"DEVELOPMENT OF PHARMACOPHORE MODEL FOR S. AUREUS DNA GYRASE INHIBITORS"
Research Article ISSN: 2321-2969 Received: 20 April 2013, Accepted: 29 April 2013 Int. J. Pharm. Biosci. Technol. To cite this Article: Click here International Journal of Pharma Bioscience and Technology; Volume 1, Issue 1, May 2013, Pg 34-49 Journal home page: www.ijpbst.com DEVELOPMENT OF PHARMACOPHORE MODEL FOR S. AUREUS DNA GYRASE INHIBITORS Arundhati C. Lele, Nilesh R. Tawari, Mariam S. Degani* Pharmaceutical Chemistry Research Laboratory, Department of Pharmaceutical Sciences and Technology, Institute of Chemical Technology, N. P. Marg, Matunga, Mumbai 400019, India Corresponding Author* E-mail address- firstname.lastname@example.org ABSTRACT: Pharmacophore mapping studies were undertaken for a series of molecules belonging to indazole, pyrazole and their congeners as inhibitors of S. Aureus DNA gyrase. A four point pharmacophore with two aromatic rings (R), one lipophilic/hydrophobic group (H) and one positive ionic feature (P) as pharmacophoric features was developed. The pharmacophore hypothesis yielded a statistically significant 3D-QSAR model, with correlation coefficient of r2 = 0.681 for training set molecules. The model so generated showed good predictive power with correlation coefficient for an external test set of fifteen molecules. The geometry and features of the pharmacophore will be useful for ligand based design of potential DNA gyrase inhibitors. Key words: Pharmacophore, 3D-QSAR, DNA gyrase, S. Aureus. INTRODUCTION negative supercoils into bacterial DNA as well as Antimicrobial resistance is a growing threat the decatenation and unknotting of DNA.  worldwide. Staphylococcus aureus infections are a major cause of morbidity and mortality in Ligand based drug design approaches like community and hospital settings. The pharmacophore mapping and QSAR can be used emergence of methicillin resistant and in drug discovery in several ways; the vancomycin resistant strains of S. Aureus rationalization of activity trend in molecules under represents an enormous threat to public health study, to predict the activity for new compounds, with MRSA accounting for as many as 50% of for database search studies in search of new hits infections. S. Aureus strains that combine and to identify important features for activity. This resistant and virulent genes have become a major paper describes the development of a robust treatment problem in Europe and the U.S. ligand based 3D- pharmacophore hypothesis Hence there is an ever increasing need for the using Pharmacophore Alignment and Scoring development of newer antibiotics to treat S. aureus Engine (PHASE) for S. aureus DNA gyrase. Ligands infections. selected for the development of the pharmacophore model and the subsequent DNA gyrase is a prokaryotic specific type ΙΙ generation of atom based 3D-QSAR were topoisomerase and has attracted considerable indazoles and pyrrazoles analogues acting as DNA attention from the pharmaceutical industry ever gyrase inhibitors and the alignment obtained from since the discovery of Nalidixic acid (DNA gyrase the generated pharmacophoric points is used to inhibitor) in the early 1960’s. Other well-known derive an atom based 3D-QSAR model. The DNA gyrase inhibitors include quinolones, pharmacophore thus developed imparts coumarins and cyclothialidines. Bacterial DNA information about important features for DNA gyrase is involved in the vital processes of DNA gyrase inhibitors and the geometry has the ability replication, transcription, and recombination and to mine 3D virtual databases of drug like catalyzes the ATP-dependent introduction of Lele et al Pg. 34 Int. J. Pharm. Biosci. Technol. Table 1(m), belonging to the family of indazole molecules. Furthermore, the contours generated and pyrrazole analogues reported as DNA gyrase from QSAR studies highlight the structural features inhibitors [6-10] were used for the pharmacophore required for DNA gyrase inhibitory activity and generation. The negative logarithm of the are useful for further design of more potent measured IC50 value as pIC50 was used in the 3D- inhibitors. QSAR study. The complete set was divided into a training set (48 compounds) and a test set (15 MATERIALS AND METHODS compounds) using randomization as well as chemical and biological diversity. Biological data A set of 63 compounds shown in Table 1(a) to Table 1. Structures and the inhibitory concemtrations of the set of molecules utilized for generation of pharmacophore model. Table 1(a)  No Structure MIC (µg/ml) 1. 64 2. 0.25 Novobiocin Table 1(b) General structure 1. No. R1 MIC (µg/ml) 3. 32 16 4. Lele et al Pg. 35 Int. J. Pharm. Biosci. Technol. H3C O 64 5. O NH HN H3C O 6. 64 O NH HN 7. 32 8. 64 9 32 10. 32 11. 16 12. 16 H3C O H 13. N 4 O Table 1(c) General structure 2  Lele et al Pg. 36 Int. J. Pharm. Biosci. Technol. Structure No. R1 MIC (µg/ml) 14. 64 15. 128 16. 32 17. 4 Table 1(d)  Structure No. Structure MIC (µg/ml) 18. 4 19. 4 20. 1 21. 1 Lele et al Pg. 37 Int. J. Pharm. Biosci. Technol. Table 1(e) General structure 3  Structure No. R3 MIC (µg/ml) 22. 4 (E Stereochemistry) 23. 4 (Z Stereochemistry) 24. 8 (E Stereochemistry) 25. 4 (Z Stereochemistry) Table 1(f) General structure 4  Structure No. R MIC (µg/ml) 26. 2 27. 8 Lele et al Pg. 38 Int. J. Pharm. Biosci. Technol. 28. 2 29. 2 30. 32 31. 16 32. 4 33. 1 34. 0.125 Sparfloxacin Table 1(g) General Structure 5  Lele et al Pg. 39 Int. J. Pharm. Biosci. Technol. Structure No. R MIC (µg/ml) 35. 2 36. 128 37. 32 38. 4 Table 1(h) General Structure 6  Structure No. R1 R2 MIC (µg/ml) 39. H 64 40. H 64 48. CH3 32 42. 4 Lele et al Pg. 40 Int. J. Pharm. Biosci. Technol. 43. 4 44. 4 45. 4 46. 1 Table 1(i) General Structure 7  Structure No. R1 R2 MIC (µg/ml) 47. CH3 64 48. 8 Lele et al Pg. 41 Int. J. Pharm. Biosci. Technol. Table 1(j)  Structure No. Structure MIC (µg/ml) 49. 64 Table 1(k) General Structure 8  Structure No. R1 H-R2 MIC (µg/ml) 50. 4 51. 2 52. 16 53. 2 54. 4 55. 64 56 2 Lele et al Pg. 42 Int. J. Pharm. Biosci. Technol. Table 1(L) General Structure 9  Structure No. R1 H-R2 MIC (µg/ml) 4 57. 58. 8 59. 4 Table 1(m) General Structure 10  MIC Structure No. X R1 H-R2 (µg/ml) 60. O 4 3- Cl 61. O 2 3- Cl 62. NH 2 3- Cl 63. NH 2 3- Cl Lele et al Pg. 43 Int. J. Pharm. Biosci. Technol. Generation of Common Pharmacophore S= WsiteSsite + WvecSvec + WvolSvol + WselSsel + Wrewm Hypothesis (CPH) Where W’s are weights and S’s are scores The pharmacophore hypothesis and alignment Ssite represents alignment score, the root mean based on it was carried out using PHASE, version square deviation in the site point position. 2.0, Schrödinger, LLC, New York, NY, 2005 installed on AMD Athelon workstation. The Svec represents vector score, and averages cosine structures were imported from the project table in of the angles formed by corresponding pairs of the “Develop Pharmacophore Hypothesis” panel vector features in the aligned structures. and geometrically refined (cleaned) using Svol represents volume score based on overlap of Ligprep. Conformations were generated by Van der Waals models of non hydrogen atoms in MCMM/LMOD method using maximum of 2000 each pair of structures. steps with distance dependent dielectric solvent model and OPLS-2005 force field. All the Ssel represents selectivity score, and accounts for conformers were subsequently minimized using what fractions of molecules are likely to match the truncated Newton conjugate gradient minimization hypothesis regardless of their activity towards up to 500 iterations. For each molecule a set of receptor. conformers with maximum energy difference of 30 Wsite, Wvec, Wvol, Wrew have default value of 1.0 kcal/mol relative to the global energy minimum while Wsel has default value of 0.0. were retained. A redundancy check of 2 Å in the heavy atom positions was applied to remove In the hypothesis generation default values have duplicate conformers. Pharmacophore features; been used. Wrewm represents reward weights hydrogen bond acceptor (A), hydrogen bond defined by m-1 where m is the number of actives donor (D), hydrophobic group (H), negatively that match the hypothesis. charged group (N), positively charged group (P), The assessment of the generated CPHs was aromatic ring (R), were defined by a set of 4 chemical structure patterns as SMARTS queries performed by correlating the observed and and assigned one of three possible geometries, estimated activity for the internal set of 48 training which define physical characteristics of the site: molecules and external set of 15 test molecules using partial least square analysis. The partial Point—the site is located on a single atom in the least square (PLS) regression was carried out SMARTS query. using PHASE with maximum of 4 PLS factors using pharmacophore based model, with grid spacing of Vector—the site is located on a single atom in the 1Å. The CPH with best predictive value and SMARTS query, and assigned directionality significant statistical data was chosen for according to one or more vectors originating from alignment of molecules and used for further 3D- the atom. QSAR studies. Group—the site is located at the centroid of a group of atoms in the SMARTS query. For aromatic QSAR Model Building rings, the site is assigned directionality, defined A training set of 48 molecules were selected by a vector that is normal to the plane of the ring. randomly, incorporating biological and chemical Active and inactive thresholds were applied to the diversity, and were used to generate atom-based dataset (Table 1.) to yield 11 actives and 12 QSAR models for all hypotheses using a grid inactives which were used for the pharmacophore spacing of 1.0 Ǻ. Models containing four or more generation and subsequent scoring. Common PLS factors tended to fit the pIC50 values beyond pharmacophoric features were then identified their experimental uncertainty and thus, only one, from a set of variants, a set of feature types that two and three factor models were considered. define a possible pharmacophore, using a tree Each of these models were validated using an based partitioning algorithm with maximum tree external test set of fifteen molecules which were depth of 4. The final size of the pharmacophore not considered during model generation. box was 1Å to optimize the number of final common pharmacophore hypotheses (CPH). RESULTS AND DISCUSSION These CPHs were examined using a scoring Pharmacophore models containing three and four function to yield the best alignment of the active sites were generated using a terminal box size of 1 ligands using overall maximum RMSD value of 1.2 Å with twelve highly active molecules selected [6- Å with default options for distance tolerance. The 10] belonging to indazole, pyrazole and their quality of alignment was measured by a survival score, which was defined using following congeners, using a tree based partition algorithm. The three featured CPHs were rejected, as they equation.  Lele et al Pg. 44 Int. J. Pharm. Biosci. Technol. were unable to define the complete binding space Hypotheses emerging from this process were of the selected molecules. A total of 10,487 subsequently scored with respect to the seven probable four featured CPHs belonging to seven inactives, using a weight of 1.0. The hypotheses types HHPR, AHPR, HHRR, AHHR, AHRR, AHHP and which survived the scoring process were used to HPRR were subjected to stringent scoring function build an atom-based QSAR model. The summary analysis, with respect to actives using default of statistical data of the best CPHs labeled CPH1 to parameters for site, vector, and volume. CPH6 with their survival scores is listed in Table 2. Table 2. Summary of QSAR results for six best CPHs with survival scores CPH1 CPH2 CPH3 CPH4 CPH5 (HPRR) (HHPR) (HHRR) (AHHR) (AHRR) Survival score 3.136 3.136 3.092 3.026 2.872 SD 0.422 0.444 0.372 0.443 0.386 r2 0.681 0.647 0.729 0.648 0.699 F 31.3 26.9 35.1 27.1 33.4 P 5.413e-11 4.949e-10 3.686e-11 4.429e-10 2.644e-11 RMSE 0.452 0.466 0.486 0.471 0.501 Q2 0.525 0.510 0.449 0.4842 0.414 Pearson-R 0.741 0.720 0.754 0.736 0.646 SD = Standard deviation of the regression, r2 = Value of r2 for the regression, F = Variance ratio P = Significance level of variance ratio, RMSE = Root-mean-square error, Q2 = Value of Q2 for the predicted activities, Pearson-R = Correlation between the predicted and observed activity for the test set. Consistent external predictivity was observed for set and test set molecules are given in Table 3. CPH1 for each combination as compared to others. Figure 1 shows the alignment of the molecules The CPH1 showed a statistically significant r2 value under study along with CPH1 while Figure 2 shows for training set (0.681), excellent predictive power the gemotry of the pharmacophore generated. The with Q2 of 0.525 and a Pearson-R value of 0.741. distances and angles between the Hence, the hypothesis CPH1 with two aromatic pharmacophoric features are depicted in Figure ring (R), one lipophilic/hydrophobic group (H) 3(a) and 3(b) respectively. Figures 4(a) and Figure and one positive ionic feature (P) as 4(b) shows the graphs of actual v/s. predicted pharmacophoric features was retained for further activity for training and test set molecules. studies. Actual and predicted values of the training Table 3. Actual and predicted values of the dataset Structure Biological Predicted Structure Biological Predicted No. activity activity No. activity activity 1 -2.283 -2.34 33 -0.359 -0.87 2 0.389 0.31 34 0.497 0.98 3 -1.995 -1.80 35 -0.669 -1.28 4 -1.626 -1.76 36 -2.599 -1.49 5 -2.166 -1.56 37 -1.994 -1.53 6 -2.166 -1.92 38 -1.094 -1.51 7 -1.894 -1.72 39 -2.302 -2.34 8 -2.195 -1.83 40 -2.363 -2.29 9 -1.866 -1.72 41 -1.964 -2.25 Lele et al Pg. 45 Int. J. Pharm. Biosci. Technol. 10 -1.866 -1.56 42 -1.013 -1.11 11 -1.551 -1.50 43 -0.969 -1.11 12 -1.577 -1.56 44 -0.99 -1.04 13 -0.949 -1.42 45 -0.966 -0.96 14 -2.211 -1.80 46 -0.359 -0.99 15 -2.474 -1.41 47 -2.265 -2.06 16 -1.874 -1.84 48 -1.267 -0.97 17 -0.937 -1.22 49 -2.283 -2.34 18 -0.949 -1.57 50 -0.969 -0.93 19 -0.937 -1.38 51 -0.668 -0.93 20 -0.359 -0.87 52 -1.571 -1.25 21 -0.345 -0.90 53 -0.68 -0.73 22 -0.991 -1.41 54 -1.011 -1.30 23 -0.991 -1.10 55 -2.138 -1.25 24 -1.273 -0.99 56 -0.654 -0.86 25 -0.972 -1.47 57 -0.969 -1.12 26 -0.57 -0.93 58 -1.312 -0.98 27 -1.267 -0.87 59 -0.955 -1.13 28 -0.646 -0.90 60 -0.968 -1.14 29 -0.66 -0.87 61 -0.653 -1.21 30 -1.867 -0.97 62 -0.668 -0.98 31 -1.561 -0.97 63 -0.654 -1.35 32 -0.966 -1.07 Fig. 1. CPH1 based alignment DNA gyrase Fig. 2. Geometry of the pharmacophore- P9 is inhibitors. the positive ionic feature represented as blue sphere; R11 and R12 are the aromatic ring features represented in orange torus and H4 is the hydrophobic feature represented in green sphere. Lele et al Pg. 46 Int. J. Pharm. Biosci. Technol. Fig. 3(a) CPH1 features and distances. Fig. 4(b). Scatter plots for the predicted and experimental pIC50 values for the DNA gyrase QSAR model applied to the test set. The pharmacophore contains two aromatic ring (R) features mapping on the pyrazole heterocycle and benzene ring, one positive ionic feature mapping on one of the nitrogens in the piperazine ring and one lipophilic/ hydrophobic feature mapping on the para-chloro atom in the distal part of molecule. When the QSAR model was visualized in the context of one of the most active molecule (Structure 21) blue cubes, favorable for the activity, were observed in context of the pharmacophoric features thus corroborating the findings of pharmacophore as observed in Fig. 5. Fig. 3(b) CPH1 features and angles. Fig. 5. QSAR model visualization for DNA gyrase B of S. aureus. CONCLUSION The current work involves development of predictive pharmacophore hypothesis (CPH) for S. Fig. 4(a). Scatter plots for the predicted and aureus DNA gyrase inhibitors using PHASE and experimental pIC50 values for the DNA gyrase performing the QSAR analysis of the model. Best QSAR model applied to the training set. CPH was selected primarily on the basis of its internal and external predictive capabilities. The Lele et al Pg. 47 Int. J. Pharm. Biosci. Technol. generated CPH consists of two aromatic rings, one ACKNOWLEDGEMENTS lipophilic/hydrophobic group and one positive Author A. C. Lele is thankful to Council of Scientific ionic feature. The pharmacophore hypothesis and Industrial Research, Delhi, India and N. R. yielded statistically significant 3D-QSAR model. Tawari is thankful to Department of Biotechnology, The developed pharmacophore model will aid in Delhi, India for financial assistance. the rational design of DNA gyrase inhibitors. REFERENCES 7. Tanitame A, Oyamada Y, Ofuji K, Kyoya Y, 1. Lyon GJ, Mayville P, Muir TW, Novick RP. Suzuki K, Ito H, Kawasaki M, Nagai K, Wachid Rational design of a global inhibitor of the M, Yamagishi J. Potent DNA gyrase inhibitors; virulence response in Staphylococcus aureus, novel 5-vinylpyrazole analogues with Gram- based in part on localization of the site of positive antibacterial activity. Bioorganic & inhibition to the receptor-histidine kinase, Medicinal Chemistry Letters. 2004; AgrC. Proceedings of the National Academy of 14(11):2863–2866. Sciences of the United States of America. 2000; 97(24):13330–13335. 8. Tanitame A, Oyamada Y, Ofuji K, Terauchi H, Kawasaki M, Nagai K, Wachid M, Yamagishi J. 2. Ginzburg E, Namias N, Brown M, Ball S, Synthesis and antibacterial activity of a novel Hameed SM, Cohn SM. Gram positive infection series of DNA gyrase inhibitors: 5-[(E)-2- in trauma patients: new strategies to decrease arylvinyl]pyrazoles. Bioorganic & Medicinal emerging Gram-positive resistance and Chemistry Letters. 2005; 15(19):4299–4303. vancomycin toxicity. International Journal of Antimicrobial Agents. 2000; 16(Supplement 9. Tanitame A, Oyamada Y, Ofuji K, Fujimoto M, 1):39-42. Iwai N, Hiyama Y, Suzuki K, Ito H, Kawasaki M, Nagai K, Wachid M, Yamagishi J. Synthesis and 3. Woods CR. Antimicrobial resistance: Antibacterial Activity of a Novel Series of Mechanisms and strategies. Paediatric Potent DNA Gyrase Inhibitors. Pyrazole respiratory reviews. 2006; 7(Supplement Derivatives. Journal of Medicinal Chemistry. 1):S128–S129. 2004; 47:3693-3696. 4. Maxwell A. DNA gyrase as a drug target. 10. Tanitame A, Oyamada Y, Ofuji K, Fujimoto M, Trends in Microbiology. 1997; 5(3):102-109. Suzuki K, Ueda T, Terauchi H, Kawasaki M, Nagai K, Wachid M, Yamagishi J. Synthesis and 5. Schechner M, Sirockin F, Stote RH, Dejaegere antibacterial activity of novel and potent DNA AP. Functionality Maps of the ATP Binding Site gyrase inhibitors with azole ring. Bioorganic & of DNA Gyrase B: Generation of a Consensus Medicinal Chemistry. 2004; 12(21):5515–5524. Model of Ligand Binding. Journal of Medicinal Chemistry. 2004; 47:4373-4390. 11. PHASE, version 2.0, Schrödinger, LLC, New York, NY (2005) User manual. 6. Tanitame A, Oyamada Y, Ofuji K, Kyoya Y, Suzuki K, Ito H, Kawasaki M, Nagai K, Wachid 12. Wang S, Folkers A, Chuckoweree I, Cockcroft M, Yamagishi J. Design, synthesis and X, Sohal S, Miller W, Milton J, Wren SP, Vicker structure–activity relationship studies of novel N, Depledge P, Scott J, Smith L, Hazel J, Mistry indazole analogues as DNA gyrase inhibitors P, Faint, R, Thompson D, Cocks S. Studies on with Gram-positive antibacterial activity. pyrrolopyrimidines as selective inhibitors of Bioorganic & Medicinal Chemistry Letters. multidrug-resistance-associated protein in 2004; 14(11):2857–2862. multidrug resistance. Journal of Medicinal Chemistry. 2004; 47:1329-1338. Lele et al Pg. 48 Int. J. Pharm. Biosci. Technol. ___________________________________________ How to cite this article APA style Lele, A. C., Tawari, N. R., & Degani, M. S. (2013). Development of pharmacophore model for S. Aureus DNA-gyrase inhibitors. International Journal of Pharma Bioscience and Technology, 1(1), 34–49. Elsevier Harvard style Lele, A.C., Tawari, N.R., Degani, M.S., 2013. Development of pharmacophore model for S. Aureus DNA-gyrase inhibitors. Int. J. Pharm. Biosci. Technol. 1, 34–49. Vancouver Style Lele AC, Tawari NR, Degani MS. Development of pharmacophore model for S. Aureus DNA- gyrase inhibitors. Int. J. Pharm. Biosci. Technol. 2013;1(1):34–49. To receive bibliographic information in RIS format (For Reference Manager, ProCite, EndNote): Send request to: email@example.com ___________________________________________ Lele et al Pg. 49