DOI: 10.1002/minf.201200113 Molecular Dynamics Investigation on the Inhibition of MDM2-p53 Interaction by Polyphenols Sharad Verma, [a] Amit Singh, [b] and Abha Mishra* [a] 1 Introduction The p53 (tumor suppressor protein) is one of the key play- ers which regulate the apoptosis, cell cycle, and DNA repair to protect cells from malignant transformation. [1–3] However, activity of p53 is regulated by MDM2. MDM2 inhibits the ability of p53 to bind to DNA. The over expressed MDM2 has been reported in tumors which led to down regulated tumor suppressor activity of p53. [4] The interaction of MDM2 and p53 governed by insertion hydrophobic resi- dues (Phe19, Trp23 and Leu26) of p53 into a deep groove of MDM2. [5] Many peptide inhibitors that mimic the MDM2– p53 interaction have been reported but these inhibitors display only modest potency due to poor membrane per- meability. [6–10] Several different small-molecule inhibitors have been designed by structure-based methods to inter- rupt the binding of p53 to MDM2 which mainly include Nutlins (based on cis-imidazolidine), [1,11] benzodiazepine- dione derivatives [12,13] and spirooxindole. [14,15] In recent years, naturally occurring polyphenolic phytochemicals such as taxifolin and quercetin (Figure 1) have remained completely ignored in this regard. The tremendous poten- tial of taxifolin and quercetin to inhibit the cancer is well understood. [16,17] In this study we tried to elucidate the effect of these polyphenols on Apo MDM2 as well as on MDM2-p53 complex with the help of molecular docking and molecular dynamic simulation along with the possible mode of action. 2 Computational Methods 2.1 Approach 1 AutoDock 4.0 suite was used as molecular-docking tool in order to carry out the docking simulations. PDB id: 1T4E, obtained from RCSB protein data bank, was used as initial structure for MDM2. The structure of ligands (quercetin and taxifolin) was generated from smile strings followed by energy minimization. Hydrogen atoms were added to pro- tein crystal structures using autodock program while all non polar hydrogen atoms were merged. Lamarckian ge- netic algorithm was used as a search parameter which is based on adaptive local search. Short range van der Waals [a] S. Verma, A. Mishra School of Biochemical Engineering, Indian Institute of Technology, Banaras Hindu University Varanasi-221005, India tel.: + 915422307070 *e-mail: [email protected][b] A. Singh Department of Pharmacology, Institute of Medical Sciences, Banaras Hindu University Varanasi-221005, India Supporting Information for this article is available on the WWW under http://dx.doi.org/10.1002/minf.201200113. Abstract : Inhibition of the MDM2-p53 interaction has become a new therapeutic strategy to activate wild type p53 in tumors. Quercetin and taxifolin bind to p53 binding hydrophobic groove of MDM2, and alter the conformation of groove as evidenced by 65 ns molecular dynamics simu- lation. Quercetin showed hydrogen bonding with Gly 16, Ser 17, Phe 55 and Val 93 along with p–p interaction with His96 and p–s with Phe 55. Taxifolin also showed similar in- teractions except p–s interaction with Phe 55. Further, we found that binding of ligands lead to the dissociation of MDM2–p53 complex. These ligands form stable hydropho- bic interactions with MDM2 which led to complete disrup- tion of MDM2-p53 hydrophobic interactions and dissocia- tion of p53 from the complex. It was found that the p–p stacking between Tyr 51 of MDM2 and ligands is the critical event in MDM2-p53 dissociation. Keywords: MDM2-p53 · Taxifolin · Quercetin · Molecular dynamics simulation · Hydrophobic groove Figure 1. Structure of quercetin and taxifolin. Mol. Inf. 2013, 32, 203 – 212 # 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 203
Transcript
DOI: 10.1002/minf.201200113
Molecular Dynamics Investigation on the Inhibition ofMDM2-p53 Interaction by PolyphenolsSharad Verma,[a] Amit Singh,[b] and Abha Mishra*[a]
1 Introduction
The p53 (tumor suppressor protein) is one of the key play-ers which regulate the apoptosis, cell cycle, and DNA repairto protect cells from malignant transformation.[1–3] However,activity of p53 is regulated by MDM2. MDM2 inhibits theability of p53 to bind to DNA. The over expressed MDM2has been reported in tumors which led to down regulatedtumor suppressor activity of p53.[4] The interaction ofMDM2 and p53 governed by insertion hydrophobic resi-dues (Phe19, Trp23 and Leu26) of p53 into a deep grooveof MDM2.[5] Many peptide inhibitors that mimic the MDM2–p53 interaction have been reported but these inhibitorsdisplay only modest potency due to poor membrane per-meability.[6–10] Several different small-molecule inhibitorshave been designed by structure-based methods to inter-rupt the binding of p53 to MDM2 which mainly includeNutlins (based on cis-imidazolidine),[1,11] benzodiazepine-dione derivatives[12,13] and spirooxindole.[14,15] In recentyears, naturally occurring polyphenolic phytochemicalssuch as taxifolin and quercetin (Figure 1) have remainedcompletely ignored in this regard. The tremendous poten-tial of taxifolin and quercetin to inhibit the cancer is well
understood.[16,17] In this study we tried to elucidate theeffect of these polyphenols on Apo MDM2 as well as onMDM2-p53 complex with the help of molecular dockingand molecular dynamic simulation along with the possiblemode of action.
2 Computational Methods
2.1 Approach 1
AutoDock 4.0 suite was used as molecular-docking tool inorder to carry out the docking simulations. PDB id: 1T4E,obtained from RCSB protein data bank, was used as initialstructure for MDM2. The structure of ligands (quercetin andtaxifolin) was generated from smile strings followed byenergy minimization. Hydrogen atoms were added to pro-tein crystal structures using autodock program while allnon polar hydrogen atoms were merged. Lamarckian ge-netic algorithm was used as a search parameter which isbased on adaptive local search. Short range van der Waals
[a] S. Verma, A. MishraSchool of Biochemical Engineering, Indian Institute of Technology,Banaras Hindu UniversityVaranasi-221005, Indiatel. : + 915422307070*e-mail : [email protected]
[b] A. SinghDepartment of Pharmacology, Institute of Medical Sciences,Banaras Hindu UniversityVaranasi-221005, India
Supporting Information for this article is available on the WWWunder http://dx.doi.org/10.1002/minf.201200113.
Abstract : Inhibition of the MDM2-p53 interaction hasbecome a new therapeutic strategy to activate wild typep53 in tumors. Quercetin and taxifolin bind to p53 bindinghydrophobic groove of MDM2, and alter the conformationof groove as evidenced by 65 ns molecular dynamics simu-lation. Quercetin showed hydrogen bonding with Gly 16,Ser 17, Phe 55 and Val 93 along with p–p interaction withHis96 and p–s with Phe 55. Taxifolin also showed similar in-
teractions except p–s interaction with Phe 55. Further, wefound that binding of ligands lead to the dissociation ofMDM2–p53 complex. These ligands form stable hydropho-bic interactions with MDM2 which led to complete disrup-tion of MDM2-p53 hydrophobic interactions and dissocia-tion of p53 from the complex. It was found that the p–p
stacking between Tyr 51 of MDM2 and ligands is the criticalevent in MDM2-p53 dissociation.
and electrostatic interactions, hydrogen bonding, entropylosses were included for energy based autodock scoringfunction.[18,19] The Lamarckian GA parameters used in thestudy were numbers of run, 30; population size, 150; maxi-mum number of eval; 25 000 000, number of generation;27 000, rate of gene mutation; 0.02 and rate of cross over;0.8. Blind docking was carried out using grid size 126, 126and 126 along the X, Y and Z axes with 0.375 � spacing.RMS cluster tolerance was set to 2.0 �. Semi-flexible dock-ing was performed which includes a flexible ligand anda rigid receptor. All the protein and ligand structuralimages were generated using PYMOL.[20]
MD simulation of the complex was carried out with theGROMACS4.5.4 package using the GROMOS96 43a1 forcefield.[21,22] The lowest binding energy (most negative) dock-ing conformation generated by Autodock was taken as ini-tial conformation for MD simulation. The topology parame-ters of proteins were created by using the Gromacs pro-gram. The topology parameters of taxifolin and quercetinwere built by the Dundee PRODRG server[23] The complexwas immersed in an octahedron box of simple pointcharge (SPC) water molecules.[24,25] The solvated system(MDM2, ligand and water) was neutralized by adding 4 Clions in all simulation. To release conflicting contacts, energyminimization was performed using the steepest descentmethod of 10 000 steps followed by the conjugate gradientmethod for 10 000 steps. MD simulation studies consist ofequilibration and production phases. To equilibrate thesystem, the solute (protein, counterions, and ligand) weresubjected to the position-restrained dynamics simulation(NVT and NPT) at 300 K for 300 ps. Finally, the full systemwas subjected to MD production run at 300 K temperatureand 1 bar pressure for 65 000 ps. For analysis, the atom co-ordinates were recorded at every 1.0 ps during the MD sim-ulation.
2.2 Calculation of Binding Free Energy
The binding free energies were calculated using molecularmechanics/Poisson-Boltzman surface area (MMPBSA) ap-proach [26–29] supplied with Amber 10 package. We choosea total number of 200 snapshots evenly from the last 10 nson the MD trajectory. The MM-PBSA method can be con-ceptually summarized as:
DGbind ¼ DGcomplex�½DGprotein þ DGlig� ð1Þ
DGbind ¼ DH�TDS, ð2Þ
where DH of the system is composed of the enthalpychanges in the gas phase upon complex formation (DEMM)and the solvated free energy contribution (DGsol), while �TDS refers to the entropy contribution to the binding.Equation 2 can then be approximated as shown in Equa-tion 3:
DGbind ¼ DEMM þ DGsol�TDS ð3Þ
where DEMM is the summation of the van der Waals (DEvdw)and the electrostatic (DEele) interaction energies.
DEMM ¼ DEvdwþDEele ð4Þ
In addition, DGsol, which denotes the solvation freeenergy, can be computed as the summation of an electro-static component (DGele,sol) and a nonpolar component(DGnonpolar,sol), as shown in Equation 5:
DGsol ¼ DGele,solþDGnonpolar,sol ð5Þ
2.3 Approach 2
In approach 2 MDM2-p53 complex was used as initial struc-ture (PDB id: 1YCQ) for performing molecular docking andmolecular dynamics simulation. Methodology used wassame as Approach 1. The lowest binding energy (most neg-ative) docking conformation generated by Autodock wastaken as initial conformation for MD simulation. The solvat-ed system (MDM2-p53, ligand and water) was neutralizedby adding 5 Cl ions in all simulation. The full system wassubjected to MD production run at 300 K temperature and1 bar pressure for 15 000 ps. 20 000 ps MD simulation ofMDM2-p53 complex was performed without ligand in samecondition as control.
3 Result and Discussion
3.1 Approach 1
Molecular docking results revealed that quercetin and taxi-folin bound to hydrophobic groove of MDM2 with bindingenergy �28.56 and �30.66 kJ/mol respectively (Fig-ure 2 a,b). Quercetin showed hydrogen bonding with Gly16, Ser 17, Phe 55 and Val 93 along with p–p interactionwith His96 and p–s with Phe 55. Taxifolin showed similarinteraction as quercetin except p–s interaction with Phe55.
The lowest binding energy (most negative) docking con-formation generated by Autodock was taken as initial con-formation for MD simulation. We have analyzed the timedependent behavior of MD trajectories for MDM2-ligandcomplex including root mean square deviation (RMSD) forall backbone atoms. Figure 3 A showed that the RMSD pro-files were always less than 0.25 nm for both quercetin andtaxifolin bound MDM2 backbone during the entire simula-tion suggesting the suitability of MD simulation run forpost analysis. Figure 3 B showed the RMSD profile of ligandsbound to MDM2 pocket. Quercetin and taxifolin showedstable profile throughout the simulation. These resultsshowed the stable binding of ligands in the hydrophobicpocket of MDM2.
Figure 2. (a) Quercetin bound to hydrophobic groove of MDM2 and 2D presentation of interaction with residues. (b) Taxifolin bound tohydrophobic groove of MDM2 and 2D presentation of interaction with residues.
Figure 3. (a) Plot of root mean square deviation (RMSD) of backbone of MDM2 complexed with quercetin (black) and with taxifolin (grey).(b) Plot of root mean square deviation (RMSD) of quercetin (black) and taxifolin (grey) in hydrophobic groove of MDM2.
Comparative analysis of final pose of MDM2-ligand com-plex after 65 ns molecular dynamics simulation with crystalstructure of MDM2 revealed that binding of the quercetinand taxifolin bring significant conformational changes inthe MDM2 structure. Figure 4 clearly indicates the distor-tion of hydrophobic groove in the ligand bound MDM2.However quercetin showed higher effect on the grooveand found to be completely overlapped by N-terminal loopin quercetin bound MDM2. All these results indicate thatthese polyphenols have potential to inhibit the MDM2binding with p53 by occupying and distorting structure ofhydrophobic groove. The binding pocket, although well de-fined in apo MDM2, undergoes profound conformationalchanges upon ligand binding.[30,31] The structure of MDM2complexed to a small molecule inhibitor of MDM2-p53 in-teraction, chromenotriazolopyrimidine (PDB: 3JZK),[32] dem-onstrates that the compound and the p53 peptide sharethe same binding cavity and induce similar conformationchanges upon binding.[33] These previous studies furthersupport the inhibitory potential of quercetin and taxifolin.These polyphenols stably occupied the p53 binding sitesimilar to chromeno-triazolopyrimidine (Figure 4). Further2D plots[34] of interaction of final pose of 65 ns MD simula-tion was generated and compared with chromeno-triazolo-
pyrimidine and another inhibitor benzodiazepine.[35] It wasfound that binding pattern of quercetin and taxifolin wasvery similar to the both known inhibitors as shown inFigure 5.
Figure 4. Comparison of last pose of 65 ns MD simulation ofMDM2 in ligands bound form with apo and chromeno-triazolopyri-midine complexed MDM2.
Figure 5. Comparison of interaction of last pose of 65 ns MD simulation of MDM2 with ligands and benzodiazepine, and chromeno-triazo-lopyrimidine complexed MDM2.
In addition, the total binding free energy for the MDM2-ligand complexes and their detailed energy contributionscalculated according to the MM-PBSA approach, are sum-marized in Table 1. The DGbind can be divided into polar
(DGele,sol +DEele) and nonpolar energies (DGnonpolar +DEvdw).The free energy of quercetin and taxifolin binding toMDM2 is primarily derived from the DGnonpolar +DEvdw ,while the DGele,sol +DEele shows a likely unfavorable contri-bution. This is due to the intermolecular van der Waals en-ergies, which is mainly achieved from the ligand bindingresidues. These results suggest that quercetin and taxifolincan bind to MDM2, dominantly, through van der Waalsenergy. As compared to previously known inhibitors[36] ,
quercetin and taxifolin showed remarkable similarity inbinding pattern and affinity for the hydrophobic groove ofMDM2.
3.3 Approach 2
Taxifolin and quercetin were found to bind at interface ofMDM2-P53 complex with lowest binding energy �34.52 kJ/mol and �34.81 kJ/mol respectively. These results indicatethat both ligands have high affinity for the MDM2-P53 in-terface. The major interactions shown in the MDM2-p53 in-terface are the important H-bonds with residues Lys 24and, Leu 26 of p53 and Tyr 51 and, Gln 55 of MDM2. Thegroups involved in H-bonding were hydroxyl (hydrogendonar) and carbonyl (hydrogen acceptor) group of taxifolinand quercetin (Supporting Information (SI) Figure SI-1).
The MDM2-p53-taxifolin and MDM2-p53-quercetin com-plex with the binding energy of �34.52 kJ/mol and�34.81 kJ/mol, respectively, obtained using Autodock wasused for carrying out MD simulation. Figure 6 a shows thatthe RMSD trajectory was always less than 0.25 nm up to~7000 ps for taxifolin bound form. A high rise in the RMSDwas observed at ~7500 ps and subsequently a constantprofile was observed with up and down for very small timeintervals. This increase in RMSD was found to be in thegood agreement with the snapshots recorded at and after7500 ps (described later) which revealed the separation ofp53 segment with MDM2. Analysis of taxifolin RMSD indi-cates that taxifolin showed remarkable stability at interface
Table 1. Calculated energy components, binding free energy (kcal/mol) of quercetin and taxifolin binding to MDM2.
Energy components (kcal mol�1) Quercetin Taxifolin
Figure 6. (a) Plot of root mean square deviation (RMSD) of backbone of MDM2-p53 complexed with taxifolin (grey), MDM2-p53 complexedwith quercetin (black) and without ligand (red). (b) Plot of root mean square deviation (RMSD) of taxifolin (grey) and quercetin (black).
of MDM2-p53 complex during MD simulation. However,some high fluctuations were during the simulation (Fig-ure 6 b). In quercetin bound form, the RMSD trajectories ofbackbone showed high increase from the value of 0.25 nmafter ~10 000 ps and subsequently a more or less constantprofile was observed (Figure 6 a). Similar to taxifolin, this in-crease of RMSD was found to be associated with the disso-ciation of MDM2-p53 complex (described later). Analysis ofquercetin RMSD indicated that quercetin showed stabilityat interface of MDM2-p53 complex during MD simulation.However, increase in RMSD was observed after ~13 000 ps(Figure 6 b). This increase in RMSD indicates the dissociationof quercetin along with p53. MDM2-p53 complex in ab-sence of any ligand showed stability as compared to ligandbound complex (Fig. 6 a). Previously, Espinoza-Fonseca and
Garc�a-Machorro (2008) showed the stability of MDM2-p53complex during a long MD simulation using same PDBstructure.[37]
Number of H-bonds (cut off 0.35 nm) which were formedduring MD simulation between ligands and MDM2-p53 wasalso calculated. A variable profile was observed which fluc-tuate between 0 to 3 with an average value of 0.15 and, 0to 4 with average value 0.25 for taxifolin and quercetin re-spectively (Figure 7). Furthermore, to identify the flexibleresidues of the protein, Root Mean Square Fluctuation(RMSF) of backbone atoms from its time averaged positionwas analyzed. All the residues of both MDM2 and p53showed marked higher fluctuation in taxifolin and querce-tin bound forms as compared to unbound form. The p53residues showed higher fluctuation increased by ~1 nm as
Figure 7. (a) Number of H-bonds formed between taxifolin and MDM2-p53 interface residues. (b) Number of H-bonds formed betweenquercetin and MDM2-p53 interface residues during 12 000 ps MD simulation.
Figure 8. RMSF of MDM2-p53 residues backbone in taxifolin bound (black), quercetin bound (grey) and ligand unbound form (light grey).
compared to ligand unbound form (Figure 8). This profileconfirmed that ligand binding induced movement in theresidues of protein complex.
Coordinates of the MDM2-p53-taxifolin system recordedat different time interval of simulation revealed that bind-ing of the MDM2-p53 remained intact for ~7000 ps. there-after, p53 segment detached from the MDM2 cleft as ob-served in the 12 000 ps and final 14 000 ps snapshots(Figure 9). These results were favored by the RMSD andRMSF profile described earlier. In case of MDM2-p53-quer-cetin major changes were observed ~10 ns. p53 segmentalong with quercetin leave the MDM2 binding site as ob-served in 14 ns snapshot while at 12 ns quercetin wasfound bound to MDM2 (Figure 10). Analysis of the MDM2-p53 complex (without ligand) during simulation revealedthat only few residues contribute to the interaction ofthese two proteins. Phe 19, Trp 23 and Leu 26 of P53 werethe residues which oriented toward the MDM2 bindingcleft. The importance of these residues was previously de-scribed several times.[5,36] In case MDM2 Lys 47, Tyr 51, Gln55 and Met 58 were found critical residues involved in in-teraction with p53. The interaction of these two proteins ismainly dependent on the hydrophobic interactions.[5,36] Tofound the effect of ligands on these residues, we analyzedsnapshots of simulation at different time interval. The 2Dplot of taxifolin and MDM2-p53 interaction at different timeinterval of MD simulation showed and confirmed that thehydrophobic interactions were dominated during simula-tion. Initially, taxifolin showed p–p interaction with both
MDM2 and p53 which finally, completely, switched to p–p
and, cation-p interaction between taxifolin aromatic ring Aand Tyr 51 and, Lys 47 respectively (Figure 11). In this wayinteraction of Trp 23 of p53 and Tyr 51 of MDM2 trans-formed into Trp 23-B-C ring and, Tyr 51-A ring interactionand finally in to A ring of taxifolin and Tyr 51and, Lys 47 ofMDM2. The 2D plot of quercetin and MDM2-p53 interactionalso showed that initially, quercetin involved in p–p interac-tion with MDM2 by C ring. At 12 000 ps B and, C-ring ofquercetin found in p–p interaction wirh Tyr 51 (Figure 12).
It is well established that interaction of MDM2 and p53 isgoverned by hydrophobic groups of residues. The presenceof aromatic groups in the ligands was found be the mainreason behind the masking of the interaction between pro-teins. Further, Tyr 51 residue was found to play lead role ininteraction with p53 as the masking of Tyr 51 hydrophobicside chain by ligands led to separation of p53. Both ap-proaches used in the study were found to be supported byeach other. Being natural polyphenols, antioxidant andpresence in large number of fruits, vegetables and beverag-es, these polyphenols may be used as lead compound forcancer prevention.
4 Conclusions
Quercetin and taxifolin bound to the hydrophobic grooveof MDM2 and alter the conformation of groove as evi-denced by 65 ns molecular dynamics simulation. Both com-
Figure 9. Snap shots at different time interval of MDM2-p53 complexed with taxifolin for 15 000 ps MD simulation showing dissociation ofcomplex.
pounds were found to inhibit the MDM2 and p53 interac-tion as evidenced by molecular dynamic simulation. The in-teraction of MDM2 and p53 is dominantly governed by thehydrophobic residues. Taxifolin and quercetin efficientlymask these interactions led to separation of p53. The hy-dropobic aromatic group system of ligands mainly contrib-uted to this action. Being the natural compounds and theirbioavailability in natural food products, these phytocemi-cals can be used to target MDM2 and p53 interaction inapo as well as complex form.
Acknowledgement
One of the authors is thankful of the Council of Scientificand Industrial Research (CSIR), India for providing a SeniorResearch Fellowship.
References
[1] L. T. Vassilev, B. T. Vu, B. Graves, D. Carvajal, F. Podlaski, Z. Fili-povic, N. Kong, U. Kammlott, C. Lukacs, C. Klein, N. Fotouhi,E. A. Liu, Science 2004, 303, 844 – 848.
[2] B. Vogelstein, D. Lane, A. J. Levine, Nature 2000, 408, 307 –310.
[3] A. J. Levine, Cell 1997, 88, 323 – 331.[4] S. G. Dastidar, D. P. Lane, C. S. Verma, BMC Bioinfo. Pro. , 2009,
10(Suppl. 15), 6.[5] J. G. Allen, M. P. Bourbeau, G. E. Wohlhieter, M. D. Bartberger,
K. Michelsen, R. Hungate, R. C. Gadwood, R. D. Gaston, B.Evans, L. W. Mann, M. E. Matison, S. Schneider, X. Huang, D.Yu, P. S. Andrews, A. Reichelt, A. M. Long, P. Yakowec, E. Y.Yang, T. A. Lee, J. D. Oliner, J. Med. Chem. 2009, 52 , 7044 –7053.
[6] A. D. Bautista, J. S. Appelbaum, C. J. Craig, J. Michel, A. Sche-partz, J. Am. Chem. Soc. 2010, 132, 2904 – 2906.
[7] R. Fasan, R. L. A. Dias, K. Moehle, O. Zerbe, J. W. Vrijbloed, D.Obrecht, J. A. Robinson, Angew. Chem. Int. Ed. 2004, 43, 2109 –2112.
[8] J. A. Kritzer, J. D. Lear, M. E. Hodsdon, A. Schepartz, J. Am.Chem. Soc. 2004, 126, 9468 – 9469.
Figure 12. 2D plots of interaction between quercetin and MDM2-p53 at different time interval of 15 000 ps MD simulation.
[9] R. Stoll, C. Renner, S. Hansen, S. Palme, C. Klein, A. Belling, W.Zeslawski, M. Kamionka, T. Rehm, P. Muhlhahn, R. Schumacher,F. Hesse, B. Kaluza, W. Voelter, R. A. Engh, T. A. Holak, Biochem-istry 2001, 40, 336 – 344.
[10] J. Zhao, M. Wang, J. Chen, A. Luo, X. Wang, M. Wu, D. Yin, Z.Liu, Cancer Lett. 2002, 183, 69 – 77.
[11] P. Chene, Nat. Rev. Cancer 2003, 2, 102 – 109.[12] G. M. Popowicz, A. Czarna, S. Wolf, K. Wang, W. Wang, A.
Dçling, T. A. Holak, Cell. Cyc. 2010, 9, 1104 – 1111.[13] H. K. Koblish, S. Zhao, C. F. Franks, R. R. Donatelli, R. M. Tomino-
vich, L. V. LaFrance, K. A. Leonard, J. M. Gushue, D. J. Parks,R. R. Calvo, K. L. Milkiewicz, J. J. Marug�n, P. Raboisson, M. D.Cummings, B. L. Grasberger, D. L. Johnson, T. Lu, C. J. Molloy,A. C. Maroney, Mol. Cancer. Therap. 2006, 5, 160 – 169.
[14] S. Shangary, S. M. Wang, Clin. Can. Res. 2008, 14, 5318 – 5324.[15] S. Shangary, S. M. Wang, Ann. Rev. Pharmacol. Toxicol. 2009,
49, 223 – 241.[16] S. B. Lee, K. H. Cha, D. Selenge, A. Solongo, C. W. Nho, Biol.
Pharm. Bull. 2007, 30, 1074 – 1079.[17] A. M. Seufi, S. S. Ibrahim, T. K. Elmaghraby, E. E. Hafez, J. Exp.
Clin. Cancer Res. 2009, 11, 80.[18] G. M. Morris, D. S. Goodsell, R. S. Halliday, R. Huey, W. E. Hart,
R. K. Belew, A. J. Olson, J. Comput. Chem. 1998, 19, 1639 –1662.
[19] G. M. Morris, R. Huey, W. Lindstrom, M. F. Sanner, R. K. Belew,D. S. Goodsell, A. J. Olson, J. Comput. Chem. 2009, 30, 2785 –2791.
[20] W. L. DeLano, DeLano Scientific LLC, San Carlos, CA, USA 2004,http://www.pymol.org.
[21] H. J. C. Berendsen, D. Van der Spoel, R. Van Drunen, Phys.Commun. 1995, 91, 43 – 56.
[22] E. Lindah, B. Hess, D. Van der Spoel, J. Mol. Model. 2001, 7,306 – 317.
[23] A. W. Schuttelkopf, D. M. F. van Aalten, Acta Crystallogr. 2004,60, 1355 – 1363
[24] W. F. Van Gunsteren, S. R. Billeter, A. A. Eising, P. H. Hunen-berger, P. K. H. C. Kruger, A. E. Mark, W. R. P. Scott, I. G. Tironi,
Biomolecular Simulation: The GROMOS96 Manual and UserGuide, Vdf Hochschulverlag AG, Z�rich, 1996.
[25] W. F. Van Gunsteren, X. Daura, A. E. Mark, “The GROMOS forcefield”, in Encyclopedia of Computational Chemistry, Vol. 2 (Ed: P.von Rague Schleyer), Wiley, Chichester, UK, 1998, 1211–1216.
[26] I. Vorontsov Ivan, O. Miyashita, J. Comput. Chem. 2011, 32,1043 – 1053.
[27] A. S. Jogalekar, S. Reiling, R. J. Vaz, Bioorg. Med. Chem. Lett.2010, 20, 6616 – 6619.
[28] J. M. J. Swanson, R. H. Henchman, J. A. McCammon, Biophys. J.2004, 86, 67 – 74.
[29] R. Thanyada, N. Nadtanet, M. Maturos, K. Nopporn, Y. Pathum-wadee, M. Arthitaya, H. Supot, J. Mol. Graph. Model. 2010, 29,347 – 353.
[30] S. Uhrinova, D. Uhrin, H. Powers, K. Watt, D. Zheleva, P. Fischer,et al. , J. Mol. Biol. 2005, 350, 587 – 598.
[31] Y. Y. Kuttner, S. Engel, J. Mol. Biol. 2012, 415, 419 – 428[32] J. G. Allen, M. P. Bourbeau, G. E. Wohlhieter, M. D. Bartberger,
K. Michelsen, R. Hungate et al. , J. Med. Chem. 2009, 52, 7044 –7053.
[33] Discovery Studio Modeling Environment, Release 3.1, AccelrysSoftware Inc. , San Diego, 2011.
[34] B. L. Grasberger, T. Lu, C. Schubert, D. J. Parks, T. E. Carver, H. K.Koblish, M. D. Cummings, L. V. LaFrance, K. L. Milkiewicz, R. R.Calvo, D. Maguire, J. Lattanze, C. F. Franks, S. Zhao, K. Rama-chandren, G. R. Bylebyl, M. Zhang, C. L. Manthey, E. C. Petrella,M. W. Pantoliano, I. C. Deckman, J. C. Spurlino, A. C. Maroney,B. E. Tomczuk, C. J. Molloy, R. F. Bone, J. Med. Chem. 2005, 48,909 – 912.
[35] U. M. Moll, O. Petrenko, Mol. Cancer Res. 2003, 1, 1001 – 1008.[36] J. Chen, J. Wang, B. Xu, W. Zhu, G. Li, J. Mol. Graph. Mod. 2011,
30, 46 – 53.[37] L. M. Espinoza-Fonseca, J. Garc�a-Machorro, Biochem. Biophys.
Res. Comm. 2008, 370, 547 – 551.
Received: September 30, 2012Accepted: December 13, 2012
