Journal of Biomolecular Structure &Dynamics, ISSN 0739-1102Volume 19, Issue Number 2, (2001)©Adenine Press (2001)

Structure Stability of Lytic Peptides During TheirInteractions With Lipid Bilayers

www.adeninepress.com

Abstract

In this work, molecular dynamics simulations were used to examine the consequences of avariety of analogs of cecropin A on lipid bilayers. Analog sequences were constructed byreplacing either the N- or C-terminal helix with the other helix in native or reverse sequenceorder, by making palindromic peptides based on both the N- and C-terminal helices, and bydeleting the hinge region. The structure of the peptides was monitored throughout the sim-ulation. The hinge region appeared not to assist in maintaining helical structure but help inmotion flexibility. In general, the N-terminal helix of peptides was less stable than the C-terminal one during the interaction with anionic lipid bilayers. Sequences with hydrophobichelices tended to regain helical structure after an initial loss while sequences with amphi-pathic helices were less able to do this. The results suggests that hydrophobic design pep-tides have a high structural stability in an anionic membrane and are the candidates forexperimental investigation.

Introduction

Resistance to chemical antibiotics in bacteria is increasing at a rate which is rais-ing concern about the possibility of an infectious disease epidemic (1,2).Antibacterial peptides, many of which have been identified from the innateimmune systems of insects (3), provide a means by which the effects of chemicalantibiotic resistance might be overcome (4). Cecropins, originally identified fromthe hemolymph of Hyalophora cecropis (5), are one group of peptides that mightprovide a basis for the design of novel antibiotics. In common with many antibac-terial peptides, they are highly cationic and they are active on gram+ and gram-

bacteria but not active on normal eukaryotic cells (6). Furthermore, they have beenshown to be active against cancer cells (7,8), thereby providing another possibletherapeutic role for these peptides.

As the cell-killing actions of antibacterial peptides can be so swift that mechanismsof action can be technically difficult to assess (9), molecular dynamics simulations(MDS) of membranes and their interactions with other molecules provide an alter-native means to gain insight into the process (10). NMR studies on cecropin A (CA)(11) and the cecropin group member, sarcotoxin IA (12), have shown a helix-hinge-helix structural motif. The N-terminal helix is amphipathic in nature, the C-termi-nal helix is largely hydrophobic and a hinge or flexible linker, Ala-Gly-Pro in CA,separates the segments. Using these known structures as a template, simulated anti-bacterial peptides can be constructed with modifications of the sequences or seg-ments and MDS can be used to examine the stability and effect of the simulatedpeptides. Further experimental work can be based on suggestions from the simula-tions.

In this work, a range of analogs of CA were constructed and their interaction with

Hueih Min Chen* and Cheng-Hao Lee+

Institute of BioAgricultural

Sciences, Academia Sinica,

Nankang, Taipei, Taiwan 115+Department of Biochemistry, Hong

Kong University of Science and

Technology, Clear Water Bay, Kowloon,

Hong Kong.

193

* Corresponding authorTel:+886-2-2651-5747; Fax:+886-2-2789-8629; E-mail: robell@gate.sinica.edu.tw

lipid bilayers was simulated. The analogs consisted of replacing the N- or C-ter-minal helix with the helix from the other terminal in either native or reversesequence and by constructing palindromic sequence based on two copies of eitherthe N- or C-terminal helices and the hinge region. In addition, the hinge region wasremoved to assess the effect of this on the peptides. During the simulation nativeCA and the analogs were examined to see how stable their structure was as theyinteracted with the lipid bilayers. From this study, the effect of the different seg-ments and their arrangements could be monitored to see which aspects contributedmost to the structural stability. If a peptide remained stable in the membrane it maybe a good candidate for pore formation or carpet like lysis through a lipid bilayer.

Methods

Design and construction of peptides and lipid bilayerAnalogs of native CA were constructed by interchanging the helical segmentsabout the flexible linker region as is shown below:

amphipathic helix hinge hydrophobic helixCA: NH2-KWKLFKKIEKVGQNIRDGIIK—AGP—AVAVVGQATQIAK-CONH2

head tail head tail

CA1th (tail-to-head biamphipathic helices)NH2-KWKLFKKIEKVGQNIRDGIIK—-AGP——KWKLFKKIEKVGQNIRDGIIK-CONH2

CA1s (symmetric biamphipathic helices)NH2-KWKLFKKIEKVGQNIRDGIIK-AGPPGA-KIIGDRINQGVKEIKKFLKWK-CONH2

CA2th (tail-to-head bihydrophobic helices)NH2-KWKLAVAVVGQATQIAK———-AGP———-AVAVVGQATQIAK-CONH2

CA2s (symmetric bihydrophobic helices)NH2-KAIQTAQGVVAVA——————-AGPPGA——AVAVVGQATQIAK-CONH2

CA3 (lack of GP)NH2-KWKLFKKIEKVGQNIRDGIIK—-A—————AVAVVGQATQIAK-CONH2

These constructed peptides were used for the investigations of the interactions withphospholipid bilayers by MDS. Three dimensional structures were constructed forthe peptides by using the reported structure of CA (11). Residues 5-21 and 25-37of CA were built as ideal α-helices with Φ,Ψ angles of –57o and –47o, respective-ly. An extended conformation was used for residues 1-4 (Φ,Ψ angles of –139o and135o) and residues 22-24 were in the form of a β-turn so that there was an angle of≈70o between the helical segments. The structure of these segments was then trans-ferred to the corresponding segments of the designed analogs. All peptide struc-tures were constructed with energy minimization by the software (Insight II sys-tem). For the lipid bilayer fragment, 15 phosphatidic acid (PA) molecules werebuilt and placed in a linear array to form one leaflet and the second leaflet wasmade by duplicating and then arrays. All molecules were built using the same sys-tem (InsightII).

Docking, energy minimization and molecular dynamicsPartial charges, van der Waal’s radii and energy parameters were taken from theConsistent Valence Force Field (CVFF) as implemented in the Discover system(Biosym/MSI). The potential function of each atom in molecule was investigat-ed such that any inappropriate action can be modified by using FIX option shownin POTENTIAL function menu. Information regarding the atoms’ charges, bonds,torsional angles and out-of-plane interactions were assigned by CVFF parameterlibrary. Peptides were docked onto the lipid bilayer by a suitable molecular orien-

194Chen and Lee

tation so that the N-terminal helix was aligned along the bilayer surface. The vander Waals interaction between peptide and lipid bilayer was properly evaluated bycomputing the intermolecular energy between two systems. Consequently, thecombined peptide and lipid bilayer were energy minimized for 100 steps by usingthe steepest descent and conjugate gradient methods. MDS were conducted at 315K, which is above the gel-to-liquid crystalline phase transition temperature. Thecutoff for non-bonded interactions was 20 Å and a time step of 1 fs was used. A200 ps time period was used for the simulation of the interaction of each peptidewith the bilayer and the state of the system was recorded every ps.

Analysis of molecular dynamics trajectoryThe stability of the peptides was monitored by calculating the backbone dihedralangles and secondary structure for each time point in the molecular dynamics tra-jectory. Secondary structure was determined using the method of Kabsch andSander (13) which was also used to calculate dihedral angles. For each trajectorythe average secondary structure of each peptide was calculated and the consisten-cy of the backbone dihedral angles was measured by determining the circular vari-ance and preferred direction of each angle (14).

Results and Discussion

Perturbation of both peptide and lipid bilayers concomitantly happens during theinteractions. Examples of the interactions of CA, CA2th and CA2s with lipid bilay-ers at different times were shown in Fig. 1. Slight disturbance of the array of lipidbilayers by CA and a loss of secondary structure on N-terminus of CA during inter-action were obtained. Whereas, other hydrophobic designed peptides, CA2th andCA2s, mostly retained their secondary structures during the interaction with lipid

bilayers. For clear view, Fig. 2 show these peptide’s retaining structures when they

195Structure Stability of Lytic

Peptides

Fig. 1: A stereo-views of the interaction of CA, CA2th and CA2s with phosphalipid bilayers by molecular dynamicssimulation at various simulated times: CA at 15 ps (a) and 25 ps (b); CA2th at 20 ps (c) and 130 ps (d); CA2s at 24 ps(e) and 172 ps (f).

interact with the lipid bilayers. CA (see Fig. 2a) begins to release its secondarystructure after 10 ps. In contrast, both CA2th (Fig. 2b) and CA2s (Fig. 2c) main-tain their structures well during the interaction with lipid bilayers. In this paper, we

focus on the investigation of the perturbation of peptide structure when it interactswith lipid bilayers. Dihedral angles (Φ,ψ) of residues were measured as demon-strated by Ramachandran plots during the simulations from 0 ps to 200 ps. Thesedata were used to estimate the peptide residues included or excluded in the helical

196Chen and Lee

Fig. 2: Conformational changes of CA, CA2th andCA2s through the dynamics simulations. (a) CAfrom 0 ps to 50 ps; (b) CA2th from 0 ps to 200 ps;(c) CA2s from 0 ps to 200 ps.

conformation upon simulations. Examples of Ramachandran plots are shown inFig. 3 for CA, CA2th, and CA2s. During the simulation, we observed that manyresidues of CA do not locate in the α-helical zone area (see Figs. 3a-d) While, mostresidues of CA2th locate in the zone of right-handed α-helix (see Figs.3e-h). Someresidues of CA2s are away from the helical zone during the simulation (Figs. 3i-l).

Based on these plots, Fig. 4 shows the results of how many peptide residues areincluded in the helical form as a function of the simulation times. During the sim-ulations, we found that: (a) For CA (open triangle in Fig. 4), both N-terminalamphipathic residues (positions 5 to 21) and C-terminal hydrophobic residues(positions 23 to 28) were unfolded after 70 ps. Only 7 C-terminal residuesremained in the helical form. (b) For CA1th analog (open square in Fig. 4), collapseof secondary structure was even significant as compared with native CA. After 30ps, a linear-coil form of entire peptide in lipid bilayer was shown. (c) For CA1s(open circle in Fig. 4), however, about 50% of residues remain in its helical form.Both termini are widely spread: e.g., a linear-partial helical form exists. Based on(a –c) above, we obtain that the original amphipathic helix is unstably fluctuatedupon approaching headgroup of anionic lipids. Significant inter-hydrogen-bondinteractions between the residues in amphipathic helices and the acyl carbonylgroups in lipids cause structural disintegration of peptides. A symmetrical designfor amphipathic peptides (hinge region acts as a symmetric center) may be helpfulto partially maintain the peptide helical conformation during the interaction withanionic lipid bilayers.

Contrast to the above designs for amphipathic peptides, the simulation results showthat (a) For peptides having two hydrophobic α-helices like CA2th (solid square inFig. 4), its secondary structure during the interaction with lipid bilayers remains,except simulations at 110 ps and 160 ps where about 25 to 35% of CA2th’s sec-ondary structure is lost. (b) For a symmetrical design peptide, CA2s (solid circle

197Structure Stability of Lytic

Peptides

Fig. 3: Ramachandran plots for all residues of CA,CA2th and CA2s during the simulations: CA (a)-(d); CA2th (e) – (h); CA2s (i) – (l) at 50 ps, 100 ps,150 ps and 200 ps, respectively.

in Fig. 4), partial unfold during simulation was observed. The peptide gains backits secondary structure at 190 ps. Based on (a-b) above, we summarize that the fac-tor of symmetrical design for hydrophobic peptides is not outstanding. Both struc-tures of CA2th and CA2s are compact. Their structures being maintained in anion-

ic lipid bilayers by reducing the interaction free energy may be because of the pep-tide residues having lowered accessible surface area to the cationic lipids. We havereported (15) that the differences in the character of antibacterial peptides affect thetypes of membranes they are effective against and their degree of efficacy. Basedon the liposome dye-leakage measurements, as compared with cecropin B1 (CB1)having biamphipathic helical segemtns, cecropin B3 (CB3) having bihydrophobichelical segments is more effective at lysing lipid bilayers, especially when lipo-somes have a higher PA content (15). CA2th having bihydrophobic helices whichis similar to that of CB3 provides the interpretation that the effective lysis of pep-tide roots on its structural stability during the interaction with lipids in bilayers. (c)For CA3 (open down-triangle in Fig. 4), more than 50% of residues remain in itshelical form. However, the mobility of CA3 was retarded in the lipid bilayers. Thisimplies that glycine-proline pair plays a role as a bridge to consolidate two helicalsegments and to supplement them having high flexibility. Suppleness of a peptidemay be one of key factors to effectively execute its function on membrane lysis.

Summary

Membrane-like phospholipid bilayers was a system to have a strong effect withcecropin peptides. The hinge region of cecropins provides the function of motionflexibility. The N-terminal amphipathic α-helices of native CA, CA1th, and CA1swere completely unfolded upon interaction with lipid bilayers. However, a sym-metric sequence design like CA1s enhances the stability of C-terminal amphipath-ic helix in anionic lipid bilayers. Designs having two hydrophobic helical seg-ments like CA2th and CA2s can be a model used to maintain their helical confor-mation in anionic lipid environment. The stable secondary structure of peptidesremains in the lipid bilayers may be a key factor that the membrane lysis can beeffectively done through either forming pore(s) or carpet-like mechanism.Hydrophobic peptide designs provides this stability during the interaction with

198Chen and Lee

Fig. 4: Plots of number of helical residues in pep-tides (CA, CA1th, CA1s, CA2th, CA2s and CA3)as a function of the simulation times. Data collectedper10 ps are obtained by the measurements of dihe-dral angles, Φ and Ψ, in α-helical conformation.

anion-rich lipid bilayer system.

Acknowledgments

The authors thank Dr. David K. Smith for his fruitful discussion. This work waspartly supported by the intramural fund from the Academia Sinica of theRepublic of China.

References and Footnotes

1. Novak, R., Henriques, B., Charpentier, E., Normark, S. and Tuomanen, E. (1999) Nature399, 590-593.

2. Hancock, R. E. W. (1997) Lancet 349, 418-422.3. Boman, H. G. and Broekaert, W. F. (1998) Immunologist 6, 234-238.4. Hancock, R. E. W. and Lehrer, R. (1998) Trends Biotech. 16, 82-88.5. Hultmark, D., Engstrom,A., Bennich, H., Kapur,R., and Boman, H.G. (1982) Eur. J.

Biochem. 127, 207-217.6. Boman, H.G., Wade, D., Boman, I. A., Wahlin, B. and Merrifield, R. B. (1989) FEBS Lett.

259, 103-106.7. Moore, A. J., Devine, D. A. and Bibby, M. C. (1994) Peptide Res. 7 265-269.8. Chen, H. M., Wang, W., Smith, D. and Chan, S. C. (1997) Biochim. Biophy. Acta 1336,

171-179.9. Boman, H. G. (1995) Annu. Rev. Immunol. 13, 61-92.10. Jakobsson, E. (1997) Trends Biochem. Sci. 17, 232-251. Holak, T. A., Engstrom, A.,

Kraulis, P. J., Lindeberg, G., Bennich, H., Jones, T.A., Gronenborn, A. M. and Clore, G. M. (1988) Biochemistry 27, 7620-7629.

12. Iwai, H., Nakajima, Y., Natori, S., Arata, Y. and Shimada, I. (1993) Eur. J. Biochem. 217, 639-644.

13. Kabsch, W. and Sander, C. (1983) FEBS Lett. 155, 179-182.14. MacArthur, M. W. and Thornton, J. M. (1993) Proteins 17, 232-251.15. Wang, W., Smith, D., Moulding, K. and Chen, H. M. (1998) J. Biol. Chem. 273, 27438-

27448.

Date Received: June 13, 2001

Communicated by the Editor Manju Bansal

199Structure Stability of Lytic

Peptides