitors. Fluorescence intensity was measured at an excitation wavelength of 390 nm and an emission wavelength of 460 nm in a microtiter plate reader (BMG POLARstar, Ortenberg, Germany).

In vitro antiproliferative study

The antiproliferative activities of the prepared compounds were evaluated using a standard in vitro MTT assay [23]. The breast cancer cell lines (MCF-7 and T47D cell lines) and a normal breast cell line (MCF10A) were seeded on 96-well plates at a density of 7500 per well and incubated in a humidified atmosphere containing 5% CO₂ at 37°C for 24 h. The cells were treated with the compounds and vorinostat (as the control) at final concentrations ranging from 0.1 to 50 μM for 48 h. Fresh MTT solution was added to the wells at a final concentration of 0.5 mg/ml and incubation was continued for another 4 h at 37°C. Then, 100 μl DMSO was added to each well to dissolve the formazan crystals and the optical density at 570 nm was measured. The percentage of cell viability was calculated using the following formula: % cell viability = 100 × [(A_Sample - A_Blank)/(A_Control - A_Blank)], where, A_Sample, A_Control and A_Blank refer to absorbances recorded for cells treated with the tested compound, untreated cells and blank wells, respectively. The experiments were performed in triplicate. The results of cell viabilities were fitted in sigmoidal dose–response curves using nonlinear regression analysis by GraphPad PRISM software (version 6.01) to calculate the IC₅₀ values.

Quantitative structure–activity relationship study

A quantitative structure–activity relationship (QSAR) study was performed on synthesized compounds using Free-Wilson regression analysis in MATLAB software (Version Minitab 19.2, LLC, USA). First, a matrix was prepared that included studied compounds and structural elements in its rows and columns such as the presence of different groups in R₁ and R₂, and length of inhibitor (n), respectively. The values in the matrix were 1 and 0 for the presence and absence of structural variables. The last column contained the IC₅₀ values for the antiproliferative activity of the derivatives on the MCF-7 cell line, which is used as a dependent variable in QSAR analysis. To calculate the constants of the independent variables (the contribution of each structural group), the partial least square method was used. The R² was obtained at 0.904, including eight components.

Docking study

To predict the mode of interactions between the designed ligands and the HDAC enzyme, a molecular docking study was performed. For performing the docking study, from class 1 HDACs, only one of them was selected. Since HDAC1, -2 and -3 are very similar to each other in terms of structure, sequence and function, one of them was enough to predict interactions in the docking study (please refer to this paper for more information: www.mdpi.com/1422-0067/21/22/8828). To enhance the validity of the docking setup, it would have been preferable to select one that co-crystallized with a HDACI. HDAC2 is the only class I member that is co-crystallized with SAHA, thus making it an optimal choice for the docking study. The HDAC2 protein (Protein Data Bank [PDB] ID: 4LXZ) was downloaded from Protein Data Bank (www.RCSB.org). The GOLD program (version 5.0; CCDC Inc., Cambridge, UK) was utilized for flexible docking of all designed ligands into the binding site of the enzyme running under the LINUX operating system. All atoms of the enzyme within 10 Å distance from the geometric center (X: 1.192; Y:-34.903; Z: -42.203) of the residues forming the binding site of the enzyme were included in the docking calculation. Based on the results of reproducing the experimentally known pose of SAHA into the binding site of HDAC2, the best scoring function (ASP) was selected by applying the default settings in GOLD. The docking solutions were filtered by the Shape-it program in which SAHA was selected as a reference ligand. The pose with the highest Tanimoto similarity score according to confirmation matching with SAHA was selected for the evaluation of interaction modes between ligand and the enzyme.

Molecular dynamics simulation

Following the molecular docking procedure, the compounds 7b, 7e, 7g and 7j, along with HDAC2, were subjected to a molecular dynamics (MD) simulation. This simulation was performed using the Assisted Model Building with Energy Refinement (AMBER) suite of programs operating on a Linux-based GPU work station. Initially, the input parameter files were generated using the ‘leap’ module including frcmode (force field parameters) and lib (library) files for ligands and zinc ion, followed by neutralizing the total system charge using a single Na⁺ ion. Using the ‘Sander’ module, the neutralized system was solvated and minimized by performing a 50-ps minimization step. Then the system was heated from 0 to 300°K at 50 ps followed by 50 ps of density equilibration with weak restraints on the complex, followed by 500 ps of constant pressure equilibration at 300°K. All simulations were performed under periodic boundary conditions, and using the SHAKE algorithm on hydrogen atoms, with a 2-fs time step and by applying a Langevin thermostat for temperature control. The equilibrated system was subjected to a 50 ns MD simulation. The coordinates were written out every 50,000 steps to obtain the trajectory of the simulation. MMPB/GBSA methods were used to calculate the binding free energy for the association of HDACIs (7b, 7e, 7j and 7g) to the enzyme (i.e., HDAC2 isoform).

Prediction of drug-likeness features

To evaluate the drug likeness of the studied compounds, the SwissADME web server was used to filter for the Lipinski Rule of Five, bioavailability criteria, and Ghose, Egan, Muegge and Veber filters.

Results & discussion

Design strategy

In the present study, a rational molecular modification approach was used to design new derivatives of previously introduced selective HDAC1–3 inhibitors, that is, 7b and 7e. Based on the relative activities of these compounds reported in our previous work, it was concluded that the length of the inhibitor is just as important as the presence of NH₂ group in the R₂ position [19]. The structures of entinostat, 7b, 7e and the designed molecules are shown in Figure 1. As a result, we have decided to expand our investigation on benzamide-based HDACIs by introducing two main structural modifications and evaluating their impact on enzyme inhibition. The two main emphases were the length of molecules and R₁ and R₂ substitutions on terminal benzene rings A and C, respectively. The NH₂ group is an already known optimum substitution at R₂ position; however, our preliminary and brief virtual evaluation of the molecules with H, NH₂ and CH₃ groups at this position by the SwissTargetPrediction server indicated that all designed structures targeted HDAC enzymes, as the first, second or third target. First, different R₂ groups (H, CH₃ and NH₂) with and without the ability to confer the formation of a seven-membered pseudoring by ZBG domain while interacting with zinc ion of the enzyme were examined. Second, the length of the inhibitors (n, the number of added methylene groups) was changed. We have also introduced different substituents to the R₁ position of the cap group to evaluate their effects on enzyme inhibition and interaction with the rim region of the receptor. In summary, various derivatives of 7b were synthesized by changing the length of the molecule (n), as well as incorporation of different R₁ and R₂ substituents in cap and zinc binding domains, respectively (Table 1).

Table 1.

Structural features of the designed compounds.

Compound	R1	n	R2
7a	H	1	H
7b	H	1	NH2
7c	H	1	CH3
7d	H	2	H
7e	H	2	NH2
7f	H	0	H
7g	F	0	NH2
7h	F	0	CH3
7i	Cl	0	H
7j	Me	0	NH2
7k	OMe	0	H
7l	OMe	0	CH3

Chemistry

The synthetic route employed to produce the new target compounds 7a–7l is depicted in Figure 2. Initially, the reaction between the carboxylic acids 1a–1g and an excess amount of thionyl chloride in solvent-free conditions at reflux temperature yielded corresponding acyl chlorides 2a–2g, which were subsequently transformed into the benzoic acid intermediates 4a–4g via a reaction with 4-methyl aminobenzoic acid 3. In the next step, the carboxylic acids 4a–4g were acylated via thionyl chloride (SOCl₂) in the presence of a catalytic amount of dry DMF to produce corresponding acyl chlorides 5a–5g. The reaction of 5a–5g with 6 (phenyl diamine) in the presence of triethylamine (Et₃N) in dry dichloromethane yielded compounds 7a–7l. The final derivatives obtained were purified using recrystallization in ethanol and ethyl acetate. The synthesized structures were characterized by IR, ¹H-NMR, ¹³C-NMR, Mass, CHNS and high performance thin layer liquid chromatography (HPTLC) analyses. Detailed information on characterization is available in the Supplementary Material.

Figure 2.

Synthesis route of target compounds 7a–7l.

Reagents and conditions: (A) SOCl₂ (excess), reflux; (B) 1M aqueous solution of potassium hydroxide, r.t.; (C) SOCl₂, dry dimethylformamide, dry dichloromethane (DCM), r.t.; (D) triethylamine (TEA), dry DCM, reflux.

r.t: Room temperature.

In vitro HDACI of the designed compounds

The results of enzymatic in vitro HDACI activity of the synthesized compounds against human class I HDACs (HDAC1, 2, 3 and 8) are shown in Table 2. Vorinostat, a wide-spectrum inhibitor, and entinostat, a class I selective inhibitor, were also evaluated as the reference substances. The synthesized compounds can be classified into two groups depending on the presence or absence of an amine functional group (NH₂) in the ortho position of the benzamide moieties. This part of the molecule works as a ZBG, which is one of the main pharmacophoric elements of HDACIs. Derivatives with the amine group (7b, 7e, 7g and 7j) exert more than 50% inhibition of HDAC1, HDAC2 and HDAC3 enzymes at 10 μM concentration. It is noteworthy that all compounds, as well as entinostat, are inactive against HDAC8. It was found that 7j had the highest inhibitory activity against HDAC1, HDAC2 and HDAC3 isoforms, with IC₅₀ values of 0.65, 0.78 and 1.70 μM, respectively. The inhibitory activity of 7j on these three HDAC isoforms was marginally stronger than that of entinostat (IC₅₀ of 0.93, 0.95 and 1.8 μM, respectively), which was used as the reference drug. Furthermore, entinostat shows the highest structural similarity to the synthesized derivatives with ortho aminobenzamide moieties, and hence may serve as a better reference compound for structure, interaction with the enzyme and activity comparison. The inhibition dose–response curves of 7j on the HDAC1, 2 and 3 isoforms is available in the Supplementary Material. As stated before, the presence of a -NH₂ functional group in the ZBG moiety, which forms hydrogen bonds with binding site residues, is an essential structural feature granting the observed selective HDACI potency. Similar to 7j, all other derivatives bearing 2-aminobenzamide (i.e., the ortho amino derivatives 7b, 7e and 7g) show inhibitory activities against HDAC1–3. In contrast, those without this structural feature (7a, 7c, 7d, 7f, 7h, 7i, 7k, and 7l) were completely devoid of class I HDACI potency. This indicates the importance of the 2-aminobenzamide group for the inhibition against HDAC, which will be discussed further in more detail below. The length of the studied inhibitors was the other structural aspect considered to be optimized. This was carried out by varying the number of carbon atoms connecting benzene ring A to the adjacent carbonyl group in the range of 0–2 atoms. The results indicate that shortening the chain length in the cap region from 2 (7e) to 1 (7b) and 0 (7g and 7j) increases the inhibitory potency, as shown in Table 2. The increase in inhibitory potency due to the decreased molecular length could be attributed to the optimum positioning of ‘cap’ and ZBG groups in favorable locations in the binding pocket of HDAC enzyme [24]. The more methylene groups, the bigger the ‘cap’ group, and the less favorable interaction of the ‘cap’ group with its specific binding site on the enzyme. Such structural modification effectively changes the distance between two terminal aromatic groups (rings A and C). Finnin et al. have shown that this distance needs to be optimum for maximum inhibitory potency [25]. For example, increasing the aliphatic chain of the ‘cap’ group in 7e may prevent the proper interaction of the ‘cap’ group with the rim of the enzyme, leading to diminished HDACI activity. Previous studies including ours have indicated that the presence of NH₂ in the R₂ position of ZBG is essential for inhibition activity. However, the results of our current study demonstrated that both of these factors, that is, the presence of NH₂ in the R₂ position of ZBG and shortening of the length of the molecule by changing the distance between rings A and B from five to three atoms, are equally important in HDACI potency. The other factor that affects HDACI potency is the nature of the R₁ group placed in the para position of the benzene ring A of the ‘cap’ group. Comparing the inhibitory potency of 7g and 7j, one may conclude that electron-donating hydrophobic substituents such as methyl (in 7j) increases HDACI potency, while the electron-withdrawing hydrophilic fluorine group in 7g decreases this potency. The size of the R₁ group may also play a role in the observed activity of these derivatives, and larger groups such as methyl may provide more optimum contact in the binding site than smaller groups such as fluorine. It is evident that the presence of a lipophilic aromatic group in the cap section is also as important as two aforementioned factors. However, for a more comprehensive understanding of the influence of the different physicochemical properties of the R₁ group on HDACI, more functional groups need to be examined.

Table 2. Inhibition potency of active compounds 7b, 7e, 7g and 7j against class I histone deacetylase.^† .

Compound	Inhibition at given concentrations (%)					ΔG binding (PBSA) (kcal/mol)
	Concentration (μM)	HDAC1	HDAC2	HDAC3	HDAC8
7b	0.1	-4.3 ± 3.4%	4.30 ± 0.47%	1.30 ± 1.9%	NI @10 μM	-0.12
	1	19.6 ± 0.4%	27.7 ± 0.50%	29.6 ± 0.90%
	10	65.5 ± 0.4%	79.5 ± 0.69%	82.5 ± 0.4%
7e	0.1	0.2 ± 4.2%	-1.7 ± 3.5%	-2.5 ± 3.6%	NI @10 μM	12.17
	1	-6.1 ± 1.4%	0.3 ± 0.9%	6.7 ± 0.9%
	10	5.2 ± 1.6%	0.6 ± 1.1%	3.8 ± 1.2%
7g	0.1	3.7 ± 1.59%	3.3 ± 2.5%	5.4 ± 8.7%	NI @10 μM	-6.95
	1	36.6 ± 1.76%	40.0 ± 1.4%	35.9 ± 3.9%
	10	70.4 ± 1.40%	58.2 ± 0.4%	76.6 ± 0.6%
7j	0.1	28.8 ± 1.54%	22.4 ± 1.8%	7.8 ± 0.5%	NI @10 μM	-10.77
	1	64.3 ± 0.48%	67.7 ± 0.7%	47.6 ± 1.0%
	10	81.3 ± 0.58%	96.4 ± 0.2%	89.6 ± 0.8%
	IC50	0.65 ± 0.07	0.78 ± 0.02	1.70 ± 0.10
Vorinostat	IC50	0.101 ± 0.007	0.43 ± 0.009	0.21 ± 0.01	ND
Entinostat	IC50	0.93 ± 0.1	0.95 ± 0.03	1.8 ± 0.1	ND

Energy binding between HDACIs and HDAC2.

^†All synthesized compounds were evaluated against HDAC1, 2, 3 and 8 isoenzymes at a 10-μM concentration, and then only those compounds that showed inhibition at 10 μM were tested at other concentrations (0.1 and 1 μM).

ND: Not determined; NI: No inhibition.

In vitro antiproliferative effect

All synthesized benzamide-based derivatives were evaluated for their antiproliferative activity against MCF-7 as a breast cancer cell line by MTT assay using vorinostat as the positive control. For the active HDAC1–3 enzyme inhibitors (7b, 7e, 7g and 7j), an MTT assay was also conducted against T47D breast cancer cells. The results are summarized in Table 3. According to the results, the synthesized derivatives were categorized into three groups based on the nature of R₂ substituents on ZBG and their antiproliferative activity. The first group consisted of compounds with R₂ of the -NH₂ group (7b, 7g and 7j) that showed potent antiproliferative activity with IC₅₀ values of 3.6, 8.9 and 0.83 μM against MCF-7, and 3.8, 3.3 and 1.4 μM against T47D cells, respectively. The most potent HDAC enzyme inhibiting compound 7j belongs to the first group and showed superior activity compared with the reference drug vorinostat (IC₅₀ values of 4.6 and 2.22 μM against MCF-7 and T47D, respectively). It is clear that the antiproliferative activities of these compounds are consistent with their HDAC enzyme inhibition potencies. In addition, compound 7e, which was almost inactive in inhibiting HDAC1–3 enzymes, again showed weak antiproliferative activity on MCF-7 cells (IC₅₀ values of 15 μM) and was unable to affect T47D cell viability. The second group of compounds are 7c, 7h and 7l, which carry a methyl functional group in ZBG (R₂: Me) and showed IC₅₀ values of 5.8, 3.18 and 2.96 μM, respectively, against MCF-7 cells. These compounds, which were inactive in inhibiting HDAC enzymes but displayed antiproliferative activity with IC₅₀ values in the submicromolar range, may exert their cytotoxic activity by affecting pathways involved in the antiproliferation of tumor cells other than the inhibition of HDACs. The third group includes derivatives with R₂ of H (7a, 7d, 7f, 7i, 7k and 7l), which are inactive as HDACIs and showed high IC₅₀ values in the MTT assay. Moreover, all synthesized compounds showed very low or no toxicity (>82% viability) against normal breast cells (MCF10A), which indicated very high selectivity of the compounds toward cancer cells over normal cells (Table 3). This selectivity was more pronounced for more active inhibitors. To understand the possible mechanism for the observed antiproliferative activity of these compounds, target prediction using the SwissTargetPrediction web server (www.swisstargetprediction.ch/) [26] was conducted. The results revealed HDAC enzymes as the first target for the first group of compounds, but this was not the case for the rest. Furthermore, different databases including RCSB, CHEMBL, PubChem and ZINC were searched against 7c, 7h and 7l compounds and related substructures based on chemical structure similarity. Different compounds (CHEMBL ID: 4216192, 1312503, 1876340 and 1372980) containing an o-tolylbenzamide substructure were found, showing an antiproliferative effect via different mechanisms. For example, one of those (PDB code: 8XR and CHEMBL ID: 4216192) inhibits BRD4, which is an epigenetic modifier similar to HDAC enzyme [27]. In addition, other targets were suggested for similar compounds including DNA polymerase, TGF-β, and so on. However, it should be noted that all of these are just indefinite probabilities, and elucidation of a more precise mechanism should wait for experimental investigations.

Table 3. In vitro antiproliferative activities (IC₅₀) of compounds 7a–7l against human breast cancer cell lines.

Compound	Structure features			IC50 (μM)†,‡		Viability @10 μM (%)		Selectivity	Activity§
	n	R1	R2	MCF-7	T47D	MCF-7	MCF-10A	MCF10A/MCF-7	Observed	Predicted¶
7a	2	H	H	48.25 ± 5.55	–	62.79	92.97	1.48	4.32	4.46
7b	2	H	NH2	3.6 ± 1.3	3.8 ± 1.81	40.15	87.48	2.17	5.44	5.11
7c	1	H	Me	5.8 ± 3.09	–	43.52	82.45	1.89	5.24	5.43
7d	1	H	H	20.45 ± 2.49	–	62.12	94.71	1.52	4.69	4.42
7e	1	H	NH2	15.97 ± 1.89	>50	75.81	100	1.31	4.80	5.07
7f	0	H	H	28.27 ± 2.12	–	100	92.17	0.92	4.55	4.55
7g	0	F	NH2	8.9 ± 1.59	3.3 ± 7.49	54.49	98.92	1.81	5.05	5.11
7h	0	F	Me	3.18 ± 1.46	–	76.94	96.99	1.26	5.50	5.43
7i	0	Cl	H	>50	–	103.65	99.42	0.95	4.30	4.30
7j	0	Me	NH2	0.83 ± 1.41	1.4 ± 1.85	34.18	92.43	2.70	6.08	6.08
7k	0	OMe	H	>50	–	102.91	93.66	0.91	4.30	4.43
7l	0	OMe	Me	2.96 ± 1.35	–	83.83	89.46	1.06	5.53	5.40
P1	1	Me	Me	–	–	–	–	–	–	6.31
P2	0	Me	Me	–	–	–	–	–	–	6.40
SAHA	–	–	–	4.61 ± 1.38	2.22 ± 1.57	–	–	–	–	–
Cisplatin	–	–	–	–	–	45.26	62.58	1.38	–	–

The selectivity index toward cancer cells (MCF7) over normal cells (MCF10A). The observed and predicted activity obtained by QSAR analysis.

^†Cells were treated with different concentrations of ^‡,§,¶the compounds for 48 h. Cell viability was measured by MTT assay, as described in the Experimental section.

^‡IC₅₀ values are indicated as mean ± standard deviation of at least three independent experiments.

^§The activity values were calculated using -log(IC₅₀), where antiproliferative IC₅₀ values against MCF-7 cells are in molar scale.

^¶The predicted antiproliferative IC₅₀ values were obtained using Free-Wilson type QSAR analysis, explained in section 3.5 QSAR study.

IC₅₀: Half-maximal inhibitory concentration.

Quantitative structure–activity relationship analysis

Considering the number of compounds, a quantitative Free-Wilson-type analysis was performed to correlate the observed antiproliferative potencies, denoted as -log (IC₅₀), of the studied derivatives to their structures. The contribution of each structural group towards the observed activity was calculated using partial least square regression analysis and the obtained values were used to calculate the activities for the synthesized compounds as well as those that could be synthesized by the structural elements considered in the analysis. The structural elements are R₁ (Me, OMe, F and Cl), R₂ (Me and NH₂) and n (n = 1 and 2), and the basic structure was defined as R₁=H, R₂=H and n = 0. The observed and predicted activity values of the basic structure (7f) were equal to 4.55. The graphical view of the contribution of each structure elements toward biological activity is shown in the Supplementary Material. The observed and predicted activity values for the synthesized derivatives are shown in Table 3. The activities for all possible derivatives using the combination of R₁, R₂ and n structural features, which have not been synthesized yet, were predicted and the results revealed that two derivatives, P₁ and P₂, may have higher antiproliferative activity values of 6.31 and 6.40, respectively than the most active compound 7j (6.08). The results of this study suggest the synthesis and biological evaluation of proposed P₁ and P₂ compounds in the future.

Molecular docking studies

To evaluate the binding mode of the synthesized compounds, we conducted molecular docking studies using crystal structures of HDAC2 (PDB ID: 4LXZ, co-crystallized with a SAHA; Supplementary Material). The GOLD program was used to perform an in silico docking study employing the ASP scoring function and predefined genetic algorithm settings. To understand more clearly, a 2D illustration of the docking results of 7j on HDAC2 and its key interactions are shown in Figure 3. Analysis of molecular docking results showed that the important amino acids in the binding site of HDAC2 interacting with 7j are Asp269, Asp181, His183, His145, Asp179, Gly154, His146, Asp186, Leu276, Tyr308, Phe155, Pro34, Gly32 and Phe155. The calculations revealed the main interactions of compound 7j bound to the active site of HDAC2 including van der Waals interactions (vdW), three hydrogen bonds, and electrostatic interactions of polar N and O atoms with Zn²⁺ oriented in trigonal bipyramid coordination geometry. The cap group and linker of 7j make multiple van der Waals contacts with hydrophobic groups lining the rim and wall of the binding pocket of the enzyme including Gly32, Pro34, Leu276 and Phe155 residues. The two hydrogens of the -NH₂ groups of 2-aminobenzamide in ZBG establish H-bonds with His145 of His145–Asp179 and His146 of His146–Asp186 charge relay systems. Zn²⁺ ion is chelated by Asp181, His183 and Asp269 residues of the enzyme and CO and NH groups of ligand ZBG (Figure 3). The interaction of 7j with the zinc ion in a bidentate fashion formed a seven-membered pseudoring structure, which was also described for HDACIs such as 2-aminobenzamide [25,28,29]. Such coordination of inhibitor to the zinc in a bidentate fashion was reported in many structural studies and was also related to enzyme inhibitory activity. Obviously, in those derivatives where the -NH₂ group was replaced with H or Me groups, the pseudoring cannot be formed. The lack of an -NH₂ group also prevents the establishment of H-bonds to charge-relay systems, which is another important structural feature of the enzyme–inhibitor complex (Figure 3). Collectively, the results of docking calculations support the proper interaction of designed compounds, particularly the most active derivative 7j, with the HDAC enzyme.

Figure 3.

Ligplot representation of compound 7j interactions with residues of enzyme binding pocket.

The black and green dashed lines are metal- and hydrogen-bonding interactions, respectively.

Correlation between enzyme-inhibitor binding free energy & enzyme inhibition potency

The best docked pose for the ligands complexed with HDAC2 was subjected to MD simulation for 50 ns using the AMBER package. Analyses of the root-mean-square deviation (RMSD) and potential energy plots during MD simulation for the complexes of HDAC2 with those ligands showing HDACI effects were indicative of the stability of the proposed docked models for the complexes. Figure 3 illustrates the results for 7j as the representative. During the simulation, coordinates at every 100 ps were written into the trajectory file and subsequently employed for the calculation of the binding free energy by applying the MM-Poisson-Boltzmann Surface Area (PBSA) method implemented in the AMBER package. Table 2 shows the binding free energy for 7b, 7e, 7g and 7j in complex with HDAC2. The results show that there is a good correlation between experimentally obtained HDAC2I potencies and MMPBSA binding energy values for the ligands (7b, 7e, 7g and 7j), with a correlation coefficient (R²) of 0.92 (Supplementary Data). These results further validate the proposed modes of interaction described in the previous section and presented in Figure 4 for 7j.

Figure 4.

Results of the RMSD and potential energy during molecular dynamics simulation.

(A) The root-mean-square deviation alterations for the 7j and HDAC2 complex during 50-ns MD simulation. (B) The total energy for 7j and HDAC2 complex during 50 ns MD simulation.

MD: Molecular dynamics; RMSD: Root-mean-square-deviation.

Drug-likeness of 7j & 7g

The drug-likeness features for 7g and 7j were evaluated by SwissADME (www.swissadme.ch/) and molsoft (https://molsoft.com/mprop/) [30] webservers, and compared with those of the lead compounds MS-275 and vorinostat (Supplementary Data). As shown in Table 4, the results indicate that compounds 7g and 7j conform to Lipinski's rule of five and some other indices proposed by Ghose et al. [31], Veber et al. [32], Egan et al. [33] and Muegge et al. [34]. Furthermore, the bioavailability of these derivatives assessed by the SwissADME webserver was predicted to be in the acceptable range. Moreover, the drug-likeness scores of 7g and 7j calculated by the molsoft webserver were close to or better than the scores calculated for the reference drugs. Based on the predictions demonstrated in Table 4, compounds 7g and 7j possess acceptable drug-like properties and deserve further investigation.

Table 4. Drug-likeness features of 7g and 7j.

Compound	Lipinski	Ghose	Veber	Egan	Muegge	Bioavailability score	Drug-likeness score
7j	Yes	Yes	Yes	Yes	Yes	0.55	0.27
7g	Yes	Yes	Yes	Yes	Yes	0.55	0.57
MS-275	Yes	Yes	Yes	Yes	Yes	0.55	0.45
Vorinostat	Yes	Yes	Yes	Yes	Yes	0.55	-0.58

‘Yes’ denotes that the compound complies with the drug-likeness rule.

Conclusion

In the present investigation, a series of novel compounds (7a-7l) were designed by applying molecular modification in three regions of previously introduced lead compound 7b. The designed compounds were synthesized and evaluated for in vitro inhibition of class I (HDAC1, 2, 3 and 8) HDAC enzymes. The results revealed that three compounds (7b, 7e, 7g and 7j) can inhibit HDAC enzymes at submicromolar concentrations. In particular, 7j demonstrated strong inhibition, with IC₅₀ values in the range of 0.65 to 1.70 μM on different HDAC isoforms. The prepared benzamide derivatives were evaluated for their antiproliferative activity against two breast cancer cell lines (MCF-7 and T₄7D), and the results showed that NH₂ and Me functional groups substituted at position 2 of the benzamide moiety (i.e., R₂) are pharmacophoric groups granting strong antiproliferative activity in the submicromolar range (IC₅₀ values of 0.83–5.8 μM) to the compounds. Again, 7j was the most antiproliferative compound, with IC₅₀ values of 0.83 and 1.4 μM in the MCF-7 and T₄7D cell lines, respectively. Quantitative structure–activity relationship analysis revealed that substitution of the Me functional group for R₁ and R₂, and adjusting the number of methylene groups, ‘n’, to 0 or 1 could lead to new molecules with improved antiproliferative activities. Molecular docking indicated that compound 7j was able to interact efficiently with the active site of the HDAC enzyme and establish the known H-bond-, vdW- and Zn²⁺-mediated coordination interactions observed experimentally for HDACIs such as 2-aminobenzamide. RMSD and potential energy plots for the complexes of HDAC2 with those ligands showing HDACI effects during MD simulation confirmed the stability of the proposed docked models. In addition, there was also a good correlation between experimentally obtained HDAC2 inhibition potencies and MMPBSA binding energy values for compounds 7b, 7e, 7g and 7j. The promising results of the current study may provide valuable information for the discovery of HDACIs and benzamide-based antiproliferative agents.

Funding Statement

The authors would like to thank the Research Office and Biotechnology Research Center of Tabriz University of Medical Sciences for providing financial support under the Postgraduate Research Grant scheme for the PhD thesis of Narges Cheshmazar (grant no. 60360).

Author contributions

N Cheshmazar: investigation, carried out the experiments, data analysis and writing the original draft of paper; Maryam Hamzeh-Mivehroud: docking and MTT assay; S Hemmati: synthesis; H Abolhasani and F Heidari: MTT assay; HN Charoudeh: biologic evaluation; M Zessin and M Schutkowski: enzyme-based biologic evaluation; W Sippl: enzyme-based biologic evaluation and writing – review and editing; S Dastmalchi: project administration, supervision, data analysis, writing – review and editing.

Financial disclosure

The authors would like to thank the Research Office and Biotechnology Research Center of Tabriz University of Medical Sciences for providing financial support under the Postgraduate Research Grant scheme for the PhD thesis of Narges Cheshmazar (grant no. 60360). The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

Competing interests disclosure

The authors have no competing interests or relevant affiliations with any organization or entity with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

Writing disclosure

No writing assistance was utilized in the production of this manuscript.