Coarse-Grained Molecular Dynamics Simulations of Cobra Cytotoxin

Dec 30, 2010 - Cardiotoxins: Functional Role of Local Conformational Changes. Anastasia G. Konshina , Nikolay A. Krylov , and Roman G. Efremov. Journa...
0 downloads 8 Views 8MB Size
ARTICLE pubs.acs.org/JPCB

Coarse-Grained Molecular Dynamics Simulations of Cobra Cytotoxin A3 Interactions with a Lipid Bilayer: Penetration of Loops into Membranes Zhi-Yuan Su*,† and Yeng-Tseng Wang*,‡ † ‡

The Department of Information Management, Chia Nan University of Pharmacy & Science, Tainan 717, Taiwan National Center for High-performance Computing, Hsin-Shi, Tainan County, Taiwan ABSTRACT: Cobra cytotoxins, which are small three-looped proteins composed of approximately 60 amino acid residues, primarily act by destroying the bilayer membranes of cells and artificial vesicles. However, the molecular mechanism governing this process is not yet completely understood. We used coarse-grained molecular dynamics (CGMD) simulations to study the mechanism underlying the penetration of cardiotoxin A3 (CTX A3), the major toxic component of Naja atra (Chinese cobra) venom, into a hydrated 1-palmitoyl-2-oleoyl-1sn-3-phosphatidylcholine (POPC) lipid bilayer. We performed CGMD simulations for three different conformations of the cobra cytotoxin; the tail, lying, and harrow conformations. The results of our simulations indicate that two of these, the tail and lying conformations, did not penetrate the bilayer system. Further, for the harrow conformation, loops 2 and 3 played important roles in penetration of CTX A3 into the bilayer system.

’ INTRODUCTION Cardiotoxins (CTXs) are small basic proteins that are abundantly present in the venom of the elapid family of snakes. CTXs have highly diverse pharmacological effects, such as hemolysis, cytotoxicity, and muscle depolarization. They are composed of 60-62 amino acid residues and 4 disulfide bridges, and belong to a large family of three-fingered toxins. The three-dimensional structures of all CTXs have a three-fingered loop-folding topology dominated by a β-sheet, with the structures differing in characteristics such as the number of secondary structures and the positions of invariant side chains.1 Their primary biological effect is considered to be their ability to decompose cell membranes. Previous studies2-5 have shown that CTXs can bind to lipid membranes, and the three fingers of CTXs play an important role in the decomposition of cell membranes. A detailed analysis of the three-fingered loop structure of the CTXs has determined that interaction between the CTXs and cell membranes involves insertion of the hydrophobic tips of the three loops into the membrane.5 The strength of the interaction depends on the nature of the toxin as well as that of the membrane lipid.2,5 Dubovskii et al.2 used nuclear magnetic resonance (NMR) and Monte Carlo (MC) techniques to measure the binding abilities of P- and S-type CTX A3 and cell membranes. Dubovskii et al. concluded that the tips of loops 1 and 2 of S-type CTX A3 are less hydrophobic than in P-type due to amino acid replacements r 2010 American Chemical Society

(Phe10fAla, Ala28fSer, and Ala29fAsp), and these residue alterations may affect the binding ability of S-type CTX A3 with cell membranes. On the basis of NMR experiments and MC simulations, Dubovskii et al. indicated that P-type CTXs containing the Pro-30 residue interact with cell membranes more strongly than do S-type CTXs containing the Ser-28 residue. Therefore, we chose P-type CTX A3 as our initial model for observing the penetration of CTX A3 loops into membranes. CTX A3, the major toxic component of Naja atra (Chinese cobra), is a P-type CTX that damages phospholipid vesicles. However, it is difficult to observe the mechanism underlying penetration of CTX A3 into lipid membranes through instrumental experimentation, as it takes place on a time scale ranging from a few microseconds to seconds.6-9 In principle, all-atom MD simulations would be useful to gain detailed insights into and obtain atomic resolution images of the interaction of CTXs with membranes.3,10,11 However, the time scale required for a thorough understanding of the entire penetration process is beyond the limits of routine modeling.5 Coarsegrained models in which small groups of atoms are represented by a single interaction site have therefore emerged as a possible Received: August 11, 2010 Revised: November 30, 2010 Published: December 30, 2010 796

dx.doi.org/10.1021/jp107599v | J. Phys. Chem. B 2011, 115, 796–802

The Journal of Physical Chemistry B

ARTICLE

alternative to overcome this limitation.12-14 Recently, CGMD simulations have been used to study the interactions involved in self-assembly of hydrophobic cyclic peptides in lipid membranes,15 particularly interactions of membrane proteins and ion channels with membranes.16-18 In the present study, three different conformations of CTX A3 (the tail, lying, and harrow conformations) and a hydrated 1-palmitoyl-2-oleoyl-1-sn-3-phosphatidylcholine (POPC) lipid bilayer were selected for a study on the penetration of CTX A3 loops into cell membranes. Considering that the time scale required for the complete penetration process is often on the order of microseconds, CGMD simulations were performed.

were created using visual molecular dynamics (VMD) software19 and are shown in Figure 1. To obtain stable conformations for CTX A3, the models were subjected to energy minimization by applying the conjugate gradient method for approximately 10 000 iterations, and then 2 ns, isothermal, constant-volume MD simulations were conducted. The latter were based on the SHAKE method for stabilizing bond lengths and CHARMM allhydrogen amino acid parameters (par_all27_ prot_lipid_na.inp),20 and were performed using the NAMD21 program. The optimized models formed the basis for the subsequent CG models. CG Models and MD Simulations. The initial POPC lipid bilayers, containing 280 POPC molecules and 20 516 TIP3 water molecules, were created using VMD software; the volume of the models was 1.2  1.2  0.56 nm. Each conformation of CTX A3 was placed on top of a POPC lipid bilayer. The TIP3 water molecules around the CTX A3 (0.24 nm) were then deleted. The preliminary CG models for the three conformations of CTX A3 were created using the CG builder module of VMD. The calculations were performed using the NAMD package with the Martini force field parameters.22,23 Unless otherwise noted, all simulations were performed in the canonical ensemble24 with a simulation temperature of 310 K, the Verlet algorithm, a time step of 0.02 ps, and the SHAKE algorithm for all covalent bonds. For electrostatic interactions, atom-based truncations were performed using the particle mesh Ewald (PME) method. In addition, a switching function was applied to the van der Waals terms with a 3.5 nm cutoff for atom pairs. The three preliminary CG models were subjected to energy minimization by applying the conjugate gradient method for approximately 200 000 iterations. The energy-minimized CG models were then subjected to 5 μs, isothermal, constant-volume MD simulations. The trajectories obtained through these simulations were used to analyze the degree to which the CTX A3 loops had penetrated the POPC lipid bilayer. Order Parameter (SPOPC lipid) for the POPC Lipid Bilayer and the Penetration Depth of CTX A3 in the Membrane. To estimate the depth to which the CTX A3 loops penetrated the POPC lipid bilayer, lipid orientations in the POPC bilayer were measured experimentally using polarized attenuated total reflection Fourier transform infrared (ATR-FTIR) spectroscopy.5 In addition, the penetration depth was estimated computationally by calculating the order parameter for the POPC lipid bilayer25,26 for the case in which the three CTX A3 loops penetrated the membrane surface, using the method applied in recent simulations.26

’ METHODS Preliminary Model of the Three Conformations of CTX A3 and the POPC Lipid Bilayer. For the initial calculations, we

used the NMR structure of CTX A3 (PDB ID: 1I02) as our fusion protein model, with 60 amino acid residues. The preliminary models of the tail, lying, and harrow conformations of CTX A3

Figure 1. Overview of POPC lipid bilayer-CTX A3 system. CTX A3 is shown as a ribbon. The three loops of CTX A3 are loop 1 (CYS3LYS12), loop 2 (TYR22-ARG36), and loop 3 (PRO43-VAL52). These correspond to the three models of CTX A3: tail, lying, and harrow.

Figure 2. Two snapshots (150 ns and 5 μs) showing penetration of harrow of CTX A3 into POPC lipid bilayer. 797

dx.doi.org/10.1021/jp107599v |J. Phys. Chem. B 2011, 115, 796–802

The Journal of Physical Chemistry B

ARTICLE

Figure 3. Two snapshots (150 ns and 5 μs) showing penetration of lying of CTX A3 into POPC lipid bilayer.

Figure 4. Two snapshots (150 ns and 5 μs) showing penetration of tail of CTX A3 into POPC lipid bilayer.

The POPC lipid bilayer order parameter was defined as follows: " # N 3 cos2 θi, j - 1 1X SPOPC lipid ¼ ð1Þ N k¼1 2 where θi,j is the angle between the two vectors along the ith and jth lipid in the POPC bilayer and N is the total number of vector pairs. To estimate the depth to which the CTX A3 loops penetrated, we used both the center of mass (CM) of the CTX A3 molecule and the three loops individually. Interaction Energy between CTX A3 and the POPC Lipid Bilayer. During our simulations, nonbonding interactions such as Coulombic (ELEC) and van der Waals (VDW) interactions between CTX A3 and the POPC lipid bilayer, which contains polar (phosphate) and nonpolar (methyl) groups, were monitored as a function of time and the interaction energy was computed.

Figure 5. Order parameter of POPC lipid bilayer estimated through 5 μs simulations for three different conformations of CTX A3.

bilayer, which reached 2.40 and 0.75 nm at 150 ns and 5 μs, respectively. At a simulation time of 150 ns, CTX A3 had adsorbed into the lipid bilayer and had begun to penetrate. The bilayer was destroyed by this conformation after 5 μs. Figure 3 shows two snapshots of the penetration depth of the lying conformation of CTX A3, which reached 2.82 and 1.92 nm at 150 ns and 5 μs, respectively. At 150 ns, CTX A3

’ RESULTS AND DISCUSSION CGMD Simulations of the POPC Lipid Bilayer-CTX A3 System. Figure 2 shows two snapshots of the penetration depth

of the harrow conformation of CTX A3 in the POPC lipid 798

dx.doi.org/10.1021/jp107599v |J. Phys. Chem. B 2011, 115, 796–802

The Journal of Physical Chemistry B

ARTICLE

Figure 6. Comparison of penetration depth of CTX A3: (A) harrow of CTX A3, (B) lying of CTX A3, and (C) tail of CTX A3. Top bilayers and bottom bilayers lines represent the highest and lowest depth of the bilayers system, respectively.

had adsorbed into the POPC lipid bilayer; however, after 5 μs of simulation time, the lying conformation had only partially penetrated the bilayer. Figure 4 shows similar results for the tail conformation, with a penetration depth of 4.19 and 3.11 nm at 150 ns and 5 μs, respectively. The penetration behaviors of the lying and tail conformations of CTX A3 were almost identical,

neither having fully penetrated the POPC lipid bilayer at 5 μs of simulation time. Order Parameter for the POPC Lipid Bilayer. At the beginning of the simulation, when the protein had not yet affected the membrane, the order parameter SPOPC lipid was 0.7182. The final SPOPC lipid was determined at 5 μs (Figure 5). For the harrow 799

dx.doi.org/10.1021/jp107599v |J. Phys. Chem. B 2011, 115, 796–802

The Journal of Physical Chemistry B

ARTICLE

shown in Figure 6C. No penetration was observed in any of these simulations. The simulation results of these three models indicate that the harrow conformation of CTX A3 penetrated the bilayer system, and in particular, loops 2 and 3 played a critical role in the penetration process. Interaction Energy between CTX A3 and the POPC Lipid Bilayer. After demonstrating that the harrow conformation of CTX A3 was able to fully penetrate the POPC lipid bilayer, we next analyzed the interaction energies between the harrow conformation and the lipid bilayer (Figures 7 and 8A). The interaction energies corresponding to the polar and nonpolar parts of the bilayer decreased quickly from 0 to 25 ns (Figures 7A and 8A). The average polar interaction affinity of the three loops was loop 2 (-300 kcal/mol) > loop 3 (-210 kcal/mol) > loop 1 (-30 kcal/mol), and the average nonpolar interaction affinity of the three loops was loop 2 (-22 kcal/mol) ≈ loop 3 (-22 kcal/ mol) > loop 1 (-11 kcal/mol). The interaction energies between the lying conformation of CTX A3 and the bilayer are shown in Figures 7 and 8B. The interaction energies decreased smoothly from 0 to 20 ns. The average polar interaction affinity of the three loops was loop 3 (-80 kcal/mol) > loop 2 (-50 kcal/mol) > loop 1 (-20 kcal/ mol), and the average nonpolar interaction affinity of the three loops was loop 3 (-22 kcal/mol) ≈ loop 2 (-22 kcal/mol) > loop 1 (-11 kcal/mol). The interaction energies between the POPC lipid bilayer and the tail conformation of CTX A3 are shown in Figures 7 and 8C. Results were similar to the lying conformation, with average polar interaction affinities of loop 3 (-40 kcal/mol) > loop 2 (-30 kcal/mol) > loop 1 (-10 kcal/ mol), and average nonpolar interaction affinities of loop 2 (-12 kcal/mol) > loop 3 (-10 kcal/mol) > loop 1 (-5 kcal/ mol). These simulation results indicate that loops 2 and 3 provide CTX A3 with the necessary interaction affinity to penetrate the polar part of the bilayer system and, furthermore, that the orientation of CTX A3 significantly affects its penetration ability. Overall, our simulation results indicate that the harrow conformation of CTX A3 is able to penetrate the bilayer system, that loops 2 and 3 are strongly associated with the penetration of CTX A3 into the bilayer system, and that loop 1 contributes only a little to the penetration mechanism. Loop 3 of the lying conformation of CTX A3 was only able to partially penetrate the bilayer system after 3.2 μs, and the strongest interaction energy was estimated as -130 kcal/mol (polar -110 kcal/mol and nonpolar -20 kcal/mol). For the tail conformation of CTX A3, no penetration was observed. Finally, the harrow conformation of CTX A3 was adsorbed into the POPC lipid bilayer by 30 ns. After 700 ns, the interaction energy between CTX A3 and the POPC lipid bilayer increased smoothly. Loop 3 invaded the lipid bilayer at 253 ns, with an SPOPC lipid of 0.4136. Thereafter, the bilayer system rapidly began to fall apart and had completely collapsed by 450 ns. Loop 2 also invaded the lipid bilayer at 753 ns, with an SPOPC lipid of 0.038 72. At this point, the greater part of CTX A3 had invaded the bilayer system, and the polar interaction energies of loops 2 and 3 exceeded -100 kcal/mol with total interaction energies exceeding -900 kcal/mol.

Figure 7. Comparison of interaction energy between POPC lipid bilayer and the three different conformations of CTX A3: (A) time scale 0-200 ns; (B) time scale 0-5000 ns.

conformation, SPOPC lipid decreased rapidly from 0 to 400 ns and remained relatively stable (0-0.1) after 400 ns. On the other hand, for the lying conformation, SPOPC lipid decreased smoothly from 0 to 900 ns and remained relatively stable (∼0.4) after 900 ns. Similarly, for the tail conformation, SPOPC lipid decreased smoothly from 0 to 870 ns and remained relatively stable (∼0.3) after 870 ns. Comparison of these results with the simulation results for the three conformations of CTX A3 indicates that the harrow conformation destroyed the POPC lipid bilayer completely, while the lying and tail conformations disordered the membrane, although it retained its bilayer nature. Penetration Depths of the CTX A3 Loops. The penetration depths of the harrow conformation of CTX A3 and its three loops into the POPC lipid bilayer were calculated based on the results of the simulations (Figure 6A). All three loops penetrated the lipid bilayer; however, loops 2 and 3 penetrated the bilayer at 220 ns, while loop 1 penetrated at 450 ns. The overall CM of CTX A3 penetrated at 220 ns. The penetration depths of the lying conformation of CTX A3 and its three loops into the POPC lipid bilayer are shown in Figure 6B. Only loop 3 had partially penetrated into the lipid bilayer by 3200 ns. No penetration was observed for the CM, loop 1, or loop 2. Finally, the penetration depths of the tail conformation of CTX A3 and its three loops into the lipid bilayer are

’ CONCLUSIONS In the present study, we used NAMD, Martini force fields, VMD, CGMD simulations, and a lipid membrane to predict the penetration depth of CTX A3 into a bilayer system 800

dx.doi.org/10.1021/jp107599v |J. Phys. Chem. B 2011, 115, 796–802

The Journal of Physical Chemistry B

ARTICLE

Figure 8. Comparison of interaction energy between POPC lipid bilayer and the three different conformations of CTX A3: (A) harrow of CTX A3, (B) lying of CTX A3, and (C) tail of CTX A3.

the bilayer were stronger than -100 kcal/mol, penetration of the loops into the membrane could occur for the harrow and lying conformations of CTX A3. Thus, the modeling approach presented here theoretically suggests that loops 2 and 3 of the harrow conformation of CTX A3 play an important role in the mechanism underlying the penetration of CTX A3 into bilayer systems.

composed of a POPC lipid bilayer and water molecules. Our simulation results indicate that the harrow conformation of CTX A3 is able to penetrate the bilayer system. Loop 3 is the first to invade the lipid bilayer, accelerating the collapse of the bilayer system. Subsequently, loop 2 invades, facilitating the penetration of CTX A3 into the bilayer system. Our observations also indicate that, when the loop interaction energies of the polar part of 801

dx.doi.org/10.1021/jp107599v |J. Phys. Chem. B 2011, 115, 796–802

The Journal of Physical Chemistry B

ARTICLE

’ ACKNOWLEDGMENT This work was supported by the National Center for Highperformance Computing and Chia Nan University of Pharmacy & Science, Taiwan.

(19) Humphrey, W.; Dalke, A.; Schulten, K. VMD - Visual Molecular Dynamics. J. Mol. Graphics 1996, 14, 33–38. (20) Brooks, B. R.; Bruccoleri, R. E.; Olafson, B. D.; States, D. J.; Swaminathan, S.; Karplus, M. CHARMM: A Program for Macromolecular Energy, Minimization, and Dynamics Calculations. J. Comput. Chem. 1983, 4, 187–217. (21) Phillips, J. C.; Braun, R.; Wang, W.; Gumbart, J.; Tajkhorshid, E.; Villa, E.; Chipot, C.; Skeel, R. D.; Kale, L.; Schulten, K. Scalable molecular dynamics with NAMD. J. Comput. Chem. 2005, 26, 1781– 1802. (22) Marrink, S. J.; Periole, X.; Tieleman, D. P.; Vries, D. Comment on using a too large integration time step in molecular dynamics simulations of coarse-grained molecular models. Phys. Chem. Chem. Phys. 2010, 12, 2254–2256. (23) Marrink, S. J.; Vries, A. H.; Mark, A. E. Coarse grained model for semiquantitative lipid simulations. J. Phys. Chem. B 2004, 108, 750– 760. (24) Ryckaert, J. P.; Ciccotti, G. Andersen’s canonical-ensemble molecular dynamics for molecules with constraints. Mol. Phys. 1986, 58, 1125–1136. (25) Han, X.; Tamm, L. K. A host-guest system to study structurefunction relationships of membrane fusion peptides. Proc. Natl. Acad. Sci. U.S.A. 2000, 97, 13097–13102. (26) Xu, D.; Budil, D. E.; Ober, C. K.; Freed, J. H. Rotational diffusion and order parameters of a liquid crystalline polymer studied by ESR: molecular weight dependence. J. Phys. Chem. 1996, 100, 15867– 15872.

’ REFERENCES (1) Jayaraman, T. K.; Kumar, C. C.; Tsai, S.; Srisailam, S. H.; Chou, C. L.; Ho, C. Elucidation of the solution structure of cardiotoxin analogue V from the Taiwan cobra (Naja naja atra), identification of structural features important for the lethal action of snake venom cardiotoxins. Protein Sci. 2000, 9, 637–646. (2) Dubovskii, P. V.; Lesovoy, D. M.; Dubinnyi, M. A.; Konshina, A. G.; Utkin, Y. N.; Efremov, R. G.; Arseniev, A. S. Interaction of threefinger toxins with phospholipid membranes: comparison of S- and P-type cytotoxins. Biochem. J. 2005, 387, 807–815. (3) Efremov, R. G.; Volynsky, P. E.; Nolde, D. E.; Dubovskii, P. V.; Arseniev, A. S. Interaction of cardiotoxins with membranes: a molecular modeling study. Biophys. J. 2002, 83, 144–153. (4) Feofanov, A. V.; Sharonov, G. V.; Astapova, M. V.; Rodionov, D. I.; Utkin, Y. N.; Arseniev, A. S. Cancer cell injury by cytotoxins from cobra venom is mediated through lysosomal damage. Biochem. J. 2005, 390, 11–18. (5) Huang, W. N.; Sue, S. C.; Wang, D. S.; Wu, P. L.; Wu, W. G. Peripheral binding mode and penetration depth of cobra cardiotoxin on phospholipidmembranes as studied by a combined FTIR and computer simulation approach. Biochemistry 2003, 42, 7457–7466. (6) Panahi, A; Feig, M. Conformational Sampling of Influenza Fusion Peptide in Membrane Bilayers as a Function of Termini and Protonation States. J. Phys. Chem. B 2010, 114, 1407–1416. (7) Chang, D. K.; Cheng, S. F. pH-dependence of intermediate steps of membrane fusion induced by the influenza fusion peptide. Biochem. J. 2006, 396, 557–563. (8) Epand, R. M.; Epand, R. F.; Martin, I.; Ruysschaert, J. M. Membrane Interactions of Mutated Forms of the Influenza Fusion Peptide. Biochemistry 2001, 40, 8800–8807. (9) Ohki, S.; Baker, G. A.; Page, P. M.; McCarty, T. A.; Epand, R. M.; Bright, F. V. Interaction of Influenza Virus Fusion Peptide with Lipid Membranes: Effect of Lysolipid. J. Membr. Biol. 2006, 211, 191–200. (10) Levtsova, O. V.; Antonov, M. Y.; Mordvintsev, D. Y.; Utkin, Y. N.; Shaitan, K. V.; Kirpichnikov, M. P. Steered molecular dynamics simulations of cobra cytotoxin interaction with zwitterionic lipid bilayer: no penetration of loop tips into membranes. Comput. Biol. Chem. 2009, 33, 29–32. (11) Konshina, A. G.; Volynskii, P. E.; Arsen’ev, A. S.; Efremov, R. G. Interaction of cardiotoxin A5 with a membrane: role of conformational heterogeneity and hydrophilic properties. Bioorg. Khim. 2003, 29, 577–88. (12) Sherwood, P.; Brooks, B. R.; Sansom, M. S. P. Multiscale methods for macromolecular simulations. Curr. Opin. Struct. Biol. 2008, 18, 630–640. (13) Ayton, G. S.; Voth, G. A. Systematic multiscale simulation of membrane protein systems. Curr. Opin. Struct. Biol. 2009, 19, 138–144. (14) Arkhipov, A.; Yin, Y.; Schulten, K. Four-Scale Description of Membrane Sculpting by BAR Domains. Biophys. J. 2008, 95, 2806–2821. (15) Khalfa, A.; Treptow, W.; Maigret, B.; Tarek, M. Self assembly of peptides near or within membranes using coarse grained MD simulations. Chem. Phys. 2009, 358, 161–170. (16) Bond, P. J.; Sansom, M. S. P. Bilayer deformation by the Kv channel voltage sensor domain revealed by self-assembly simulations. Proc. Natl. Acad. Sci. U.S.A. 2007, 104, 2631–2636. (17) Treptow, W.; Marrink, S. J.; Tarek, M. Gating Motions in Voltage-Gated Potassium Channels Revealed by Coarse-Grained Molecular Dynamics Simulations. J. Phys. Chem. B 2008, 112, 3277–3282. (18) Chetwynd, A. P.; Scott, K. A.; Mokrab, Y.; Sansom, M. S. P. CGDB: A database of membrane protein/lipid interactions by coarsegrained molecular dynamics simulations. Mol. Membr. Biol. 2008, 25, 662–669. 802

dx.doi.org/10.1021/jp107599v |J. Phys. Chem. B 2011, 115, 796–802