Biochimica et Biophysica Acta 1417 (1999) 211^223Di¡erential scanning calorimetric study of the e¡ect of the antimicrobial peptide gramicidin S on the thermotropic phase behavior of phosphatidylcholine, phosphatidylethanolamine and phosphatidylglycerol lipid bilayer membranes Elmar J. Prenner a , Ruthven N.A.H. Lewis a , Leslie H. Kondejewski aYb , Robert S. Hodges aYb , Ronald N. McElhaney aY * a Department of Biochemistry, University of Alberta, Edmonton, Alta. T6G 2H7, Canada b Protein Engineering Network of Centres of Excellence, University of Alberta, Edmonton, Alta. T6G 2H7, Canada Received 22 September 1998; received in revised form 21 December 1998; accepted 7 January 1999Abstract We have studied the effects of the antimicrobial peptide gramicidin S (GS) on the thermotropic phase behavior of largemultilamellar vesicles of dimyristoylphosphatidylcholine (DMPC), dimyristoylphosphatidylethanolamine (DMPE) anddimyristoyl phosphatidylglycerol (DMPG) by high-sensitivity differential scanning calorimetry. We find that the effect of GSon the lamellar gel to liquid-crystalline phase transition of these phospholipids varies markedly with the structure and chargeof their polar headgroups. Specifically, the presence of even large quantities of GS has essentially no effect on the main phasetransition of zwitterionic DMPE vesicles, even after repeating cycling through the phase transition, unless these vesicles areexposed to high temperatures, after which a small reduction in the temperature, enthalpy and cooperativity of the gel toliquid-crystalline phase transitions is observed. Similarly, even large amounts of GS produce similar modest decreases in thetemperature, enthalpy and cooperativity of the main phase transition of DMPC vesicles, although the pretransition isabolished at low peptide concentrations. However, exposure to high temperatures is not required for these effects of GS onDMPC bilayers to be manifested. In contrast, GS has a much greater effect on the thermotropic phase behavior of anionicDMPG vesicles, substantially reducing the temperature, enthalpy and cooperativity of the main phase transition at higherpeptide concentrations, and abolishing the pretransition at lower peptide concentrations as compared to DMPC. Moreover,the relatively larger effects of GS on the thermotropic phase behavior of DMPG vesicles are also manifest without cyclingthrough the phase transition or exposure to high temperatures. Furthermore, the addition of GS to DMPG vesicles protectsthe phospholipid molecules from the chemical hydrolysis induced by their repeated exposure to high temperatures. Theseresults indicate that GS interacts more strongly with anionic than with zwitterionic phospholipid bilayers, probably because Abbreviations: GS, gramicidin S; PC, phosphatidylcholine; PE, phosphatidylethanolamine; PG, phosphatidylglycerol; DPG, diphos-phatidylglycerol; PS, phosphatidylserine ; SpM, sphingomyelin; DMPC, dimyristoylphosphatidylcholine ; DPPC, dipalmitoylphosphatid-ylcholine ; DMPE, dimyristoylphosphatidylethanolamine ; DMPG, dimyristoylphosphatidylglycerol ; DSC, di¡erential scanning calorim-etry; ESR, electron spin resonance; NMR, nuclear magnetic resonance; IR, infrared; NBD-GS, N-4-nitrobenz-2-oxa-1,3-diazole H Hgramicidin S; Lc or Lc , lamellar crystalline phase with untilted or tilted hydrocarbon chains, respectively; LL or LL , lamellar gel phase Hwith untilted or tilted hydrocarbon chains, respectively; PL , lamellar rippled gel phase with tilted hydrocarbon chains; LK , lamellar liquid-crystalline phase * Corresponding author. Fax: +1-403-492-0095; E-mail: firstname.lastname@example.org / 99 / $ ^ see front matter ß 1999 Elsevier Science B.V. All rights reserved.PII: S 0 0 0 5 - 2 7 3 6 ( 9 9 ) 0 0 0 0 4 - 8
212 E.J. Prenner et al. / Biochimica et Biophysica Acta 1417 (1999) 211^223of the more favorable net attractive electrostatic interactions between the positively charged peptide and the negativelycharged polar headgroup in such systems. Moreover, at comparable reduced temperatures, GS appears to interact morestrongly with zwitterionic DMPC than with zwitterionic DMPE bilayers, probably because of the more fluid character of theformer system. In addition, the general effects of GS on the thermotropic phase behavior of zwitterionic and anionicphospholipids suggest that it is located at the polar/apolar interface of liquid-crystalline bilayers, where it interacts primarilywith the polar headgroup and glycerol-backbone regions of the phospholipid molecules and only secondarily with the lipidhydrocarbon chains. Finally, the considerable lipid specificity of GS interactions with phospholipid bilayers may prove usefulin the design of peptide analogs with stronger interactions with microbial as opposed to eucaryotic membrane lipids. ß 1999Elsevier Science B.V. All rights reserved.1. Introduction bilayers which is required to design GS analogs with enhanced activity for bacterial membranes and di- Gramicidin S (GS)1 is a cyclic decapeptide[cyclo- minished activity against the plasma membranes of(Val^Orn^Leu^D^Phe^Pro)2 ] ¢rst isolated from Ba- human and animal cells.cillus brevis (see ) and is one of a series of anti- Although considerable evidence exists that themicrobial peptides produced by this microorganism principle target of GS is the lipid bilayer of cell sur-(see ). This peptide exhibits appreciable antibiotic face membranes, only a limited number of studies ofactivity against a broad spectrum of both Gram-neg- the interaction of this peptide with lipid bilayer mod-ative and Gram-positive bacteria as well as against el membranes have been published to date. More-several pathogenic fungi [3,4]. In aqueous solution over, the conclusions reached from these studies doGS forms an amphiphilic, two-stranded, antiparallel not always agree with one another. Pache et al. L-sheet structure in which the four hydrophobic Leu reported that even at a low lipid/peptide ratio (10:1),and Val residues project from one side of this disk- GS had almost no e¡ect on the gel to liquid-crystal-shaped peptide molecule and the two basic Orn res- line phase transition temperature, enthalpy or coop-idues project from the other [2,5^8]. In general, the erativity of DPPC multilamellar vesicles, nor was theantimicrobial and hemolytic activity of GS analogs organization the DPPC bilayer signi¢cantly per-increase with the degree of hydrophobicity and am- turbed by the presence of the peptide, as monitoredphiphilicity of the peptide up to some optimal value by NMR or ESR spectroscopy. The only signi¢cantand the presence of two positively charged amino e¡ect of GS addition observed was a decrease in theacids are essential for maximal activity [3,9,10]. DPPC P^O stretching frequency as monitored by IRAlthough the Orn residues of GS may be replaced spectroscopy. Pache and coworkers thus concludedby Lys or Arg residues, the presence of two posi- that GS is bound to the surface of the DPPC bilayertively charged amino acid residues on the same face and interacts only with the lipid polar headgroups. Aof the peptide molecule are also known to be re- similar conclusion was reached by Datema et al. ,quired for maximal antimicrobial activity [2,4,6,11]. based on their observations of the very small reduc- Considerable evidence exists that the primary tar- tion of the temperature and cooperativity of theget of GS is the lipid bilayer of cell surface mem- main phase transition of DPPC observed by DSCbranes and that this peptide kills cells by destroying even at low lipid/peptide ratios (6:1), the insensitivitythe structural integrity of the lipid bilayer by induc- of the motional characteristics of the GS molecule toing the formation of pores or other localized defects the DPPC gel to liquid-crystalline phase transition,[12^17]. Unfortunately, GS is rather nonspeci¢c in its and the ability of GS to induce an isotropic compo-actions and exhibits appreciable hemolytic as well as nent in the 31 P-NMR spectrum at higher tempera-antimicrobial activity [18^20]. The therapeutic uti- tures. However, based on the generally appreciablelization of GS has therefore been limited to topical hydrophobicity of the GS molecule, the sensitivity ofapplications . A major goal of our current studies the lateral di¡usion coe¤cient of this peptide to theon the interaction of GS with model and biological host bilayer chain-melting phase transition, and themembranes is to provide the fundamental knowledge ability of cholesterol to induce lateral aggregation ofof the mechanism of action of this peptide on lipid GS in DPPC and particularly in DMPC bilayers, Wu
E.J. Prenner et al. / Biochimica et Biophysica Acta 1417 (1999) 211^223 213et al.  suggested that GS must penetrate at least polar headgroups. This research is intended to pro-partially into the hydrophobic core of the host lipid vide basic information on the sensitivity of variousbilayer. This conclusion was later supported by Zi- membrane lipid classes to the action of GS and ondovetzki et al. , who demonstrated by 2 H-NMR how these are a¡ected by variations in the structurespectroscopy that GS alters the orientational order of the peptide molecule. Because the lipid composi-parameter pro¢le of the hydrocarbon chains of tions of bacterial and eucaryotic cell membranes areDMPC, at least at low lipid/protein ratios (5.5:1), quite di¡erent and because antimicrobial and hemo-and by 31 P-NMR spectroscopy, which suggested lytic activities can be at least partially dissociated inthat the structure of the bilayer is abolished at even GS analogs , such information could potentially belower lipid/peptide ratios (2.7:1). Thus these early used in the design of more therapeutically useful GSstudies, which were done exclusively on zwitterionic derivatives. In fact, our recent work has shown forPC systems, failed to resolve either the issue of the the ¢rst time that GS does exhibit an appreciablelocation of GS in lipid model membranes or the na- lipid polar headgroup speci¢city, at least as regardsture of its interactions with the host lipid bilayer. the lamellar/nonlamellar phase behavior of various We have recently studied the e¡ect of GS on the glycerophospho- and sphingophospholipids .thermotropic phase behavior of a variety of synthetic The present high-sensitivity DSC study of the e¡ectphospholipids using primarily 31 P-NMR spectros- of GS on the thermotropic phase behavior ofcopy supplemented by DSC and X-ray di¡raction. DMPC, DMPE and DMPG bilayer membranes rep-We found that at physiologically relevant concentra- resents a continuation of this research program.tions of GS (lipid/protein ratios of 25:1), GS does Since PCs are virtually absent in bacterial mem-not a¡ect the lamellar phase preference of the zwit- branes but are generally the most abundant phos-terionic lipids PC and SpM nor of the anionic lipids pholipid in eucaryotic plasma membranes, DMPCDPG and PS, even at high temperatures. However, can serve as a model for the surface membrane ofGS was found to potentiate inverted cubic phase the cells of higher animals, particularly so since PCsformation in zwitterionic PE and, to a lesser extent, are typically found primarily in the outer monolayerin anionic PG dispersions, as well as in total polar of the lipid bilayer of such membranes. Similarly,lipid extracts from the glycolipid-based membranes PGs are absent in eukaryotic plasma membranesof Acholeplasma laidlawii and from the phospholip- but are ubiquitous and often abundant in bacterialid-based membranes of Escherichia coli. Moreover, membranes, so that DMPG can serve as a goodthe ability of GS to induce nonlamellar phase for- model for the bacterial membrane. Finally, PEs aremation was found to increase with the intrinsic non- also major components of virtually all eukaryoticlamellar phase-forming propensity of the PE molec- plasma membranes but are also abundant in mostular species studied. We suggested that the ability of Gram-negative bacteria and in members of theGS to induce localized regions of high curvature Gram-positive Bacillus genus of bacteria as well.stress in the lipid bilayers of biological membranes However, in eukaryotic cell membranes PEs are usu-may be relevant to the mechanism by which this ally enriched in the inner monolayer of the lipid bi-peptide disrupts cell membranes. layer. Thus information gained from a study of It thus seems likely that interactions between GS DMPE may be applicable to both bacterial and ani-and biological membranes will vary markedly with mal plasma membranes, but especially to the former.the physical properties, and thus with the lipid com-position, of the target membrane. Also, because lipidpolar headgroups essentially determine the surface 2. Materials and methodsproperties of biological membranes, interactions be-tween GS and biological membranes should be par- DMPC, DMPE and DMPG were purchased fromticularly sensitive to membrane lipid polar head- Avanti Polar Lipids (Alabaster, AL) and were usedgroup composition. We are thus studying the without further puri¢cation. Gramicidin S was ob-nature of the interactions of GS with lipid bilayer tained from Sigma (St. Louis, MO) and dissolvedmodel membranes which contain a wide range of in pure ethanol and stored at 320³C. The lipid stock
214 E.J. Prenner et al. / Biochimica et Biophysica Acta 1417 (1999) 211^223solutions were dissolved in redistilled methanol. Lip- phase behavior in a system exhibiting highly cooper-id/peptide mixtures were prepared by mixing appro- ative lipid phase transitions and to minimize the ther-priate amounts of the lipid and peptide stock solu- mal history dependence of the system, lipid^peptidetions, drying under N2 , and exposure of the lipid/ complexes were prepared as described above andpeptide ¢lms to high vacuum overnight. The multi- then these complexes were hydrated to produce largelamellar GS-containing vesicles were then prepared multilamellar vesicles in which the GS was presum-by vortexing in excess bu¡er (10 mM Tris^HCl, ably uniformly distributed among the various bi-100 mM NaCl and 2 mM EDTA) normally at tem- layers. In this way we could accurately measure theperatures above the main phase transition tempera- e¡ects of very small amounts of GS on phospholipidture of the phospholipid. All chemicals were pur- thermotropic phase behavior while minimizing,chased from BDH (Toronto, Ont.). although not entirely eliminating, any thermal his- We found that the e¡ects of GS on the thermo- tory dependence of the sample. This latter e¡ecttropic phase behavior of the phospholipid vesicles was obviated by performing three heating and cool-studied depended signi¢cantly on the method of ing cycles in the DSC instrument before collectingpreparation of these peptide^phospholipid complexes data.and on their thermal history. In order to mimic the The calorimetry was performed on a Microcalin vivo e¡ects of GS as closely as possible, we ini- MC-2 high-sensitivity Di¡erential Scanning Calorim-tially added peptide to preformed large unilamellar eter (Microcal, Northampton, MA). For all samplesvesicles at temperatures above the phospholipid gel a scan rate of 10³C/h was used. Sample runs wereto liquid-crystalline phase transition temperature. repeated at least three times to ensure reproducibil-However, the chain-melting phase transition of these ity. Data acquisition and analysis was done usinglarge unilamellar vesicles was so broad, even in the Microcals DA-2 and Origin software (Microcal).absence of peptide, that the e¡ects of the addition of The total lipid concentrations used for the DSC anal-smaller quantities of GS on phospholipid thermo- yses were about 0.5 mg/ml, providing for full hydra-tropic phase behavior could not be accurately deter- tion for the GS^phospholipid mixtures. Samples con-mined. Moreover, the e¡ects of the addition of larger taining GS alone, dissolved in bu¡er at peptideamounts of GS to these vesicles varied considerably concentrations corresponding to those of the highestwith the number of heating and cooling scans per- peptide/lipid molar ratios studied (1:10), exhibitedformed, presumably because GS can cause vesicular no thermal events over the temperature range 0^lysis and fusion at higher concentrations, allowing 90³C. This indicates that GS does not denaturethe peptide access to both surfaces of the lipid bi- over this temperature range and that the endother-layer. Alternatively, when small amounts of GS are mic events observed in this study arise exclusivelyadded to preformed large multilamellar vesicles, from phase transitions of the phospholipid vesicles.which exhibit much more cooperative phase transi-tions in the absence of peptide, only very small ef-fects on phospholipid thermotropic phase behavior 3. Resultswere observed, probably because GS was interactingonly with the outer monolayer of the single bilayer 3.1. Thermotropic phase behavior of GS/DMPCon the surface of these multilamellar vesicles. Again, mixtureswhen larger amounts of peptide were added to thesesystems, the characteristic e¡ects of GS on the ther- DSC heating endotherms illustrating the e¡ects ofmotropic phase behavior of the multilamellar phos- GS on the thermotropic phase behavior of large,pholipid vesicles became increasingly more marked multilamellar vesicles of DMPC alone are presentedwith each DSC heating and cooling cycle, presum- in Fig. 1. Aqueous dispersions of DMPC alone,ably because the peptide was progressively gaining when not extensively annealed at low temperatures,access to the surfaces of bilayers which were initially exhibit only two endothermic events, a less energeticinaccessible. Thus, in order to maximize the observ- pretransition near 14³C and a more energetic mainable e¡ects of GS on phospholipid thermotropic transition near 24³C. Under these conditions a sub-
E.J. Prenner et al. / Biochimica et Biophysica Acta 1417 (1999) 211^223 215 Fig. 2. A plot of variations in the midpoint temperatures for the pretransition and main transition of DMPC multilamellar vesicles as a function of GS concentration. E, pretransition tem- perature; b, overall temperature for main phase transition; k, temperature of the sharp component of the main phase transi- tion; a, temperature of the broad component of the main phase transition.Fig. 1. High-sensitivity DSC heating scans illustrating the e¡ectof the addition of increasing quantities of GS on the thermo- temperature endotherm. The temperature (Fig. 2),tropic phase behavior of DMPC multilamellar vesicles. The topscan is of DMPC alone, and the lipid/peptide molar ratios of enthalpy (Fig. 3) and cooperativity (Fig. 1) of boththe lower scans are indicated on the ¢gure itself. components of the chain-melting phase transition also decrease with increases in peptide concentration relative to DMPC alone. However, the GS-inducedtransition centered at 18³C is not observed, since decreases in the phase transition temperature of the Hthere is insu¤cient time for the formation of an Lcphase and this phase was not nucleated by exposureto low temperatures (320³C). The pretransition H Harises from the conversion of the LL to the PL phaseand the main or chain-melting phase transition from Hthe conversion of the PL to the LK phase. For a moredetailed discussion of the thermotropic phase behav-ior of DMPC and other members of the homologousseries of linear saturated PCs, see Lewis et al. . The interaction of GS with DMPC vesicles clearlyalters the thermotropic phase behavior of the latter,as illustrated in Fig. 1. Small amounts of GS de-crease the temperature (Fig. 2) and enthalpy (Fig.3) of the pretransition and the pretransition is abol-ished entirely in DMPC vesicles having lipid/peptideratios of 100:1 or less (Fig. 1). Moreover, the pres-ence of increasing quantities of GS also results in the Fig. 3. A plot of the transition enthalpies for the pretransitioninduction of a two-component main phase transition, and main transition of DMPC multilamellar vesicles as a func-with a more cooperative, lower temperature endo- tion of GS concentration. The symbols used are the same astherm superimposed over a less cooperative, higher for Fig. 2.
216 E.J. Prenner et al. / Biochimica et Biophysica Acta 1417 (1999) 211^223two main phase transition components are modest(2^3³C) and the total enthalpy of chain melting isreduced by only 10^15%, even at a phospholipid/pep-tide ratio of 10:1. These results indicate that thepresence of GS produces only a modest destabiliza-tion of the gel state of DMPC bilayers, even aftermaximizing GS^DMPC interactions by multiple cy-cling through the gel to liquid-crystalline phase tran-sition. Generally similar although less detailed resultswere reported previously for GS/DMPC and GS/DPPC mixtures studied by low-sensitivity DSC[22,23].3.2. Thermotropic phase behavior of GS/DMPE mixtures Fig. 5. High-sensitivity DSC heating and cooling scans illustrat- DSC heating scans illustrating the e¡ects of GS ing the e¡ect of the exposure to high temperature on the ther-addition on the thermotropic phase behavior of motropic phase behavior of DMPE multilamellar vesicles hav-large, multilamellar vesicles of DMPE are presented ing a lipid/peptide molar ratio of 25:1. Heating and coolingin Figs. 4 and 5. Aqueous dispersions of DMPE thermograms of DMPE alone are illustrated in A and B, re- spectively, for comparison. The heating scan of GS-containingalone, in the absence of extensive incubation at low DMPE vesicles not previously exposed to high temperature istemperature, exhibit a single, relatively energetic LL / shown in C, while heating and cooling scans of the sameLK phase transition near 50³C (see , and referen- vesicles exposed brie£y to a temperature of 75³C are shown inces therein for a more complete description of the D and E, respectively.thermotropic phase behavior of DMPE and othermembers of the homologous series of linear saturated atures above but near to the gel to liquid-crystallinePEs). If GS is added to DMPE vesicles at temper- phase transition temperature of DMPE, essentially no e¡ect on the thermotropic phase behavior is ob- served upon heating (Fig. 4), even at high GS con- centrations. However, if GS-containing DMPE vesicles are exposed to temperatures well above the gel to liquid-crystalline phase transition temperature, increasing quantities of GS lower the temperature, enthalpy and cooperativity of the chain-melting phase transition of DMPE slightly when DSC cool- ing curves are run (Fig. 5). Moreover, upon subse- quent reheating, the characteristic e¡ects of the pres- ence of the peptide on the phase behavior of DMPE vesicles are retained (Fig. 5). However, the e¡ects of GS addition on the temperature, enthalpy and coop- erativity of the main phase transition observed upon cooling and subsequent reheating are very small, even at the highest peptide^phospholipid ratio tested.Fig. 4. High-sensitivity DSC heating scans illustrating the e¡ect These ¢ndings indicate that the presence of GS pro-of the addition of increasing quantities of GS on the thermo- duces only a very slight destabilization of the geltropic phase behavior of DMPE multilamellar vesicles not ex-posed to high temperatures (i.e., temperatures above 65^70³C). state of DMPE bilayers and then only if the GS-The top scan is of DMPE alone, and the lipid/peptide molar containing vesicles are ¢rst exposed to high temper-ratios of the lower scans are indicated on the ¢gure itself. atures. However, the interaction of GS with DMPE
E.J. Prenner et al. / Biochimica et Biophysica Acta 1417 (1999) 211^223 217Fig. 6. High-sensitivity DSC heating scans illustrating the e¡ect of the addition of relatively large or relatively small quantities of GSon the main phase transition of DMPG multilamellar vesicles are presented in panels A and B, respectively. In each panel the topscan is of DMPG alone, and the lipid/peptide molar ratios of the lower scans are indicated on the ¢gure panels. The insets of panelB are blow-ups of the DSC scan in the temperature range just above the sharp component of the main phase transition, illustratingthe behavior of the minor broad endothermic peak present in these samples. High-sensitivity DSC scans illustrating the e¡ect of theaddition of small quantities of GS on the pretransition of DMPG are presented in panel C. The DMPG concentration is threefoldthat of the samples utilized in panels A and B, and only the pretransition regions of the DSC thermograms are illustrated.
218 E.J. Prenner et al. / Biochimica et Biophysica Acta 1417 (1999) 211^223bilayers at high temperatures can also induce the LKphase of this phospholipid to form an inverted cubicphase , indicating that structurally signi¢cant in-teractions between GS and a su¤ciently £uid phaseof DMPE can occur, and that these interactions dopersist after cooling to temperatures below the mainphase transition temperature.3.3. Thermotropic phase behavior of GS/DMPG mixtures DSC heating scans illustrating the e¡ect of GSaddition on the thermotropic phase behavior oflarge, multilamellar vesicles of DMPG are presentedin Fig. 6A,B. Aqueous dispersions of DMPG, not Fig. 8. A plot of the transition enthalpies of the main phaseextensively annealed at low temperatures, also exhib- transition of DMPG multilamellar vesicles as a function of GSit two endothermic events upon heating, a less ener- concentration. The symbols used are the same as for Fig. 2.getic pretransition near 14³C and a more energeticmain near 24³C. Again, under these conditions asubtransition is not observed. The pretransition anionic phospholipid. Speci¢cally, the presence of H Harises from the conversion of the LL gel to the PL increasing quantities of GS reduces the temperaturegel phase and the main transition or chain-melting and enthalpy of the pretransition, which appears to Hphase transition from the conversion of the PL to the be abolished entirely at lipid/peptide ratios of 500:1LK phase. For a more detailed discussion of the ther- or less. Moreover, at low concentrations of peptide,motropic phase behavior of DMPG and other mem- two endotherms are clearly present in the DSC heat-bers of the homologous series of linear saturated ing thermograms, a sharp, relatively energetic endo-PGs, see Zhang et al. . therm centered at 22³C and a broad, less energetic The interaction of GS with DMPG has a major endotherm centered near 27³C (Fig. 6B). As the GSe¡ect on the thermotropic phase behavior of this concentration increases, the initially higher temper- ature endotherm decreases in temperature and coop- erativity but becomes relatively more energetic as compared to the lower temperature, more coopera- tive endotherm (Fig. 6A). However, at higher peptide concentrations, both endotherms decrease in temper- ature (Fig. 7) but the less cooperative transition, which becomes increasingly more prominent, is now centered at a lower temperature than the more cooperative transition. Moreover, the total enthalpy associated with both transitions decreases with in- creasing peptide concentration (Fig. 8), particularly at higher peptide concentrations. If one assumes that the sharp endotherm is due to the chain-melting of DMPG domains relatively poor in peptide and the broad endotherm is due to the presence of domains of DMPG enriched in peptide, then it appears that atFig. 7. A plot of variations in the midpoint temperatures of themain phase transition of DMPG multilamellar vesicles as a lower concentrations GS stabilizes the gel state offunction of GS concentration. The symbols used are the same DMPG bilayers, as might be expected for a posi-as for Fig. 2. tively charged peptide. However, at higher GS con-
E.J. Prenner et al. / Biochimica et Biophysica Acta 1417 (1999) 211^223 219Fig. 9. DSC heating scans illustrating the e¡ect of repeated heating to high temperature on the thermotropic phase behavior ofDMPG multilamellar vesicles alone (A), or containing smaller (lipid/peptide molar ratio 250:1) (B) or larger (lipid/peptide molar ratioof 25:1) (C) quantities of GS. The ¢rst heating scan terminating at 95³C is shown as the solid line and the fourth such scan is shownas the dotted line.centrations, the presence of GS appears to destabilize much less pronounced, with the DMPE andboth the relatively peptide-rich and peptide-poor do- DMPC vesicles. However, GS-containing DMPGmains of DMPG vesicles. Note that the e¡ects of GS vesicles exhibit thermotropic phase behavior whichon the temperature, enthalpy and cooperativity of is much more similar to GS-containing DMPGthe gel to liquid-crystalline phase transition of vesicles not exposed to high temperatures, and inDMPG vesicles are much larger than those observed fact the presence of increasing quantities of GS re-in DMPE or DMPC vesicles at comparable peptide sults in a progressively greater inhibition of heat-cat-concentrations. alyzed chemical hydrolysis of DMPG vesicles (Fig. 9). These results suggest that the binding of GS to3.4. E¡ect of GS on the thermal stability of DMPG DMPG bilayers at high temperatures is relatively bilayers strong and that the phospholipid^peptide complex formed provides protection from heat-catalyzed The e¡ect of exposure to high temperature on the chemical degradation. Unfortunately, the intrinsi-thermotropic phase behavior of DMPG vesicles, re- cally strong resistance of DMPC and DMPE bilayersconstituted with and without GS, is illustrated in Fig. to hydrolysis upon exposure to high temperatures9. In the absence of peptide, the DMPG vesicles ex- precluded an analysis of the e¡ect of the presencehibit complex DSC endotherms after repeated heat- of GS on the chemically stability of these phospho-ing to 95³C and subsequent cooling. We have shown lipids.by TLC chromatographic analysis (data not pre-sented) that the complex thermotropic phase behav-ior of the DMPG vesicles observed after repeated 4. Discussionheating to high temperatures is due to the progres-sive chemical hydrolysis of these phospholipids and The major new ¢nding of this work is that thethat such hydrolysis is not observed, or at least is e¡ect of GS on the thermotropic phase behavior of
220 E.J. Prenner et al. / Biochimica et Biophysica Acta 1417 (1999) 211^223phospholipid bilayers varies markedly with both the bilayers, respectively. However, the shielding of thestructure and charge of the lipid polar headgroup. positively charged quaternary nitrogen of the cholineSpeci¢cally, the addition of GS has essentially no moiety in DMPC by the three methyl groups at-detectable e¡ect on the thermotropic phase behavior tached to it may result in a somewhat stronger (butof zwitterionic DMPE bilayers employed in this still rather weak) net electrostatic attractive interac-study, even at very high peptide levels and even after tion between GS and DMPC as compared to DMPEmultiple cycling through the gel to liquid-crystalline bilayers, perhaps accounting in part for the slightlyphase transition temperature. Only after exposure of stronger interactions of GS with the former lipidGS-containing DMPE vesicles to high temperatures observed in the present study. The fact that the pres-is a slight decrease in the phase transition temper- ence of two ornithine or other positively chargedature, enthalpy and cooperativity of this phospholip- amino acid residues at speci¢c positions in the GSid observed (see also ). Similarly, GS has only a molecule are required for antimicrobial activitysmall e¡ect on the thermotropic phase behavior of [2,4,11] suggests that electrostatic interactions be-zwitterionic DMPC bilayers, inducing a small de- tween this peptide and anionic phospholipids, ofcrease in the temperature and enthalpy, and a mod- the type observed here between GS and DMPG,erate decrease in the cooperativity, of the main phase are of functional signi¢cance in vivo.transition of this phospholipid. Moreover, under However, other di¡erences between the physicalthese conditions the pretransition of DMPC is abol- properties of DMPC and DMPE bilayers may alsoished at moderate GS concentrations. However, GS account for the apparently stronger interactions ofhas a much more pronounced e¡ect on the thermo- GS with DMPC and compared to DMPE vesicles.tropic phase behavior of DMPG, substantially reduc- For example, GS may interact preferentially withing the temperature, enthalpy and cooperativity of liquid-crystalline DMPC rather than DMPE bilayersthe main phase transition at higher peptide concen- because of the greater £uidity and decreased packingtrations and abolishing the pretransition at lower GS density of the former phospholipid as compared toconcentrations than are required for DMPC. These the latter (see ). This suggestion is supported byresults indicate that GS interacts much more strongly our present ¢ndings that GS interacts less stronglywith anionic rather than zwitterionic lipids, and, in with gel as compared to liquid-crystalline bilayers ofthe latter category, more strongly with DMPC than both of these zwitterionic phospholipids and morewith DMPE at comparable reduced temperatures in strongly with DMPC than with DMPE at compara-the liquid-crystalline state. ble reduced temperatures in the liquid-crystalline The stronger interaction of GS with anionic state, and by our previous observation that signi¢-DMPG bilayers in comparison with zwitterionic cant GS^DMPE interactions are only observed afterDMPC and DMPE bilayers is almost certainly due such exposure to high temperatures . This sugges-in part to the more favorable electrostatic interac- tion is also supported by the fact that the incorpo-tions between the peptide and lipid, and speci¢cally ration of cholesterol into DMPC vesicles attenuatesto the strong attractive interactions between the two the e¡ect of GS addition on both their thermotropicpositively charged ornithine residues of the peptide phase behavior and permeabilization (unpublishedand the negatively charged phosphate moieties of the observations from this laboratory). It is also possiblephosphorylglycerol headgroups of the former lipid. that the capacity for hydrogen bond formation be-Although attractive electrostatic interactions between tween GS and the polar headgroup and interfacialthe positively charged ornithine residues of GS and regions of the host phospholipid bilayer may alsothe negatively charged phosphate moieties of the be important in determining the strength and naturephosphorylcholine and phosphorylethanolamine po- of peptide^phospholipid interactions. Additionallar headgroups probably also exist, the overall studies on a wider variety of lipids employing variousstrength of the electrostatic interactions between pep- spectroscopic techniques will clearly be required totide and phospholipid will be markedly attenuated by elucidate the nature of GS^phospholipid interactionsthe presence of the positive charges on the choline in greater detail.and ethanolamine moieties of the DMPC and DMPE The comparative e¡ects of the incorporation of
E.J. Prenner et al. / Biochimica et Biophysica Acta 1417 (1999) 211^223 221GS on the thermotropic phase behavior of zwitter- ability of the anionic phospholipid vesicles withionic and anionic phospholipid bilayers can also be which they interact, while the incorporation ofused to deduce both the general location of this pep- Group II and Group III proteins increases vesicletide relative to the lipid bilayer and the nature of the permeability. The results of the present study indi-lipid^peptide interactions involved. For example, Pa- cate that overall GS behaves most like a Group IIpahadjopoulos et al.  and McElhaney  have protein. Thus it is probably located in the interfacialproposed that many membrane-associated proteins region of phospholipid bilayers, where it interactscan be classi¢ed into one of three groups with regard primarily with the polar headgroup and glycerol-to their interactions with phospholipid bilayers. backbone region of the phospholipid molecules byGroup I proteins are typically positively charged, electrostatic and hydrogen-bonding interactions, itswater soluble, peripheral membrane proteins that in- modest perturbation of hydrocarbon chain packingteract much more strongly with anionic than with in the gel state being a secondary e¡ect of its inter-zwitterionic lipids. The interactions of such proteins actions primarily with the polar region of the phos-with anionic phospholipid bilayers typically increases pholipid molecules. We also suggest, however, thatboth the temperature and enthalpy of the main phase the hydrophobic leucine and valine residues of thetransition temperature while decreasing its coopera- GS molecule may have limited interactions with thetivity only slightly. Group I proteins are localized on phospholipid hydrocarbon chains near the interfacialthe bilayer surface where they interact only with the region of the bilayer, and the somewhat amphiphilicphospholipid polar headgroups primarily by electro- phenylalanine residues may interact with both thestatic interactions. Group II proteins are also typi- glycerol backbone and hydrocarbon chains at thecally positively charged at neutral pH but are some- bilayer polar/nonpolar interface. However, the exactwhat less water soluble and also interact more location of GS molecules in liquid-crystalline lipidstrongly with anionic than with zwitterionic phos- bilayers may vary somewhat with the charge andpholipid bilayers. The interactions of these proteins structure of the lipid polar headgroup (unpublishedwith anionic phospholipid bilayers usually decreases data from this laboratory). An interfacial location ofthe temperature, enthalpy and cooperativity of the the GS molecules would also explain the protectionmain phase transition moderately, at least at rela- a¡orded to DMPG bilayers from the thermally in-tively high protein concentrations. Group II proteins duced hydrolysis of the fatty acyl ester linkages ofare localized at the bilayer interface where they in- these phospholipid molecules by this peptide.teract primarily with the polar headgroups and glyc- We believe that our suggestion for the generallyerol backbone region of the phospholipid molecules interfacial localization of the GS molecule in lipidby both electrostatic and hydrogen bonding interac- bilayers is actually compatible with most of thetions, although some hydrophobic interactions with data on GS^PC interactions previously published.the region of the hydrocarbon chains near the bilayer For example, the failure of Pache et al.  to detectinterface also occurs. Finally, Group III proteins any e¡ect of GS on the thermotropic phase behaviorhave a range of charges but are water-insoluble, in- of DPPC bilayers was almost certainly due to thetegral membrane proteins that interact equally well insensitivity of the calorimeter employed, since sub-with anionic and zwitterionic phospholipid bilayers. sequent workers [22,23] including ourselves (, andThe e¡ect of the incorporation of such proteins into the present work) have clearly shown a modest butphospholipid bilayers is usually to reduce the temper- signi¢cant e¡ect of GS incorporation on the pretran-ature only slightly but to decrease the enthalpy and sition and main transition of DPPC and DMPC bi-cooperativity of the main phase transition markedly. layers. Thus, this piece of calorimetric evidence forGroup III proteins penetrate into or through phos- an exclusively surface location is not valid. More-pholipid bilayers and interact with the phospholipid over, a careful inspection of the ESR spectrum pub-hydrocarbon chains by hydrophobic and van der lished by Pache et al.  reveals a small but signi¢-Waals interactions, as well as with the phospholipid cant degree of immobilization of the 12-nitroxylpolar headgroups. The incorporation of Group I stearate spin-labeled PC molecule in the presence ofproteins also usually has little e¡ect on the perme- admittedly high levels of GS, indicating some e¡ect
222 E.J. Prenner et al. / Biochimica et Biophysica Acta 1417 (1999) 211^223on the lipid hydrocarbon chains. Moreover, the DSC GS on phospholipid phase preference at tempera-and lateral di¡usion and lateral aggregation meas- tures above the main phase transition temperatureurements of Wu et al. , which suggested penetra- , we found that the major e¡ect of GS on lamel-tion of the peptide into the bilayer, utilized NBD- lar/nonlamellar phase preference was on DMPE bi-GS, a £uorescent derivative in which one or both layers, with a weaker e¡ect on DMPG and no e¡ectof the ornithine residues in the GS molecules were at all on DMPC bilayers. Moreover, in general GSchemically modi¢ed. Such a modi¢cation will reduce induced inverted cubic phases more readily in zwit-the positive charge distribution on the GS molecule terionic as opposed to anionic phospholipids. In con-as well as altering its amphiphilicity, hydrophilicity trast, our present DSC studies on the e¡ect of GS onand size. Thus the interactions of NBD-GS with PC the main phase transition of these phospholipids in-bilayer may well not accurately re£ect the behavior dicate that GS interacts more strongly with theof the underivatized peptide. Moreover, it seems to anionic DMPG vesicles than with the zwitterionicus that the sensitivity of the lateral di¡usion coe¤- DMPC and DMPE vesicles, and more stronglycient of NBD-GS to the phase transition of the PC with the former than the latter. Moreover, both ofbilayer, and the lateral aggregation of this derivat- these types of e¡ects of GS on phospholipid thermo-ized peptide induced by the addition of cholesterol, tropic phase behavior may be relevant to the actionwould both be expected from a peptide molecule of GS on natural membranes. For example, anioniclocalized at the bilayer interface as well as one lipids such as PG may increase the rate or extent ofpenetrating partially or fully into the lipid bilayer GS binding to the bilayer surface, thus facilitatinghydrocarbon core. Moreover, the 2 H-NMR results the interaction of this peptide with zwitterionic lipidsof Zidovetzki et al. , which show that GS incor- such as PE or nonionic lipids such as monoglycosylporation slightly reduces overall hydrocarbon chain diacylglycerols, which can in turn form localized re-disorder in £uid DMPC bilayers and decreases signal gions of nonlamellar structure which disrupt the lipidintensity primarily in the methylene segments closest bilayer of the target membrane. We are thus continu-to the glycerol backbone, are fully compatible with ing to study the e¡ects of GS on a variety of lipidthe postulated interfacial location of this peptide. systems using additional physical techniques.The only information which does not seem compat-ible with a generally interfacial location of the GSmolecule in liquid-crystalline PC bilayers is the re- Acknowledgementsport of Datema et al. , which suggested that GSmolecular motion is insensitive to the phase state of This work was supported by operating and majorDPPC bilayers. However, this result is at any rate at equipment grants from the Medical Research Coun-variance with the report of Wu et al. , who found cil of Canada (R.N.M. and R.S.H.), by major equip-that the lateral di¡usion coe¤cient of at least NBD- ment grants from the Alberta Heritage FoundationGS is markedly altered by the phase state of DMPC for Medical Research (R.N.M. and R.S.H.), and bybilayers. Clearly, additional work will be required to the Protein Engineering Network of Centres ofresolve this latter discrepancy, but in our view the Excellence (R.S.H.). E.J.P. and L.H.K. were sup-preponderance of evidence supports a location of ported by postdoctoral fellowships from the AlbertaGS in £uid lipid bilayers in which the primary inter- Heritage Foundation for Medical Research and theactions are between the peptide molecule and the Protein Engineering Network of Centres of Excel-polar headgroups and the interfacial regions of the lence, respectively.phospholipid molecule. In closing, we note the importance of applying anumber of physical techniques to studying various Referencesaspects of GS^phospholipid interactions in order toarrive at a more complete understanding of these  G.F. Gause, M.G. Brazhnikova, Nature 154 (1944) 703.systems. Thus, in our previous 31 P-NMR spectro-  N. Izumiya, T. Kato, H. Aoyaga, M. Waki, M. Kondo,scopic and X-ray di¡raction studies of the e¡ect of Synthetic Aspects of Biologically Active Cyclic Peptides:
E.J. Prenner et al. / Biochimica et Biophysica Acta 1417 (1999) 211^223 223 Gramicidin S and Tyrocidines, Halsted Press, New York,  T. Katsu, M. Kuroko, T. Morikawa, K. Sanchicka, Y. Fu- 1979. jita, H. Yamamura, M. Uda, Biochim. Biophys. Acta 983  L.H. Kondejewski, S.W. Farmer, D. Wishart, C.M. Kay, (1989) 135^141. R.E.W. Hancock, R.S. Hodges, Int. J. Pept. Protein Res.  S.H. Portlock, M.J. Clague, R.J. Cherry, Biochim. Biophys. 47 (1996) 460^466. Acta 1030 (1990) 1^10.  L.H. Kondejewski, S.W. Farmer, D. Wishart, C.M. Kay,  J.A. Midez, R.L. Hopfer, G. Lopez-Berestein, R.T. Mehta, R.E.W. Hancock, R.S. Hodges, J. Biol. Chem. 271 (1996) Antimicrob. Agents Chemother. 33 (1989) 152^155. 25261^25268.  H.P. Lambert, F.W. OGrady, Antibiotic and Chemother-  D.C. Hodgkin, B.M. Oughton, Biochem. J. 65 (1957) 752^ apy, Churchill Livingstone, Edinburgh, 1992, pp. 232^233. 756.  T. Katsu, S. Nakao, S. Iwanaga, Biol. Pharm. Bull. 16  Y.A. Ovchinnikov, T. Ivanov, Tetrahedron 31 (1975) 2177^ (1993) 178^178. 2209.  W. Pache, D. Chapman, R. Hillaby, Biochim. Biophys. Acta  S.E. Hull, R. Karlsson, P. Main, M.M. Woolfson, Nature 255 (1972) 358^364. 275 (1978) 206^207.  K.P. Datema, K.P. Pauls, M. Bloom, Biochemistry 25 (1986)  S. Rackovsky, H.A. Scheraga, Proc. Natl. Acad. Sci. U.S.A. 3796^3803. 77 (1980) 6965^6967.  E.S. Wu, K. Jacobson, F. Szoka, A. Portis, Biochemistry 17  M. Tamaki, M. Takimoto, S. Nozaki, I. Muramatsu, (1978) 5543^5550. J. Chrom. Biomed. Appl. 413 (1987) 287^292.  R. Zidovetzki, U. Banerjee, D.W. Harrington, S.I. Chan, T. Katayama, K. Nakao, M. Akamatsu, T. Ueno, T. Fumi- Biochemistry 27 (1988) 5686^5692. ta, J. Pharm. Sci. 83 (1994) 1357^1362.  E.J. Prenner, R.N.A.H. Lewis, K.C. Newman, S.M. Gruner, Y.A. Ovchinnikov, V.T. Ivanov, in: H. Neurath, R.L. Hill L.H. Kondejewski, R.S. Hodges, R.N. McElhaney, Bio- (Eds.), The Proteins, vol. V, Academic Press, New York, chemistry 36 (1996) 7906^7916. 1982, pp. 391^398.  R.N.A.H. Lewis, N. Mak, R.N. McElhaney, Biochemistry R.E.W. Hancock, P.G.W. Wong, Antimicrob. Agents 26 (1987) 6118^6126. Chemother. 26 (1984) 48^52.  R.N.A.H. Lewis, R.N. McElhaney, Biophys. J. 64 (1993) T. Katsu, H. Kobayashi, Y. Fujita, Biochim. Biophys. Acta 1081^1096. 860 (1986) 608^619.  Y.P. Zhang, R.N.A.H. Lewis, R.N. McElhaney, Biophys. J. T. Katsu, H. Kobayashi, T. Hirota, Y. Fujita, K. Sato, U. 72 (1997) 779^793. Nagai, Biochim. Biophys. Acta 899 (1987) 159^170.  D. Papahadjoupolos, M. Moscarello, E.H. Eylar, T. Issac, T. Katsu, C. Ninomiya, M. Kuroko, H. Kobayashi, T. Biochim. Biophys. Acta 401 (1975) 317^335. Hirota, Y. Fumita, Biochim. Biophys. Acta 939, (1988)  R.N. McElhaney, Biochim. Biophys. Acta 864 (1986) 361^ (1988) 57^63. 421.