18
X-ray studies on the interaction of the antimicrobial peptide gramicidin S with microbial lipid extracts: evidence for cubic phase formation Erich Staudegger a , Elmar J. Prenner b , Manfred Kriechbaum a , Ga ¤bor Degovics a , Ruthven N.A.H. Lewis b , Ronald N. McElhaney b , Karl Lohner a ; * a Institut fu «r Biophysik und Ro «ntgenstrukturforschung, O º sterreichische Akademie der Wissenschaften, Steyrergasse 17/VI, A-8010 Graz, Austria b Department of Biochemistry, University of Alberta, Edmonton, AB, Canada T6G 2H7 Received 21 March 2000; received in revised form 29 May 2000; accepted 7 June 2000 Abstract We have investigated the effect of the interaction of the antimicrobial peptide gramicidin S (GS) on the thermotropic phase behavior of model lipid bilayer membranes generated from the total membrane lipids of Acholeplasma laidlawii B and Escherichia coli. The A. laidlawii B membrane lipids consist primarily of neutral glycolipids and anionic phospholipids, while the E. coli inner membrane lipids consist exclusively of zwitterionic and anionic phospholipids. We show that the addition of GS at a lipid-to-peptide molar ratio of 25 strongly promotes the formation of bicontinuous inverted cubic phases in both of these lipid model membranes, predominantly of space group Pn3m. In addition, the presence of GS causes a thinning of the liquid-crystalline bilayer and a reduction in the lattice spacing of the inverted cubic phase which can form in the GS-free membrane lipid extracts at sufficiently high temperatures. This latter finding implies that GS potentiates the formation of an inverted cubic phase by increasing the negative curvature stress in the host lipid bilayer. This effect may be an important aspect of the permeabilization and eventual disruption of the lipid bilayer phase of biological membranes, which appears to be the mechanism by which GS kills bacterial cells and lysis erythrocytes. ß 2000 Elsevier Science B.V. All rights reserved. Keywords : Gramicidin S ; Antimicrobial peptide ; Phospholipid ; Microbial lipid ; Lipid^peptide interaction ; Non-lamellar phase 1. Introduction Gramicidin S (GS) is a cyclic decapeptide of pri- mary structure (cyclo-(Val-Orn-Leu-D-Phe-Pro) 2 ) (see Fig. 1) ¢rst isolated from Bacillus brevis [1]. This peptide shows potent antibiotic activity against a broad spectrum of both Gram-negative and Gram- positive bacteria as well as several pathogenic fungi (see [2^5]). Unfortunately, GS is rather non-speci¢c in its actions and exhibits appreciable hemolytic as well as antimicrobial activity, thus restricting the use of GS as an antibiotic to topical applications (see [2,3]). However, recent work has shown that struc- tural analogs of GS can be designed with markedly reduced hemolytic activity and enhanced antimicro- bial activity, suggesting the possibility that appropri- ate GS derivatives may be used as potent oral or injectable broad-spectrum antibiotics [4,5]. GS has been extensively studied by a wide range of physical techniques (see [2,3]) and its three-dimen- sional structure is shown in Fig. 1. In this minimum 0005-2736 / 00 / $ ^ see front matter ß 2000 Elsevier Science B.V. All rights reserved. PII:S0005-2736(00)00260-1 * Corresponding author. Fax: +43-316-812367; E-mail : [email protected] Biochimica et Biophysica Acta 1468 (2000) 213^230 www.elsevier.com/locate/bba

X-ray studies on the interaction of the antimicrobial …...X-ray studies on the interaction of the antimicrobial peptide gramicidin S with microbial lipid extracts: evidence for cubic

  • Upload
    others

  • View
    7

  • Download
    0

Embed Size (px)

Citation preview

Page 1: X-ray studies on the interaction of the antimicrobial …...X-ray studies on the interaction of the antimicrobial peptide gramicidin S with microbial lipid extracts: evidence for cubic

X-ray studies on the interaction of the antimicrobial peptide gramicidin Swith microbial lipid extracts: evidence for cubic phase formation

Erich Staudegger a, Elmar J. Prenner b, Manfred Kriechbaum a, Gabor Degovics a,Ruthven N.A.H. Lewis b, Ronald N. McElhaney b, Karl Lohner a;*

a Institut fu«r Biophysik und Ro«ntgenstrukturforschung, Oë sterreichische Akademie der Wissenschaften, Steyrergasse 17/VI,A-8010 Graz, Austria

b Department of Biochemistry, University of Alberta, Edmonton, AB, Canada T6G 2H7

Received 21 March 2000; received in revised form 29 May 2000; accepted 7 June 2000

Abstract

We have investigated the effect of the interaction of the antimicrobial peptide gramicidin S (GS) on the thermotropic phasebehavior of model lipid bilayer membranes generated from the total membrane lipids of Acholeplasma laidlawii B andEscherichia coli. The A. laidlawii B membrane lipids consist primarily of neutral glycolipids and anionic phospholipids, whilethe E. coli inner membrane lipids consist exclusively of zwitterionic and anionic phospholipids. We show that the addition ofGS at a lipid-to-peptide molar ratio of 25 strongly promotes the formation of bicontinuous inverted cubic phases in both ofthese lipid model membranes, predominantly of space group Pn3m. In addition, the presence of GS causes a thinning of theliquid-crystalline bilayer and a reduction in the lattice spacing of the inverted cubic phase which can form in the GS-freemembrane lipid extracts at sufficiently high temperatures. This latter finding implies that GS potentiates the formation of aninverted cubic phase by increasing the negative curvature stress in the host lipid bilayer. This effect may be an importantaspect of the permeabilization and eventual disruption of the lipid bilayer phase of biological membranes, which appears tobe the mechanism by which GS kills bacterial cells and lysis erythrocytes. ß 2000 Elsevier Science B.V. All rights reserved.

Keywords: Gramicidin S; Antimicrobial peptide; Phospholipid; Microbial lipid; Lipid^peptide interaction; Non-lamellar phase

1. Introduction

Gramicidin S (GS) is a cyclic decapeptide of pri-mary structure (cyclo-(Val-Orn-Leu-D-Phe-Pro)2)(see Fig. 1) ¢rst isolated from Bacillus brevis [1].This peptide shows potent antibiotic activity againsta broad spectrum of both Gram-negative and Gram-positive bacteria as well as several pathogenic fungi

(see [2^5]). Unfortunately, GS is rather non-speci¢cin its actions and exhibits appreciable hemolytic aswell as antimicrobial activity, thus restricting the useof GS as an antibiotic to topical applications (see[2,3]). However, recent work has shown that struc-tural analogs of GS can be designed with markedlyreduced hemolytic activity and enhanced antimicro-bial activity, suggesting the possibility that appropri-ate GS derivatives may be used as potent oral orinjectable broad-spectrum antibiotics [4,5].

GS has been extensively studied by a wide range ofphysical techniques (see [2,3]) and its three-dimen-sional structure is shown in Fig. 1. In this minimum

0005-2736 / 00 / $ ^ see front matter ß 2000 Elsevier Science B.V. All rights reserved.PII: S 0 0 0 5 - 2 7 3 6 ( 0 0 ) 0 0 2 6 0 - 1

* Corresponding author. Fax: +43-316-812367;E-mail : [email protected]

BBAMEM 77920 22-9-00

Biochimica et Biophysica Acta 1468 (2000) 213^230www.elsevier.com/locate/bba

Page 2: X-ray studies on the interaction of the antimicrobial …...X-ray studies on the interaction of the antimicrobial peptide gramicidin S with microbial lipid extracts: evidence for cubic

energy conformation, the two tripeptide sequencesVal-Orn-Leu form an antiparallel L-sheet terminatedon each side by a type IIP L-turn formed by the twoD-Phe-Pro sequences. Four intramolecular hydrogenbonds, involving the amide protons and carbonylgroups of the two Leu and two Val residues, stabilizethis rather rigid structure. Note that the GS moleculeis amphiphilic, with the two polar and positivelycharged Orn side chains and the two D-Phe ringsprojecting from one side of this molecule, and thefour hydrophobic Leu and Val side chain projectingfrom the other side. Moreover, the amphiphilic na-ture of GS is crucial for the manifestation of itsantimicrobial activity [6,7]. A number of studieshave shown that this conformation of the GS mole-cule is maintained in water, in protic and aproticorganic solvents of widely varying polarity, and indetergent micelles and phospholipid bilayers, even

at high temperatures and in the presence of agentswhich often alter protein conformation (see [2,3]).

Considerable evidence exists that the principal tar-get of GS is the lipid bilayer of bacterial or erythro-cyte membranes (see [2,3,8]). Therefore, a number ofbiophysical studies have been carried out on the ef-fect of GS on the structure and physical properties ofphospholipid bilayer model and biological mem-branes (for a review, see [8]). There is a general con-sensus that GS partitions strongly into the biologi-cally relevant liquid-crystalline (LK) phase of lipidbilayers and is located primarily near the glycerolbackbone region of the phospholipid molecule,where it interacts with the lipid polar headgroupsand the upper regions of the lipid hydrocarbonchains. The presence of GS seems to further disorderliquid-crystalline bilayers and, at su¤ciently highconcentrations, may destroy the lipid bilayer struc-ture. The presence of GS at lower concentrationsalso increases the permeability of model and biolog-ical membranes and at higher concentration causesmembrane lysis and solubilization. Finally, GS seemsto interact more strongly with anionic than withzwitterionic or uncharged phospho- or glycolipid bi-layers and the interaction of GS with these modelmembranes is reduced by the presence of cholesterol[8].

We have recently shown that GS can also alter thelamellar/non-lamellar phase preferences of somephospholipid and glycolipid molecular species [9].Speci¢cally, we examined the interactions of GS (lip-id/peptide ratio 25:1) with a variety of single-compo-nent lipid bilayers, and with membrane polar lipidextracts of Acholeplasma laidlawii B and Escherichiacoli, by 31P-NMR spectroscopy and X-ray di¡rac-tion. For mixtures of GS with lipids such as phos-phatidylcholine (PC), phosphatidylserine (PS), cardi-olipin, and sphingomyelin, axially symmetric 31P-NMR lamellar phase powder patterns are observedthroughout the entire temperature range examined(0^90³C). However, with mixtures of GS with eitherphosphatidylethanolamine (PE), phosphatidylglycer-ol (PG), or a non-lamellar phase-forming phosphati-dylcholine, axially symmetric 31P-NMR powder pat-terns are also observed at low temperatures.However, at high temperatures, an isotropic compo-nent is observed in their 31P-NMR spectra, and therelative intensity of this component increases signi¢-

Fig. 1. The structure and conformation of GS. The upper panelis a view of the GS molecule perpendicular to the plane of thering, illustrating the peptide backbone structure and the posi-tions of the hydrogen bonds in the antiparallel L-sheet region.The lower panel is a view of the GS molecule in the plane ofthe ring, indicating the disposition in space of the hydrophobicVal and Leu residues (top) and the basic Orn residues (bottom)relative to the peptide ring.

BBAMEM 77920 22-9-00

E. Staudegger et al. / Biochimica et Biophysica Acta 1468 (2000) 213^230214

Page 3: X-ray studies on the interaction of the antimicrobial …...X-ray studies on the interaction of the antimicrobial peptide gramicidin S with microbial lipid extracts: evidence for cubic

cantly with temperature and with GS concentration.Once formed at high temperatures, this isotropiccomponent exhibits a marked cooling hysteresisand in most cases disappears only when the sampleis recooled to temperatures well below the lipid hy-drocarbon chain-melting phase transition tempera-ture. We also showed that GS induces the formationof isotropic components in the 31P-NMR spectra ofheterogeneous lipid mixtures such as occur in A. laid-lawii B and E. coli membranes. These observationssuggest that GS induces the formation of cubic (Q)or other three-dimensionally ordered inverted non-lamellar phases when it interacts with some typesof lipid bilayers, a suggestion strongly supported byour X-ray di¡raction studies. Moreover, we foundthat the capacity of GS to induce the formation ofsuch phases increases with the intrinsic non-lamellarphase-preferring tendencies of the lipids with which itinteracts.

The appearance of an isotropic signal in the 31P-NMR spectrum of aqueous phospholipid dispersionsis observed for fast tumbling lipid aggregates, likesmall unilamellar vesicles and micelles, or for Q orother three dimensionally ordered inverted non-la-mellar phases. As 31P-NMR spectroscopy cannot dis-tinguish between these putative lipid^peptide aggre-gates due to their fast tumbling in respect to the 31P-NMR time scale (e.g. [10,11]), these lipid aggregateswere further studied by X-ray techniques. Prenner etal. [9] found that isotropic 31P-NMR spectra are ob-served at high temperatures. Moreover, a mixture ofdimyristoylphosphatidylethanolamine^GS, exhibitedvery weak re£ections in addition to strong lamellarBragg re£ections at temperatures higher than 85³C.The probable existence of an inverted Q phase in theform of two gyroid lattices with a basis of 13.1 and15.3 nm were proposed from these supplementaryre£ections, which could not be indexed on a lamellarphase. Thus in order to obtain information on thestructural basis of the isotropic component of lipidextracts from the plasma membranes of E. coli andA. laidlawii B, we performed a much more extensiveX-ray di¡raction study. We clearly show here thatthe interaction of GS with these lipid extracts fromnatural bacterial membranes markedly enhancestheir tendencies to form non-lamellar isotropicphases, which could be identi¢ed as a bicontinuouscubic lipid phase.

2. Materials and methods

2.1. Peptides and lipids

The cyclic peptide gramicidin S (HCl salt) wasobtained from Sigma (St. Louis, MO) and puri¢edby the HPLC methodology described by Kondejew-ski et al. [4]. The A. laidlawii B membrane lipids usedwere the polar lipid extracts of cells grown in avidin-containing media supplemented with an equimolarmixture of palmitic and oleic acid. The polar lipidextracts were obtained by silicic acid chromatogra-phy of the total membrane lipid extracts [12]. E. colitotal and polar lipid extracts were obtained fromAvanti Polar Lipids (Alabaster, AL).

2.2. Sample preparation

Lipid^peptide mixtures (lipid^peptide molar ratio,R = 25) were prepared in glass tubes by codissolvingthe lipid extracts in chloroform/methanol (2:1 v/v)with appropriate amounts of GS stock solutions(chloroform/methanol 1:2 v/v) and subsequent evap-oration of the solvent under a stream of nitrogen.After removal of any residual traces of solvent invacuo overnight, the dried lipid or lipid^peptide ¢lmswere hydrated by vigorous vortexing at 40³C with abu¡er consisting of 50 mM Tris, 100 mM NaCl,5 mM EDTA and 1 mM sodium azide (pH 7.4).The lipid concentration of the X-ray samples was5^10 mg/100 Wl.

2.3. Small- and wide-angle X-ray (SWAX)experiments

SWAX di¡raction experiments were performed ona modi¢ed Kratky compact camera (HECUS-MBraun-Graz, Graz, Austria), which allows simulta-neous recording of di¡raction data in both the small-and wide-angle region, as described elsewhere [13].Ni-¢ltered CuKK-radiation (V= 0.154 nm) originatingfrom a Philips X-ray generator with a Cu-anode op-erating at 50 kV and 40 mA was used. The camerawas equipped with a Peltier-controlled, variable-tem-perature cuvette (temperature precision = þ 0.1³C)and linear, one-dimensional, position-sensitive detec-tors OED 50-M (MBraun, Garching, Germany).Calibration in the small-angle region was performed

BBAMEM 77920 22-9-00

E. Staudegger et al. / Biochimica et Biophysica Acta 1468 (2000) 213^230 215

Page 4: X-ray studies on the interaction of the antimicrobial …...X-ray studies on the interaction of the antimicrobial peptide gramicidin S with microbial lipid extracts: evidence for cubic

with silver stearate and in the wide-angle region withp-bromo-benzoic acid standards, respectively. Tem-perature control and data acquisition was achievedby programmable temperature control equipment(MTC-2.0, HECUS-MBraun-Graz, Graz, Austria).A temperature cycle consisted of a ramp from 25to 90³C and then back to 25³C in steps of 5³C.The samples were equilibrated at the respective tem-peratures for 10 min before sampling the X-ray dif-fractograms. Multiplexed exposure times of 2000 sfor the small-angle and 1000 s for the wide-angleregion were chosen.

Synchrotrone X-ray measurements were performedat the high-£ux Austrian SAXS beamline (station5.2L) of the 2 GeV electron storage ring ELETTRA,Trieste [14] using the 8-keV X-rays which corre-sponds to a wavelength of 0.1542 nm. The sampleswere placed in Mark glass capillaries with a diameterof 1 mm, sealed and mounted in a rotating samplestage. The resolution range was from 20 to 2 nm andexposure times of 120 s were chosen. Di¡ractograms

were recorded by means of a linear position-sensitivedetector. While the rectangular slit geometry, de¢n-ing the pro¢le of the beam, of the SWAX camera inthe laboratory yields X-ray di¡raction pattern con-voluted with the beam pro¢le, this e¡ect is negligibleat the SAXS beamline at ELETTRA.

SAX-data arising from particle scattering were fur-ther analyzed by Indirect Fourier Transformation. Inthe case of £at particles, such as extended lamellarstructures, where the axial thickness of the lamella ismuch smaller than its surface area, the one-dimen-sional distance distribution function pt(r) can be de-rived from the scattering data. This represents theautocorrelation function of the electron density nor-mal to the bilayer plane, which yields information onthe cross-bilayer distance between the electron-densephosphate groups (dPÿP), i.e. of the phosphategroups located in the opposing monolayers of thebilayer. Thereby, dPÿP can be determined from theposition of the outer maximum of the pt(r)-function(see also Fig. 4a). Brie£y, pt(r) were computed after

Fig. 2. Small- and wide-angle (inset) X-ray di¡ractograms of an aqueous dispersion of A. laidlawii B membrane polar lipid extract re-corded at 5, 10, 15, 20, 25 and 30³C (bottom to top); s = 1/d = 2sin(3)/V, where V is the wavelength of the X-ray beam and 23 thescattering angle. Di¡ractograms are displayed vertically for better visualization.

BBAMEM 77920 22-9-00

E. Staudegger et al. / Biochimica et Biophysica Acta 1468 (2000) 213^230216

Page 5: X-ray studies on the interaction of the antimicrobial …...X-ray studies on the interaction of the antimicrobial peptide gramicidin S with microbial lipid extracts: evidence for cubic

background subtraction and normalization of the re-spective bu¡er blank curve and data-point reductionof the di¡erence curve by a funneling routine. Fur-thermore, the data were corrected for instrumentalbroadening by using the program ITP of Glatter[15] to yield desmeared data, which were interpretedin real space in terms of their pair distance distribu-tion function [16,17].

3. Results

A. laidlawii B membrane lipid dispersions. Repre-sentative SWAX di¡raction patterns for aqueous dis-persions of the membrane polar lipids derived fromA. laidlawii B are shown in Fig. 2. At temperaturesbetween 5 and 15³C, the wide-angle di¡ractogramsexhibit a single symmetric peak centered at 0.42 nm(s = 2.38 nm31), whose scattering intensity decreaseswith increasing temperature, vanishing around 20³C

(inset Fig. 2). Such a di¡raction pattern is character-istic of hexagonally packed all-trans lipid acyl chainsoriented normal to the bilayer plane as observed forthe lamellar gel (LL) phase [18]. At temperaturesabove 20³C, only a di¡use pattern is detected, whichdemonstrates that all the lipid hydrocarbon chainsare in a melted state, characteristic of the lamellarliquid-crystalline (LK) phase. This LL to LK phasetransition is also re£ected in the small-angle X-rayexperiments (Fig. 2). Below 20³C, these di¡racto-grams show two clearly resolvable re£ections super-imposed upon a broad scattering background, whileat 25³C, the di¡raction pattern is characterized bybroad side maxima and minima. Moreover, a strongscattering intensity in the innermost part, i.e. at verylow angles (s6 0.05 nm31), typical for particle scat-tering, is observed in this temperature range. Thislatter feature suggests the existence of unilamellarvesicles, most likely resulting from the high contentof negatively charged lipids in this microbial lipid

Fig. 3. Small- and wide-angle (inset) X-ray di¡ractograms of an aqueous dispersion of a mixture of A. laidlawii B membrane polar lip-id extract and GS at a lipid-to-peptide molar ratio of 25:1 recorded at 5, 10, 15, 20, 25 and 30³C (bottom to top); s = 1/d = 2sin(3)/V,where V is the wavelength of the X-ray beam and 23 the scattering angle. Di¡ractograms are displayed vertically for better visualiza-tion.

BBAMEM 77920 22-9-00

E. Staudegger et al. / Biochimica et Biophysica Acta 1468 (2000) 213^230 217

Page 6: X-ray studies on the interaction of the antimicrobial …...X-ray studies on the interaction of the antimicrobial peptide gramicidin S with microbial lipid extracts: evidence for cubic

extract. On the other hand, the presence of two re-£ections may be attributed to ¢rst- and second-orderBragg re£ections of a lamellar lattice of 11.3 nm,which suggests the existence of multilamellar vesiclesas well. However, the broad re£ections are indicativethat these liposomes have only a small number oflamellae [19] and/or exhibit a strong distortion ofthe long range order of the lattice [20]. In this tem-perature range, generally similar SWAX di¡ractionpatterns were obtained for mixtures of the A. laidla-wii B membrane polar lipids with GS (R = 25) (Fig.3). However, a closer inspection of the wide-angledi¡ractograms (inset Fig. 3) reveals a slightly smaller(V15%) integral peak intensity below 20³C as com-

pared to the integral peak intensities of the polarlipids without GS. However, the peak position isnot a¡ected. This is indicative of partially perturbedhydrocarbon side chains, suggesting an interaction ofthe peptide with some fraction of the lipids. Addi-tionally, in the small-angle region of the GS-contain-ing sample (Fig. 3), a third Bragg re£ection is ob-served at 7.1 nm (s = 0.14 nm31), residing betweenthe two re£ections which are characteristic for theA. laidlawii B lipid extract, also indicating an inter-action of the peptide with this lipid extract below thelamellar gel to liquid-crystalline phase transition re-gion. The limited number of Bragg re£ections did notallow any assignment of a non-lamellar phase. Onthe other hand, these three Bragg re£ections cannotbe attributed to one single lamellar phase either. Aswe are dealing with a complex lipid mixture here, theadditional re£ection could arise from a stronger in-teraction of GS with the negatively charged lipidspresent [8], which may give rise to a new lamellarstructure.

Information on the cross-bilayer distance betweenthe electron dense phosphate groups of the opposinglipid lea£ets was derived from the one-dimensionaldistance distribution function pt(r) as described inSection 2. Such functions are shown in Fig. 4 forlipid extracts of A. laidlawii B in the absence (panela) and in the presence (panel b) of GS at 25³C. We¢nd that in the presence of GS, dPÿP is reduced sig-ni¢cantly by about 0.25 nm in the presence of thepeptide, i.e. from 4.1 to 3.85 nm. Moreover, in thepresence of GS, the outer maximum exhibits anasymmetry towards larger distances as well as ashoulder around 5.3 nm, which may be due to pep-tide being bound to the membrane surface.

Heating aqueous dispersions of A. laidlawii Bmembrane polar lipids up to 90³C reveals a secondphase transition centered around 70³C (Fig. 5a).Above 35³C the scattering intensity in the innermostpart of the di¡ractogram (s6 0.05 nm31), as well asthe intensity of the broad side maxima, decreasesdrastically. Concomitantly, the formation of Braggre£ections is observed being superimposed upon theside maximum around 7 nm (s = 0.143 nm). Theseadditional peaks increase in intensity with increasingtemperature. This higher temperature phase shows astrong hysteresis upon subsequent cooling to 25³Cand is stable upon repeated heating and cooling

Fig. 4. Computed one-dimensional thickness distance distribu-tion function pt(r) of A. laidlawii B membrane polar lipid ex-tract (a) and mixtures of these lipids with GS at a lipid-to-pep-tide molar ratio of 25:1 (b) derived from experimentalscattering data at 25³C. Arrow in the upper panel indicatesdPÿP ; for details see Section 2.

BBAMEM 77920 22-9-00

E. Staudegger et al. / Biochimica et Biophysica Acta 1468 (2000) 213^230218

Page 7: X-ray studies on the interaction of the antimicrobial …...X-ray studies on the interaction of the antimicrobial peptide gramicidin S with microbial lipid extracts: evidence for cubic

cycles, a common feature of three-dimensional iso-tropic phases. This new supramolecular phase seemsto coexist with the lamellar phase under these exper-imental conditions, as evidenced by the broad back-

ground scattering and by the weak, but still signi¢-cant, side maxima at higher angles arising from theparticle scattering. Our previous 31P-NMR data alsoindicated a coexistence of an isotropic phase with a

Fig. 5. Small-angle X-ray di¡ractograms of an aqueous dispersion of A. laidlawii B membrane polar lipid extract (a) and a mixture ofthese lipids with GS at a lipid-to-peptide molar ratio of 25:1 (b) recorded from the ¢rst heating and cooling cycle (frames from topto bottom: 25, 35, 45, 60, 70, 75, 80, 85, 90, 80, 70, 60, 50, 40, 25³C; selected temperatures are indicated in the panels) ; s =1/d = 2sin(3)/V, where V is the wavelength of the X-ray beam and 23 the scattering angle.

BBAMEM 77920 22-9-00

E. Staudegger et al. / Biochimica et Biophysica Acta 1468 (2000) 213^230 219

Page 8: X-ray studies on the interaction of the antimicrobial …...X-ray studies on the interaction of the antimicrobial peptide gramicidin S with microbial lipid extracts: evidence for cubic

lamellar one in this same range of temperatures [9].Sometimes, the resolution of the re£ections arisingfrom such phases in the presence of the intrinsicallystronger scattering from the lamellar phase can bedi¤cult in a mixed phase system, even when the for-mer is the dominant component in the mixture.However, we were able to unequivocally assign theexistence of two cubic phases (Fig. 6). The predom-inant one consists of up to eight orders of di¡ractionspace in the ratio of k2:k3:k4:k6:k8:k9:k10:k11which were indexed as (110), (111), (200), (211),(220), (221), (310), and (311) re£ections on a three-dimensional cubic phase of space group Pn3m. Thereciprocal spacing (s) of cubic phases is related to thelattice spacing (a) by

s �hkl� � �h2 � k2 � l2�1=2=a �1�where h, k, and l are the Miller indices [21]. There-fore, the lattice spacing can be calculated from thereciprocal gradient of the plot s vs. (h2+k2+l2)1=2.From this plot a lattice spacing of 13.2 nm was cal-

culated for the Pn3m phase (see inset Fig. 6). Thisvalue is in the same range as previously reportedones (12.5^14 nm) for lipids adopting the Pn3m orPn3 space group [22]. A second cubic phase can bedetected from additional weak re£ections which donot belong to the Pn3m space group. These re£ec-tions are in ratios of k2:k4:k6:k8:k10:k12:k14, whichwere indexed as (110), (200), (211), (220), (310),(222), and (321) re£ections on a three-dimensionalcubic phase of space group Im3m (inset Fig. 6). Alattice spacing of 18.9 nm was calculated. It shouldbe noted that an assignment of other cubic phasesfailed. Coexisting Pn3m and Im3m phases were alsoreported from aqueous dispersions of dioleoylphos-phatidylethanolamine (DOPE) which were cycled1400 times through the LK-inverted hexagonal (HII)phase transition range [23]. Moreover, for dielaidoylPE a cooperative Im3mCPn3m transition was ob-served at elevated temperatures and transforms thelatter phase into a mixture of coexisting Pn3m andIm3m phases [24]. Our results are also in accordance

Fig. 6. Phase assignment of an aqueous dispersion of A. laidlawii B membrane polar lipid extract. Position of hkl-re£ections corre-sponding to a cubic phase of space group Pn3m (solid line) and of space group Im3m (broken line), respectively, are indicated in thedi¡ractogram. The inset shows the indexing of these two cubic lattices from the plot s vs. (h2+k2+l2)1=2 ; open circles represent re£ec-tions of space group Pn3m, while open squares represent re£ections of space group Im3m. Miller indices (hkl) are indicated in thepanel. Data were recorded at 25³C after the ¢rst heating/cooling cycle up to 90³C.

BBAMEM 77920 22-9-00

E. Staudegger et al. / Biochimica et Biophysica Acta 1468 (2000) 213^230220

Page 9: X-ray studies on the interaction of the antimicrobial …...X-ray studies on the interaction of the antimicrobial peptide gramicidin S with microbial lipid extracts: evidence for cubic

with observations by Sen et al. [25], who reported thepresence of Pn3m or Pn3 space groups for syntheticmonoglucosyldiacylglycerol. Moreover, Lindblomand coworkers [26] recently characterized the phasebehavior of single lipid components of A. laidlawii

(strain A) by NMR and X-ray di¡raction. The stron-gest non-lamellar propensity was reported for mono-glucosyldiacylglycerol and a monoacylmonoglucosyl-diacylglycerol derivative, for which a Q phase of thespace group Ia3d was assigned at 20 wt% of water in

Fig. 7. Small-angle X-ray di¡ractograms of an aqueous dispersion of E. coli membrane polar lipid extract (a) and a mixture of theselipids with GS at a lipid-to-peptide molar ratio of 25:1 (b) recorded from the ¢rst heating and cooling cycle (frames from top to bot-tom: 25, 35, 45, 60, 70, 75, 80, 85, 90, 80, 70, 60, 50, 40, 25³C; selected temperatures are indicated in the panels); s, see legend toFig. 5.

BBAMEM 77920 22-9-00

E. Staudegger et al. / Biochimica et Biophysica Acta 1468 (2000) 213^230 221

Page 10: X-ray studies on the interaction of the antimicrobial …...X-ray studies on the interaction of the antimicrobial peptide gramicidin S with microbial lipid extracts: evidence for cubic

the sample. These ¢ndings are not in contradiction toour present results, but can be explained by the dif-ferent levels of hydration studied, as Lindblom andRilfors [27] presented experimental and theoreticalevidence that the sequence of formation of di¡er-ent Q phases with increasing water content isIa3dCPn3mCIm3m.

In the presence of GS (R = 25) the onset of Qphase formation is lowered by about 10³C as indi-cated by the ¢rst appearance of Bragg peaks (Fig.5b). More interestingly, the Bragg re£ections arisingfrom the isotropic phase are less resolved as com-pared to the pure A. laidlawii B membrane polarlipids. In addition, the di¡ractograms for the lipid^peptide mixture are characterized by a much lowerintegral scattering intensity, as usually found for lessordered systems. An unequivocal assignment of theBragg re£ections in respect of a Q phase is moredi¤cult for this microbial lipid extract in the pres-ence of GS owing to a lack of a su¤cient number ofre£ections allowing unambiguous assignment. None-theless, a Q phase of the same symmetry as found forthe peptide-free lipid can be deduced from the dif-fractogram obtained at 25³C. The lattice spacing of12 nm is about 1 nm smaller as compared to the purelipid mixture (Table 1).

These X-ray di¡raction data support the ¢ndingsof our earlier 31P-NMR experiments, where the over-all axially symmetric powder pattern was shown toremain the dominant feature of the 31P-NMR spec-trum upon heating the A. laidlawii B polar lipid ex-tract to temperatures near 90³C [9]. However, a mi-nor sharp component centered near 2 ppm down¢eldwas also observed, suggesting that A. laidlawii B po-lar lipid extracts form non-lamellar phases at veryhigh temperatures. The 31P-NMR spectra of the mix-ture of A. laidlawii polar lipids with GS di¡ered sig-ni¢cantly from those exhibited by the lipid extractalone at all temperatures above the LL/LK phasetransition temperature, which is in accordance withthe X-ray data. Speci¢cally, the up¢eld componentsof the 31P-NMR powder patterns were considerably

Fig. 8. Phase assignment of an aqueous dispersion of E. colimembrane polar lipid extract in the absence and presence ofGS at a lipid-to-peptide molar ratio of 25:1. Re£ex positionsfor two cubic lattices vs. (h2+k2+l2)1=2 are shown; circles repre-sent re£ections of space group Pn3m, while squares representre£ections of space group Im3m (closed symbols with and opensymbols without GS). Miller indices (hkl) are indicated in thepanel. Re£ections were taken from di¡ractograms recorded at25³C after the ¢rst heating/cooling cycle up to 90³C.

Table 1Lattice spacing of the Pn3m cubic phase for microbial lipid extracts at 25³C after the ¢rst heating/cooling cyclea

Microbial lipid extract Gramicidin S (R = 25)b Lattice spacing (nm)

A. laidlawii B, 3 13.2polar + 12.0c

E. coli, 3 15.3polar + 14.3E. coli 3 14.9total + 13.4aA minor fraction of lipids was found to adopt a cubic phase of space group Im3m characterized by an enlarged lattice spacing by afactor of 1.29 þ 0.1 as compared to the cubic phase of space group Pn3m.bR, lipid-to-peptide molar ratio.cNo unequivocal phase and lattice assignment.

BBAMEM 77920 22-9-00

E. Staudegger et al. / Biochimica et Biophysica Acta 1468 (2000) 213^230222

Page 11: X-ray studies on the interaction of the antimicrobial …...X-ray studies on the interaction of the antimicrobial peptide gramicidin S with microbial lipid extracts: evidence for cubic

sharper than observed in the absence of the peptide,and the relative intensities of the down¢eld compo-nents of the powder pattern were smaller than isobserved with the polar lipid extract alone. Finally,at temperatures near 90³C, the major feature of the31P-NMR spectrum of this lipid^GS mixture was asharp peak centered near 2 ppm, suggesting that theinteraction of GS with this particular mixture of lip-ids potentiates the formation of non-lamellar phases.

3.1. E. coli membrane lipid dispersions

Wide-angle X-ray di¡raction patterns of polarmembrane lipid extracts from E. coli are character-ized by a di¡use re£ection (data not shown), demon-strating that the hydrocarbon chains are in a meltedstate in the temperature range investigated (5^90³C).The small-angle X-ray data are typical for particlescattering as deduced from the broad intensive side

maxima at 6.9 nm (s = 0.14 nm31) and very weak sidemaxima at 2.1 nm (0.48 nm31) and 1.4 nm (0.71nm31) (25³C). Above 70³C, the intensity of the sidemaxima decreases strongly and the onset of Braggre£ections is detectable between 85 and 90³C, whichare clearly resolved upon cooling (Fig. 7a). Indexingof these Bragg re£ections again revealed that thenewly formed lipid structures belong to the Pn3mand Im3m space group (Fig. 8) characterized by lat-tice spacings of 15.3 (Table 1) and 18.3 nm, respec-tively. Interestingly, the (110) re£ection of both Qphases exhibited similar peak intensities suggestingthat in the case of E. coli polar lipid extract, a largerfraction of Im3m as compared to A. laidlawii B lipidextract does occur. Although the respective Braggre£ections increase in intensity upon cooling, broadside maxima and minima are still observed, againindicating the coexistence of a Q phase with a lamel-lar lipid phase. This observation is in accordance

Fig. 9. Small-angle X-ray di¡ractogram of an aqueous dispersion of E. coli membrane total lipid extract recorded at 25³C after heat-ing up the sample to 90³C; s see legend to Fig. 5. Data taken at the SAXS-beamline at the Synchrotrone ELETTRA, Trieste. Posi-tion of hkl-re£ections corresponding to a cubic phase of space group Pn3m (solid line) and of space group Im3m (broken line), respec-tively, are indicated in the di¡ractogram. The inset shows the indexing of two cubic lattices from the plot s vs. (h2+k2+l2)1=2 ; opencircles represent re£ections of space group Pn3m, while open squares represent re£ections of space group Im3m. Miller indices (hkl)are indicated in the panel.

BBAMEM 77920 22-9-00

E. Staudegger et al. / Biochimica et Biophysica Acta 1468 (2000) 213^230 223

Page 12: X-ray studies on the interaction of the antimicrobial …...X-ray studies on the interaction of the antimicrobial peptide gramicidin S with microbial lipid extracts: evidence for cubic

with our 31P-NMR data [9], showing that above 65^75³C E. coli lipids exhibit axially symmetric 31P-NMR powder patterns typical of the lamellar liq-uid-crystalline phase coexisting with an isotropiccomponent, which indicates non-lamellar phase for-mation.

Similar di¡raction patterns were obtained for mix-tures of E. coli membrane polar lipids and GS(R = 25) (Fig. 7b). However, the peptide decreasesstrongly the onset of non-lamellar phase formationby about 25³C. Phase assignment revealed the pres-ence of the same two cubic phases as found for thepure E. coli polar lipid extract (Fig. 8). However, theIm3m phase represents only a minor fraction. In ad-dition, a signi¢cant reduction of the Pn3m cubic lat-tice is observed in the presence of the peptide asdeduced from the smaller lattice spacing parameterof 14.3 nm at 25³C after the ¢rst temperature cycling(Fig. 8, Table 1), which was less pronounced for theIm3m phase (lattice spacing of 18.0 nm). This is

again in agreement with our recent 31P-NMR data,which showed a comparable decrease of the onsettemperature of the non-lamellar phase formation[9]. Once formed at high temperatures, these struc-tures form a stable and transparent gel inside theX-ray capillary. Long-lived Q phases were also re-ported for phosphatidylethanolamines when theywere cycled through temperatures bracketing LK/HII

phase transition range [28,29]. However, the seem-ingly stable Q phases formed under such conditionsare believed to be metastable structures which havebeen kinetically trapped by rapid cycling through theLK/HII phase transition [28]. One-dimensional dis-tance distribution functions pt(r) were also deter-mined for these samples to gain information on thebilayer structure. Again, as already observed forA. laidlawii B, the cross-bilayer distance betweenthe phosphate groups of the opposing monolayersis diminished in the presence of the peptide from3.9 to 3.7 nm.

Fig. 10. Small-angle X-ray di¡ractogram of a mixture of E. coli membrane total lipid extract of these lipids with GS at a lipid-to-pep-tide molar ratio of 25:1 recorded at 25³C after heating up the sample to 90³C; s, see legend to Fig. 5. Data taken at the SAXS-beam-line at the Synchrotrone ELETTRA, Trieste. Position of hkl-re£ections corresponding to a cubic phase of space group Pn3m (solidline) and of space group Im3m (broken line), respectively, are indicated in the di¡ractogram. The inset shows the indexing of two cu-bic lattices from the plot s vs. (h2+k2+l2)1=2 ; closed circles represent re£ections of space group Pn3m, while closed squares representre£ections of space group Im3m. Miller indices (hkl) are indicated in the panel.

BBAMEM 77920 22-9-00

E. Staudegger et al. / Biochimica et Biophysica Acta 1468 (2000) 213^230224

Page 13: X-ray studies on the interaction of the antimicrobial …...X-ray studies on the interaction of the antimicrobial peptide gramicidin S with microbial lipid extracts: evidence for cubic

Finally, the SAX pattern of aqueous dispersions ofE. coli membrane total lipids in the presence andabsence of GS closely resembled that from the polarlipid extract. For these samples, both conventionaland Synchrotrone X-ray radiation experiments werecarried out yielding identical results. In contrast to E.coli membrane polar lipids, the onset of Q phaseformation was already found between 40 and 45³C,which was further reduced in the presence of thepeptide. In the latter case, Bragg re£ections were al-ready observed at 25³C. Phase assignment is shownfor the SAX di¡raction patterns recorded from Syn-chrotrone experiments at 25³C after the ¢rst heating/cooling cycle using rotating X-ray capillaries (Figs. 9and 10). The relatively large number of Bragg re£ec-tions allowed an unequivocal phase assignment (insetFigs. 9 and 10). Again it was found that the latticespacing of the Pn3m phase decreased in the presenceof GS from 14.9 to 13.4 nm (Table 1). Moreover,from these di¡ractograms it is evident that the coex-isting Im3m phase only represents a minor fraction.Lattice spacings of 19.5 and 17.2 nm were calculatedfrom the reciprocal gradient of s vs. (h2+k2+l2)1=2 inthe absence and presence of the peptide (inset inFigs. 9 and 10). Our results are in excellent agree-ment with data from Morein et al. [30] who studiedlipid extracts of the inner membrane as well as anextract of the total lipids from both the inner and theouter membrane of E. coli and characterized theirphase behavior by NMR spectroscopy. Under vary-ing conditions, they found both HII and isotropicphase formation. Furthermore, they also investigatedby X-ray di¡raction the total membrane lipid ex-tracts of E. coli K12 grown at 17³C and found thatthe isotropic phase belonged to the Pn3m spacegroup with a lattice spacing of 14.8 nm.

4. Discussion

Although the maintenance of stable lamellar struc-tures is essential to normal membrane function, it iswell known that cell membranes contain substantialquantities of so-called `non-lamellar' phase-forminglipids. The importance of the proper balance betweenlamellar and non-lamellar phase-forming lipids hasbeen widely discussed [31^34]. The presence of thelatter signi¢cantly increases membrane monolayer

curvature stress, thereby conferring upon cell mem-branes a degree of non-lamellar-forming propensity.This is believed to be essential for normal membranefunction (for discussions of the probable biologicalroles of non-lamellar phase-forming lipids, see e.g.[35^37]). The X-ray study presented here clearlydemonstrates that vesicles, which are formed atroom temperature from lipid extracts of the plasmamembrane of A. laidlawii B or E. coli, are prone tothe formation of three-dimensionally ordered struc-tures. Speci¢cally, inverted cubic lipid phases of thespace group Pn3m and Im3m are formed upon heat-ing the lipid extracts and, once formed, these phasespersist upon cooling to physiologically relevant tem-peratures. Unequivocal phase assignment of cubicphases strongly depends on the number of re£ectionsobserved. This requirement was met for the Pn3mphase. On the other hand, the Im3m space groupdi¡er from the Pn3n space group only at the 11thre£ection, which corresponds to the k22 re£ection ofthe former phase and to the k21 re£ection of thelatter [38]. In the case of A. laidlawii B membranelipids, this re£ection could not be resolved clearlywith our experimental setup. However, the latticespacing ratio of 1.29 þ 0.1 found for our systems (Ta-ble 1) is close to the value of 1.28, a theoretical valuewhich is expected when both inverse bicontinuouscubic phases are in equilibrium in excess water [39].The interaction of GS with these lipid extracts pro-motes the formation of these isotropic phases as evi-denced by decreasing the temperature of the onset ofthis Q phase formation. The combination of X-raydi¡raction (present study) and 31P-NMR spectro-scopic data [9] acquired with these microbial lipidextracts is consistent with the existence of a polydo-main structure in which a lamellar phase coexistswith a three-dimensionally ordered phase, i.e. an in-verted Q phase.

A recent study has shown that alamethicin is alsoable to induce the formation of an inverted cubicphase of the same space group (Pn3m) in dielaidoylPE bilayers, which coexists with the HII phase [40]. Itwas suggested that alamethicin may induce suchphases by changing the thickness and/or £exibilityof the lipid bilayer. This suggestion is supported bythe observation that adsorption of alamethicin ontodiphytanoyl PC bilayer causes bilayer thinning,thereby inducing chain disorder over a large area

BBAMEM 77920 22-9-00

E. Staudegger et al. / Biochimica et Biophysica Acta 1468 (2000) 213^230 225

Page 14: X-ray studies on the interaction of the antimicrobial …...X-ray studies on the interaction of the antimicrobial peptide gramicidin S with microbial lipid extracts: evidence for cubic

[41,42]. Furthermore, it was shown that magaininbehaves in many aspects similar to alamethicin, caus-ing membrane thinning in PC/PS bilayers below thecritical concentration for peptide insertion [43],which roughly correlates with the concentration re-quired for cytolytic activity [44]. Very recently, Helleret al. [45] demonstrated that the L-sheet antimicro-bial peptide protegrin-1 also decreases the thicknessof diphytanoyl PC bilayers. Huang and coworkersproposed that the decrease of bilayer thickness iscompensated by an increase of the hydrophobiccross-sectional area of the lipid acyl chains. Thiswould also be consistent with the smaller cross-bi-layer distances of the phosphate groups found forthe GS^lipid mixtures in this study. In case of e.g.phosphatidylethanolamine, a major phospholipidcomponent of the E. coli lipid extracts, this lateralexpansion will further enhance the mismatch betweenthe cross-sectional areas of the smaller, morestrongly interacting headgroups and hydrocarbonside chains, inducing the lipid monolayer to curl.Although there are di¡erent molecular mechanismsthat may lead to formation of non-lamellar phasesby amphipathic peptides, a signi¢cant increase inmonolayer curvature stress is likely to be of majorimportance (for review see e.g. [46,47]), and may wellbe key to their membrane-disruptive properties (seebelow). However, it is important to note that otherantimicrobial peptides like magainin [48,49] or mas-toparan [50], as well as hemolytic peptides, such as N-lysin [51], impose `positive' curvature strain on non-lamellar-prone phospholipids, thereby destabilizingthe bilayer structure and hence perturbing membraneintegrity by di¡erent mechanisms, such as formingpores or exhibiting detergent-like action [47].

Our observations, however, clearly indicate thatGS may disrupt the structural integrity of lipid mem-branes by promoting the formation of inverted (typeII) non-lamellar lipid phases. We suggest that thelimited £exibility of the L-turn of GS, in particularwhen exposed to di¡erent environmental conditions,as well as the clustered location of the ornithine sidechains, might facilitate an accommodation of thepeptide in the lipid membrane that favors Q phaseformation. Moreover, the fact that the peptide-con-taining Q phases exhibit smaller lattice spacings thanfound for the pure lipid extracts further supports theidea that GS destabilizes the bilayer by increasing the

membrane curvature stress. A similar observationwas reported from X-ray di¡raction studies, whichshowed that the lattice parameter of the inverse hex-agonal phase in the presence of the fusion peptide ofsimian immunode¢ciency virus (SIV) was slightly lessas compared to the peptide-free lipid system [52].These authors demonstrated that peptides resemblingthe N-terminus of the SIV fusion peptide only exhib-ited fusogenic activity when capable of inducing neg-ative curvature stress in a set of three di¡erent PEmatrices. In contrast, a peptide of the same aminoacid composition, but an altered sequence, inducedpositive curvature stress and was thus not fusogenic.In addition, a variety of amphiphilic peptides or pro-teins that raised the LK/HII phase transition temper-ature were shown to be inhibitors of viral fusion [53].The potential of GS to facilitate or induce phasechanges that can lead to fusion was described byLegendre and Szoka [54], who showed that a 1:1complex of GS and DNA was mixed with a HII

phase-competent phosphatidylethanolamine and suc-cessfully used for cell transfection [55,56]. The activeinvolvement of transient lipid rearrangements inmembrane fusion was previously proposed by Siegel[57,58]. These observations are in agreement withdata described and discussed in this paper and o¡erintriguing interpretations for the biological activityof GS.

That the membrane-disrupting capacity of GS mayactually be a function of the non-lamellar phase-forming propensity of the target cell membrane hassome interesting implications for the mechanism ofaction of this peptide. The non-lamellar-forming pro-pensities of lipid bilayers are largely determined bytheir monolayer curvature stress [59,60], which in ourcase will be di¡erent for each lipid or mixtures there-of. Nevertheless, given that GS promotes the forma-tion of highly curved structures, such as Q phaseswhen it interacts with lipid bilayers, we can concludethat the monolayer curvature stress in bilayers de-rived from lipid^GS mixtures signi¢cantly exceedsthat which occurs in the corresponding peptide-freemembrane. We suggest that the occurrence of local-ized domains of high curvature stress are probablykey to the cytolytic activities of GS. This is becausedefects occurring at the boundaries of these domainsmay function as sites through which the leakage ofcellular contents may either occur. Moreover, be-

BBAMEM 77920 22-9-00

E. Staudegger et al. / Biochimica et Biophysica Acta 1468 (2000) 213^230226

Page 15: X-ray studies on the interaction of the antimicrobial …...X-ray studies on the interaction of the antimicrobial peptide gramicidin S with microbial lipid extracts: evidence for cubic

cause of the localized e¡ects of such lipid^peptideinteraction, these defects may begin to accumulateat low peptide-to-lipid molar ratios and will be max-imal when the intrinsic monolayer curvature stress ofthe membrane is high or when large disparities be-tween the curvature stresses in peptide-rich and pep-tide-poor membrane domains exist. Therefore, onecan envisage how the degradation of membrane bar-rier properties and other forms of membrane desta-bilization can readily occur once the number of suchdefects exceeds some critical value.

Previous work has also shown that GS promotesthe release of small phospholipid particles from E.coli and it was suggested that this process may beresponsible for the leakage of intracellular compo-nents and, ultimately, cell death [61^63]. It was alsodemonstrated that marked changes in cell morphol-ogy occur when GS binds to the erythrocyte mem-brane [62]. The results of these studies are all com-patible with our suggestion of GS induced membranedestabilization being mediated by highly localizeddomains of high membrane monolayer curvaturestress. Another aspect of this suggestion is that theprobability of signi¢cant peptide-induced membranedisruption should increase with the increases in theintrinsic curvature stress of the membrane. Since sig-ni¢cant amounts of non-lamellar phase-forming lip-ids are present in virtually all cell membranes, it fol-lows that all membranes should be susceptible todisruption by GS, though their susceptibility to itsaction should vary with amount of non-lamellarphase-forming lipids present as well as their intrinsicnon-lamellar-forming propensity.

In conclusion, our data suggests that GS has con-siderable potential for destabilizing lipid bilayerswith respect to highly curved and/or inverted non-lamellar phases. It should also be noted that unlikemammalian membranes, glycolipids are the predom-inant structural components of A. laidlawii B mem-branes. Interestingly, the ¢rst micellar-preferring bac-terial membrane lipid has also been described in thismycoplasma, namely glycerylphosphoryldiglucosyl-diacylglycerol [64]. Diglucosyldiacylglycerol andphosphatidylglycerol, the other major lipid compo-nent of the A. laidlawii B membrane, were shownto form only lamellar phases in the temperaturerange of 5^80³C, whereas monoglucosyldiacylglycer-ol exists in a lamellar phase at lower temperatures,

but does form non-lamellar phases at higher temper-atures [25,26,33]. The ¢nding from a recent mono-layer study, that the lipid mixture of A. laidlawii Apack more closely than the individual components,further emphasizes the capability of such lipid mix-tures to form non-lamellar structures owing to sup-posedly increased hydrogen bonding between the po-lar head group of PG and the sugar groups of theglycolipids [65]. While FTIR data suggest moderateinteraction of GS with single glycolipids [66], stron-ger interaction with phosphatidylglycerol has beenreported [67]. Nevertheless, the demonstration thatGS can promote the formation of non-lamellarphases in these membranes indicates that this prop-erty is not limited to phospholipid-based lipid bi-layers. We note that the suggestion that the cytolyticactivities of GS may actually be directed at the non-lamellar phase-forming tendencies of its membranetargets does not rule out the possibility that electro-static interactions between the basic residues of GSand the negatively charged phosphate groups ofphospholipid molecules may initiate or otherwise fa-cilitate lipid^GS interactions. The basic residues ofGS seem to be intimately linked to its antimicrobialactivity (see [2,5,68,69]) and negatively charged lipidsform a signi¢cant fraction of virtually all bacterialcell membranes. Thus, it seems likely that at thevery least, electrostatic interactions between the basicresidues of GS and negatively charged moieties at thesurfaces of cell membranes may promote antibioticactivity through the kinetic facilitation of lipid^pep-tide interaction. Moreover, we have recently shownthat the presence of cholesterol reduces the interac-tions of GS with PC model membranes and reducesthe propensity of GS to induce inverted non-lamellarphases in PE model membranes [70]. We also notethat the mechanism proposed above neither impliesnor requires the actual formation of cubic phasesunder the conditions where the cytolytic activity ofGS is commonly observed. The amount of GS re-quired in our experiments to detect signi¢cant Qphase formation may exceed that which was reportedfor antibiotic and hemolytic activities (for examplesof the latter, see [4,5] and references cited therein).However, the formation of Q phases or other three-dimensionally ordered structures can provide valua-ble information about the general mechanism of ac-tion of this antimicrobial peptide. This, in turn, may

BBAMEM 77920 22-9-00

E. Staudegger et al. / Biochimica et Biophysica Acta 1468 (2000) 213^230 227

Page 16: X-ray studies on the interaction of the antimicrobial …...X-ray studies on the interaction of the antimicrobial peptide gramicidin S with microbial lipid extracts: evidence for cubic

be helpful for the development of more speci¢c andpotent GS derivatives.

Acknowledgements

This work was supported by a National Grant ofthe Austrian Ministry of Science and Transportation(K.L.) as well as from the Jubila«umsfonds der Oë ster-reichischen Nationalbank (Project 7190, K.L.) andby an operating grant from the Medical ResearchCouncil of Canada and the Protein Engineering Net-work of Centers of Excellence (R.N.M.), and by apostdoctoral fellowship to E.J.P. from the AlbertaHeritage Foundation for Medical Research. The au-thors are indebted to Dr. Heinz Amenitsch for histechnical assistance with the Synchrotrone experi-ments at ELETTRA, Trieste, Italy.

References

[1] G.F. Gause, M.G. Brazhnikova, Gramicidin S and its use inthe treatment of infected wounds, Nature 154 (1944) 703.

[2] N. Izumiya, T. Kato, H. Aoyaga, M. Waki, M. Kondo(Eds.), Relationship between the primary structure and ac-tivity of gramicidin S and tyrocidines, in: Synthetic Aspectsof Biologically Active Cyclic Peptides: Gramicidin S andTyrocidines, Halsted Press, New York, 1979, pp. 49^70.

[3] M. Waki, N. Izumiya, Recent advances in the biotechnologyof L-lactams and microbial bioactive peptides, in: H. Klein-haug, H. van Doren (Eds.), Biochemistry of Peptide Anti-biotics, de Gruyter, Berlin, 1990, pp. 205^244.

[4] L.H. Kondejewski, S.W. Farmer, D.S. Wishart, C.M. Kay,R.E.W. Hancock, R.S. Hodges, Gramicidin S is activeagainst both Gram-positive and Gram-negative bacteria,Int. J. Pept. Protein Res. 47 (1996) 460^466.

[5] L.H. Kondejewski, S.W. Farmer, D.S. Wishart, C.M. Kay,R.E.W. Hancock, R.S. Hodges, Modulation of structure andantibacterial and hemolytic activity by ring size in cyclicgramicidin S analogs, J. Biol. Chem. 271 (1996) 25261^25268.

[6] R. Schwyzer, in: Amino Acids and Peptides with Antimeta-bolic Activity, Churchill, New York, 1958, p. 171.

[7] T. Kato, N. Izumiya, Conformations of di-N-methylleucinegramicidin S and N-methylleucine gramicidin S compatiblewith the sidedness hypothesis, Biochim. Biophys. Acta 493(1977) 235^238.

[8] E.J. Prenner, R.N.A.H. Lewis, R.N. McElhaney, The inter-action of the antimicrobial peptide gramicidin S with lipidbilayer model and biological membranes, Biochim. Biophys.Acta 1462 (1999) 201^221.

[9] E.J. Prenner, R.N.A.H. Lewis, K.C. Neuman, S.M. Gruner,L.H. Kondejewski, R.S. Hodges, R.N. McElhaney, Nonla-mellar phases induced by the interaction of gramicidin Swith lipid bilayers. A possible relationship to membrane-dis-rupting activity, Biochemistry 36 (1997) 7906^7916.

[10] J. Seelig, 31P Nuclear magnetic resonance and the headgroupstructure of phospholipid membranes, Biochim. Biophys.Acta 515 (1978) 105^140.

[11] C.P.S. Tilcock, P.R. Cullis, S.M. Gruner, On the validity of31P-NMR determinations of phospholipid polymorphicphase behavior, Chem. Phys. Lipids 40 (1986) 47^56.

[12] J.R. Silvius, N. Mak, R.N. McElhaney, Lipid and proteincomposition and thermotropic lipid phase transitions in fattyacid-homogeneous membranes of Acholeplasma laidlawii B,Biochim. Biophys. Acta 597 (1980) 199^215.

[13] P. Laggner, H. Mio, SWAX ^ a dual detector camera forsimultaneous small- and wide-angle X-ray di¡raction inpolymer and liquid crystal research, Nucl. Inst. MethodsPhys. Res. A 323 (1992) 86^90.

[14] H. Amenitsch, S. Bernstor¡, M. Kriechbaum, D. Lombardo,H. Mio, M. Rappolt, P. Laggner, Performance and Firstresults of the ELETTRA high-£ux beamline for small-angleX-ray scattering, J. Appl. Crystallogr. 30 (1997) 872^876.

[15] O. Glatter, A new method for the evaluation of small-anglescattering data, J. Appl. Crystallogr. 10 (1977) 415^421.

[16] O. Glatter, Data Treatment, in: O. Glatter, O. Kratky(Eds.), Small Angle X-Ray Scattering, Academic Press, Lon-don, 1982, pp. 119^166.

[17] O. Glatter, Interpretation, in: O. Glatter, O. Kratky (Eds.),Small Angle X-Ray Scattering, Academic Press, London,1982, pp. 167^196.

[18] A. Tardieu, V. Luzatti, F.C. Reman, Structure and polymor-phism of the hydrocarbon chains of lipids: a study of lec-ithin^water phases, J. Mol. Biol. 75 (1973) 711^733.

[19] J.A. Bouwstra, G.S. Gooris, W. Bras, H. Talsma, Smallangle X-ray scattering: possibilities and limitations in char-acterization of vesicles, Chem. Phys. Lipids 64 (1993) 83^98.

[20] A.E. Blaurock, Evidence of bilayer structure and of mem-brane interactions from X-ray di¡raction analysis, Biochim.Biophys. Acta 650 (1982) 167^207.

[21] J.M. Seddon, Structure of the inverted hexagonal (HII)phase, and non-lamellar phase transitions of lipids, Biochim.Biophys. Acta 1031 (1990) 1^69.

[22] S.M. Gruner, M.W. Tate, G.L. Kirk, P.T. So, D.C. Turner,D.T. Keane, X-ray di¡raction study of the polymorphic be-havior of N-methylated dioleoylphosphatidylethanolamine,Biochemistry 27 (1988) 2853^2866.

[23] J. Erbes, C. Czeslik, W. Hahn, R. Winter, M. Rappolt, G.Rapp, On the existence of bicontinuous cubic phases in di-oleoylphosphatidylethanolamine, Ber. Bunsenges. Phys.Chem. 98 (1994) 1287^1293.

[24] B. Tenchov, R. Koynova, G. Rapp, Accelerated formationof cubic phases in phosphatidylethanolamine dispersions,Biophys. J. 75 (1998) 853^866.

[25] A. Sen, S.W. Hui, D.A. Mannock, R.N.A.H. Lewis, R.N.McElhaney, Physical properties of glycosyl diacylglycerols.

BBAMEM 77920 22-9-00

E. Staudegger et al. / Biochimica et Biophysica Acta 1468 (2000) 213^230228

Page 17: X-ray studies on the interaction of the antimicrobial …...X-ray studies on the interaction of the antimicrobial peptide gramicidin S with microbial lipid extracts: evidence for cubic

2. X-ray di¡raction studies of a homologous series of 1,2-di-O-acyl-3-O-(alpha-D-glucopyranosyl)-sn-glycerols, Biochem-istry 29 (1990) 7799^7804.

[26] A.-S. Andersson, L. Rilfors, G. Ora«dd, G. Lindblom, Totallipids with short and long acyl chains from Acholeplasmaform nonlamellar phases, Biophys. J. 75 (1998) 2877^2887.

[27] G. Lindblom, L. Rilfors, Cubic phases and isotropic struc-tures formed by membrane lipids ^ possible biological sig-ni¢cance, Biochim. Biophys. Acta 988 (1989) 221^256.

[28] E. Shyamsunder, S.M. Gruner, M.W. Tate, D.C. Turner,P.T.C. So, C.P.S. Tilcock, Observation of inverted cubicphase in hydrated dioleoylphosphatidylethanolamine mem-branes, Biochemistry 27 (1988) 2332^2336.

[29] J.A. Veiro, R.G. Khalifah, E.S. Rowe, P-31 nuclear mag-netic resonance studies of the appearance of an isotropiccomponent in dielaidoylphosphatidylethanolamine, Biophys.J. 57 (1990) 637^641.

[30] S. Morein, A.-S. Andersson, L. Rilfors, G. Lindblom, Wild-type Escherichia coli cells regulate the membrane lipid com-position in a `window' between gel and non-lamellar struc-tures, J. Biol. Chem. 271 (1996) 6801^6809.

[31] L. Rilfors, G. Lindblom, A. Wieslander, A. Christiansson,Membrane Fluidity, in: M. Kates, L.A. Mason (Eds.), Bio-membranes, Vol. 12, Plenum Press, New York, 1984, pp.205^245.

[32] L. Rilfors, A. Wieslander, G. Lindblom, Regulation andphysicochemical properties of the polar lipids in Acholeplas-ma laidlawii, in: S. Rottem, I. Kahane (Eds.), SubcellularBiochemistry, Vol. 20, Plenum Press, New York, 1993, pp.109^166.

[33] P.J. Foht, Q.M. Tran, R.N.A.H. Lewis, R.N. McElhaney,Quantitation of the phase preferences of the major lipids ofthe Acholeplasma laidlawii B membrane, Biochemistry 34(1995) 13811^13817.

[34] R.N. McElhaney, in: J. Manilo¡, R.N. McElhaney, L.R.Finch, J.B. Baseman (Eds.), Mycoplasma: Molecular Biol-ogy and Pathogenesis, American Society for Microbiology,Washington, DC, 1992, pp. 113^155.

[35] S.-W. Hui, A. Sen, E¡ects of lipid packing on polymorphicphase behavior and membrane properties, Proc. Natl. Acad.Sci. USA 86 (1989) 5825^5829.

[36] S.M. Gruner, Non-lamellar lipid phases, in: P.L. Yeagle(Ed.), Structure of Biological Membranes, CRC Press,Boca Raton, FL, 1992, pp. 211^250.

[37] K. Lohner, Is the high propensity of ethanolamine plasmal-ogens to form non-lamellar lipid structures manifested in theproperties of biomembranes?, Chem. Phys. Lipids 81 (1996)167^184.

[38] J.S. Kasper, K. Lonsdale, International Tables for X-rayCrystallography, Vol. 2, Riedel, Dordrecht, 1985.

[39] S. Hyde, S. Andersson, B. Ericsson, K. Larsson, A cubicstructure consisting of a lipid bilayer forming an in¢niteperiodic minimum surface of the gyroid type in theglyceromonooleate^water system, Z. Krystallogr. 168(1984) 213^219.

[40] S.L. Keller, S.M. Gruner, K. Gawrisch, Small concentra-

tions of alamethicin induce a cubic phase in bulk phospha-tidylethanolamine mixtures, Biochim. Biophys. Acta 1127(1996) 241^246.

[41] Y. Wu, K. He, S.J. Ludtke, H.W. Huang, X-ray di¡ractionstudy of lipid bilayer membranes interacting with amphi-philic helical peptides: diphytanoyl phosphatidylcholinewith alamethicin at low concentrations, Biophys. J. 68(1995) 2361^2369.

[42] K. He, S.J. Ludtke, W.T. Heller, H.W. Huang, Mechanismof alamethicin insertion into lipid bilayers, Biophys. J. 71(1996) 2669^2897.

[43] S.J. Ludtke, K. He, H.W. Huang, Membrane thinningcaused by magainin 2, Biochemistry 34 (1995) 16764^16769.

[44] S.J. Ludtke, K. He, Y. Wu, H.W. Huang, Cooperative mem-brane insertion of magainin correlated with its cytolytic ac-tivity, Biochim. Biophys. Acta 1190 (1994) 181^184.

[45] W.T. Heller, A.J. Waring, R.I. Lehrer, T.A. Harroun, T.M.Weiss, L. Yang, H.W. Huang, Membrane thinning e¡ect ofthe beta-sheet antimicrobial protegrin, Biochemistry 39(2000) 139^145.

[46] R.M. Epand, Lipid polymorphism and protein^lipid interac-tions, Biochim. Biophys. Acta 1376 (1998) 353^368.

[47] K. Lohner, E.J. Prenner, Di¡erential scanning calorimetryand X-ray di¡raction studies of the speci¢city of the inter-action of antimicrobial peptides with membrane-mimetic sys-tems, Biochim. Biophys. Acta 1462 (1999) 141^156.

[48] T. Wieprecht, M. Dathe, R.M. Epand, M. Beyermann, E.Krause, W.L. Maloy, D.L. MacDonald, M. Bienert, In£u-ence of the angle subtended by the positively charged helixface on the membrane activity of amphipathic, antibacterialpeptides, Biochemistry 36 (1997) 12869^12880.

[49] K. Matsuzaki, K. Sugishita, N. Ishibe, M. Ueha, S. Nakata,K. Miyajima, R.M. Epand, Relationship of membrane cur-vature to the formation of pores by magainin 2, Biochemis-try 37 (1998) 11856^11863.

[50] E.M. Tytler, J.P. Segrest, R.M. Epand, S.-Q. Nie, R.F.Epand, V.K. Mishra, Y.V. Venkatachalapathi, G.M. Anan-tharamaiah, Reciprocal e¡ects of apolipoprotein and lyticpeptide analogs on membranes ^ cross-sectional molecularshapes of amphipathic alpha helixes control membranestability, J. Biol. Chem. 268 (1993) 22112^22118.

[51] E. Staudegger, Interaction of the Hemolytic Bacteriotoxin N-Lysin with Model Membranes, Thesis, 1998, Graz Universityof Technology.

[52] A. Colotto, I. Martin, J. Ruysschaert, A. Sen, S.W. Hui,R.M. Epand, Structural study of the interaction betweenthe SIV fusion peptide and model membranes, Biochemistry35 (1996) 980^989.

[53] R.M. Epand, in: R.C. Aloia, C.C. Curtain (Eds.), Mem-brane Interactions of HIV: Implications for Pathogenesisand Therapy in Aids, Wiley-Liss, New York, 1992, pp. 99^112.

[54] J.Y. Legendre, F.C. Szoka, Cyclic amphipathic peptide-DNA complexes mediate high-e¤ciency transfection of ad-herent mammalian cells, Proc. Natl. Acad. Sci. USA 90(1993) 893^897.

BBAMEM 77920 22-9-00

E. Staudegger et al. / Biochimica et Biophysica Acta 1468 (2000) 213^230 229

Page 18: X-ray studies on the interaction of the antimicrobial …...X-ray studies on the interaction of the antimicrobial peptide gramicidin S with microbial lipid extracts: evidence for cubic

[55] J.Y. Legendre, A. Supersaxo, Short-chain phospholipids en-hance amphipathic peptide-mediated gene transfer, Biochem.Biophys. Res. Commun. 217 (1995) 179^185.

[56] T. Hara, H. Kuwasawa, Y. Aramaki, S. Takada, K. Koike,K. Ishidate, H. Kato, S. Tsuchiya, E¡ects of fusogenic andDNA-binding amphiphilic compounds on the receptor-medi-ated gene transfer into hepatic cells by asialofetuin-labeledliposomes, Biochim. Biophys. Acta 1278 (1996) 51^58.

[57] D.P. Siegel, Energetics of intermediates in membrane fusion:comparison of stalk and inverted micellar intermediatemechanisms, Biophys. J. 65 (1993) 2124^2140.

[58] D.P. Siegel, The modi¢ed stalk mechanism of lamellar/in-verted phase transitions and its implications for membranefusion, Biophys. J. 76 (1999) 291^313.

[59] G.L. Kirk, S.M. Gruner, D.E. Stein, A thermodynamicmodel of the lamellar (LK) to inverse hexagonal (HII) lipidmembrane^water systems, Biochemistry 23 (1984) 1093^1102.

[60] S.M. Gruner, Intrinsic curvature hypothesis for biomem-brane lipid composition: a role for nonbilayer lipids, Proc.Natl. Acad. Sci. USA 82 (1985) 3665^3669.

[61] T. Katsu, H. Kobayashi, Y. Fujita, Mode of action of gram-icidin S on Escherichia coli membrane, Biochim. Biophys.Acta 860 (1986) 608^619.

[62] T. Katsu, C. Ninomiya, M. Kuroko, H. Kobayashi, T. Hir-ota, Y. Fujita, Action mechanism of amphipathic peptidesgramicidin S and melittin on erythrocyte membrane, Bio-chim. Biophys. Acta 939 (1988) 57^63.

[63] T. Katsu, M. Kuroko, T. Morikawa, K. Sanchicka, Y. Fu-jita, H. Yamamura, M. Uda, Mechanism of membrane dam-age induced by the amphipathic peptides gramicidin S andmelittin, Biochim. Biophys. Acta 983 (1989) 135^141.

[64] R.N.A.H. Lewis, R.N. McElhaney, Acholeplasma laidlawii Bmembranes contain a lipid (glycerylphosphoryldiglucosyldi-

acylglycerol) which forms micelles rather than lamellar orreversed phases when dispersed in water, Biochemistry 34(1995) 13818^13824.

[65] A.-S. Andersson, R.A. Demel, L. Rilfors, G. Lindblom, Lip-ids in total extracts from Acholeplasma laidlawii A packmore closely than the individual lipids. Monolayers studiedat the air^water interface, Biochim. Biophys. Acta 1369(1998) 94^102.

[66] R.N.A.H. Lewis, E.J. Prenner, L.H. Kondejewski, C.R.Flach, R. Mendelsohn, R.S. Hodges, R.N. McElhaney,Fourier transform infrared spectroscopic studies of the inter-action of the antimicrobial peptide gramicidin S with lipidmicelles and with lipid monolayer and bilayer membranes,Biochemistry 38 (1999) 15193^15203.

[67] E.J. Prenner, R.N.A.H. Lewis, L.H. Kondejewski, R.S.Hodges, R.N. McElhaney, Di¡erential scanning calorimetricstudy of the e¡ect of the antimicrobial peptide gramicidin Son the thermotropic phase behavior of phosphatidylcholine,phosphatidylethanolamine and phosphatidylglycerol lipid bi-layer membranes, Biochim. Biophys. Acta 1417 (1999) 211^223.

[68] Y.A. Ovchinnikov, V.T. Ivanov, Conformational states andbiological activity of cyclic peptides, Tetrahedron 31 (1975)2177^2209.

[69] Y.A. Ovchinnikov, V.T. Ivanov, The cyclic peptides: struc-ture, conformation, and function, in: H. Neurath, R.L. Hill(Eds.), The Proteins, Vol. V, Academic Press, New York,1982, pp. 391^398.

[70] E.J. Prenner, R.N.A.H. Lewis, M. Jelokhani-Niaraki, R.S.Hodges, R.N. McElhaney, Cholesterol attenuates the inter-action of the antimicrobial peptide gramicidin S with phos-pholipid bilayer membranes, Biochim. Biophys. Acta, sub-mitted for publication.

BBAMEM 77920 22-9-00

E. Staudegger et al. / Biochimica et Biophysica Acta 1468 (2000) 213^230230