Structure Based Pharmacophore Modeling and Virtual Screening for Identification of Novel Inhibitor for Cyclooxygenase-2

Saranyah K., Sukesh.K., and Dinesh Kumar K. Department of Bioinformatics,

School of Bioengineering, SRM University, Kattankulathur, Tamil Nadu. India 603203

Email:ksaranyah.biotech@gmail.com

Lilly M. Saleena Assistant Professor, Department of Bioinformatics,

School of Bioengineering, SRM University, Kattankulathur, Tamil Nadu. India 603203

E-mail: lmsaleena@ktr.srmuniv.ac.in

Abstract–Chronic inflammation is a first-rate example of a disease that symbolizes turmoil in normal host defense systems. COX-2, an isoform of COX family, is highly inducible in response to proinflammatory stimuli, cytokines and consequential in exaggerated prostaglandin release. The pharmacophore of known inhibitor S-58 screened with more than 100,000 ligands from Asinex Phase Database using Schrodinger. The best five ligands showed good orientation with the known inhibitor S-58 and further taken for the drug likeness analysis.

Keywords-Cyclooxygenase, Inhibitors, Structure based Pharmacophore

I. INTRODUCTION Cyclooxygenase (COX) enzymes catalyses the first march

in the biogenesis of prostaglandins and are the pharmacological targets of non-steroidal anti inflammatory drugs (NSAIDs) [1]. COX appends two oxygen molecules to arachidonic acid, and leads the commencement a set of reactions that will eventually creates inflammatory. Cyclooxygenase survives as two isoforms, the constitutively produced COX-1 and the inducible COX-2 [2].

COX-2 is expressed in response to inflammatory stimuli and is principally responsible for the striking increase in prostaglandin production at sites of inflammation [3]. COX-2 is induced in response to growth factors, cytokines, and thus it is suggested to be concerned in the production of prostaglandin in inflammatory diseases [4]. A casual relationship between the activity of COX and cancer has been detected thus COX emerged as a molecular target for chemoprevention and a suitable inhibitor need to be designed [5].

COX 2 inhibitors helps in alleviating the epithelial malignancies [6]. COX-1 and COX-2 are inhibited by the NSAIDs randomly and this leads to the ulceration in the gastric parts and additionally affects renal functions [7]. People can procure hypersensitivity to these inhibitor molecules as there are sulpha groups used to develop the inhibitor drugs. Clinically available inhibitors affect COX-2 dependent physiological systems; eases the advance of cardiovascular diseases [8]. To avoid the adverse effects of the prevailing inhibitor drugs efforts are taken for the designing of novel drugs alleviating the ill effects.

II. MATERIALS AND METHODS

A. Protein preparation The structure of the protein COX-2 (1CX2) was retrieved

from the Protein Databank (PDB). In Schrodinger [9], the structure was imported and the protein preparation wizard was used to optimize and minimize the protein [10]. This wizard was executed to help 1CX2 to ensure structural correctness at the outset and equipping the protein with an ideal high-confidence structure. The minimized protein was further taken for the docking and pharmacophore analysis.

B. Grid generation and ligand preparation As the crystallized protein structure from PDB already had

the sites for S58, the grid was generated for the particular S58 ligand considering it a reference ligand. The amino acids binding to the S58 were Ser353, Gln192, Leu352 and Phe518.

The ligand structure of S58 was retrieved from the Protein Databank akin to the COX-2 protein structure. The ligand preparation was then done by the Ligprep module in Schrodinger.

C. Docking studies After the preparation wizard of the protein and the ligand,

they were taken for the docking studies. Using Glide option, the ligand docking was performed. “Write XP descriptor information” was chosen during the docking run which deduces energy terms such as hydrogen-bond interactions, electrostatic interaction, hydrophobic enclosure, and pi–pi stacking interactions and the rest parameters were kept default.

D. Energy based pharmacophore Schrödinger module of energy based pharmacophore (E-

Pharmacophore) was used for the structure based screening of unknown ligands. Using ‘Scripts’ option, the docking post processing was selected and E-Pharmacophore option was specified. There were two choices namely ligand and fragment. In this the ligand mode was performed. The Xpdes result file was then given as an input which contains the information about protein ligand interactions. The result of E- pharmacophore contains the functional groups which are involved in their bioactivity towards target protein.

373

2011 International Conference on Bioscience, Biochemistry and Bioinformatics IPCBEE vol.5 (2011) © (2011) IACSIT Press, Singapore

E. Database screening The above result was then used to screen the Asinex

database containing 100,000 unique structure records [11]. The obtained S-58 pharmocophoric features were then exported using find matches to hypotheses option in the ‘Phase’ tab. The unknown ligands with similar pharmacophoric features were identified and used for virtual screening.

III. RESULTS AND DISCUSSION

A. Glide xp docking for known inhibitor S-58 Docking was performed for COX-2 and S58 ligand. The

glide energy obtained was -52.327 with a Glide score (K cal mol-1 was -11.864 (Fig. 1). There were three hydrogen bond interactions with Arg120, Ser353 and Arg513. The results of this known ligand were further used for the selection of equivalently best ligand molecules matching its structure and functional groups using E- Pharmacophore.

The binding mode of the S-58 ligand to the COX-2 was defined by the interaction pattern of its pharmacophore. The docking result (Xpdes) was then imported to find the structure based pharmacophoric features, which help in finding the best featured functional groups. The results of the E- pharmacophore rank the known ligand to identify the best pharmacophoric features given in below table. The features involved two aromatic rings, two donor molecules and two hydrophobic groups (Table 1 and Fig. 2).

The ranking was not taken as a sole foundation for the selection of the vital functional groups but also the Glide XP docking results were considered in support of the E pharmacophore generated. The hydrophobic group 6 and Donor 4 were found not to hold a vital interaction with COX-2. In case of Donor 5, it had a sturdy interaction in the form of amino group, donating two hydrogen bonds to Ser353 and Leu352. The other features like aromatic group 9 and 10 and also hydrophobic group 7 hold a hydrophobic interaction with protein. Omitting H6 and D4, the other four features were considered for the virtual screening and ligand docking using Asinex database.

The pharmacophoric features were then taken for the screening of known compounds with the Asinex database containing 100,000 compounds. The option implying ‘find matches to the hypothesis in Maestro were used. Thousand hits with pharmacophoric features similar to the known ligand were then obtained.

B. Virtual screening The structurally matched thousand ligands were then taken

for the virtual screening. The Virtual Screening Workflow option is then performed which helps in prefiltering ligands. The virtual screening options for HTVS (High Throughput Virtual Screening), SP (Standard Precision) and Glide XP (extra precision) docking were all checked to be executed. The best scoring ligand obtained from Glide XP docking was the known ligand, S58 itself. Then a total of 18 ligands from SP and 95 ligands from HTVS docking were obtained.

C. Glide xp docking HTVS does primary screening and then secondary is the SP

docking. The survivors of these preliminary screening would progress to Glide XP docking. The results of the above virtually screened ligands with the known ligand (S-58) were further given for Glide XP docking to find hydrogen-bond interactions, electrostatic interaction, hydrophobic enclosure, and pi–pi stacking interactions and root mean square deviation (RMSD). The finest 5 ligands and their hydrogen bond interactions are depicted in Fig. 3. The RMSD and fitness score of the best-scoring pose to the known ligand were reported. The RMSD values were found to be below 1.0A° [12] (Fig. 4) (Table: 2).These hits were further given as an input to MolSoft online software to check their drug like properties [13]. (Table: 3).

IV. CONCLUSION Cyclooxygenases have an atypical orientation within the

membrane and they possess a hydrophobic pocket that projecting inward from the membrane surface of the enzyme [14]. And this hydrophobic pocket self-possesses His 90, Arg 120, Gln 192, Ser 353, Tyr 355, Tyr 385, Arg 513 and Gln 524 and they act as channel for the inhibitor. The known inhibitor S-58 was used to screen the unknown ligands from the Phase database. The hydrogen bond interactions were found between the accurate hydrophobic pocket residues of COX-2. The best hits were then checked for their orientation with the known ligand by using RMSD and these hits were further given as an input to MolSoft online software to check their drug like properties. Hence, compared to the known ligand, the ligands we report were found to accept the Lipinski’s rule of five which further can be taken for in vitro analysis.

ACKNOWLEDGEMENT The authors thank the SRM University for their constant

support and funding.

REFERENCES [1] M. C. Walker, R. G. Kurumbail, J. R. Kiefer, K. T. Moreland, C. M.

Koboldt, P. C. Isakson, K. Seibert and J. K. Gierse, “A three-step kinetic mechanism for selective inhibition of cyclo-oxygenase-2 by diarylheterocyclic inhibitors,” Biochem. J; vol. 357, 2001, pp.709-718.

[2] Chien-Shu Chen, Chen-Ming Tan, Chiung-Hua Huang, Ling-Chu Chang, Jih-Pyang Wang, Fong-Chi Cheng, Ji-Wang, “Discovery of 3-(4-bromophenyl)-6-nitrobenzo [1.3.2] dithiazolium ylide 1, 1-dioxide as a novel dual cyclooxygenase/5-lipoxygenase inhibitor that also inhibits tumor necrosis factor-a production,” Chern. Bioorganic & Medicinal Chemistry; vol. 18, 2010, pp. 597–604.

[3] D. W. Gilroy, P. R. Colville-Nash, “New insights into the role of COX 2 in inflammation,” J Mol Model, vol. 78, 2000, pp. 121–129.

[4] E. Stachowska, B. Dolegowska, V. Dziedziejko, M. Rybicka1, M. Kaczmarczyk, J. Bober, M. Rac, B. Machalinski, D. Chlubek, “Prostaglandin E2 (PGE2) and Thromboxane A2 (TXA2) synthesis is regulated by conjugated linoleic acids (CLA) in human macrophages’” Journal Of Physiology And Pharmacology, vol. 60, 2009, pp. 77–85.

[5] Min Yao, Wei Zhou, Simren Sangha, Andrew Albert, Albert J. Chang, Thomas C. Liu, and M. Michael Wolfe, “Effects of Nonselective Cyclooxygenase Inhibition with Low-Dose Ibuprofen on Tumor Growth, Angiogenesis, Metastasis, and Survival in a Mouse Model of Colorectal Cancer,” Clinical Cancer Research, vol. 11, 2005, pp. 1618–1628.

374

[6] Jiuxiang Zhu, Jui-Wen Huang, Ping-Hui Tseng, Ya-Ting Yang, Joseph Fowble, Chung-Wai Shiau, Yeng-Jeng Shaw, Samuel K. Kulp, and Ching-Shih Chen, “From the Cyclooxygenase-2 Inhibitor Celecoxib to a Novel Class of 3-Phosphoinositide-Dependent Protein Kinase-1 Inhibitors,” Cancer Research, vol. 64, 2004, pp. 4309–4318.

[7] J. R. Vane,Y. S. Bakhle, and R. M. Botting, “Cyclooxygenases 1 and 2,” Annu. Rev. Pharmacol. Toxicol, vol. 38, 1998, pp. 97–120.

[8] S. B.Terrachet, A.T. Maouche, B. Maouche, S. T. Kellou, “Modeling the binding modes of stilbene analogs to cyclooxygenase-2: a molecular docking study,” J Mol Model, vol. 10, 2010, pp. 0679-0677.

[9] N. K. Salam, R. Nuti, W. Sherman, “Novel Method for Generating Structure-Based Pharmacophores Using Energetic Analysis,” J. Chem. Inf. Model, vol. 49, 2009, pp. 2356-2368.

[10] Chenzhong Liao, Rajeshri G. Karki, Christophe Marchand, Yves Pommier, Marc C. Nicklaus, “Virtual screening application of a model of full-length HIV-1 integrase complexed with viral DNA,” Bioorganic & Medicinal Chemistry Letters, vol. 19, 2007, pp. 5361-5365.

[11] ASINEX, ASINEX Platinum Collection, ASINECorp., Winston-Salem, NC, USA, http://www.asinex.com/

[12] Loving, N. K. Salam and W. Sherman, “Energetic analysis of fragment docking and application to structure-based pharmacophore hypothesis generation,” Journal of Computer-Aided Molecular Design, vol. 49, 2009, pp. 2356-68.

[13] Mukesh Yadav, Anuraj Nayarisseri, Girish Singh Rajput, Aditya Jain, Ankit Verma, Priyanka Gupta, “Comparative modeling of 3-oxoacyl-acyl-carrier protein synthase I/II in Plasmodium falciparum– A potent target of malaria,” vol. 2, 2009, pp. 100-103.

[14] David l. Dewitt, “Cox-2-Selective Inhibitors: The New Super Aspirins,” Molecular Pharmacology, vol. 55, 1999, pp. 625–631.

TABLE I. SHOWS THE ENERGY BASED PHARMACOPHORIC FEATURES OF KNOWN INHIBITOR S-58

S.No FEATURE LABEL

SCORE (Rank)

1. R9 1.48

2. R10 1.16

3. D5 0.53

4. H7 0.34

5. H6 0.30

6. D4 0.03

TABLE II. SHOWS THE HYDROGEN BOND INTERACTION AND THE RMSD ORIENTATION VALUES OF BEST FIT LIGANDS

Ligan

d

H bond Interaction

H Bond Distance

Glide Score (K cal mol-1)

Rmsd Value

FitnessScore

66899 ARG 513 HIS 90

2.211 2.436

-8.240 0.10 1.870

34652 SER 353

2.694 2.439

-8.743 0.004 1.831

66883 ARG 513 TYR 385

1.988 1.808

-8.665 0.012 1.970

17847 LEU 352GLN 192

1.795 2.259

-8.312 0.001 1.729

34665 SER 353 2.064 -8.280 0.002 1.710

TABLE III. THE DRUG LIKE PROPERTIES OF UNKNOWN LIGANDS WITH LIPINSKI’S RULE OF FIVE

Ligand Molecular weight

HBA1 HBD2 LogP3

66899 387.1 7 1 0.76

34652 328.16 5 3 0.93

66883 381.17 7 1 1.07

17847 338.12 4 2 3.36

34665 336.13 5 3 2.24

HBA: Hydrogen bond acceptor,

HBD: Hydrogen bond donor,

Log P: Octanol-water partition coefficient

Figure 1. shows hydrogen bond interaction of S-58 (purple) with COX-2.

(H-bonds are shown in black and the ligand is represented in balls and sticks and amino acids as lines.)

375

Figure 2. show the Energy based Pharmacophoric features of S-58. Ring aromatic groups are represented in orange color, green – Hydrophobic and

blue – Hydrogen bond acceptor.

(i) 66899

(ii)34652

(iii) 66883

(iv) 17487

(v) 34665

Figure 3. The following depiction shows the hydrogen bond interaction (purple) of the ligands, and their distances with COX-2

376

Figure 4. shows RMSD (root mean square deviation) orientation of the Unknown Ligands with S-58

377