molecular dynamics simulations on pars sandeep kaushik ... · out multiple expli.it solvent...

905

Journal of Biomolecular Structure & Dynamics, ISSN 0739-1102 Volume 29, Issue Number 5, (2012) ©Adenine Press (2012)

*Corresponding authors:Debasisa MohantyAvadhesha SuroliaPhone: 191-11-26703749Fax: 191-11-26742125E-mail: [email protected]

[email protected]

Sandeep KaushikDebasisa Mohanty*Avadhesha Surolia*

National Institute of Immunology, Aruna

Asaf Ali Marg, New Delhi-110067

Molecular Dynamics Simulations on Pars Intercerebralis Major Peptide-C (PMP-C) Reveal the

Role of Glycosylation and Disulfide Bonds in its Enhanced Structural Stability and Function

http:::www..bsdonline..om

Abstract

Fu.osylation of Thr 9 in pars intercerebralis major peptide-C (PMP-C) enhan.es its stru.tural stability and fun.tional ability as a serine protease inhibitor. In order to understand the role of disulfide bonds and gly.osylation on the stru.ture and fun.tion of PMP-C, we have .arried out multiple expli.it solvent mole.ular dynami.s (MD) simulations on fu.osylated and non-fu.osylated forms of PMP-C, both in the presen.e and absen.e of the disulfide bonds. Our simulations revealed that there were no signifi.ant stru.tural .hanges in the native disulfide bonded forms of PMP-C due to fu.osylation. On the other hand, the non-fu.osylated form of PMP-C without disulfide bonds showed larger deviations from the starting stru.ture than the fu.osylated form. However, the stru.tural deviations were restri.ted to the terminal regions while .ore β-sheet retained its hydrogen bonded stru.ture even in absen.e of disulfide bonds as well as fu.osylation. Interestingly, fu.osylation of disulfide bonded native PMP-C led to a de.reased thermal flexibility in the residue stret.h 29-32 whi.h is known to intera.t with the a.tive site of the target proteases. Our analysis revealed that disulfide bonds .ovalently .onne.t the residue stret.h 29-32 to the .entral β-sheet of PMP-C and using a novel network of side .hain intera.tions and disulfide bonds fu.osylation at Thr 9 is altering the flexibility of the stret.h 29-32 lo.ated at a distal site. Thus, our simulations explain for the first time, how presen.e of disulfide bonds between .onserved .ysteines and fu.osylation enhan.e the fun.-tion of PMP-C as a protease inhibitor.

Key words: Mole.ular dynami.s simulations; Serine protease inhibitors; Gly.osylation; Fu.osylation; Disulfide bonds and PMP-C.

Introduction

Pars intercerebralis major peptide-C (PMP-C) is a ‘small’ serine protease inhibitor .omposed of 36 amino a.id residues. PMP-C is extra.ted from the brain and fat bodies of the migratory lo.ust, Locusta migratoria (1). PMP-C belongs to a large group of regulatory proteins, known as “serine protease inhibitors” or serpins whi.h a.t as substrates for serine proteases (2). Serpins regulate the a.tivity of various proteases for whi.h they a.t as a substrates. Serine proteases are a ubiquitous and well explored .lass of enzymes .onstituting over one third of all known proteolyti. enzymes (3, 4). They .atalyze peptide bond hydrolysis through a nu.leophili. atta.k on the target peptide bond using a uniquely rea.tive serine (5, 6). The atta.k o..urs between residues known as P1 and P1 generating an a.yl-enzyme .omplex and a peptide with P1 residue as the new amino terminus (4). Later, the a.yl-enzyme .omplex is .leaved releasing a peptide with P1 as the C-terminus. They are involved in primary pro.esses like digestion, protein maturation and turnover, hemostasis, immune responses etc. Their dire.t or indire.t malfun.tion leads to several .lini.al

Open Access ArticleThe authors, the publisher, and the right holders grant the right to use, reproduce, and disseminate the work in digital form to all users.

906

Kaushik et al.

syndromes (3, 7-10). An imbalan.e between proteases and their inhibitors .ould affe.t the survival of the organism (11). Several hereditary diseases like emphysema, .irrhosis and angioedema arise due to defi.ien.y of su.h inhibitors (11). Viruses and tumor or virus-infe.ted .ells survive against .ytotoxi. responses of the immune sys-tem using several intra.ellular inhibitors of granzymes to es.ape .ell death (12, 13). Therefore, re.ently serine protease inhibitors are being used as antiviral agents (14). They have been found to be prote.tive against tissue damage o..urring through free radi.al produ.tion or apoptosis by inhibiting pro.esses like adhesion and migration (11). In view of the biologi.al importan.e of the serine protease inhibitors, several studies have attempted to understand stru.tural features of serine protease inhibitors and intri.a.ies of their intera.tion with proteases (2, 11, 15-19).

PMP-C belongs to the pas.ifastin family of serine protease inhibitors found in arthropods. A unique feature of pas.ifastin family is the presen.e of .onserved .ysteines whi.h are involved in disulfide bonds. PMP-C has been found to exist in both fu.osylated (FU-PMPC) and non-fu.osylated (NF-PMPC) forms. Fu.o-sylation o..urs on a threonine (Thr-9) residue, making it an O-linked gly.opep-tide. Both, FU-PMPC and NF-PMPC are .apable of inhibiting human leuko.yte elastase and α-.hymotrypsin but NF-PMPC .an inhibit high voltage-a.tivated Ca21 .urrents whereas FU-PMPC does not. Fu.osylation of PMP-C has been observed to enhan.e its stru.tural stability .ompared to its non-fu.osylated form (20). The three dimensional stru.ture (Figure 1A) of PMP-C .onsists of a β-sheet formed by 3 anti-parallel β-strands, two loop regions .onne.ting the β strands and small flexible regions at N- and C-terminus (19, 20). The three β-strands are .on-ne.ted by 3 disulfide bonds between three pairs of .ysteines i.e., Cys 4-Cys 19, Cys 14-Cys 33 and Cys 17-Cys 28. A hydrophobi. .ore exists on one side of the β-sheet in a manner similar to other serpins (21-24). NMR studies on FU-PMPC as well as NF-PMPC indi.ate that fu.osylation does not alter the stru.ture of PMP-C signifi.antly and stru.tural differen.es arising from fu.osylation are lo.al-ized to the site of fu.osylation (20). However, FU-PMPC exhibited an enhan.ed stru.tural stability .ompared to NF-PMPC. The proton ex.hange rate for several hydrophobi. .ore residues was found to be de.reased in FU-PMPC whi.h is a sign of a .ompa.t hydrophobi. .ore relative to NF-PMPC. Based on these observations it has been proposed that there is an overall de.rease in the dynami. flu.tuations in FU-PMPC and this might be the reason for its enhan.ed stability. Apart from enhan.ing stru.tural stability fu.osylation is also known to enhan.e the fun.-tional properties of PMP-C. The FU-PMPC is known to be a more potent protease

Figure 1: (A) A .artoon depi.tion of PMP-C with all three disulfide bonds inta.t has been shown. The starting, disulfide linked and ending residues have been labeled. The site of fu.osylation Thr-9, Thr-16, Arg-18 and Fu.ose residues are shown in ball and sti.ks representation. (B) A s.hemati. representation of fu.osylation at Thr-9 of PMP-C. The gly.osidi. dihedral angle (θg) has been depi.ted.

907

MD Simulations on Pars Intercerebralis Major

Peptide-C

inhibitor .ompared to its non-fu.osylated .ounterpart. Therefore, it is ne.essary to understand the mole.ular basis of the effe.t of fu.osylation on the stru.tural stability and .onformational flexibility of this fun.tionally important peptide.

It is well known that gly.osylation affe.ts stru.tural and dynami. features of many proteins and peptides, and .onsequently, has signifi.ant effe.t on their fun.tional properties. However, la.k of atomi.ally detailed .rystal:NMR stru.tures of gly-.oproteins and gly.opeptides along with their gly.an moieties have often been a ma.or bottlene.k in understanding of the effe.t of gly.osylation on stru.ture and dynami.s. Even though .rystallographi. and NMR studies (19, 20, 25) have given novel insights on the stru.ture of PMP-C, effe.ts of fu.osylation on stru.ture and stability of PMP-C and its intera.tion with proteases, the .rystal and NMR stru.-tures of PMP-C la.k .oordinates of the gly.an moiety . Therefore, it is ne.essary to .omplement results from experimental studies with .omputer simulations for .omprehensive understanding of the effe.t of fu.osylation on stru.ture and fun.-tion. In re.ent years, .omputer simulations of gly.ans and gly.oproteins are being in.reasingly used to address biologi.al problems, be.ause of easy availability of high performan.e .omputing resour.es and enormous progress in development of reliable for.efield parameters. In our earlier work, we have used MD simulations to understand dynami. interplay between ion binding and substrate re.ognition by Con.anavalin A (ConA) (26), and effe.t of gly.osylation on stru.ture and stability of the Erythrina corallodendron le.tin (E.orL) (27). Similarly, MD simulations have been used by several other groups to study various aspe.ts of gly.oproteins and obtain insights on the subtleties of gly.oproteins (28-36). MD simulations have been used to analyze serine proteases as well (37). These studies prompted us to use MD simulations for understanding the role of fu.osylation on stru.tural stability and .onformational flexibility of PMP-C. There has been one report on unfolding MD simulations on PMP-C (38). However, the stru.tural model of PMP-C used in this simulation study did not in.orporate disulfide bonds between the .onserved .ysteines whi.h impart stability to both FU-PMPC and NF-PMPC.

In this work, we have .arried out multiple expli.it solvent MD simulations on FU-PMPC and NF-PMPC with and without disulfide bonds to understand rela-tive .ontributions of disulfide bonds and fu.osylation towards stru.tural stability and .onformational flexibility of this novel protease inhibitor. The main ob.e.tive of the study has been to explore the .onformations sampled by the fu.ose moiety and identify the residues of PMP-C with whi.h fu.ose moiety is involved in stable intera.tions over long time s.ale. We have also tried to understand, whether fu.o-sylation indeed lowers the flexibility of the PMP-C as seen in NMR studies. Sin.e, residues 28-32 of PMP-C are known to intera.t with the protease inhibitor, we have also attempted to understand how exa.tly fu.osylation at Thr 9 whi.h is away from the protease intera.ting segment affe.ts its fun.tion as an inhibitor.

Methods

Starting Structures for MD Simulations on NF-PMPC and FU-PMPC

The .oordinates for the NF-PMPC were obtained from PDB. The PDB entry 1PMC (25) had the NMR derived .oordinates for 36 models of PMP-C. Sin.e, the 36 stru.tural models were very similar to ea.h other and had maximum ba.kbone (N, Cα and C) RMSD of 2.26 Å, one of them was .hosen as the starting stru.ture for MD simulations on NF-PMPC. The stru.ture of the FU-PMPC is not available in PDB. However, NMR studies (20) on FU-PMPC have indi.ated that stru.ture of FU-PMPC is similar to NF-PMPC and these studies have also determined the dihe-dral angles whi.h define the orientation of fu.osyl moiety with respe.t to the Thr 9 side .hain. Based on these dihedral values the fu.osylated-Thr 9 was modeled in the starting stru.ture for FU-PMPC using XLEAP module of AMBER 9 (39), while all other amino a.ids had .oordinates identi.al to those in NF-PMPC. In order to

908

Kaushik et al.

build the fu.osylated Thr, fu.ose residue was added as the last residue (Fu. 37) and relevant bonds, angles and dihedrals were defined between Thr 9 and Fu. 37. Figure 1B shows the gly.osidi. dihedral angle (θg) whi.h defines the orientation of the fu.osyl moiety of Fu. 37 with respe.t to the Thr 9. The AMBER ff03 (40) and Gly.am06 (41) for.efields were used for assigning atom types, .harges, equi-librium bond lengths, bond angles, dihedral and van der Waals parameters for the simulations. Both NF-PMPC and FU-PMPC stru.tures were solvated using TIP3P (42) water box whi.h extended 10 Å from the outer most atom of the peptide along X, Y and Z axes. The periodi. boundary .onditions (PBC) were used to negate the surfa.e effe.ts at the box boundaries. Parti.le-Mesh Ewald (PME) summation method (43) was used for the .al.ulation of ele.trostati. potential. The native form of PMP-C .ontains 3 disulfide bonds between six .ysteine residues. Apart from disulfide bonded PMP-C, simulations were also .arried out on redu.ed forms of NF-PMPC and FU-PMPC whi.h la.ked the disulfide bonds.

Molecular Dynamics

Both the solvated PMP-C stru.tures were minimized using the SANDER mod-ule of AMBER9 (39) till the RMS gradient of potential energy rea.hed a value of 0.001 K.al:mole:Å. It may be noted that, for preparing the starting stru.tures for MD simulations we have used unrestrained minimizations and the minimized stru.tures also remained .lose to the starting NMR stru.tures. However, some of the re.ent studies have used su..essive steps of restrained minimizations to avoid problems arising from lo.al minima (44-46). Minimization was followed by equili-bration dynami.s whi.h was .arried out in two steps. The first step involved a 20 ps .onstant volume (NVT) equilibration to bring the system to the desired temperature of 300 K, followed by the se.ond step involving a 200 ps .onstant pressure (NPT) equilibration to bring the system to the desired pressure i.e., 1 atm. After equili-brating the system to the desired temperature and pressure, MD simulation was .ontinued under NPT .onditions for 12 ns. A step size of 1 fs was used during the entire equilibration and produ.tion phases of the simulations. Sin.e the .onforma-tions sampled during a parti.ular MD run is often dependent on the initial velo.ities assigned to the atoms in the mole.ule, for effi.ient sampling of the .onformational spa.e it is ne.essary to .arry out multiple MD runs with different starting velo.ity assignments. Cal.ulations of dynami. averages of various stru.tural and thermo-dynami. parameters from multiple tra.e.tories are statisti.ally more meaningful than parameters .omputed from single simulations. Therefore, for ea.h of the two stru.tures, four different simulations were run using different seeds for the random number generators for initial velo.ity assignments. The .oordinates were stored at an interval of 1000 steps of MD integration, thus, generating 12000 stru.tures for ea.h MD tra.e.tory. Thus, the four tra.e.tories provided us with a total of 48000 .onformations for both the fu.osylated as well as non-fu.osylated PMP-C.

Analysis of MD Trajectories

PyMol (47) software was used for depi.tion analysis of three dimensional stru.-tures. Ptra. module of AMBER9 (39) and in-house PERL s.ripts were used to analyze various .onformational features of the stru.tures sampled during the MD simulations. For ea.h of the stru.tures sampled during the simulations, root mean square deviation (RMSD) of the .oordinates with respe.t to the starting stru.ture was .omputed to analyze .onformational .hanges during the MD simulation. Apart from RMSD, radius of gyrations (Rg) of the whole mole.ule and theo-reti.al B-fa.tors (BF) for ea.h residue of the fu.osylated and non-fu.osylated PMP-C were also .omputed. Radius of gyration (Rg) indi.ated the .hange in overall shape of the mole.ule, while theoreti.al B-fa.tor (BF) values were a mea-sure of the mobility of the various residues. The hydrogen bonds were .al.ulated using a .ut off value of 3.5 Å for the donor-a..eptor distan.e and 120° for the

909


Peptide-C

donor-hydrogen-a..eptor angle. The per.entage o..upan.y of hydrogen bonds was .al.ulated using a window size of 100 ps and only those hydrogen bonds whi.h persisted for at least 10% of the time in ea.h window were .onsidered as stable during dynami.s. The hydrogen bonds were divided into two groups, namely hydrogen bonds involving main .hain atoms only (M-M) and hydrogen bonds involving side .hains (S-M:S). During analysis of inter residue .onta.ts, the distan.es between methyl groups of Fu. 37 (Fu.37-CH3), Thr 16 (T16-CH3) and δ-methylene group of Arg 18 (R18-CH2) were .al.ulated using the .enters of masses of the respe.tive groups. The mean values for various parameters were .al.ulated by averaging over all values obtained from a parti.ular tra.e.tory. Thus, there were four mean values obtained for a parti.ular parameter ea.h rep-resenting four different tra.e.tories. The .ombined mean values were also .al.u-lated .onsidering all the values of the parameter from all four tra.e.tories.

Figure 2: The RMSD plot has been shown for 3ds and 0ds forms of both fu.osylated and non-fu.osylated PMPC. (A) Individual RMSD plots for various simulations with different tra.e.tories being .olored differ-ently. (B) Average RMSD values been plot-ted for the four types of simulations.

910

Kaushik et al.

Table I RMSD and Radius of Gyration Values The RMSD and Rg values from different simulations representing different .onditions of simulations have been tabulated. The peak, mean and

standard deviation values for ea.h of the four simulations have been tabulated as: peak (mean 6 std_devn). The last .olumn of the Table (combined mean) tabulates the averages .al.ulated using all the values from all four simulations.

Simulation 1 Simulation 2 Simulation 3 Simulation 4 Combined

RMSD

FU-3ds 3.2 (2.060.4) 3.4 (1.760.5) 4.0 (2.460.6) 3.8 (2.060.6) 4.0 (2.060.6)NF-3ds 3.3 (2.360.4) 4.0 (2.260.6) 3.6 (2.260.4) 4.0 (2.360.6) 4.0 (2.260.5)FU-0ds 5.4 (2.961.1) 4.2 (2.860.5) 6.5 (3.961.1) 5.5 (3.261.1) 6.5 (3.261.1)NF-0ds 4.3 (2.860.6) 4.7 (3.260.5) 5.8 (3.460.9) 8.3 (5.161.3) 8.3 (3.661.3)

Rg

FU-3ds 11.5 (10.760.2) 11.4 (10.760.2) 11.9 (10.960.3) 12.0 (11.060.4) 12.0 (10.860.3)NF-3ds 11.9 (11.060.3) 11.7 (10.760.4) 11.7 (10.960.2) 11.3 (10.660.2) 11.9 (10.860.3)FU-0ds 11.5 (10.560.3) 11.8 (10.860.4) 12.3 (11.160.4) 12.6 (10.860.5) 12.6 (10.860.5)NF-0ds 11.1 (9.760.4) 12.0 (10.560.6) 12.5 (10.760.5) 12.6 (11.260.5) 12.6 (10.560.7)

Results and Discussions

Analysis of the Structural Changes During Dynamics Simulation

MD simulations were .arried out on both FU-PMPC and NF-PMPC. Both forms .ontained three disulfide bonds between six .ysteine residues; therefore, we refer to the simulations of FU-PMPC and NF-PMPC forms as FU-3ds and NF-3ds, respe.-tively. However, presen.e of these .ovalent disulfide bonds between β strands is likely to signifi.antly restri.t the main .hain .onformational spa.e a..essible to these mole.ules. This stru.tural rigidity might overshadow the stru.tural stability arising from fu.osylation and thus, de.iphering the effe.t of fu.osylation on stru.-ture of PMP-C from FU-3ds and NF-3ds would be a diffi.ult task. Therefore, we have performed another set of simulations as well on FU-PMPC and NF-PMPC whi.h la.ked disulfide bonds. We refer to these simulations as FU-0ds and NF-0ds. Simulations on ea.h of these 4 forms of PMP-C were repeated 4 times using dif-ferent random number seeds for velo.ity assignments. This resulted in a total of 16 MD tra.e.tories of 12 ns length ea.h with an effe.tive simulation time of 48 ns for ea.h of the four forms of PMP-C. Multiple MD tra.e.tories with different initial velo.ity assignments, not only enhan.es .onformational sampling, but also helps in ensuring .onsisten.y of the results obtained from the analysis of these tra.e.tories. Figure 2A shows the variation of ba.kbone RMSD (for N, Cα, and C atoms) for the four different forms of PMP-C with respe.t to their respe.tive starting stru.-tures and for ea.h form results from four different tra.e.tories obtained by different random number seeds are shown. Figure 2B shows similar RMSD plots for four different forms of PMP-C whi.h have been obtained by .al.ulating the averages from the four simulations with different random number seeds. Table I shows the peak RMSD for ea.h tra.e.tory and also mean and standard deviation values for RMSDs from ea.h tra.e.tory. As .an be seen from Figure 2 and Table I, the mean RMSD for native FU-PMPC with disulfide bonds varies between 1.7-2.4 Å and the peak RMSD ranged between 3.2-4.0 Å. On the other hand, the mean RMSD for native NF-PMPC (NF-3ds) varied between 2.2-2.3 Å while the peak values ranged between 3.3-4.0 Å. The mean RMSD values obtained from averaging of four tra-.e.tories for both FU-3ds and NF-3ds were found to be 2.0 and 2.2 Å, respe.tively (Table I). These results indi.ate that, native stru.tures of both FU-PMPC and NF-PMPC have a rigid ba.kbone .onformation be.ause of the presen.e of disulfide bonds. These results also suggest that, ba.kbone .onformations of native forms of FU-PMPC and NF-PMPC were very similar. In .ontrast to native PMP-C with dis-ulfide bonds, non-native PMP-C whi.h la.ked disulfide bonds showed signifi.ant differen.es between FU-PMPC and NF-PMPC. As .an be seen from Figure 2A and Table I, for the four tra.e.tories obtained for FU-0ds the mean RMSD varied between 2.8 and 3.9 Å whereas peak RMSD values ranged between 4.2 and 6.5 Å. On the other hand, mean RMSD for NF-0ds varied between 2.8-5.1 Å while the peak values ranged between 4.3-8.3 Å. These results indi.ate that in absen.e of the disulfide bonds the .onformational flexibilities are higher and hen.e, differ-

911


Peptide-C

ent simulations with different random number generators sample different regions of the available .onformational spa.e. Therefore, even at room temperature these stru.tures unfold, deviating as high as 8 Å from the native stru.ture, but they again refold to stru.tures whi.h are within 3 to 4 Å of the native state. Interestingly, the degree of unfolding seen in various NF-PMPC forms la.king disulfide bonds is higher than their FU-PMPC .ounterparts. This suggested that fu.osylation helps in in.reasing the stru.tural stability of FU-PMPC. It may be noted that, be.ause of the presen.e of the disulfide bonds in FU-3ds and NF-3ds simulations the stability indu.ed by fu.osylation was not prominent be.ause of the rigid stru.ture adopted by the polypeptide ba.kbone. The effe.t of fu.osylation on the stability of non- native PMP-C la.king disulfide bonds is more prominent in Figure 2B, whi.h shows the mean RMSD values obtained from averaging over multiple tra.e.tories. As .an be seen from Figure 2B for the disulfide bonds .ontaining native forms, FU-3ds and NF-3ds, the two tra.e.tories almost superimpose indi.ating that fu.osylation does not signifi.antly alter the ba.kbone stru.ture or the stability of the native PMP-C. However, in .ase of non-native PMPC the tra.e.tories for FU-0ds and NF-0ds .onverge to an RMSD of 4 Å, while the .orresponding RMSD values for native PMP-C simulation is 2 Å. This indi.ates that loss of disulfide bonds makes PMP-C more flexible. Comparison of the FU-0ds and NF-0ds tra.e.tories indi.ate that, while FU-0ds tra.e.tory show an RMSD lower than 4 Å during most part of the 12 ns simulation, NF-PMPC attains a peak RMSD of 5 Å. These results .learly indi.ate that, in absen.e of disulfide bonds the fu.osylation imparts additional sta-bility to the PMP-C stru.ture. Figure 3 shows the variation of the radius of gyration in ea.h of the tra.e.tories for the four different forms of PMP-C. As .an be seen from Figure 3, there are no appre.iable differen.es between the Rg values for the four different forms of PMP-C ex.ept that higher flu.tuations are seen in the Rg

values for non-fu.osylated PMP-C la.king disulfide bonds.

Sin.e, RMSD plots indi.ated that maximum deviations from the native PMP-C NMR stru.ture was observed in absen.e of fu.osylation and disulfide bonds, we

Figure 3: The variation of Rg values over the 12 ns tra.e.tories have been plotted for 3ds and 0ds forms of both fu.osylated and non-fu.osylated PMPC. The four different tra.e.tories for ea.h form of PMP-C have been .olored differently.

912

Kaushik et al.

extra.ted the relevant stru.tures from the NF-0ds tra.e.tory and superposed it on the NMR stru.ture of native PMP-C. As .an be seen from Figure 4 most striking .hanges o..urred in the terminal regions of the peptide whereas the .ore β-sheet regions were found to be similar in both stru.tures. This indi.ates that presen.e of fu.ose moiety and disulfide bonds help in redu.ing the flexibility of the terminal region and make it more ordered, while the β-sheet .ore of PMP-C maintains the similar stru.ture be.ause of the presen.e of inter strand hydrogen bonds. Therefore, we pro.eeded to analyze in detail the intra mole.ular hydrogen bonds in PMP-C.

Analysis of Hydrogen Bonds

The NMR stru.ture of PMP-C whi.h was the starting .onformation for our MD simulations, was found to have eight β-sheet .arbonyl oxygen and nitrogen atom pairs within a distan.e .ut off of 3.5 Å and therefore, .apable of forming inter-strand hydrogen bonds. We monitored the breaking and formation of these eight hydrogen bonds in all the MD tra.e.tories. Table II shows the average period of o..urren.e for these eight hydrogen bonds .al.ulated from ea.h of the four tra.e.tories. As .an be seen only five out of the eight hydrogen bonds persisted for longer than 7 ns in the disulfide bonded PMP-C stru.tures, while six out of those eight hydrogen bonds persisted in non-native PMP-C stru.ture la.king disulfide bonds. The extra hydro-gen bond present in non-native forms of PMP-C involved the .arbonyl oxygen of Thr 16 and NH group of Thr 29. Interestingly, this hydrogen bond was present for longer than 7 ns in all eight of simulation of PMP-C la.king disulfide bonds. This result was surprising as one would a priori expe.t the disulfide bonded β-sheet .ore

Figure 4: A .artoon presentation of superposition of native PMP-C (green) with the most deviated stru.ture (.yan) of the non-native PMP-C (NF-0ds) extra.ted from the simulations. The disulfide bonds in the native (NF-3ds) PMP-C stru.ture have been shown as sti.ks. The terminal residues have been labeled to indi.ate their relative spatial positions.

913


Peptide-C

to have more number of hydrogen bonds than the non-native form la.king disulfide bonds. Sin.e, the disulfide bond between Cys 17 and Cys 28 is immediately ad.a-.ent to this hydrogen bond (Figure 5), it is possible that presen.e of this disulfide bond alters the ba.kbone .onformation of the ad.a.ent residues whi.h leads to the disappearan.e of this hydrogen bond between Thr 16 and Thr 29. The two hydrogen bonds whi.h were missing in both the native as well as non-native forms of PMP-C involved residues Lys 8, Lys 11 and Thr 29 whi.h are present towards terminal regions of the β strands and flexibility of the terminal regions .ould be the reason for absen.e of these two hydrogen bonds during the dynami.s simulation. Thus, our inter-strand hydrogen bond analysis indi.ated that, neither fu.osylation nor the presen.e of disulfide bonds affe.ted the inter-strand hydrogen bonds in PMP-C stru.ture in a signifi.ant way. All the four different forms of PMP-C have similar inter-strand hydrogen bonds, ex.ept that presen.e of Cys 17-Cys 28 disulfide bond leads to disappearan.e of hydrogen bond between NH of Thr 29 and O of Thr 16. Figure 6 shows the variation of all the different types of intra mole.ular hydrogen bonds involving main .hain and side .hain atoms in the .ombined MD tra.e.tories of various forms of PMPC. The peak, mean and standard deviations in the numbers of different types of hydrogen bonds for individual tra.e.tories are listed in Table III. The detailed analysis of all these intra mole.ular hydrogen bonds in various forms of PMP-C also indi.ated that all four forms of PMP-C had very similar number of inter mole.ular hydrogen bonds. Here also, it was observed that upon loss of the disulfide bonds there was an in.rease in both hydrogen bonds involving main .hain (2 H-bonds) and hydrogen bonds involving side .hains (1 hydrogen bond) leading to a differen.e of 3 hydrogen bonds in total intra mole.ular hydrogen bonds. Thus, our analysis suggests that both fu.osylated as well as non-fu.osylated forms of non-native PMP-C la.king disulfide bonds have higher number of hydro-gen bonds .ompared to their native .ounter parts having disulfide bonds.

Glycosidic Dihedral Angle and Fucose-Peptide Interactions

We also analyzed the variation of the gly.osidi. dihedral angle (θg) in various simulations for fu.osylated native (FU-3ds) PMP-C and fu.osylated non-native (FU-0ds) PMP-C whi.h la.k disulfide bonds. Figure 7A shows the variations of θg over the various tra.e.tories for FU-3ds and Fu-0ds, while distributions of the .orresponding gly.osidi. dihedral angles are shown in Figure 7B. For FU-3ds, the mean value of θg ranged between 274.5° and 273.5° in various simulations giving rise to a .ombined mean value of 274.0 6 12.1°. For FU-0ds the mean of θg ranged between 278.9° to 286.4° with a .ombined mean value of 283 6 14° (Figure 7B). The plot for θg values and their distribution (Figure 7) .learly showed that fu.ose moiety in both native and non-native PMP-C had similar orientations and the mean values obtained from simulations were .lose to the experimental value of 257.9 6 0.6° (20).

Table II Hydrogen Bonds in the Core Sheet The average o..urren.e of the hydrogen bonds, .al.ulated to be present in the β-sheet of the NMR stru.ture of PMPC using a distan.e .ut off of 3.5 Å (ref. Figure 5), during all four types of simula-tions i.e., FU-3ds, NF-3ds, FU-0ds and NF-0ds.

Sr. No.Donor

residue (O)

A..eptor residues

(NH)

NMR stru.ture distan.e

Averageduration (ns)

FU-3ds NF-3ds FU-0ds NF-0ds

1 8 19 2.7 11 11.8 12 11.52 10 17 2.9 10.7 9.2 11.3 11.93 11 17 3.5 0.0 0.0 0.1 0.04 16 29 2.7 0.0 0.0 7.1 7.25 17 10 3.1 11.9 11.2 11.9 11.96 18 27 2.9 8.5 8.2 8.8 10.87 27 18 2.5 8.5 6.2 9.3 10.38 29 16 3.5 0.0 0.0 0.0 1.6

Figure 5: The intra-sheet (inter-strand) ba.kbone hydrogen bonds, .al.ulated using a distan.e .ut off of 3.5 Å for the ba.kbone .arbonyl oxygen (O) and nitrogen atoms (N), have been depi.ted for PMPC. Seven of the hydrogen bonds are .learly visible whereas the hydrogen bond between Lys 11 (O) and Cys 17 (NH) appears hidden behind the .artoon depi.tion of the PMP-C ba.kbone.

914

Kaushik et al.

Table III Intra-Peptide Hydrogen Bonds Cal.ulated numbers for various types of intra-peptide hydrogen bonds i.e., total, M-M, S-M:S and S-S hydrogen bonds from different simulations have been tabulated. The peak, mean and standard deviation values for ea.h of the four simulations have been tabulated as: peak (mean 6 std_devn). The last .olumn of the table (combined mean) tabulates the averages .al.ulated using all the values from all four simulations.


HbTOT

FU-3ds 37.0 (26.2 6 4.3) 38.0 (28.0 6 4.0) 38.0 (27.2 6 3.4) 40.0 (28.2 6 4.5) 40.0 (27.4 6 4.2)NF-3ds 35.0 (26.0 6 3.2) 38.0 (28.3 6 4.0) 45.0 (27.6 6 5.2) 38.0 (27.0 6 3.9) 45.0 (27.2 6 4.2)FU-0ds 38.0 (26.8 6 3.9) 42.0 (33.6 6 4.3) 39.0 (28.9 6 4.2) 39.0 (28.0 6 3.9) 42.0 (29.3 6 4.8)NF-0ds 44.0 (32.9 6 4.0) 49.0 (36.0 6 4.4) 38.0 (29.5 6 3.4) 42.0 (30.3 6 4.4) 49.0 (32.2 6 4.8)

HbM-M


HbM:S-S


HbS-S


Figure 6: The average values for total intra-peptide (top panel), total intra ba.kbone (M-M; se.ond from top), total side-.hain (S-S:M; third from top) and intra side-.hain (S-S; bottom panel) hydrogen bonds obtained from various simulations have been plotted for ea.h of the fu.osylated and non-fu.osylated forms of PMPC. Different tra.e.tories have been .olored differently.

Sin.e, the gly.osidi. dihedral angle in fu.osylated PMP-C was restri.ted to a value around 280°, we wanted to investigate if intera.tions between the fu.ose moiety and spe.ifi. residues of the peptide lo.ked the fu.ose moiety in an essentially fixed orientation. The starting NMR stru.ture of PMP-C used in our simulation had hydrophobi. intera.tions between methyl groups of Thr-16 and fu.ose ring (20). Therefore, we have monitored the distan.e between the .enter of masses (COM) of the methyl groups of Thr-16 (T16-CH3) and fu.ose (Fu.37-CH3) in various .onfor-mations sampled in FU-3ds tra.e.tories. As .an be seen from Figure 8, the initial distan.e between the methyl groups was more than 8 Å, but after 2 ns of simulation the two methyl groups .ame as .lose as 6 Å and remained in .onta.t during the most part of the remaining simulation period. This suggests that the hydrophobi.

915


Peptide-C

intera.tion between these methyl groups was stable during our simulations. In FU-0ds simulations, the distan.e between these two methyl groups was about 10 Å throughout the simulation period in three of the four simulations. However, in one of the simulations, this distan.e remained 10 Å for initial 7 ns but dropped to 4.5 Å for the later part. We also analyzed the distan.e between the COM of R18-CδH2 and Fu.37-CH3 (Figure 8). It was observed that, in FU-3ds simulations starting with a distan.e of 6Å, the two groups moved further apart as the simulations progressed. Within 2 ns of simulation, they moved 12 Å apart and stayed there for the rest of the simulation period. The mean distan.e ranged between 10 and 11 Å with a peak of 15 Å. This indi.ates that in fu.osylated PMP-C the fu.ose moiety prefers to inter-a.t with Thr 16 rather than Arg 18. Table IV lists peak, mean and standard deviation values for the variation of the distan.e between fu.ose methyl group and Thr 16 as well as Arg 18. As .an be seen from Figure 8 and Table IV, in some of the FU-0ds simulations a reverse trend is observed i.e., Arg 18 to fu.ose methyl distan.e was lower .ompared to Thr 16 to fu.ose distan.e (Figure 8).

Figure 7: (A) The gly.osidi. dihedral angle (Cδ-O1-C1-O5) values for FU-3ds and FU-0ds tra.e.tories of PMPC have been plotted. The dihedral angles from different tra.e.tories (obtained with different starting velo.ities) have been .olored differently. (B) The gly.osidi. dihedral angle distribution plot showing the population of ea.h dihedral value. It .an be seen that in FU-3ds simulations, the values from the four simulations are highly similar. This is not the .ase in FU-0ds simulations of PMPC.

916

Kaushik et al.

Effect of Fucosylation on Flexibility of PMP-C

In order to identify regions of PMP-C whi.h show higher mobility, we .omputed theoreti.al B-fa.tors for ea.h residue of various forms of PMP-C from different MD tra.e.tories. The analysis of these B-fa.tors showed .ertain features with inter-esting impli.ations for fun.tion of PMP-C. Figure 9 shows B-fa.tor plots for the four different forms of PMP-C obtained by averaging over multiple simulations for the same stru.tures. As .an be seen from Figure 9, terminal residues 1-2 and 35-36 have very high B-fa.tors ex.eeding 100 Å2 in all four forms of PMP-C. For all the residues of PMP-C, the non-native forms la.king disulfide bonds have higher B-fa.tors .ompared to native forms whi.h .ontain three disulfide bonds between the beta strands (Figure 9). This result is .onsistent with the earlier observation from RMSD and Rg plots whi.h suggested higher flexibility for non-native forms of PMP-C. Comparison of the B-fa.tor plots for the fu.osylated and non-fu.osylated forms of PMP-C shows .ertain interesting features. As .an be seen from Figure 9, both the native forms, FU-3ds and NF-3ds, have very similar B-fa.tors for all the residues ex.ept for the amino a.id stret.h 29-32. The fu.osylated form has lower B-fa.tor for the residue stret.h 29-32 .ompared to the non-fu.osylated form. Interestingly, this region intera.ts with the a.tive site of the protease when PMP-C

Table IV Distance Between Center of Mass of Methyl Groups (Fuc vs. Thr16 and Arg18)Distan.e between the COM (.enter of mass) of methyl(-ene) groups of Thr16 and Arg18 with respe.t to the COM of methyl group of fu.ose from different simulations have been tabulated. The peak, mean and standard deviation values for ea.h of the four simulations have been tabu-lated as: peak (mean 6 std_devn). The last .olumn of the Table (combined mean) tabulates the averages .al.ulated using all the values from all four simulations.


Thr16-Fu. methyl distan.e

FU-3ds 11.7 (7.7 6 1.1) 12.4 (7.3 6 1.4) 12.4 (7.3 6 1.4) 12.8 (7.4 6 1.5) 12.8 (7.4 6 1.4)FU-0ds 14.1 (9.8 6 0.7) 12.8 (9.8 6 0.6) 14.3 (10.1 6 0.8) 12.3 (8.0 6 2.6) 14.3 (9.4 6 1.7)

Arg18-Fu. methyl distan.e

FU-3ds 13.9 (10.0 6 2.2) 14.1 (10.9 6 1.4) 14.4 (10.9 6 2.0) 15.4 (9.8 6 2.3) 15.4 (10.4 6 2.1)FU-0ds 10.9 (7.4 6 0.8) 9.5 (6.8 6 0.9) 10.1 (6.3 6 1.0) 12.1 (8.6 6 1.4) 12.1 (7.3 6 1.3)

Figure 8: The distan.e between the .enter of masses (COM) of methyl groups of fu.ose and Thr16 (top panel) and δ-methylene group of Arg-18 (bottom panel) have been plotted for the FU-3ds and FU-0ds tra.e.tories. The distan.e values from different tra.e.tories (obtained with different starting velo.ities) have been .olored differently.

917


Peptide-C

fun.tions as a protease inhibitor (48). A..ording to Kellenberger et al., (48), Leu30 most probably intera.ts with the P1 po.ket of the protease as L30V mutation abro-gated α-.hymotrypsin inhibition a.tivity of the mutant peptide. Thus, the lower mobility of the residue stret.h 29-32 in the fu.osylated PMP-C .ompared to the non-fu.osylated .ounterpart .orrelates well with the fun.tional differen.es between fu.osylated and non-fu.osylated PMP-C. Hen.e, our simulations provide a stru.-tural rationale for the stability and fun.tional differen.es between FU-PMPC and NF-PMPC.

We also analyzed how exa.tly the fu.osylation at Thr 9 results in a mobility dif-feren.e in the amino a.id stret.h 29-32 whi.h is far apart from the site of fu.o-sylation both in sequen.e and also in three dimensional stru.ture. As .an be seen from Figure 10 the disulfide bonds Cys17-Cys28 and Cys14-Cys33 .ovalently

Figure 9: The average B-fa.tors for both fu.osylated and non-fu.osylated forms of PMPC have been plotted. It .an be .learly seen that the native (3ds) forms of PMP-C differ in their average B-fa.tor values in the 29-32 region.

Figure 10: Cartoon depi.tion of the NMR stru.ture of PMP-C and the stru.ture obtained after 12 ns of simulations for the fu.osylated native (FU-3ds) form of PMP-C. Thr-16, Arg-18, and Fu.ose-37 residues have been shown in ball and sti.k representation using PyMol. The in.reased distan.e between Arg-18 and Fu.ose-37 methyl groups .an be seen while the methyl groups of Thr-16 and Fu.ose-37 .an be seen to have moved .loser.

918

Kaushik et al.

.onne.t the residue stret.h 29-32 to the .entral β sheet of PMP-C and side .hains of Thr-16 and Arg-18 of the .entral β sheet are in .onta.t with the fu.ose moiety. Thus, using a novel network of side .hain intera.tions and disulfide bonds, fu.osy-lation at Thr 9 is altering the flexibility of the stret.h 29-32 lo.ated at a distal site. It may be noted that FU-0ds and NF-0ds simulations do not show any signifi.ant differen.es in the mobility of the stret.h 29-32 (Figure 9). This further establishes the role of disulfide bonds in the network of intera.tions whi.h .onne.t the site of fu.osylation to the residue stret.h 29-32 whi.h intera.ts with the P1 site of the protease when PMP-C fun.tions as a protease inhibitor. Thus our MD simulation studies have provided a mole.ular basis for the fun.tion of PMP-C as a protease inhibitor by explaining how the fu.osylation and disulfide bonds help in enhan.-ing its fun.tion as a protease inhibitor.

Conclusions

We have .arried out multiple expli.it solvent mole.ular dynami.s simulations for 12 ns on fu.osylated and non-fu.osylated forms of PMP-C both in pres-en.e and absen.e of disulfide bonds. Analysis of the various MD tra.e.tories have indi.ated that disulfide bonded FU-PMPC and NF-PMPC had very simi-lar .onformations. The differen.es between FU-PMPC and NF-PMPC be.ome prominent only in the absen.e of disulfide bonds. In absen.e of disulfide bonds NF-PMPC exhibited larger stru.tural deviations than FU-PMPC indi.ating an enhan.ed stru.tural stability for the latter. Mostly, the stru.tural deviations were restri.ted to the terminal regions while .ore β-sheet retained its hydro-gen bonded stru.ture even in absen.e of both disulfide bonds and fu.osylation. Fu.osylation helps in lowering the flexibility of the terminal regions and hen.e fu.osylated form of non-native PMP-C does not show large deviations from the stru.ture of the disulfide bonded native form. Sin.e, disulfide bonds .an form only after the formation of β sheet .ore, based on the results of our simulations it is tempting to spe.ulate that fu.osylation helps in folding as well. On.e the mole.ule is folded, the disulfide bonds lo.k it in a rigid .onformation whi.h is optimal for its fun.tion.

Analysis of theoreti.al B-fa.tors or mobility of different regions of PMP-C in various forms showed .ertain features with interesting impli.ations for fun.tion of PMP-C. The residue stret.h 29-32 intera.ts with the a.tive site of the protease when PMP-C fun.tions as a protease inhibitor. We have observed that FU-PMPC had lower B-fa.tors for this residue stret.h .ompared to NF-PMPC. On the other hand, fu.osylation did not affe.t mobility of this stret.h in non-native PMP-C la.king disulfide bonds. Our analysis revealed that fu.osylation at Thr 9 alters the flexibility of the residue stret.h 29-32 lo.ated at a distal site through a novel network of side .hain intera.tions and disulfide bonds namely, Cys17-Cys28 and Cys14-Cys33. These disulfide bonds .ovalently .onne.t the residue stret.h 29-32 to the .entral β sheet of PMP-C. Thus, our simulations have provided for the first time a mole.ular basis for the role of fu.osylation in stru.ture and fun.tion of PMP-C as a protease inhibitor and explain how presen.e of disulfide bonds between .onserved .ysteines and fu.osylation help in enhan.ing its fun.tion as a protease inhibitor.

Acknowledgements

This work was supported by .ore grants to the National Institute of Immunology and pro.e.t grants to D. M. and A. S. from the Department of Biote.hnology (DBT), Government of India. Computational resour.es provided under the BTIS pro.e.t of DBT, India are gratefully a.knowledged. S. K. a.knowledges CSIR, India for Senior Resear.h Fellowship. A. S. also a.knowledges a J. C. Bose Fellowship from Department of S.ien.e and Te.hnology, Govt. of India.

919


Peptide-C

References

N. Nakakura, H. Hietter, A. Van Dorsselaer, and B. Luu. 1. Eur J Biochem 204, 147-53 (1992).M. I. Plotni.k, L. Mayne, N. M. S.he.hter, and H. Rubin. 2. Biochemistry 35, 7586-90 (1996).M. J. Page and E. Di Cera. 3. Cell Mol Life Sci 65, 1220-36 (2008).E. Zakharova, M. P. Horvath, and D. P. Goldenberg. 4. Proc Natl Acad Sci U S A 106, 11034-9 (2009).C. S. Craik, S. Ro.zniak, C. Largman, and W. J. Rutter. 5. Science 237, 909-13 (1987).J. Kraut. 6. Annu Rev Biochem 46, 331-58 (1977).G. Menas.he, J. Feldmann, A. Fis.her, and G. de Saint Basile. 7. Immunol Rev 203, 165-79 (2005).J. A. Trapani. 8. Genome Biol 2, Reviews 3014 (2001).M. Bots and J. P. Medema. 9. J Cell Sci 119, 5011-4 (2006).P. Megyeri, K. M. Pabst, and M. J. Pabst. 10. Immunology 86, 629-35 (1995).H. Rubin. 11. Nat Med 2, 632-3 (1996).M. Bots, I. G. Kolfs.hoten, S. A. Bres, M. T. Rademaker, G. M. de Roo, M. Kruse, 12. K. L. Franken, M. Hahne, C. J. Froeli.h, C. J. Melief, R. Offringa, and J. P. Medema. Blood 105, 1153-61 (2005).J. P. Medema, J. de Jong, L. T. Peltenburg, E. M. Verdegaal, A. Gorter, S. A. Bres, 13. K. L. Franken, M. Hahne, J. P. Albar, C. J. Melief, and R. Offringa. Proc Natl Acad Sci U S A 98, 11515-20 (2001).M. Reiser and J. Timm. 14. Expert Rev Anti Infect Ther 7, 537-47 (2009).A. Wei, H. Rubin, B. S. Cooperman, and D. W. Christianson. 15. Nat Struct Biol 1, 251-8 (1994).D. A. Lawren.e, D. Ginsburg, D. E. Day, M. B. Berkenpas, I. M. Verhamme, J. O. Kvassman, 16. and J. D. Shore. J Biol Chem 270, 25309-12 (1995).N. R. Matheson, H. van Halbeek, and J. Travis. 17. J Biol Chem 266, 13489-91 (1991).P. Gettins, P. A. Patston, and M. S.hapira. 18. Bioessays 15, 461-7 (1993).A. Roussel, M. Mathieu, A. Dobbs, B. Luu, C. Cambillau, and C. Kellenberger. 19. J Biol Chem 276, 38893-8 (2001).G. Mer, H. Hietter, and J. F. Lefevre. 20. Nat Struct Biol 3, 45-53 (1996).W. Bode, E. Papamokos, D. Musil, U. Seemueller, and H. Fritz. 21. Embo J 5, 813-8 (1986).C. A. M.Phalen, I. Svendsen, I. Jonassen, and M. N. James. 22. Proc Natl Acad Sci U S A 82, 7242-7246 (1985).C. A. M.Phalen, H. P. S.hnebli, and M. N. James. 23. FEBS Lett 188, 55-8 (1985).M. Fu.inaga, R. J. Read, A. Siele.ki, W. Ardelt, M. Laskowski, Jr., and M. N. James. 24. Proc Natl Acad Sci U S A 79, 4868-72 (1982).G. Mer, H. Hietter, C. Kellenberger, M. Renatus, B. Luu, and J. F. Lefevre. 25. J Mol Biol 258, 158-71 (1996).S. Kaushik, D. Mohanty, and A. Surolia. 26. Biophys J 96, 21-34 (2009).S. Kaushik, D. Mohanty, and A. Surolia. 27. Protein Sci 20, 465-481 (2011).S. J. Marrink, H. J. Risselada, S. Yefimov, D. P. Tieleman, and A. H. de Vries. 28. J Phys Chem B 111, 7812-24 (2007).L. Janosi, A. Prakash, and M. Doxastakis. 29. Biophys J 99, 284-92 (2010).B. Hess, C. Kutzner, D. van der Spoel, and E. Lindahl. 30. Journal of Chemical Theory and Computation 4, 435-447 (2008).F. Chiappori, I. Merelli, L. Milanesi, and E. Rovida. 31. Proteins 78, 2679-90 (2010).Vorontsov, II and O. Miyashita. 32. Biophys J 97, 2532-40 (2009).N. S. Gandhi and R. L. Man.era. 33. Glycobiology 19, 1103-15 (2009).D. Ganguly and C. Mukhopadhyay. 34. Biopolymers 83, 83-94 (2006).G. Gupta, S. Vishveshwara, and A. Surolia. 35. IUBMB Life 61, 252-60 (2009).P. Hansia, S. Dev, A. Surolia, and S. Vishveshwara. 36. Proteins 69, 32-42 (2007).M. Topf, P. Varnai, C. J. S.hofield, and W. G. Ri.hards. 37. Proteins 47, 357-69 (2002).Y. Choi, H. Kim, J. H. Lee, S. Park, K. Jeong, and S. Jung. 38. J Mol Graph Model 28, 487-94 (2010).D. A. Case, T. A. Darden, T. E. Cheatham, III, C. L. Simmerling, J. Wang, R. E. Duke, 39. R. Luo, K. M. Merz, D. A. Pearlman, M. Crowley, R. C. Walker, W. Zhang, B. Wang, S. Hayik, A. Roitberg, G. Seabra, K. F. Wong, F. Paesani, X. Wu, S. Brozell, V. Tsui, H. Gohlke, L. Yang, C. Tan, J. Mongan, V. Hornak, G. Cui, P. Beroza, D. H. Mathews, C. S.hafmeister, W. S. Ross, and P. A. Kollman. AMBER 9, University of California, San Fran.is.o (2006).Y. Duan, C. Wu, S. Chowdhury, M. C. Lee, G. Xiong, W. Zhang, R. Yang, P. Cieplak, R. Luo, 40. T. Lee, J. Caldwell, J. Wang, and P. Kollman. J Comput Chem 24, 1999-2012 (2003).K. N. Kirs.hner, A. B. Yongye, S. M. Ts.hampel, J. Gonzalez-Outeirino, C. R. Daniels, 41. B. L. Foley, and R. J. Woods. J Comput Chem 29, 622-55 (2008).W. L. Jorgensen, J. Chandrasekhar, J. D. Madura, R. W. Impey, and M. L. Klein. 42. The Journal of Chemical Physics 79, 926-935 (1983).

920

Kaushik et al.

T. Darden, D. York, and L. Pedersen. 43. The Journal of Chemical Physics 98, 10089-10092 (1993).A. S. Gonçalves, E. R. Caffarena, and P. G. Pas.utti. 44. J Braz Chem Soc 20, 1227 (2009).A. P. Guimaraes, A. A. Oliveira, E. F. da Cunha, T. C. Ramalho, and T. C. Fran.a. 45. J Biomol Struct Dyn 28, 455-69 (2011).A. S. Gonçalves, T. C. C. França, J. D. Figueroa-Villar, and P. G. Pas.utti. 46. J Braz Chem Soc 22, 155 (2011).W. L. DeLano. 47. The PyMOL Molecular Graphics System., San Carlos, CA, USA (2002).C. Kellenberger, C. Boudier, I. Bermudez, J. G. Bieth, B. Luu, and H. Hietter. 48. J Biol Chem 270, 25514-9 (1995).

Date Received: December 22, 2011

Communicated by the Editor Ramaswamy H. Sarma

molecular dynamics simulations on pars sandeep kaushik ... · out multiple expli.it solvent...

Documents