3D-QSAR and Molecular Docking Studies of N-(2-Aminophenyl)-Benzamide Derivatives as Inhibitors of HDAC₂

 

Navjot Kaur¹, Monika¹, Kulwinder Singh²*

 

ABSTRACT:

Inhibitors of HDACs are an important emerging class of drugs for the treatment of cancers. 3-Dimensional quantitative structure-activity relationship (3D-QSAR) studies were performed on a series of N-(2-aminophenyl)-benzamide derivatives by using Scigress Explorer software. The multiple linear regression analysis was used to correlate the physicochemical descriptors with the anti-HDAC2 activity of 20 training set of compounds and the best QSAR model was developed. The best model was validated using leave-one-out method and found to be statistically significant, with coefficient of determination (r2) of 0.735696. This model was further used to predict the anti-HDAC2 activity of 29 test set of compounds. We also performed docking of these 29 test set of compounds using Molegro Virtual Docker software and found that most of the compounds formed H-bond interactions with amino acid residues. Predicted pIC50 value of one of the test compounds was 7.358 and it showed H-bond interactions with HDAC2 protein (PDB ID: 3MAX). This study shall help in rational drug design and synthesis of new selective HDAC2 inhibitors with predetermined affinity and activity and provides valuable information for the understanding of interactions between HDAC2 and the novel compounds.

 

 

INTRODUCTION:

Histone deacetylase (HDAC) is the enzyme, which catalyzes the histone deacetylation within eukaryotes. Deacetylation is a release of the acetyl group from the histone tail, and it causes the histone to be twisted around the DNA, disrupting gene transcription, by blocking the pathway of transcription factor binding¹. Human histone deacetylases can be divided into 4 classes based on structure, sequence homology and domain organization. Class I consists of HDACs 1, 2, 3 and 8. Class I HDACs are nuclear and play roles in cell proliferation and apoptosis. Class II includes HDACs 4, 5, 6, 7, 9 and 10². These enzymes are characterized by a large NH2-terminal domain or a second catalytic site and their expression is more restricted, suggesting roles in cellular differentiation and development³. Class III enzymes include the sirtuins and are nicotinamide adenine dinucleotide-dependent (NAD-dependent) deacetylases⁴. HDAC11 is designated as class IV based on phylogenetic analysis⁵.

 

HDACs are present in a variety of organisms and participate in a myriad of biological functions, including post-translational modifications of various non-histone substrates. The aberrant activity of HDACs has been connected to numerous diseases, such as inflammation⁶ and neurosis^7-9, as well as regulation of tumour suppressor genes and oncogenes, contributing to cancer pathogenesis^{4, 10}. HDAC inhibitors are potent antiproliferative agents in cell based studies, where they exhibit striking effects on tumour cell proliferation^11-15. Consequently, there has been great interest in developing HDAC inhibitors as new anticancer agents, and an extensive number of clinical trials are underway in which HDAC inhibitors either alone or in combination with other agents are being evaluated¹⁴. To date, vorinostat or suberoylanilide hydroxamic acid (SAHA) and more recently romidepsin are the only HDAC inhibitor based therapies that have achieved regulatory approval, being marketed for the treatment of advanced and refractory cutaneous T-cell lymphoma (CTCL)¹⁵.

 

HDAC2 is a potential target for anticancer drug discovery¹⁶. Several classes of HDAC2 inhibitors are under research, which are hydroxamic acid, cyclic tetra peptides and benzamide derivatives. Different types of hydroxamic acid and benzamide derivatives are in clinical trials¹⁷. N-(2-aminophenyl)-benzamide derivatives entinostat (MS-275) and mocetinostat (MGCD0103) are in phase II clinical studies for the treatment of various cancers^18-21. Thus, there is a need for novel, selective, and potent HDAC inhibitors with good potency.

 

Quantitative structure-activity relationship (QSAR) is a type of analysis where some measures of chemical properties are correlated with biological activity to derive a mathematical illustration of the underlying structure activity relationship (SAR)²². QSAR studies are unquestionably of great importance in modern chemistry and biochemistry. To get an insight into the SAR we need molecular descriptors that can effectively characterize molecular size, molecular branching or the variations in molecular shapes, and can influence the structure and its activities²³.

 

Design and development of new drugs is simplified and made more cost-effective because of the advances in the concepts of QSAR studies. A methodology of QSAR studies is one of the approaches to the rational drug design²⁴. The introduction of Hansch model, in early 1960, enabled chemists to describe the structure activity relationships in quantitative terms and check those using statistical methods²⁵. QSAR are statistically derived models that can be used to predict the biological activity of untested compounds from their molecular structures^{26, 27}. This concept helps to understand the role of physicochemical descriptors of compounds in determining the biological activity and in estimating the characteristics of the new and potent compounds, without the chemical synthesis of the compounds²⁵.

 

Docking various ligands to the protein of interest followed by scoring to determine the affinity of binding and to reveal the strength of interaction has also become increasingly important in the context of drug discovery²⁸. Thus, the objective of the present work was to develop various QSAR models by multiple linear regression (MLR) methods and to use the best QSAR model for the prediction of anti-HDAC activity of newly designed compounds by using Scigress Explorer software²⁹. We also performed the molecular docking of the newly designed compounds against HDAC protein, 3MAX (PDB ID) with bound ligand N-(4-aminobiphenyl-3-yl)-benzamide extracted from protein data bank (PDB), by utilizing a fast, exhaustive docking software Molegro Virtual Docker³⁰.

 

MATERIALS AND METHODS:

Data set for 3D QSAR

The first step in developing QSAR equations was to compile a list of compounds for which the experimentally determined inhibitory activity was known. The histone deacetylase inhibitory activity data and chemical structures of substituted N-(2-aminophenyl)-benzamide derivatives were retrieved from literature³¹. The biological activity (IC50) of the molecules were converted to their corresponding pIC50 values and used as dependent variables in the QSAR calculations. The data set was divided into training set for model generation (Table 1), and a test set for model validation (Table 2), containing 20 and 29 compounds respectively.

 

Fig 1. Structure of parent compound used for QSAR analysis

 

Chemical structure construction and optimization

The molecules were drawn using chemical drawing software ‘ACD/ChemSketch’³² and 3D optimization of molecules was done by ‘ACD/3D viewer’³³. The molecules were first optimized to their lowest energy state using Merck molecular force field-3 (MMFF3) method³⁴ using Scigress explorer software. To avoid the local stable conformations of the compounds, geometry optimization was run many times with different starting points of each molecule, and conformation with the lowest energy was considered for the calculation of the molecule descriptors.

 

 

Table 1. Molecular structures of 20 training set of compounds and their pIC50 values

Compound	Structure	pIC50	Compound	Structure	pIC50
Compound (1)		7.853	Compound (11)		6.853
Compound (2)		7.309	Compound (12)		6.698
Compound (3)		7.221	Compound (13)		6.568
Compound (4)		7.154	Compound (14)		6.504
Compound (5)		7.148	Compound (15)		6.443
Compound (6)		7.142	Compound (16)		6.301
Compound (7)		7.045	Compound (17)		6.107
Compound (8)		6.978	Compound (18)		6.096
Compound (9)		6.939	Compound (19)		6.045
Compound (10)		6.886	Compound (20)		5.42

 

Table 2. Molecular structures of 29 test set of compounds and their substituents at R1 & R2 positions

 

 

Calculation of physicochemical descriptors

The structure of a molecule is expressed quantitatively in terms of its physicochemical descriptors, which are lipophilic, electronic and steric in nature. The aligned molecules were selected for calculation of the descriptors after inserting the biological activity as a dependent variable and the descriptors generated were selected as independent variables. List of physicochemical descriptors used in this study are summarised in Table 3.

 

Table 3. List of physicochemical descriptors selected for this study

 

Development and validation of QSAR models

The QSAR studies were carried out to correlate physicochemical descriptors of 20 derivatives from the training set with their anti-HDAC activity. The physicochemical descriptors were taken as the independent variables and the anti-HDAC activity was taken as the dependent variable. Various QSAR models were developed by correlating more than one (stepwise MLR analysis implemented in Scigress explorer's “Project Leader” program) physicochemical descriptors at a time, with anti-HDAC activity of the compounds. Validation parameter, predictive r² (r² pred) was calculated for evaluating the predictive capacity of the models. The models were then cross-validated by the ‘leave one out’ scheme ³⁵ where a model was built with n-1 compounds and the nth compound was predicted. Each compound was left out of the model derivation and predicted in turn. An indication of the performance of the model was obtained from the cross-validated r²_CV (or predictive q²) coefficient which is defined as:

q² = (SD-PRESS/SD)

Where, SD is the sum of squares deviation for each activity from the mean. PRESS (or predictive sum-of-squares) is the sum of the squared difference between the actual and that of the predicted values when the compound is omitted from the fitting process. Cross-validation coefficient q² is considered as an indicator of the predictive performance and stability of a model. For a reliable model, the square of cross-validation coefficient q² should be ≥ 0.5³⁶. The anti-HDAC activity of 20 compounds in the training set and 29 compounds in the test set was predicted using the best QSAR model (Equation 1). For further validation of the accuracy of the predicted values by the best QSAR model, the experimental anti-HDAC activity of the 20 training set compounds was correlated with their predicted anti-HDAC activity.

 

Graphical analysis

Graphical analysis was performed using Scigress explorer's plotting facilities to display molecules that were outliers in the database. Through scatter plot there was evaluation of regression in the graph. By plotting the actual activities along X-axis versus the predicted activities along Y-axis, the predicted ability of the model was assessed. From the regression line it was easy to predict the number of molecules lie on and away from regression line.

 

Receptor X-ray structure

The 3D coordinates of the crystal structure of HDAC2 in complex with N-(4-aminobiphenyl-3-yl)-benzamide (PDB code: 3MAX) extracted from the protein data bank (www.rcsb.org/‎) was selected as the receptor model for docking experiments.

 

Docking analysis

We used the template docking available in Molegro Virtual Docker software and evaluated MolDock, Rerank and protein-ligand interaction scores from MolDock and MolDock [GRID] options. Template docking is based on extracting the chemical properties like the pharmacophore elements of a ligand bound in the active site and using that information for docking structurally similar analogs. We used the default settings, including a grid resolution of 0.30 Å for grid generation and a 15 Å radius from the template as the binding site. We used the MolDock optimizer as a search algorithm, and the number of runs was set to 10. A population size of 50, maximum iteration of 2000, scaling factor of 0.50, crossover rate of 0.90 and a variation based termination scheme for parameter settings were used. The maximum number of poses was set to a default value of 5.

 

RESULTS AND DISCUSSION:

QSAR analysis

Various physicochemical descriptors listed in Table 3 were calculated for the training set of molecules using the Scigress explorer's “Project Leader” program. Anti-HDAC activity (experimental activity) of all the training compounds was added manually in the data set (from Table 1) and was correlated with the different physicochemical descriptors by stepwise MLR analysis and QSAR models were generated. The best model (equation 1) was validated using leave-one-out method and found to be statistically significant, with coefficient of determination (r² pred) of 0.735696 and cross-validated r²CV (or predictive q²) coefficient of 0.643793.

 

Equation 1 (Model 1):

M=0.0687994*SE-0.590428*HF-0.388229*HOMO-0.205929*POL+0.0368046*SASA-0.0556615*DP +0.11061*TE-4.10224*IP-0.0160347*MR+1.62358*¹X +32.9088              r²_CV=0.643793              r²=0.735696

 

Table 4. Predicted activity values of 29 test set of compounds calculated from the best QSAR model (equation 1)

 

Ten QSAR models were generated and equation 1 was considered as the best model to predict the activities of 29 test set of molecules (Table 4). In order to validate our results we have correlated the predicted activities of 20 molecules of the training set using the model expressed by equation 1 and compared with the experimental values. Predicted and the experimental activities were very close to each other evidenced by low values of residual activity (difference between experimentally observed activity and QSAR predicted activity) (Table 5).

 

Table 5. Values of actual, predicted & residual activities of 20 training set of compounds

 

The graph between predicted and experimental activity of training set compounds by using model 1 is illustrated in Figure 2. Through this scatter plot, the compounds aligned on and around the regression line shows good correlation level between the predicted and experimental activity and compounds which are deviated from the regression line shows low correlation level between the predicted and experimental activity of training set of compounds.

 

Results of docking experiments

Before the docking experiments, the protocol was validated. 3MAX (PDB ID) bound ligand N-(4-aminobiphenyl-3-yl)-benzamide was docked into the binding pocket of HDAC2 protein to obtain the docked pose and the RMSD (Root Mean Square Deviation) of all atoms between these two conformations indicating that the parameters for docking simulation were good in reproducing the X-ray crystal structure. Therefore, N-(4-aminobiphenyl-3-yl)-benzamide derivatives (29 set of molecules) were docked into the binding pocket of HDAC2 protein. 3MAX co-crystallized N-(4-aminobiphenyl-3-yl)-benzamide ligand resulted in MolDock score of -149.417kcal/mol. Therefore, any molecule from the dataset which shows a score lower than -149.417kcal/mol would be regarded as ligand with higher binding affinity and would act as inhibitor against HDAC2 protein. Our approach identified seven compounds from the test set of molecules with better energy scores than the 3MAX bound co-crystallized ligand. The docked energies (Moldock score) and H-bond interaction data of the 7 best compounds from the 29 test set of molecules are given in Table 6.

 

Out of 29 test set of molecules, the best one was molecule 27^th with predicted pIC50 value of 7.358 and binding energy score of -166.262kcal/mol. This compound was docked with the binding pocket of HDAC2 protein (PDB ID: 3MAX) forming H-bond interactions with His183, His145, His146, Asp181 & Gly154. Interaction parameters of HDAC2 with 29th test compound are illustrated in Figure 3.

 

Table 6. Interaction parameters of 3MAX with the 7 best test set of compounds and co-crystallized N-(4-aminobiphenyl-3-yl)-benzamide (reference ligand)

*H-Bond stands for Hydrogen Bond interaction score.

 

 

Fig 2. Graph between predicted (vertical axis) and experimental activity (horizontal axis) of training set of compounds by using equation 1. Compounds 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 12, 16, 17, 18 & 19 are aligned on and around the regression line showing good correlation level between the predicted and experimental activity

 

Fig 3. Interactions between the HDAC2 (pdb id: 3MAX) and test compound 29. Blue dashed lines - hydrogen bonds (image generated using Molegro Virtual Docker software)

 

 

CONCLUSION:

Finding novel compounds at starting points for lead optimization is a major challenge in drug discovery. The number of methods and softwares which use the QSAR and docking approaches are increasing at a rapid pace. It has been clearly demonstrated that the approach utilized in this study was successful in finding novel HDAC2 inhibitors from the data set developed by computational methods. The model generated from various physicochemical descriptors corresponds to the essential structural features of N-(2-aminophenyl)-benzamide derivatives and found to have significant correlation (coefficient of determination (r²) of 0.735696) with HDAC2 inhibiting activity. N-(2-aminophenyl)-benzamide derivatives designed by using computational approaches also showed good interactions with HDAC2 protein. Compound (27), in particular, showed high binding affinity with MolDock score of -166.262kcal/mol against 3MAX (PDB ID) in docking analysis and predicted pIC50 value of 7.358 in QSAR analysis. The ligand was docked deeply within the binding pocket region forming hydrogen bond interactions with His183, His145, His146, Asp181 & Gly154. This study shall help in rational drug design and synthesis of new selective HDAC2 inhibitors with predetermined affinity and activity and provides valuable information for the understanding of interactions between HDAC2 and the novel compounds and might pave the way towards discovery of novel HDAC2 inhibitors. The physicochemical descriptors used in QSAR analysis in this study were important in further lead optimization of the N-(2-aminophenyl)-benzamide derivatives.

 

ACKNOWLEDGMENT:

Authors would like to extend our heartfelt thanks to Molegro ApS for giving us a fully functional version of Molegro Virtual Docker software for a period of 30 days during which all the in-silico docking work was carried out.

 

REFERENCES:

1.        Subha K, Kumar GR. Assessment for the identification of better HDAC inhibitor class through binding energy calculations and descriptor analysis. Bioinformation 2008;3:218–222.

2.        Imai S, et al. Transcriptional silencing and longevity protein Sir2 is an NAD-dependent histone deacetylase. Nature 2000;403:795–800.

3.        Laherty CD, et al. Histone deacetylases associated with the mSin3 corepressor mediate mad transcriptional repression. Cell 1997;89:349–356.

4.        De Ruijter AJ, et al. Histone deacetylases (HDACs): characterization of the classical HDAC family. Biochem J 2003;370:737–749.

5.        Gregoretti IV, Lee YM, Goodson HV. Molecular evolution of the histone deacetylase family: functional implications of phylogenetic analysis. J Mol Biol 2004;338:17–31.

6.        Leoni F, et al. The antitumor histone deacetylase inhibitor suberoylanilide hydroxamic acid exhibits antiinflammatory properties via suppression of cytokines. Proc Natl Acad Sci USA 2002;99:2995–3000.

7.        Dompierre JP, et al. Histone deacetylase 6 inhibition compensates for the transport deficit in Huntington's disease by increasing tubulin acetylation. Neurosci 2007;27:3571–3583.

8.        Minucci S, Pelicci PG. Histone deacetylase inhibitors and the promise of epigenetic (and more) treatments for cancer. Nat Rev Cancer 2006;6:38–51.

9.        Suzuki T. Explorative study on isoform-selective histone deacetylase inhibitors. Chem Pharm Bull (Tokyo) 2009;57:897–906.

10.     Tan J, et al. Novel histone deacetylase inhibitors in clinical trials as anti-cancer agents. J Hematol Oncol 2010;3:5.

11.     Marks PA, Breslow R. Dimethyl sulfoxide to vorinostat: Development of this histone deacetylase inhibitor as an anticancer drug. Nat Biotechnol 2007;25:84–90.

12.     Richon VM, et al. Histone deacetylase inhibitors: Development of suberoylanilide hydroxamic acid (SAHA) for the treatment of cancers. Blood Cells Mol Dis 2001;27:260–264.

13.     Richon VM, Garcia-Vargas J, Hardwick JS. Development of vorinostat: Current applications and future perspectives for cancer therapy. Cancer Lett 2009;280:201–210.

14.     Khan O, La Thangue NB. Drug insight: Histone deacetylase inhibitor-based therapies for cutaneous T-cell lymphomas. Nat Clin Pract Oncol 2008;5:714–726.

15.     Boumber Y, Issa JP. Epigenetics in cancer: what’s the future? Oncology (Williston Park) 2011;25:220–226.

16.     Kramer OH. HDAC2: a critical factor in health and disease. Trends Pharmacol Sci 2009;30:647–655.

17.     Khan N, et al. Determination of the class and isoform selectivity of small-molecule histone deacetylase inhibitors. Biochem J 2008;409:581–589.

18.     Witter DJ, et al. Optimization of biaryl selective HDAC1&2 inhibitors (SHI-1:2). Bioorg Med Chem Lett 2008;18:726–731.

19.     Tessier P, et al. Diphenylmethylenehydroxamic acids as selective class IIa histone deacetylase inhibitors. Bioorg Med Chem Lett 2009;19:5684–5688.

20.     Yardley DA, et al. Randomized Phase II, Double-Blind, Placebo-Controlled Study of Exemestane With or Without Entinostat in Postmenopausal Women With Locally Recurrent or Metastatic Estrogen Receptor-Positive Breast Cancer Progressing on Treatment With a Nonsteroidal Aromatase Inhibitor. J Clin Oncol 2013;31:2128–2135.

21.     Younes A, et al. Mocetinostat for relapsed classical Hodgkin's lymphoma: an open-label, single-arm, phase 2 trial. Lancet Oncol 2011;12:1222–1228.

22.     Testa B, Kier LB. The concept of molecular structure in structure-activity relationship studies and drug design. Med Res Rev 1991;11:35–48.

23.     Thakur A, et al. Application of topological and physicochemical descriptors: QSAR study of phenylamino-acridine derivatives. ARKIVOC 2004;14:36–43.

24.     Mahajan S, et al. QSAR analysis of benzophenone derivatives as antimalarial agents. Indian J Pharm Sci 2012;74:41–47.

25.     Wolf M, Kubiyini H. Burger’s Medicinal chemistry and drug discovery. A Wiley Interscience, New York.1995.

26.     Roy PP, et al. Two novel parameters for validation of predictive QSAR models. Molecules 2009;14:1660–1701.

27.     Konovalov DA, et al. Robust cross-validation of linear regression QSAR models. Chem Inf Model 2008;48:2081–2094.

28.     Sudha KN, et al. Virtual screening for novel COX-2 inhibitors using the ZINC database. Bioinformation 2008;2:325–329.

29.     Scigress Explorer, FQS Poland, Krakow, Poland; software available at http://www.fqs.pl/chemistry_materials_life_science/products/scigress

30.     Thomsen R, Christensen MH. MolDock: a new technique for high-accuracy molecular docking. J Med Chem 2006;49:3315–3321.

31.     Kandakatla N, et al. QSAR Studies of N-(2-aminophenyl)-benzamide derivatives as histone deacetylase2 Inhibitors. Int J PharmTech Res 2012;4:1110–1121.

32.     ACD/ChemSketch Freeware 11.01, Advanced Chemistry Development, Inc., Toronto, Ontario, Canada; software available at http://www.acdlabs.com/resources/freeware/chemsketch/.

33.     ACD/3D Viewer, Advanced Chemistry Development, Inc., Toronto, Ontario, Canada; software available at http://www.acdlabs.com/products/draw_nom/draw/chemsketch/3dviewer.php.

34.     Halgren TA. Merck molecular force field: I. Basis, form, scope, parameterization, and performance of MMFF94. J Comput Chem 1996;17:490–519.

35.     Agrawal VK, et al. QSAR study on 5,6-Dihydro-2-pyrones as HIV-1 protease inhibitors. ARKIVOC 2006;2:162–177.

36.     Eriksson L, et al. Methods for reliability and uncertainty assessment and for applicability evaluations of classification and regression based QSARs. Environ Health Persp 2003;111: 1361–1375.

 

 

 

Received on 13.05.2014          Modified on 02.06.2014

Accepted on 10.06.2014          © RJPT All right reserved