A molecular docking study of anticancer drug paclitaxel and its analogues

Transcription

1 Indian Journal of Biochemistry & Biophysics Vol. 48, April 2011, pp A molecular docking study of anticancer drug paclitaxel and its analogues Ruma Sinha, Ambarish Sharan Vidyarthi and Shankaracharya* Department of Biotechnology, Birla Institute of Technology, Mesra, Ranchi , India Received 16 August 2010; revised 04 March 2011 Present study was aimed at finding a better alternative to paclitaxel, an anticancer chemotherapeutic drug. Two targets, tubulin β-1 chain and apoptosis regulator Bcl-2 protein (2O2F) were used in the study. Of these, structure of tubulin β-1 chain is not known and that of Bcl-2 was taken from protein data bank with ID 2O2F. Tertiary structure model of tubulin β-1 chain was predicted and validated. The validated 3D structure of tubulin β-1 chain and Bcl-2 protein was taken to study their interaction with paclitaxel. Molecular docking of paclitaxel and its analogues was performed with these targets separately. Results showed that out of 84 analogues taken from PubChem, CID_ had glide score of -9.62, as compared to of paclitaxel with tubulin β-1 chain. It was also observed that CID_ had glide score of -9.0, as compared to of paclitaxel with Bcl-2 protein. However, further experimental and clinical verification is needed to establish these analogues as drug. Keywords: Paclitaxel, Tubulin β-1 chain, Bcl-2, Homology modeling, Docking Cancer is a leading cause of death worldwide and projected to continue rising with an estimate of 12 million deaths in The last decade witnessed the introduction of exciting new chemotherapeutic agents. Among them, paclitaxel, a plant alkaloid (a taxane derivative) obtained from Yew tree (Taxus brevifolia) has emerged as one of the most popular anti-cancer drugs (antineoplastic or cytotoxic) used against a broad spectrum of tumour types, including breast, ovarian, lung, head and neck cancers. It is also active against AIDS-associated Kaposi's sarcoma as an antimicrotubule agent 2-4. Tubulin β-1 chain and Bcl-2 protein are the two known targets of paclitaxel. Paclitaxel binds to the β subunit of tubulin, hence blocking microtubules in place 5. It checks dynamic instability by stabilizing guanosine diphosphate (GDP) bound tubulin in the microtubule. The binding of paclitaxel with GDP hyperstabilizes microtubule structure by destroying cells ability to use its cytoskeleton in a flexible manner 6. Paclitaxel also binds to an apoptosis stopping protein Bcl-2, thus inducing apoptosis. Structure of tubulin β-1 chain is not available till date. Earlier, the homology modeling has been used for generation of 3-D structures of vaccine related *Corresponding author shankaracharya@bitmesra.ac.in shankaracharya2222@gmail.com Tel: , Fax: kinases1 (VRK1) protein 7, cyclin dependent kinase 4 (CDK4) protein 8, tubulin β-1 protein 9 and cell division control protein 2 (CDCP2) protein 10. The quality of the homology model depends on the quality of sequence alignment, template structure 11 and extent of identity between the template and target sequences. In this study, we have used the homology modeling to construct an atomicresolution model of the "target" protein from its amino acid sequence based on an experimental 3-D structure of a related homologous protein (the "template"). Docking program GLIDE has been used to show the interaction between modeled tubulin β-1 chain protein and Bcl-2 target proteins with paclitaxel analogs to find better drug molecules compared to existing drug paclitaxel. Methodology Data source and structures The information on drug paclitaxel was collected from the drugbank database ( and about template sequences of target proteins of paclitaxel from RCSB protein data bank. Analogue structures of paclitaxel were collected from NCBI PubChem database. Generation of 3-D structure through homology modeling Homology modeling was done using Modeller 9v This requires one sequence of known 3D structure with significant similarity with the target

2 102 INDIAN J. BIOCHEM. BIOPHYS., VOL. 48, APRIL 2011 sequence and Python 2.5 script files containing Modeller commands. The co-ordinate file of template from protein data bank was used as such. Evaluation and validation of 3-D structure All predicted five models were evaluated by Procheck 14, What check 15, Errat 16 and Verify_3D 17. Procheck was used to perform full geometric analysis as well as stereochemical quality of a protein structure by analyzing residue-by-residue geometry and overall structure geometry. Ramachandran plot statistics was used to evaluate the stability of the model. Gnuplot was finally used to plot the profiles generated by Modeller ( for substantial validation of the structure. Virtual screening of paclitaxel analogues through molecular docking Paclitaxel and its analogues were taken from NCBI Pubchem database in SDF format and converted to 3D structures using Weblab viewer lite program. The 3D structure of modeled tubulin β-1 chain and Bcl-2 (2O2F) proteins were used for molecular docking with 3D structures of paclitaxel analogues using GLIDE program 18. Results and Discussion In oder to find out a new and effective drug for cancer treatment, molecular modeling of the human tubulin β-1 chain protein was performed and validated. Predicted model of this protein along with Bcl-2 protein was subjected to molecular docking study with 84 paclitaxel analogues and also with original drug molecule of paclitaxel 19 (PubChem compound ID CID_36314). The interaction results of both the proteins and paclitaxel analogues were compared. Search for the template for 3D modeling of tubulin β-1 chain protein resulted into 49 hits. Among them, 1SAO was selected on the basis of lowest E value (0.0), highest resolution (3.58 Å) and maximum identity (77%) 20. Total five models were generated and their discrete optimize potential energy (DOPE) score were calculated using model-single.py modeller script. The model-3 (tub.b pdb, Fig. 1) having minimum score was considered as the best model of tubulin β-1 chain protein (Table 1). Rasmol was used to find the number of helices, strands and turns in the protein. Maximum numbers of helices strands and turns of model tub.b pdb suggested that selected model was more compact than other models, suggesting the better quality of predicted Fig. 1 Modeled structure of tubulin β-1 chain as viewed in visual molecular dynamics (VMD) [Color coding is as follows: alpha helix purple, 3_10_helix blue, extended yellow, coil white and turn cyan] Fig. 2 Gnuplot between the profiles of target (tub.b pdb) and template protein (1SAO) showing the similarity between the patterns of both profiles model (Table 1). Further residues-wise DOPE score of model (tub.b pdb) and template (1SAO.pdb) was plotted using the program gnuplot and result showed that the DOPE energy profile of both proteins was similar to each other (Fig. 2). Further validation of predicted model was performed using a series of programs like Procheck, Verify_3D and Errat. After running Procheck 14, Ramachandran plot (data not shown) showed that for

3 SINHA et al.: A MOLECLAR DOCKING STUDY OF PACLITAXEL AND ITS ANALOGUES 103 Table 1 Comparative study of DOPE score of models, number of H-bonds, strands and turns in five models predicted through modeller and overall quality factor determination through Errat Sl. No. Models predicted by Modeller DOPE score H-bonds Strands Turns Overall quality factor (ERRAT) 1 tub.b pdb tub.b pdb tub.b pdb tub.b pdb tub.b pdb Table 2 Comparative Ramachandran statistics for the validation of the structure of five models generated through modeller Predicted structure Most favored Additional allowed Ramachandran statistics No. of residues in Generously allowed the model tub.b pdb, out of total 398 residues, 350 (87.9%) were in the most favored, 39 (9.8%) in the additional allowed, 7 (1.8%) in the generously allowed and only 2 (0.5%) residues were found in the disallowed, making this model more acceptable, as compared to other predicted models (Table 2). The selected model passed the validation step of Verify_3d 17 program having 81.42% of the residues with averaged 3D-1D score >0.2. Overall quality factor of all predicted models was determined by the program Errat 16. Result confirmed that the model tub.b pdb had the overall quality factor of , which was better than other predicted models (Table 1). What check result had predicted better Ramchandran Z-score, bond angle Z-score and bond length Z-score for model tub.b pdb than other models (Table 2). Further docking study of paclitaxel analogues with the modeled protein tub.b pdb and Bcl-2 Disallowed Percentage in most favored Rama ch-andran Whatif check result Bond angle Bond length tub.b pdb % tub.b pdb % tub.b pdb % tub.b pdb % tub.b pdb % Table 3 Result of docking studies between modeled protein of tubulin β-1 chain and paclitaxel analogues [GLIDE score of analogue CID_ was better than the paclitaxel analogues CID_36314 (indicated as *)] Title G Score vdw Coul Emodel Cvdw Intern Pose CID_ CID_ CID_ CID_ CID_ CID_36314* was performed separately. In the first study, when paclitaxel analogues were docked with the predicted 3D model of the protein tub.b pdb, result showed that analogue CID_ ranked one with the minimum glide score of in comparison to other analogues, as well as paclitaxel (CID_36314) having the glide score of (Table 3). Other binding parameters like van der waals interaction energy (-38.1), Coulomb interaction energy (-13.3), Emodel (-59.6) and cumulative van-der-waals interaction energy (-51.3) of CID_ were found higher than the other analogues, as well as paclitaxel molecule. This suggested the greater binding affinity of the said analogue than others. Similarly in the second docking study, both paclitaxel analogues and paclitaxel were docked with the other target Bcl-2 protein (pdb id 2O2F) and the drug itself. Results showed that analogue CID_ at pose 7 had the minimum glide score of -9.00, as compared to other analogues as well as

4 104 INDIAN J. BIOCHEM. BIOPHYS., VOL. 48, APRIL 2011 Table 4 Result of docking studies between Bcl-2 protein (2O2F.pdb) and paclitaxel analogues [GLIDE score of analogue CID_ at pose 7 was better than the paclitaxel analogue CID_36314 (indicated as *)] Title GScore vdw Coul Emodel Cvdw Intern Pose CID_ CID_ CID_ CID_ CID_ CID_36314* Fig. 3 Docked result showing the interaction between active site residues of tubulin β-1 chain and ligand CID_ the drug paclitaxel (glide score -8.24). Also, other binding parameters such as van der waals interaction energy (-22.6), Emodel (-46.3) cumulative van der waals interaction energy (-32.5) and internal energy (18.0) of CID_ were found to be higher than the other analogues, as well as paclitaxel molecule (Table 4), suggesting the greater binding affinity of the said analogue than others. In the predicted model of the protein tub.b pdb, active site residues included Gly15, Asp177, Ile16, Asn226, Leu225, Val169, Thr143, Ala97, Gln11, Glu69 and Ser138. Hydrogen bonds play an important role for structure and function of biological molecules, especially for the enzyme catalysis 21. In binding mode, two hydrogen bonds of length Å and Å were detected between the ligand and protein residues at Gln15 and Ile10 respectively (Fig. 3). Whereas in the second docking study between Bcl-2 and paclitaxel analogues, it was observed that the protein s active site residues included Phe150, Trp141, Asn140, Asp109, Gly142, Ala146, Leu134, and Glu133. Total three hydrogen bonds of length Å, Å and Å were formed between the ligand and protein residues at Asp109, Glu133 and Asn140, respectively (Fig. 4). Fig. 4 Docked result showing the interaction between active site residues of Bcl-2 protein and ligand CID_ Earlier, similar study was reported for the structure modeling of VRK1 protein and molecular docking study with ribavirin analogues, wherein validation of model resulted in 97.5% residues in most favored and additionally allowed, whereas in our result 97.75% residues were in this 5. Conclusion In the present study, two target proteins tubulin β-1 chain and Bcl-2 of an anticancer chemotherapeutic drug paclitaxel was used for drug target interaction study. 3D Structure of tubulin β-1 chain was predicted and validated by using bioinformatics tools. Predicted model had 97.7% residues in favored and allowed of Ramachandran plot along with overall quality factor of Further, modeled 3D structure of tubulin β-1 chain and Bcl-2 (PDB 202F) was used to dock paclitaxel and its analogues using a flexible docking method of GLIDE. Result showed that among 84 analogues, CID_ had glide score of in comparison to of paclitaxel and CID_ had glide score of -9.0 in comparison to of paclitaxel. Thus, the analogues CID_ and CID_ were found to be more suitable than the existing drug paclitaxel

5 SINHA et al.: A MOLECLAR DOCKING STUDY OF PACLITAXEL AND ITS ANALOGUES 105 (CID_36314). This study may be the subject of experimental validation and clinical trial to establish these said analogues as more potent drug for the treatment of different cancers. Acknowledgement The authors are thankful to the Sub-Distributed Information Center (BTISnet SubDIC), Department of Biotechnology (No. BT/BI/04/065/04), New Delhi, India for financial support and to the Departments of Biotechnology and Pharmaceutical Sciences, Birla Institute of Technology, Mesra, Ranchi for providing the access of the softwares. References 1 Chang Y & Walkom E (2008) 17 th Expert Committee on the Selection and Use of Essential Medicines, Geneva, pp Wani M, Taylor H, Wall M, Coggon P & McPhail A (1971) J Am Chem Soc 93, Goodman J & Walsh V (2001) The Story of Taxol: Nature and Politics in the Pursuit of an Anti-Cancer Drug, 1 st edn, pp , Cambridge University Press, London 4 Hanahan D & Weinberg R A (2000) Cell 100, Parness J, Horwitz S B (1981) J Cell Biol 91, Cheung C H, Chen H H, Kuo C C, Chang C Y, Coumar M S, Hsieh H P & Chang J Y (2009) Mol Cancer 3, Shankaracharya, Srivastava B & Vidyarthi A S (2010) Int J Pharm Bio Sci 1, Shankaracharya, Priyamvada & Vidyarthi A S (2010) PharmBIT, 21, Shankaracharya, Sharma P & Vidyarthi A S (2010) Int J Pharm Tech 3, Shankaracharya, Priyamvada & Vidyarthi A S (2010) Int J Pharm Sci Rev Res 4(1), Martí-Renom M A, Stuart A C, Fiser A, Sánchez R, Melo F & Sali A (2000) Annu Rev Biophys Biomol Struct 29, Fiser A & Sali A (2003) Meth Enzymol 374, Sali A & Blundell T L (1993) J Mol Biol 234, Laskoswki R A, MacArthur M W, Moss D S & Thorton J M (1993) J Appl Cryst 26, Vriend G (1990) Protein Eng 4, Colovos C & Yeates T O (1993) Protein Sci 2, Eisenberg D, Lüthy R & Bowie J U (1997) Meth Enzymol 277, Friesner R A, Banks J L, Murphy R B, Halgren T A, Klicic J J, Mainz D T, Repasky M P, Knoll E H, Shelley M, Perry J K, Shaw D E, Francis P & Shenkin P S (2004) J Med Chem 47, Kaiser J (2005) Science 308, Sinha R (2010) A molecular docking study of paclitaxel and its analogues, Master s thesis, Birla Institute of Technology, Mesra, Ranchi, India 21 Crabtree R H (1998) Science 282, 5396