Molecular Dynamics Study of Secretory Phospholipase A2 of Russell's

Oct 5, 2010 - Centre of Advanced Study in Crystallography and Biophysics, University of Madras, Guindy Campus, Chennai 600025, India, and Central ...
1 downloads 0 Views 8MB Size
Download PDF
J. Phys. Chem. B 2010, 114, 13463–13472

13463

Molecular Dynamics Study of Secretory Phospholipase A2 of Russell’s Viper and Bovine Pancreatic Sources C. Ramakrishnan,† V. Subramanian,‡ and D. Velmurugan*,† Centre of AdVanced Study in Crystallography and Biophysics, UniVersity of Madras, Guindy Campus, Chennai 600025, India, and Central Leather Research Institute, Adyar, Chennai 600020, India ReceiVed: March 8, 2010; ReVised Manuscript ReceiVed: August 25, 2010

A comparative molecular dynamics simulation of free and inhibitor-bound form of secretory phospholipase A2 (sPLA2) of Russell’s viper discloses the sort of restrictions in active site for inhibitor binding and implies suitable sites for further design of inhibitors based on active site scaffold. This enzyme belongs to group II PLA2s and dimerize asymmetrically with difference in orientation of W31 at the gateway of the active site of both the subunits. Hence, the active site of subunit A is open and that of subunit B is inaccessible to monodispersed inhibitors. PLA2 enzymes are active at solvent-lipid interface and their action could be inhibited at the solvent environment before it reacts with aggregated substrates. Some sPLA2s, especially of different venom sources, undergo aggregation in a concentration-dependent manner, associate symmetrically into dimeric or trimeric form, and attain functional monomeric form during their interaction with the aggregated substrate. All sPLA2s exhibit catalysis with similar mechanism and show considerable differences in its way of inhibition. This necessitates conformational analysis on asymmetric dimer viper PLA2 and its comparison with bovine pancreatic sPLA2 (BPsPLA2) which belongs to group IB. BPsPLA2 exists in monomeric form and does not have W31 at the gateway of hydrophobic pocket. In general, both monomeric and dimeric forms possess conserved active site with six subsites including the residues H48 and D49, and calcium-binding and surface loops. In the PLA2 inhibitor complexes, the presence of calcium in monomer and W31 in dimer form is the unique feature and it makes the difference only in inhibitory mechanism without altering the catalytic mechanism. With this context, molecular dynamics simulation is performed for monomer and dimer form of sPLA2s in both native and complex forms. Comparison of trajectories with respect to fluctuation and deviation discloses the dynamics of surface and calcium-binding loops as well as the difference in dynamics of active site residues of group IB and II sPLA2. Further, principal component and conformational cluster analyses are performed to substantiate the results. Introduction Inflammation is initiated by hydrolytic cleavage of phospholipids at the C2 position and production of the corresponding lysophospholipids. PLA2 plays a key role in the initial step of cascade mechanism by cleaving ester bonds of phospholipids at the sn-2 position and releasing free arachidonic acids. This further digested by cyclooxygenases leads to synthesis of proinflammatory compounds known as eicosanoids. Eicosanoids are involved in several disease pathways such as inflammation, asthma, platelet aggregation, and arthritis. So, the compounds that will inhibit PLA2 can be potent antiinflammatory agents. The PLA2 family is broadly classified into cytosolic isoforms (cPLA2s) involved in signal transduction and secretory isoforms (sPLA2s) involved in inflammation and microbicidal pathways. Secretory PLA2s catalyze the reaction at a lipid-water interface and its activity is high in severalfold against the aggregated substrates than the monodispersed.1 The catalytic mechanism of sPLA2 is similar throughout the family with conserved tertiary structure and the differences can be observed at the sequence level.2,3 These enzymes exist in monomeric, dimeric, or trimeric forms depending upon the nature of the source4 and its concentration. The structure of * To whom correspondence should be addressed. Phone: +91-4422300122. Fax: +91-44-22300122. E-mail: [email protected]. † University of Madras. ‡ Central Leather Research Institute.

sPLA2s consists three helices (H1, H2, and H3), two short helices (SH4 and SH5), and one β-wing with strands β1 and β2 in antiparallel fashion4-6 and possesses coordination with calcium ion for its catalytic activity. The viper sPLA2s do not have calcium ion in their native and inhibitor bound forms.8,9 It means that it acquires calcium coordination only when it reacts with aggregated substrates such as membrane phospholipids. It possesses the calcium-binding loop made up of the residues 25-34 with tryptophan at position 31 and its orientation aids binding of inhibitor. Further, there are seven conserved disulfide bridges affording rigidity. The crystal structures of viper sPLA2s are dimer and asymmetric with 0.71 Å rmsd8,9 between subunits. The W31 is present in the calcium-binding loop at the gateway of the active site of both subunits. In subunit A, it possesses hydrophobic contacts with interfacial residues of subunit B, while in subunit B it is buried inward due to hydrophilic environment. Thus, it acquires different conformations between subunits A and B. This makes the active site open in molecule A and closed in molecule B.8 In addition, the calcium-binding loop and the β-wing with two antiparallel β-strands also show conformational difference between the subunits. Prominently, the stability of the dimer is crucial in the inhibitory mechanism of viper PLA2. The dimer is formed through the long interface (Figure S1 in Supporting Information), which includes the key residues R43(B), F46(B), and V47(B) nearby W31(A). In addition, L2, L3, and L119 of

10.1021/jp102073f  2010 American Chemical Society Published on Web 10/05/2010

13464

J. Phys. Chem. B, Vol. 114, No. 42, 2010

subunit A and A101, Q108, and K131 of subunit B make hydrophobic interface through weak intermolecular hydrogen bonds and tough hydrophobic contacts.1,9 BPsPLA2s exist in the form of monomer with coordination between protein, calcium ion, and inhibitor. The hydrophobic channel comprises conserved amino acid residues, which include H48, D49, Y52, and D99. The calcium-binding loop and D49 are important for calcium coordination and the catalytic activity of enzyme.7 Another important feature is the conserved water molecule10,11 at the active site, which binds to the residues H48 and D49 and inhibitor through hydrogen bond interactions. The N-terminal helix is rich in hydrophobic residues thereby involving in membrane binding.5,12 Studies on kinetics of wildtype and mutant structures revealed the key residues13,14 responsible for specificity toward different substrate analogues. Based on available complex structures, there are six subsites reported so far.15 For example, the natural inhibitor vitamin E (R-tocopherol) occupies the subsites 1, 2, 4, and 5. Similarly, n-tridecanoic acid16 (subsites 1, 2, 4, and 6), indomethacin15 (subsites 4, 5, and 6), designed peptide LAIYS9 (subsites 1-5), etc., bind at the active site through interactions with specific subsites as mentioned. This gives rise to the further designing of more potent and specific inhibitors which will interact with most of the active site residues. PLA2s carry out interfacial catalysis at the boundary of water and condensed lipid phase. Interfacial binding of sPLA2 is characterized using both biochemical and computational studies. Interfacial activation necessitates the correlation between membrane surface properties and membrane-induced structural changes in sPLA2.17 Moreover, comparison of i-face or interfacial recognition site (IRS) of groups IA, IB, IIA, V, and X shows that the aromatic amino acids have high affinity toward zwitterionic phase and both the enzyme and nature of solvent-lipid interface together control the interfacial binding and function of sPLA2.18-20 The catalytic reaction at the interface is guided by H48 and D49 along with F5,19,106, D99, and A102 around the active site. Kinetics of substitutions for these residues revealed the mode of interfacial binding, catalysis and activation.6,21 Moreover, native and bound forms of sPLA2s were investigated using molecular dynamics simulation technique in both solvent and lipid environments.17,22-31 Particularly, comparative study of calcium-dependent and -independent forms of sPLA2 reveals the differences in stability of the catalytic network and its inhibitory mechanism. In addition, the possible dissociation of dimer Crotalus atrox venom sPLA2s into two functional monomers is also studied.22 Among the flexible parts of sPLA2, the N-terminal helix shows a twist motion by which Ala located at first position is involved in increasing the rate of reaction toward micelles and membrane phospholipids.24 Earlier MD studies on comparison of monomer with dimer sPLA2 and its mutants dealt with the stability of Ca2+ ion, conformation of flexible loops and W31 with various time scale simulations between 200 ps and 10 ns. In the present study, the comparison is made in such a way that the differences in conformational changes between the sPLA2 belonging to group IB and II using MD simulations with reasonable time scale of 20 ns each. Hence, simulations of PLA2s from bovine pancreas (monomer and calcium dependent) and viper (dimer in the absence of calcium) are performed for both apo and complex forms. Trajectories were analyzed to understand the difference in inhibitory mechanism between these two groups of PLA2. Hence, the important structural changes upon inhibitor binding to monomer and dimer forms give a clue to design the subtypespecific inhibitors as antiinflammatory compounds. Moreover,

Ramakrishnan et al. it will explore difference in the active site scaffold between subunits of viper sPLA2 dimer and help to design potent inhibitors with improved inhibitory activity. Experimental Methods Choosing Protein Structures. The structure of the native form of BPsPLA2 at the atomic resolution of 0.97 Å (1g4i) and complex form at a resolution of 2.6 Å (1o2e)11 belonging to group IB and native (1fb2, 1.9 Å) and bound (1kpm, 1.8 Å)8 forms belonging to group II are downloaded from Protein Data Bank (PDB).54 As explained in the Introduction, inhibitors make coordination with calcium ions present in active site of the PLA2 belonging to group IB whereas in group II it directly binds without the aid of a cofactor. The complex form of group IB PLA211 which has been chosen is a triple mutant (K53,56,120M). These mutations switch the preference of anionic interface to zwitterionic interface, which is crucial for interfacial binding and catalysis.11,21 Also, it enhances catalysis by making the loop of residues 58-75 well ordered and shows no large conformational changes. Hence, the comparative dynamics of wild-type native and inhibitor bound triple mutant not only reveals the inhibitory mechanism of the BPsPLA2 but also explores the suitable conformation acquired with zwitterionic preference. Similarly, comparison of the native and bound forms of dimer PLA2 addresses the problem of inhibitor binding to subunit B while it binds to subunit A without hindrance. Consequently, it will explore the differences between the inhibitory mechanisms between these two groups of PLA2. System Preparation. Each system is named to avoid ambiguity. 1G4I and 1O2E denote native and inhibitor (4methoxybenzoic acid) bound form of BPsPLA2, respectively. Similarly, AB and AIB denote native and inhibitor (R-tocopherol abbreviated as R-TP) bound form of viper sPLA2. Here, A, B, and I denote subunits (A and B) and inhibitor (R-TP), respectively. The possibility for inhibition of the viper sPLA2 is up to 50% only. It shows that the dimer may get dissociated and subunit B is free to interact with membrane phospholipids. To portray the influence of subunit B upon inhibitor binding to subunit A, system AI is prepared by eliminating the subunit B and is compared with AIB. All the crystal structures were systematically prepared using Leap module of the AMBER10 suite32 in the following way. At first, all the charged residues were protonated and seven disulfide bonds were created manually, which are found similar in these two subtypes of PLA2. To bring total charge to zero, the system was neutralized by adding adequate numbers of Na+ or Cl- ions appropriately. For systems 1G4I and 1O2E, the library files were created for calcium ion by assigning full +2 charges.33 The residue and atom names of calcium were modified as CAL and Ca2, respectively, to avoid the conflict with standard CR atoms of amino acids. Force field parameters of anisic acid and RTP were created using general Amber force field (GAFF)34 embedded with antechamber module. AM1-BCC35 charge method was used to assign the atomic charges. Molecular mechanical force field ff03 of AMBER was used for system preparation and simulations.36,37 Finally, the protein was set at the center of a TIP3P box of three-point charged triangulated water molecules with minimum distance of 10 Å between the wall and any atom of the solute. Initial coordinates and corresponding parameter and topology files were created for subsequent steps. To relax each of the five systems, energy minimization was performed in two steps with nonbonded cutoff of 10 Å. In the first step, the solute atoms including calcium ions were restrained by a harmonic potential with force constant of 100 kcal/(mol · Å2).

Secretory Phospholipase A2 (sPLA2) of Russell’s Viper The water molecules were relaxed using 500 cycles of steepest descent and 2000 cycles of conjugate gradient algorithms. Finally, the whole system was relaxed using 2500 cycles of conjugate gradient without restraints. Molecular Dynamics Simulation. Prior to simulations, each system was equilibrated in three phases with 2 fs of integration time step. In the first phase, the temperature was brought to 300 K using Berendsen temperature coupling38 with time constant 2 ps. Similarly, in the second phase, the system was set with constant pressure using isotropic position scaling. In these two stages, all nonsolvent atoms were restrained. Finally, equilibration was extended at constant temperature and pressure without restraints. Consequently, simulations were performed in explicit solvent environment using NPT ensemble with 1 fs integration time step. Bonds involving hydrogen atoms were constrained using SHAKE algorithm.39 The initial velocities were assigned from a Maxwell distribution at a given temperature. Trajectories were computed for 20 ns each on an 8-node Linux cluster with Pentium Core2 Duo processors. Equilibration and simulation processes were validated as a function of potential energy of each system. Trajectory Analysis. Each simulation produced a trajectory with snapshots at 1 ps interval. Analysis was started with examination of the potential of each of five independent trajectories. The variations in total, potential and kinetic energies, temperature, and pressure were estimated as a function of simulation time to confirm whether the systems obey NTP ensemble throughout the simulation. Root mean square deviation (rmsd) and atomic positional fluctuation were calculated to find out dynamical behavior of all the forms PLA2. Hydrophobic interactions play an important role in both binding of inhibitors to viper PLA2 and dimerization. In order to understand the difference in structural and catalytic behavior between viper and BPsPLA2, the hydrophobic and hydrogen bond occupancies were calculated using HBPLUS40 and LIGPLOT41 with cutoff ranges (Å) 2.6-3.9 and 2.6-3.0, respectively. For further support of results, additional relative parameters such as radius of gyration (Rg) and solvent-accessible surface area (SASA) were also calculated. The Rg was calculated for every frame of the trajectory using MMTSB tools.42 Similarly, SASA of whole protein was calculated for coordinates at every 1 ps using MSMS program43 with probe radius 1.4 Å. The important role of W31 in PLA2 inhibition and dimerization has been discussed in many articles dealing with structural biology of sPLA2. Hence, to focus on the residual behavior of W31, the dihedral angles φ and ψ (for main chain) and χ1 and χ2 (for side chain) were calculated. The interatomic distances were calculated for atom pairs each from W31, L2, L3, and L119 of subunit A and residues of subunit B at the dimer interface. In addition, accessible surface was computed for W31 and K69 of viper sPLA2 and Y69 of BPsPLA2 using NACCESS program in order to get accessible surface for every individual residue of the protein. In-house scripts were written using Perl programming language to fetch desired data from the output files of tools used for various purposes. The plots are made using Gnuplot.44 To substantiate the MD results, essential dynamics was performed for all the trajectories using principal component analysis with the diagonalization of 3N × 3N covariance matrix of atomic coordinates. The backbone atoms of the protein were considered for constructing the covariance matrix. The eigenvalues of maximum magnitude and the corresponding eigenvectors were calculated using PTRAJ45 module. To monitor the modes of internal motions of the system, trajectories were projected onto the principal modes. Among the 25, the first two

J. Phys. Chem. B, Vol. 114, No. 42, 2010 13465 significant eigenvectors (principal components PC1 and PC2) with largest eigenvalues were used to make 2D projection for each of five independent trajectories. The anharmonic and largescale motions in the essential subspace could be correlated with the important functions of the proteins. Further, conformational clustering was performed for all the snapshots projected on the three-dimensional space spanned by the first two eigenvectors of the covariance matrix with the largest eigenvalues and potential energy. The Perl program cluster.pl of mmtsb tools

Figure 1. (a) Multiple sequence alignment of group IB and II sPLA2 sequences using clustalW49 and Jalview.50 Six subsites are underlined (black), viz., residues 2-10, 17-23, 28-32, 48-52, 68-70, and 98-106 which form the subsites 1-6, respectively. (b) Six subsites of PLA2 represented in order with (1) red, (2) yellow, (3) green, (4) cyan, (5) violet, and (6) pink colors using Chimera.51 Inhibitor (anisic acid) located at active site is shown in ball and stick.

Figure 2. Atomic positional fluctuations for viper PLA2 (a) and BPsPLA2 (b).

13466

J. Phys. Chem. B, Vol. 114, No. 42, 2010

Ramakrishnan et al.

Figure 3. Root mean square deviation of backbone atoms. Deviations decrease in the order of AB > AIB > A[AB] > B[AB] > B[AIB] > others.

was used. This performs clustering using k-means algorithm46 that finds the centroid (random seed at the center of cluster) of each conformational family by an iterative search through minimizing sum of distances or square Euclidean distances to cluster members. The number of amino acid residues is varying between different groups of sPLA2 due to natural mutations. Totally, there are 121 residues in the former per subunit and 123 residues in the latter PLA2. The structure-based sequence alignment between these two groups of PLA2 gives 131 residue numbers (including gaps) as available in the literature. The trajectory analysis is performed based on PLA2 group specific numbering scheme. In order to understand the functional importance of key residues, it has been labeled using alignment-based numbering, so that the results presented here could be compared with literature. For example, tryptophan residue located at the 30 position of viper PLA2 is labeled as W31 in the literature due to its position at 31 in the global multiple sequence alignment with PLA2 sequence of other groups. Similarly, lysine residue is labeled as K69 as per alignment; however, it is located at the 60th position of viper PLA2. Results and Discussion Molecular Dynamics Simulation. For each of five systems (AB, AIB, AI, 1G4I, and 1O2E), temperature of 300 K and pressure of 1 atm were attained during first and second equilibration phases, respectively. Then the simulations with NPT ensemble were performed for 20 ns to search for energy landscape and to understand the mode of binding of inhibitor to viper (dimer). It is compared with BPsPLA2s (monomer) to reveal differences in inhibition mechanism and dimerization between these two subtypes. Analysis of each of five 20 ns trajectories depicts the structural behavior these two groups of sPLA2. The kinetic and total energies show small rms fluctuation. In addition, the correlation between potential energy and temperature for each trajectory is ideal. Flexibility of Viper sPLA2. The conformational changes in viper PLA2 with respect to inhibitor binding were analyzed in terms of fluctuations and deviations about backbone atoms (Figure S5 in the Supporting Information). Abundance of helix and seven disulfide bonds control the flexibility of the enzyme. In the graph plotted for residual fluctuations, apo sPLA2s (AB and 1G4I) show more fluctuation than the complex form. The residual fluctuation in subunit A coincides with its unique conformational feature for inhibitor binding is explained below (Figure 2). Comparison of the fluctuations between the subunit A in all viper sPLA2 simulations confirm that the surface loop (residues 59-61) movement is arrested in the presence of inhibitor. Particularly, K69 is stabilized in both AIB and AI and fluctuates more in AB. The role of surface loop in inhibitor

binding is apparent. The subunit B of both AB and AIB shows similar fluctuations and show no difference with respect to binding of inhibitor to subunit A. Due to the absence of subunit B, AI shows much less fluctuation compared to the subunit A of both AB and AIB (Figure 2a). Hence, it is proved that the hydrophobic residues of subunit A are buried inside in the absence of subunit B as well as the subunit B may induce the suitable conformation in subunit A by which the active site is open for inhibitor binding. In the case of BPsPLA2, binding of inhibitor (here, anisic acid) arrests the fluctuation of both surface and calcium-binding loops (Figures 1b and 2b). The main difference between bovine and viper sPLA2 is that in the complex form both the loops are stable in the former and only surface loop is stable in the latter. This confirms the calciumbinding loop of viper sPLA2 showing fluctuation in the inhibitor (RTP) bound form. Figure 3 shows the backbone rmsd of all the trajectories and this is well correlated with atomic positional fluctuations depicted in Figure 2. The viper sPLA2 shows the rmsd at maximum of 1.0-4.0 Å which is reasonably higher than that of BPsPLA2 (0.5-2.0 Å) due to abundance of helices compared to the former. Since the deviations are much less between the apo and complex forms, with reference to the literature as well as the results observed so far, further analysis is focused only on flexible regions (loops) and the residues responsible for inhibitor binding. Hence, the backbone rmsd for surface and calcium-binding loops is plotted (Figure 4). This clearly depicts the actual behavior of loops with respect to inhibitor binding and discriminates the dynamics of viper PLA2 from bovine source. As observed in atomic positional fluctuation, both the loops show more rms deviation in apo form than complex form of BPsPLA2 (Figure 4a). Similarly, in viper sPLA2, the deviation is less in the surface loop of apo form than the complex. The calcium-binding loop of subunit A shows similar deviation in both complex (AIB) and apo form (AB). Strikingly, unlike the surface loop, it is highly mobile in the absence of subunit B (AI). This again proves that the orientation of calcium-binding loop and W31 is mostly controlled by subunit B. In addition, binding of RTP to subunit A makes the active site more hydrophobic and decreases the number of W31 involved hydrophobic contacts at the interface. Role of Hydrophobicity. Hydrophobicity plays an essential role in the function of sPLA2 including its interaction with inhibitors, membrane phospholipids, micelle, concentration dependent aggregation4 etc. In addition, sPLA2 needs hydrophobic residues at surface called ‘i-face’10 for its activation at the solvent-membrane interface through hydrophobic interactions.47 Hence, the analysis is extended with the focus on the influence of hydrophobic contacts in sPLA2 inhibition using parameters like solvent-accessible surface area (refer to Figure 5b, and Figure S3 in the Supporting Information), radius of

Secretory Phospholipase A2 (sPLA2) of Russell’s Viper

J. Phys. Chem. B, Vol. 114, No. 42, 2010 13467

Figure 5. (a) Number of hydrophobic contacts in viper and BPsPLA2s using HBPLUS, (b) solvent-accessible surface area using MSMS, and (c) radius of gyration using mmtsb toolset. All three plots are well correlated to each other. The SASA and Rg are increasing with increasing number of hydrophobic contacts.

Figure 4. Root mean square deviation of calcium-binding (a) and surface (b) loops. In viper PLA2, it shows that the movement of calciumbinding loop of subunit A is mostly controlled by residues of subunit B at the dimer interface than RTP, whereas the movement of the surface loop is controlled by inhibitor alone and not by subunit B. BPsPLA2 shows minimum deviations in surface and calcium-binding loops between apo (1G4I) and complex form (1O2E).

gyration (Rg), and life of nonbonded interaction (Figure 5). It shows very clearly that binding of inhibitor increases the number of hydrophobic contacts in both types of PLA2. Interestingly, the difference between subunit A of apo (AB) and complex (AIB) denotes that the number of hydrophobic contacts is increased only in the subunit to which inhibitor binds (Figure 5a). Other equivalent measures such as SASA and Rg have also been calculated (Figure 5b,c) and confirm the same. Hence,

binding of an inhibitor increases the number of hydrophobic contacts and reduces the solvent-accessible surface in both viper and BPsPLA2s. This directs us to start investigation specifically on the hydrophobic residues which are involved in dimerization and inhibitor binding. (Please refer to the Supporting Information for residue-based SASA and hydrophobic contacts.) Thus, in viper sPLA2, the residues at both the gateways of hydrophobic pocket and dimer interface (L2, L3, L119, and W31 of subunit A and R43, F46, V47, A101, Q108, and K131 of subunit B) were considered for further analysis. Figure 6a depicts the differences in number of hydrophobic contacts between native and complex form. In the native form (AB), W31 of subunit A is involved in more number of hydrophobic contacts with residues at the dimer interface in contrast to complex form (AIB). Similarly, L2, L3, and L119 also show a similar pattern. Hence, binding of inhibitor (here, RTP) disturbs

13468

J. Phys. Chem. B, Vol. 114, No. 42, 2010

Ramakrishnan et al.

Figure 6. (a) Number of hydrophobic contacts between key residues (L2, L3, L119, and W31) in the hydrophobic pocket of subunit A and the residues of subunit B at the dimer interface. (b) Solvent-accessible surface area of W31, K69 of viper PLA2, and Y69 of BPsPLA2 is calculated using NACCESS52 with 1.4 Å probe sphere.

TABLE 1: Interatomic Distance (Å) for Atom Pairs Involved in Nonbonded Contacts (Including Hydrophobic and Ionic Interactions) Stabilizing the Dimer Form of Viper sPLA2 with the Distance Cutoff 2.6-3.9 Å atom pairs

mean and SD

chain A

chain B

AB

AIB

CD1(W31) CD2(W31) CG(W31) CE2(W31) NE1(W31) NE1(W31) CZ2(W31) CZ2(W31) N(L2) CB(L2) N(L3) CD2(L3) CB(L3) O(L119) CD2(L119)

CD(R43) CD(R43) CD(R43) CG(R43) CG(R43) CG2(V47) CD2(F46) CE2(F46) O(K131) CB(K131) O(K131) CG(K131) CG(K131) NE2(Q108) C(A101)

5.2(7) 4.1(4) 4.7(6) 4.8(8) 5.5(10) 5.9(18) 6.5(19) 6.7(23) 10.4(40) 13.5(42) 8.4(33) 8.9(28) 9.8(31) 3.1(4) 4.8(9)

6.4(11) 7.0(10) 6.0(10) 8.0(10) 7.9(11) 7.6(10) 12.2(10) 13.1(11) 5.5(11) 8.0(9) 7.2(19) 10.3(23) 9.5(20) 3.1(4) 4.3(4)

more than 50% of hydrophobic contacts at the dimer interface and shows the sign of dissociation of subunit B from the complex into active monomer (Table 1). It is also confirmed by dynamic profile of dihedral angles (φ, ψ, χ1, and χ2) of W31 (Figure 7) and could be correlated with the same finding reported in crystallographic studies.8 The ψ angle of W31 in subunit A

stays at around -10° throughout the simulations regardless of the presence of inhibitor and subunit B, whereas the φ angle stays around -80° in AIB (similar to crystal structure) and fluctuates more in the absence of subunit B (AI) and in the absence of inhibitor (AB). Similarly, the other dihedral angles (χ1 and χ2) also show more fluctuations in subunit A which is in the absence of inhibitor and subunit B. Thus, all the dihedral angles are very consistent in subunit A with inhibitor and subunit B (AIB) and confirm that the position of W31 is fixed without translation and internal rotation. In the case of subunit B, φ and ψ angles do not show any difference in fluctuation; however, χ1 and χ2 show (Figure 7c,d) fluctuations. The fluctuation patterns for χ1 and χ2 depicted in Figure 7d are the conformation with which subunit B restricts the binding of inhibitor and it is well correlated with the essential dynamics results shown in Figure 9, b and d. The same fluctuation pattern is observed very clearly in AI which lacks subunit B. This gives a strong base to confirm that the hydrophobic contacts at the dimer interface induce the conformational change suitable in subunit A by altering the dihedral angles of W31 for inhibitor binding. Precisely, the changes in χ1 and χ2 are induced by only the residues belong to subunit B at the dimer interface, whereas the changes in φ angles are induced by both inhibitor and subunit B. The same feature could be visually investigated by comparing clusters a, c, and e of Figure 9. Since the conformation of W31 is important for the inhibitory process and it is induced by the neighboring residues at the dimer interface, it is necessary to display the interatomic distance between residues belongs to subunit A and B at the interface. The difference between the complex and native form with respect to the W31 involving interatomic distance (Table 1) strongly suggests that W31 possesses stable hydrophobic contacts with R43, F46, and V47 of subunit B at the dimer interface in the native form and it is considerably disturbed in the presence of inhibitor (RTP). The interatomic distance (averaged over 20 ns) for other residues involved in dimerization (Table S6 in the Supporting Information) shows that the residues L2 and L3 of N-terminal helix lose nonbonded interactions at the interface in both native and complex forms. The distance plot shows the variations in interatomic distances as a function of simulation time (Figure S2 in the Supporting Information). These distances increased up to 20 Å in the native form after 5 ns and up to 13-14 Å in complex form after 4 ns. This is due to flexibility of the N-terminal helix in native form than that in the presence of inhibitor. Hence, the inhibitory mechanism of viper sPLA2 depends on the conformation of W31 and is confirmed by molecular dynamics through the above results as well as it is correlating well with the results of biochemical studies available in the literature. Principal Component Analysis and Conformational Clusters. The MD snapshots at every 1 ps were projected onto the first two eigenvectors, the two most significant principal components out of 25 calculated. The 2D plots of two principal components (PC1 and PC2 with largest eigen values) for both viper sPLA2 and BPsPLA2 are shown in Figure 8, a and b, respectively. More the distribution of dots, more are the conformational changes that occur. As observed in the rmsd plot (Figure 3), it clearly depicts that the distribution for native sPLA2s is large compared to ligand bound forms. It is obviously due to the rigidity introduced by binding of inhibitor. Consequently, the distribution of 1O2E is similar to that of AI and is less than that of 1G4I. Also, the distribution of AIB is less than that of AB. The corresponding contour maps of PC1 and PC2 along with the potential energies are calculated (refer to Figure

Secretory Phospholipase A2 (sPLA2) of Russell’s Viper

J. Phys. Chem. B, Vol. 114, No. 42, 2010 13469

Figure 7. φ, ψ, χ1, and χ2 angles of W31 in subunit A of AB, AIB, and AI (a,b) and subunit B (c,d) of AB and AIB.

S4 in the Supporting Information) to enable the clustering analysis for conformation belonging to lowest potential energies. In order to explore the structural characteristics and to support essential dynamics results, it is necessary to mine the meaningful conformations that represent the reaction coordination. The representative structure for each of conformational families was generated using k-means clustering algorithm embedded with mmtsb tools42,46 only for the coordinates at the potential energy landscape covered by the first two significant principal modes. The number of clusters correlates well with the corresponding distribution density as shown in Figure 8. Since the number of cluster to be formed is not obvious, the cutoff is adjusted manually to get a reasonable number. The detailed comparative analysis of cluster representatives (centroids) suggests that most of the conformational changes occurred at the flexible loops (calcium-binding and surface loops) and N-terminal helix of both viper and BPsPLA2 (Figure 9). Subunit A in the apo form (AB) of viper sPLA2 shows conformational changes at the N-terminal helix and surface loop very prominently, and at the

calcium-binding loop it is observed little with respect to the orientation of W31 (Figure 9a). The same for subunit B shows no considerable difference with respect to binding of inhibitor to subunit A (Figure 9b,d). All favorable conformational changes were arrested in subunit a-inhibitor complex with subunit B (Figure 9c). The calcium-binding loop is not showing much movement with respect to the binding of inhibitor. However, it is fluctuating more in the absence of subunit B (Figure 9e). In contrast, the surface loop of subunit A is very stable in the absence of subunit B. This shows that subunit B provides hydrophobic environment to subunit A so that the surface loop and N-terminal helix are free from hydrophilic environment and are free to move. At the same time the calcium-binding loop is controlled through the hydrophobic contacts between the W31 and neighbor residues of subunit B. This could be confirmed by comparing conformation of W31at the dimer interfaces (AIB and AB) with the conformation of the same in the absence of subunit B (AI). Strikingly, in the absence of subunit B, the movement of subunit A is almost similar to the subunit B and

13470

J. Phys. Chem. B, Vol. 114, No. 42, 2010

Ramakrishnan et al. positions of viper sPLA2. The result of hydrophobic interaction profile shows that the overall hydrophobicity of bovine source is lesser than that of viper sPLA2 and it may be the reason for the failure in its dimerization in solution. On the other hand, the lesser hydrophobicity makes the active site scaffold suitable for binding of even monodispersed inhibitors directly. Accordingly, the results of SASA, Rg (Figure 5), are well correlated with the conformational changes observed in principal component analysis (Figures 8 and 9). Conclusion

Figure 8. Principal component analysis of all MD trajectories: (a) and (b) represent the 2D projections of five individual trajectories with their first two eigenvectors with largest eigenvalues. Corresponding contour maps show the potential energy (×103 kcal/mol) along the first two principal components (refer to the Supporting Information).

it could be confirmed by comparing clusters b and d of Figure 9 with cluster e. It is noteworthy that dynamics of subunit B does not favor the binding of inhibitors. The result shows no conformational changes in subunit B with respect to inhibitor binding to subunit A. However, comparison within subunit A of these three simulations (AB, AIB, and AI) shows clearly that the residues from N-terminal helix and surface and calciumbinding loops form comfortable scaffold in subunit A for inhibitor binding. Particularly, conformation of active site scaffold of subunit A completely depends on the presence of subunit B. Hence, unlike subunit A, the subunit B lacks such kind of scaffold due to absence of consecutive subunit at its surface. Instead, it is surrounded by the water molecules and restricts the movement of surface loop and N-terminal helix predominantly. In addition, the calcium-binding loop shows movement with uncomfortable orientation of W31 for inhibitor binding as described earlier using dihedral angle (Figure 7). The conformational change observed in BPsPLA2 is almost similar to that in the viper source. The important difference observed in the dynamics of bovine source is stability of surface loop irrespective of inhibitor binding, whereas the calciumbinding loop is flexible only in the absence of inhibitor (Figure 9, f and g). The residues involved in dimerization of viper sPLA2 (L2, L3, W31, L119 of subunit A and R43, F46, V47, A101, Q108, and K131 of subunit B) are not identical to the residues at the corresponding position of bovine source (Figure 1a). Moreover, the scaffold of active site is formed by the hydrophobic residues of the N-terminal helix and L31 and Y69 are less hydrophobic compared to the residues at the corresponding

The comparative molecular dynamics of groups IB and II sPLA2s disclose their similar and unique structural features. The results clearly show that the N-terminal helix and surface loops are flexible in both groups of PLA2s with respect to binding of inhibitors. In viper sPLA2, the movement of calcium-binding loop of subunit A is controlled by hydrophobic contacts between its W31 and residues of subunit B (mainly R43, F46, and V47). Binding of inhibitor to subunit A does not restrict the movement of the calcium-binding loop while that of N-terminal and surface loop is restricted. This makes the difference in dynamics between viper and BPsPLA2s. The reason for the differential behavior of the calcium-binding loop is the presence of W31 and its contact with residues at the interface of viper sPLA2. On the other hand, movements of surface loop and N-terminal helix are completely arrested in the subunit B (Figure 9b,d). Comparison between subunit A of all AB, AIB, and AI clearly shows the proper orientation of all hydrophobic residues which favor inhibitor binding. Hence, in order to get such a suitable scaffold in subunit B with solvent on its surface, the inhibitor must possess sufficient hydrophobicity to induce favorable conformational changes in residues mainly W31, K69, L2, and L3. In the case of subunit A, it is complemented by the subunit B so that the active site is open to access for inhibitors. Unlike viper sPLA2, the BPsPLA2 (group IB) is monomer and its active site is open to access for any inhibitors without restrictions. At positions 3, 31, and others, the amino acids responsible for inhibitor binding and stability of dimer in viper sPLA2 are not identical to amino acids at corresponding positions of BPsPLA2. Consequently, the hydrophobic contacts between L2 and W31 responsible for closing of active site in viper sPLA2 are not an issue in BPsPLA2. Thus, the present study confirms that the inhibitor with sufficient hydrophobicity to induce conformational changes particularly in W31, K69, L2, and L3 and to interact with the conserved active site residues H48 and D49 through hydrogen bonding is essential for inhibition of subunit B as well. The number of hydrophobic interactions and SASA (Figure 6), dihedral angle profile (Figure 7), and interatomic distance plot for W31 substantiate the same. As per interatomic distance (Table 1), the orientation of W31 is dictated by its hydrophobic contacts with neighboring residues of subunit B (R43, F46, and V47) and the inhibitor. So the inhibitor must possess sufficiently equivalent hydrophobicity to face W31 at the gateway. Obviously, a small molecule inhibitor with much hydrophobicity equivalent to that from hydrophobic residues of subunit B at dimer interface or aggregated substrates like micelles/membrane satisfying druggable properties is questionable. Hence, our results narrow down path for further study on structure-based drug design for viper sPLA2s with more effectiveness. Further, it confirms that an inhibitor showing binding preference to subsites 4, 5, and 6 will show improved activity by binding with both the subunits of viper sPLA2 dimer. Similarly, in the recently deposited crystal structure of monomeric form of viper sPLA2-inhibitor complex, the inhibitor (indomethacin) binds

Secretory Phospholipase A2 (sPLA2) of Russell’s Viper

J. Phys. Chem. B, Vol. 114, No. 42, 2010 13471

Figure 9. Centroids of two distant conformational families selected based on backbone rmsd and superimposed. Subunit A of native (a) and complex (c) and subunit B of native (b) and complex (d) forms of viper PLA2 dimer. (e) Subunit A without subunit B (AI). Similarly, native (f) and complex (g) forms BPsPLA2 are shown. The inhibitor is not shown for transparency in viper PLA2. Green and red arrows point toward surface and calcium-binding loops, respectively.

at the subsites other than 1, 2, and 315,48 by making interactions with the active site amino acids H48 and D49 and K69. With these molecular features of enzyme and ligand, further study

on designing inhibitors will help to discover the drugs to treat the inflammation and its related diseases such as asthma, arthritis, neurodegenerative diseases, etc.

13472

J. Phys. Chem. B, Vol. 114, No. 42, 2010

Acknowledgment. The authors gratefully acknowledge the UGC for financial support. Supporting Information Available: Details on the dimer form of viper PLA2 including stability of dimer, and the variations in hydrophobic contacts and SASA between the two distinct conformations calculated using principal component analysis and k-means cluster are available. It also explains the variations in the same between native and inhibitor bound form of both viper and bovine sources. In addition, the tables for details on rmsd and residue contact map are presented using MMTSB-tools for better understanding of interactions involved in conformational changes as shown in Figure 9a-g. The contour maps of first two principal components (PC1 and PC2) along with potential energy are also shown. This explains the conformational space occupied by each type of sPLA2s during the 20 ns simulation time. This material is available free of charge via the Internet at http://pubs.acs.org. References and Notes (1) Wei, Y.; Quinn, D. M.; Sigler, P. B.; Gelb, M. H. Biochemistry 1990, 29, 6082. (2) Ward, R. J.; Alves, A. R.; Neto, J. R.; Arni, R. K.; Casari, G. Protein Eng. 1998, 11, 285. (3) Jeyaseelan, K.; Armugam, A.; Donghui, M.; Tan, N. Mol. Biol. EVol. 2000, 17, 1010. (4) Fremont, D. H.; Anderson, D. H.; Wilson, I. A.; Dennis, E. A.; Xuong, N. Proc. Natl. Acad. Sci. U.S.A. 1993, 90, 342. (5) Dijkstra, B. W.; Kalk, K. H.; Hol, W. G. J.; Drenth, J. J. Mol. Biol. 1981, 147, 97. (6) Dennis; et al. J. Biol. Chem. 1994, 269, 13057. (7) Rajakannan, V.; Yogavel, M.; Poi, M.; Arockia, J. A.; Jeyakanthan, J.; Velmurugan, D.; Tsai, M.-D.; Sekar, K. J. Mol. Biol. 2002, 324, 755. (8) Chandra, V.; Jasti, J.; Kaur, P.; Betzel, C.; Srinivasan, A.; Singh, T. P. J. Mol. Biol. 2002, 320, 215. (9) Chandra, V. s.; Jasti, J.; Kaur, P.; Dey, S.; Srinivasan, A.; Betzel, C.; Singh., T. P. Acta Crystallogr. Sect. D 2002, 58, 1813. (10) Steiner, R. A.; Rozeboom, H. J.; de Vries, A.; Kalk, K. H.; Murshudov, G. N.; Wilson, K. S.; Dijkstra, B. W. Acta Crystallogr., Sect.D 2001, 57, 516. (11) Sekar, K.; Vaijayanthi, M. S.; Yogavel, M.; Velmurugan, D. J. Mol. Biol. 2003, 333, 367. (12) Dijkstra, B. W.; Kalk, K. H.; Drenth, J.; de Haas, G. H.; Egmond, M. R.; Slotboom, A. J. Biochemistry 1984, 23, 2759. (13) Kumar, A.; Chandra, S.; Boopathy, R. Protein Sci. 1994, 3, 2082. (14) Hains, P. G.; Sung, K.-L.; Tseng, A.; Broady, K. W. J. Biol. Chem. 2000, 275, 983. (15) Singh, N.; Somvanshi, R. K.; Sharma, S.; Dey, S.; Kaur, P.; Singh, T. P. Curr Top Med Chem 2007, 7, 757. (16) Singh, G.; Jasti, J.; Saravanan, K.; Sharma, S.; Kaur, P.; Srinivasan, A.; Singh, T. P. Protein Sci. 2005, 14, 395. (17) Ling, W. C.; Balali-Mood, K.; Gavaghan, D.; Sansom, M. S. P. Biophys. J. 2008, 95, 1649. (18) Winget Jason, M.; Pan Ying, H.; Bahnson Brian, J. Biochim. Biophys. Acta 2006, 1761, 1260. (19) Arni, R. K.; Ward, R. J. Toxicon 1996, 34, 827. (20) Ramirez, F.; Jain, M. K. Proteins 1991, 9, 229. (21) Yu, B.-Z.; Rogers, J.; Tsai, M.-D.; Pidgeon, C.; Jain, M. K. Biochemistry 1999, 38, 4875. (22) Demaret, J. P.; Brunie, S. Protein Eng., Des. Sel. 1990, 4, 163.

Ramakrishnan et al. (23) Tiago, C. d. O.; Hermes Luı’s, N. d. A.; Jorge, A. G. J. Mol. Struct.: THEOCHEM 2007, 818, 31. (24) Jean-Philippe, D.; Simone, B. J. Mol. Graphics 1992, 10, 257. (25) Jones, S. T.; Ahlstro¨m, P.; Berendsen, H. J. C.; Pickersgill, R. W. Biochim.Biophys. Acta (BBA) Protein Struct. Mol. Enzymol. 1993, 1162 (5), 135. (26) Sessions, R. B.; Dauber-Osguthorpe, P.; Osguthorpe, D. J. J. Mol. Biol. 1989, 210, 617. (27) Jones, S. T.; Ahlstro¨m, P.; Berendsen, H. J. C.; Pickersgill, R. W. Biochim.Biophys. Acta (BBA) Protein Struct. Mol. Enzymol. 1993, 1162, 135. (28) Kanaujia, S. P.; Sekar, K. Acta cryst. D 2008, 64, 1003. (29) Tomoo, K.; Ohishi, H.; Ishida, T.; Inoue, M.; Ikeda, K.; Sumiya, S.; Kitamura, K. Proteins 2004, 19 (4), 330. (30) Tatulian, S. A. Biophys. J. 2001, 80, 789. (31) Hyvonen, M. T.; Oorni, K.; Kovanen, P. T.; Ala-Korpela, M. Biophys. J. 2001, 80, 565. (32) Pearlman, D. A.; Case, D. A.; Caldwell, J. W.; Ross, W. S.; Cheatham, T. E.; DeBolt, III, S.; Ferguson, D.; Seibel, G.; Kollman, P. Comput. Phys. Commun. 1995, 91, 1. (33) Bradbrook, G. M.; Gleichmann, T.; Harrop, S. J.; Habash, J.; Raftery, J.; Kalb, J.; Yariv, J.; Hillier, I. H.; Helliwell, J. R. Journal of the Chemical Society-Faraday Transactions 1998, 94, 1603. (34) Wang, J.; Wolf, R. M.; Caldwell, J. W.; Kollamn, P. A.; Case, D. A. J. Comput. Chem. 2004, 25, 1157. (35) Jakalian, A.; Bush, B. L.; Jack, B. D.; Bayly, C. I. J. Comput. Chem. 2000, 21, 132. (36) Ponder, J. W.; Case, D. A. AdV. Protein Chem. 2003, 66, 27–85. (37) Duan; et al. J. Comput. Chem. 2003, 24, 1999. (38) Berendsen, H. J. C.; Postma, J. P. M.; van Gunsteren, W. F.; Dinola, A.; Haak, J. R. J. Chem. Phys. 1984, 81, 3684. (39) Ryckaert, J.-P.; Ciccotti, G.; Berendsen, H. J. C. J.Comput. Phys. 1977, 23, 327. (40) McDonald, I. K.; Thornton, J. M. J. Mol. Biol. 1994, 238, 777– 793. (41) Wallace, A. C.; Laskowski, R. A.; Thornton, J. M. Protein Eng. 1995, 8, 127. (42) Michael, F.; John, K.; Charles, L.; Brooks, I. I. I. Journal of Molecular Graphics and Modeling 2004, 22, 377. (43) Sanner, M. F.; Olson, A. J.; Spehner, J. C. Biopolymers 1996, 38, 305. (44) Thomas W., Colin K., 2008, http://www.gnuplot.info/. (45) Shao, J.; Tanner, S. W.; Thompson, N.; Cheatham, T.E., III. J.Chem. Theory Comp. 2007, 3, 2312. (46) Hartigan, J. A.; Wong, M. A. Appl., Stat. 1979, 28, 100. (47) Jain, M. K.; Rogers, J.; Jahagirdar, D. V.; Marecek, J. F.; Ramirez, F. Biochim. Biophys. Acta 1986, 860, 435. (48) Singh, N.; Kumar, R. P.; Kumar, S.; Sharma, S.; Mir, R.; Kaur, P.; Srinivasan, A.; Singh, T. P. J.Mol.Recognit. 2009, 22, 43. (49) Larkin, M. A.; Blackshields, G.; Brown, N. P.; Chenna, R.; McGettigan, P. A.; McWilliam, H.; Valentin, F.; Wallace, I. M.; Wilm, A.; Lopez, R.; Thompson, J. D.; Gibson, T. J.; Higgins, D. G. Bioinformatics 2007, 23, 2947. (50) Clamp, M.; et al. Bioinformatics 2004, 20, 426. (51) Pettersen, E. F.; Goddard, T. D.; Huang, C. C.; Couch, G. S.; Greenblatt, D. M.; Meng, E. C.; Ferrin, T. E. J. Comput. Chem. 2004, 25, 1605. (52) Hubbard, S. J.; Thornton, J. M. Department of Biochemistry and Molecular Biology, University College London, 1993. (53) DeLano, W. L. http://www.pymol.org, 2002. (54) Berman, H. M.; Westbrook, J.; Feng, Z.; Gilliland, G.; Bhat, T. N.; Weissig, H.; Shindyalov, I. N.; Bourne, P. E. Nucleic Acids Res. 2000, 28, 235.

JP102073F