2424
Biophysical Journal Volume 108 May 2015 2424–2426
Biophysical Letter The Structure of a Melittin-Stabilized Pore John M. Leveritt III,1 Almudena Pino-Angeles,1 and Themis Lazaridis1,* 1
Department of Chemistry, The City College of New York, New York, New York
ABSTRACT Melittin has been reported to form toroidal pores under certain conditions, but the atomic-resolution structure of these pores is unknown. A 9-ms all-atom molecular-dynamics simulation starting from a closely packed transmembrane melittin tetramer in DMPC shows formation of a toroidal pore after 1 ms. The pore remains stable with a roughly constant radius for the rest of the simulation. Surprisingly, one or two melittin monomers frequently transition between transmembrane and surface states. All four peptides are largely helical. A simulation in a DMPC/DMPG membrane did not lead to a stable pore, consistent with the experimentally observed lower activity of melittin on anionic membranes. The picture that emerges from this work is rather close to the classical toroidal pore, but more dynamic with respect to the configuration of the peptides.
Received for publication 21 January 2015 and in final form 9 April 2015. *Correspondence:
[email protected]
Melittin is a 26-residue lytic peptide found in bee venom (1), which bears a resemblance to antimicrobial peptides and has antimicrobial activity (2). The peptide’s interaction with membranes has been extensively studied (3). Oriented circular dichroism and neutron scattering experiments suggested that melittin forms toroidal pores in zwitterionic membranes under certain conditions dependent on peptide concentration, lipid composition, temperature, and hydration (4). Leakage experiments with varying lipid compositions also support the toroidal pore model (5). However, the precise structure of this pore is still a subject of debate. The classical toroidal pore is envisioned as a lipidic pore with peptides embedded in a transmembrane orientation (4). This picture was challenged by simulations of spontaneous pore formation, which showed a more disordered picture (6,7). Other simulations starting from a transmembrane bundle (8) or a preformed pore (9–11) were too short to provide an equilibrated pore structure. In this work, long all-atom MD simulations were performed on the ANTON Supercomputer (Pittsburgh Supercomputing Center, Carnegie Mellon University, University of Pittsburgh, Pittsburgh, PA) to determine a putative melittin pore structure. Our initial system is a closely packed tetramer in a membrane with a peptide/lipid ratio of 1:18. Two lipid compositions were considered: the zwitterionic DMPC and a 3:1 DMPC/DMPG mixture, with 25% anionic character. Technical details can be found in the Supporting Material. Representative snapshots from the DMPC simulation are shown in Fig. 1. After 1 ms, a state is found with the peptides highly tilted or bent, lining a small pore along with lipid headgroups. At this point the pore radius begins to expand, keeping the pore open for the remainder of the 9-ms simulation. The melittin tetramer is highly dynamic and transitions between different states where one or two peptides undergo
a transition from the transmembrane (I) state to a surface (S) state. This is shown in Fig. 1 B, where even with one of the monomers in the S state the pore remains open. Near the end of the simulation the tetramer returns to a classical toroidal pore structure where the peptides line the pore and the headgroups come together from each of the leaflets (Fig. 1 C). Thus, the pore is highly stable but also highly dynamic. The frequent transitions suggest that the simulation has reached a state of at least metastable equilibrium. Throughout the simulation each of the monomers undergoes at least a small amount of tilting (Fig. S2). Two of the monomers remain pretty stable throughout the simulation with tilt angles 21–28 , consistent with the solid-state NMR values of 10–30 in DMPC at high peptide/lipid ratio (12). One monomer is dynamic enough to fluctuate from its initial I state to an S state (tilt angle 75–90 ), as shown in Fig. 1, B and C. This transition first occurs at ~2 ms and then two more times at 4 and 7.5 ms. A second monomer transitions briefly to the S state at ~1.5, 6, and 8 ms (Fig. S2). The pore radius was calculated at different membrane depths as a function of time. The minimum pore radius, which was found near the center of the membrane, is shown in Fig. 2. After the first microsecond, the simulation clearly converges to a stable, open pore. The calculated radius is close to the radii found in dye leakage experiments (12.5– ˚ ) (13–15) but is significantly smaller than those deter15 A mined by neutron scattering (4). The observed pore size in our simulations may be limited by the relatively small size of the simulation box.
Editor: Emad Tajkhorshid. Ó 2015 by the Biophysical Society http://dx.doi.org/10.1016/j.bpj.2015.04.006
Biophysical Letters
2425
FIGURE 1 Snapshots of the formation and stabilization of a toroidal pore caused by melittin. (Colored ribbons) Melittin peptides; (orange spheres) phosphates; (gray spheres) lipids; (red and gray sticks) water. (A) Structure of the preformed pore at 1 ms where it can be seen that phosphates have entered the membrane, and the peptides are disordered. (B) Structure of the pore after 2.25 ms; the widening of the pore is clear with the presence of an S-state monomer. (C) Final structure at 9 ms with a neatly formed classical toroidal pore and all of the monomers in the I state. (D) Final structure at 9 ms for the melittin tetramer in DMPC/DMPG where no stable pore is present. To see this figure in color, go online.
Fig. 3 shows the average pore radius at different membrane depths over the last microseconds. The shape of the pore is roughly parabolic and symmetric, indicating equal contribution of headgroups from each leaflet. The distribution of the lipid headgroups as a function of membrane depth over the last microseconds, which mirrors the pore radius plot, is also shown in Fig. 3. Previous all-atom simulations of ideal toroidal pores found that the ratio of headgroup density in the pore center relative to that on the flat membrane (the homogeneity factor h) is close to 0.6 (16). A similar analysis on this pore structure gives h ¼ 0.5. The slightly lower value could be simply due to the excluded volume effect by the peptides. Previous simulation results on melittin and a magainin derivative using the GROMOS force field (www.gromos. net) found that the peptides are largely unfolded (6,7,10). In this simulation using CHARMM, Ver. 36 (www. charmm.org) (17,18), helicities calculated over the last microseconds give 69% for the melittin monomer that undergoes the I to S state transition most frequently and 77, 79, and 83% for the other three monomers. The loss of helicity is restricted at the termini, in particular the initial and final three residues. The remaining central portion of the peptides is very helical, with values of 86% for the I-S transitioning monomer, and 96, 99, and 99% for the other three monomers. High helicity was also observed in a recent simulation of melittin on the membrane surface (19). To determine the contribution to pore formation of the peptide that transitions frequently to the S state, we ran an additional 2 ms after deleting this peptide and reequilibrating the system. The pore was found to shrink, and one peptide separated from the other two (see the Supporting Material). Thus, it seems that the fourth peptide stabilizes the pore substantially, whether it is in the I or the S orientation. The melittin tetramer in 3:1 DMPC/DMPG undergoes very different dynamics compared to DMPC alone. In this membrane a stable pore is not formed and one peptide separates from the other three (Fig. 1 D). Throughout the simula-
tion the peptides remain in the I state and the lipid headgroups remain on the membrane surface (see the Supporting Material for details). Interestingly, several groups reported that melittin causes leakage preferentially in zwitterionic vesicles (20,21), while other leakage experiments with anionic lipid vesicles suggest a detergentlike mechanism that opposes pore formation (22). In addition, recent electron paramagnetic resonance experiments in 7:3 DPPC/PG failed to find evidence of pores (23). On the other hand, toroidal pores have been recently detected in a POPC/POPG mixture (24). Clearly, it is not possible to reach a conclusion on the basis of a single simulation. Future work will need to systematically address the effect of different lipid properties. It would also be useful to determine the intrinsic free energy of pore formation in anionic versus zwitterionic membranes. The picture that emerges from this work differs from that proposed in previous simulations (6,7) in a number of ways: 1) the peptides are much more helical, 2) they are not aggregated (average Ca-Ca distances between peptide pairs are ˚ over the last microseconds), and 3) at least some 19–30 A of them have a roughly transmembrane orientation. This picture is closer to the classical toroidal pore model, but also has elements of the disordered one in that one or two peptides can be parallel to the membrane surface while
˚ ngstroms) as a function of time (in FIGURE 2 Pore radius (in A microseconds). Biophysical Journal 108(10) 2424–2426
2426
Biophysical Letters 3. Raghuraman, H., and A. Chattopadhyay. 2007. Melittin: a membraneactive peptide with diverse functions. Biosci. Rep. 27:189–223. 4. Yang, L., T. A. Harroun, ., H. W. Huang. 2001. Barrel-stave model or toroidal model? A case study on melittin pores. Biophys. J. 81:1475– 1485. 5. Allende, D., S. A. Simon, and T. J. McIntosh. 2005. Melittin-induced bilayer leakage depends on lipid material properties: evidence for toroidal pores. Biophys. J. 88:1828–1837. 6. Leontiadou, H., A. E. Mark, and S. J. Marrink. 2006. Antimicrobial peptides in action. J. Am. Chem. Soc. 128:12156–12161. 7. Sengupta, D., H. Leontiadou, ., S.-J. Marrink. 2008. Toroidal pores formed by antimicrobial peptides show significant disorder. Biochim. Biophys. Acta. 1778:2308–2317. 8. Lin, J. H., and A. Baumgaertner. 2000. Stability of a melittin pore in a lipid bilayer: a molecular dynamics study. Biophys. J. 78:1714–1724. 9. Mihajlovic, M., and T. Lazaridis. 2010. Antimicrobial peptides in toroidal and cylindrical pores. Biochim. Biophys. Acta. 1798:1485–1493.
FIGURE 3 Plot of average pore radius and phosphate distribution as a function of membrane depth (z ¼ 0 at the center of the membrane). To see this figure in color, go online.
lining the rim of the pore (6). This may not be true for other peptides; similar simulations of magainin and PGLa gave peptide configurations that are much more tilted (A. PinoAngeles, J. M. Leveritt III, and T. Lazaridis, unpublished). In this work, a tetramer was considered based on available experimental evidence (25,26). It is necessary to perform similar simulations of larger systems and other oligomeric states to determine the optimal aggregate and its physical determinants. SUPPORTING MATERIAL Supporting Material, Supporting Results, and five figures are available at http://www.biophysj.org/biophysj/supplemental/S0006-3495(15)00384-7.
10. Irudayam, S. J., and M. L. Berkowitz. 2011. Influence of the arrangement and secondary structure of melittin peptides on the formation and stability of toroidal pores. Biochim. Biophys. Acta. 1808:2258–2266. 11. Mihajlovic, M., and T. Lazaridis. 2012. Charge distribution and imperfect amphipathicity affect pore formation by antimicrobial peptides. Biochim. Biophys. Acta. 1818:1274–1283. 12. Naito, A., T. Nagao, ., H. Saitoˆ. 2000. Conformation and dynamics of melittin bound to magnetically oriented lipid bilayers by solid-state 31P and 13C NMR spectroscopy. Biophys. J. 78:2405–2417. 13. Matsuzaki, K., S. Yoneyama, and K. Miyajima. 1997. Pore formation and translocation of melittin. Biophys. J. 73:831–838. 14. Ladokhin, A. S., M. E. Selsted, and S. H. White. 1997. Sizing membrane pores in lipid vesicles by leakage of co-encapsulated markers: pore formation by melittin. Biophys. J. 72:1762–1766. 15. Kokot, G., M. Mally, and S. Svetina. 2012. The dynamics of melittininduced membrane permeability. Eur. Biophys. J. 41:461–474. 16. He, Y., L. Prieto, and T. Lazaridis. 2013. Modeling peptide binding to anionic membrane pores. J. Comput. Chem. 34:1463–1475. 17. Best, R. B., X. Zhu, ., A. D. MacKerell, Jr. 2012. Optimization of the additive CHARMM all-atom protein force field targeting improved sampling of the backbone 4, j and side-chain c1 and c2 dihedral angles. J. Chem. Theory Comput. 8:3257–3273.
AUTHOR CONTRIBUTIONS
18. Klauda, J. B., R. M. Venable, ., R. W. Pastor. 2010. Update of the CHARMM all-atom additive force field for lipids: validation on six lipid types. J. Phys. Chem. B. 114:7830–7843.
J.M.L., A.P.-A., and T.L. designed research; J.M.L. and A.P.-A. performed research and analyzed the data; and J.M.L. and T.L. wrote the article.
19. Andersson, M., J. P. Ulmschneider, ., S. H. White. 2013. Conformational states of melittin at a bilayer interface. Biophys. J. 104:L12–L14.
ACKNOWLEDGMENTS The ANTON machine at the Pittsburgh Supercomputing Center was generously made available by D.E. Shaw Research. This work was supported by the National Science Foundation (Molecular and Cellular Biosciences grant No. 1244207). Infrastructure support was provided in part by Research Centers in Minority Institutions grant No. 8G12MD007603 from the National Institutes of Health. ANTON computer time was provided by the National Center for Multiscale Modeling of Biological Systems through grant No. P41GM103712-S1 from the National Institutes of Health and the Pittsburgh Supercomputing Center.
REFERENCES 1. Dempsey, C. E. 1990. The actions of melittin on membranes. Biochim. Biophys. Acta Rev. Biomembr. 1031:143–161. 2. Javadpour, M. M., M. M. Juban, ., M. L. McLaughlin. 1996. De novo antimicrobial peptides with low mammalian cell toxicity. J. Med. Chem. 39:3107–3113. Biophysical Journal 108(10) 2424–2426
20. Benachir, T., and M. Lafleur. 1995. Study of vesicle leakage induced by melittin. Biochim. Biophys. Acta. 1235:452–460. 21. Hincha, D. K., and J. H. Crowe. 1996. The lytic activity of the bee venom peptide melittin is strongly reduced by the presence of negatively charged phospholipids or chloroplast galactolipids in the membranes of phosphatidylcholine large unilamellar vesicles. Biochim. Biophys. Acta. 1284:162–170. 22. Ladokhin, A. S., and S. H. White. 2001. ‘Detergent-like’ permeabilization of anionic lipid vesicles by melittin. Biochim. Biophys. Acta. 1514:253–260. 23. Gordon-Grossman, M., H. Zimmermann, ., D. Goldfarb. 2012. Investigation of model membrane disruption mechanism by melittin using pulse electron paramagnetic resonance spectroscopy and cryogenic transmission electron microscopy. J. Phys. Chem. B. 116:179–188. 24. Lee, M.-T., T.-L. Sun, ., H. W. Huang. 2013. Process of inducing pores in membranes by melittin. Proc. Natl. Acad. Sci. USA. 110: 14243–14248. 25. Tosteson, M. T., and D. C. Tosteson. 1981. The sting. Melittin forms channels in lipid bilayers. Biophys. J. 36:109–116. 26. Vogel, H., and F. Ja¨hnig. 1986. The structure of melittin in membranes. Biophys. J. 50:573–582.
