effect of point mutations on the secondary structure and membrane interaction of antimicrobial...

Published: February 22, 2011

r 2011 American Chemical Society 2371 dx.doi.org/10.1021/jp108343g | J. Phys. Chem. B 2011, 115, 2371–2379

ARTICLE

pubs.acs.org/JPCB

Effect of Point Mutations on the Secondary Structure and MembraneInteraction of Antimicrobial Peptide AnoplinAmy Won, Stahs Pripotnev, Annamaria Ruscito, and Anatoli Ianoul*

Department of Chemistry, Carleton University, 1125 Colonel By Drive, Ottawa, ON K1S 5B6, Canada

’ INTRODUCTION

Antimicrobial peptides (AMPs) are relatively short, cationicamphipathic molecules, constituting a part of the innate defenseof plants and various organisms and exhibiting a broad range ofantibacterial, antifungal, and antitumor properties.1-3 Because ofgenerally nonspecific mechanisms of action, they are consideredas promising candidates to help battle infectious diseases affectedby antibiotic resistance. Structurally, AMPs can be roughlydivided into R-helical and β-sheet/turn peptides often stabilizedby disulfide bridges. Efforts to improve peptides selective toxicityhave led to numerous studies on AMPs' structure/functionrelationship, particularly forR-helical peptides.1-3 Modificationsof peptide primary structure (amino acid substitutions) involv-ing changes in the character of the side chain (charge, hydro-phobicity, and surface area) affect other peptide propertiessuch as hydrophobic moment, hydrophobicity, and size ofthe polar/hydrophobic domains, as well as charge and chargedistribution.1-4 Effect of variation in these properties on theactivity of AMPs has previously been extensively studied using anumber of natural and model peptides. It is well established thatelectrostatic interactions between the cationic peptides andmostly anionic bacterial cell membranes are responsible forinitial interactions between the molecules and therefore forselective toxicity. Hydrophobicity, amphipathicity, and helicityappear to affect hemolytic activity more significantly than anti-bacterial activity. At the same time, the effect of those parameterson antibacterial activity is not direct and is superimposed withpeptide/lipid electrostatic interactions.1-4 The modification of

one parameter (for example, charge), however, often leads to thechange in the other (such as hydrophobic moment), thus makingdirect correlation between certain physicochemical property andpeptide activity challenging.1-4 In addition, structural studies ofAMPs rely heavily on the circular dichroism (CD) techniquewhich is known to have limited sensitivity for short helicalsequences.5 Finally, the hydrophobic moment of a peptide isusually calculated assuming 100% helicity, which cannot beassumed for short peptides.2,6,7 Therefore, the structure/activityrelationship of short antimicrobial peptides requires alternativeapproaches. In this respect UV resonance Raman spectroscopyoffers significant advantage as a technique that allows probingconformation of short peptide sequences and selective unfoldingof the peptide interior versus terminal parts.8

Short AMPs attract attention due to the potentially lower costof their production. A number of such short antimicrobialpeptides have recently been isolated.9-14 Among these, anoplinis one of the shortest.15 It has broad spectrum activity againstboth gram-positive and gram-negative bacteria and exhibits lowhemolytic activity toward human erythrocytes and thus has agreat potential for further development.15 Correlation betweenthe structure and the activity of anoplin was previouslystudied15,16 and resulted in two potent derivatives with improvedantibacterial and weak hemolytic activity.16 However, detailed

Received: September 1, 2010Revised: January 10, 2011

ABSTRACT: Anoplin (GLLKRIKTLL-NH2) is the smallest linearR-helical antimicrobial peptide found naturally to date. Antibacterialand hemolytic properties of anoplin depend strongly on physico-chemical properties. Two anoplin derivatives, anoplin-8K (Ano8K,GLLKTIKKLL-NH2) and anoplin-1K5 V8K (Ano1K5 V8K,KLLKVIKLL-NH2), were found to have increased bacterial andlow hemolytic activity. In the present work physicochemical proper-ties of these three peptides were studied by UV resonance Raman(UVRR) spectroscopy, Langmuir-Blodgett monolayer technique,and carboxyfluorescein (CF) leakage assay. UVRR data indicatedthat all three peptides adopt predominantly unordered conformation in aqueous buffer solution. In membrane-mimickingtrifluoroethanol, the R-helical content increases for all three peptides with Ano1K5 V8K having the highest R-helix percentage,followed by Ano8K and anoplin. Critical micelle concentrations were found to be similar for all three peptides, and the saturationpressure decreases in the sequence Ano1K5 V8K, anoplin, Ano8K. Critical pressure of insertion was found to be greater for anioniclipid monolayer DPPG than for zwitterionic lipid DPPC indicating preferential adsorption of all three peptides to DPPG. Finally,membrane lytic activities of all three peptides toward various model lipid vesicles were compared through CF leakage assay. Overallthe data indicate that antimicrobial activity of anoplin increases with charge, whereas membrane lytic activity correlates withpeptides helicity and amphipathicity.

2372 dx.doi.org/10.1021/jp108343g |J. Phys. Chem. B 2011, 115, 2371–2379

The Journal of Physical Chemistry B ARTICLE

structural studies of these derivatives were unsuccessful, likelydue to the limitations of CD. Therefore, in this work we appliedUV resonance Raman spectroscopy in combination with Lang-muir monolayer technique and vesicle leakage assay to correlatethe activities of three anoplin derivatives with their secondarystructure and surface activity.

’MATERIALS AND METHODS

Chemicals. 1,2-Dipalmitoyl-sn-glycero-3-phosphocholine (DPPC),1,2-dipalmitoyl-sn-glycero-3-[phospho-rac-(1-glycerol)] (DPPG),1,2-dioleoyl-sn-glycero-3-phosphoethanolamine (DOPE), 1,2-dio-leoyl-sn-glycero-3-phospho-(10-rac-glycerol) (sodium salt)(DOPG), 1,10,2,20-tetramyristoyl cardiolipin (sodium salt) (CL),1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC), and choles-terol (Chol, bovine wool, >98%) were purchased from AvantiPolar Lipids Inc. Anoplin-NH2 (>95% purity), Anoplin-8K-NH2 (98.9% purity), and Anoplin-1K5 V8K-NH2 (98.9%purity) were synthesized by GenScript Corporation. Phosphatebuffer saline (PBS, 0.01 M, 138 mM NaCl, 2.7 mM KCl,pH 7.4), sodium perchlorate (98%), 2,2,2-trifluoroethanol(TFE, >99.0%), and ascorbic acid (>99.0%) were purchasedfrom Sigma-Aldrich. Carboxyfluorescein (6-carboxyfluorescein)was purchased from ACROS Organic. Ammonium molybdate,ethylenediaminetetraacetic acid (EDTA), and triton X-100 werepurchased from Bioshop.Methods. UV Raman Spectroscopy. Each peptide stock solu-

tion was dissolved to 5 mg/mL in pH 7.4 aqueous 0.01 M PBSwith 0.1 MNaClO4 as an internal standard and frozen at-20 �Cuntil used. Prior to immediate UVRR spectra collection, peptidesamples were diluted to 2 mg/mL with pH 7.4 aqueous 0.01 MPBS with 0.1 M NaClO4 as an internal standard or diluted to 2mg/mL in 50% (v/v) TFE. The volume difference was made upwith pH 7.4 aqueous 0.01 M PBS with 0.1 M NaClO4.All experiments were carried out using a Coherent 90C FreD

argon ion laser with β barium borate intracavity second harmonicgeneration, producing continuous wave light at 229 nm with anominal power of 8 mW. The laser was focused on a 3 mm � 3mm quartz cuvette with the temperature controlled by a NeslabRTE-7 circulating water bath with magnetic stirring (Starna).The Raman signal was collected in a 135� backscatteringgeometry using a JobinYvon 1000 mm single optical spectro-meter and Spec-10 1340 � 400 B/LN CCD camera with UVcoating (Roper Scientific). Multiple spectra were collected inorder to assess sample integrity, and the sum was used for thespectra in the figures. Total accumulation time was 45 min foreach sample in PBS and TFE. Solvent contributions to samplespectra were minimized by subtraction of pure solvent withinternal standard spectra after correcting all spectra to the samestandard intensity value for accuracy. Cyclohexane was used forRaman shift calibration.All spectra were processed and the peak fitting was performed

using the GRAMS/AI software (Galactic Industries, Inc.). Fourpeaks were chosen in the amide II17 and five in the amide Iregions18 to achieve the best fit to the original trace whileminimizing χ2 values. Using an alternative number of peaksincreased the χ2 undesirably or did not provide as successful of afit. A Gaussian-Lorentzian mix was employed for the peakcharacteristic shape with a linear baseline and a maximumbandwidth at half height (BWHH) restriction of 25 cm-1.Monolayer Experiments. Stock solutions of the peptides

(1.65 mg/mL) were prepared in PBS. PBS (∼20 mL) was added

to a round poly(tetrafluoroethylene) dish along with a magneticstirring bar. A small amount of peptide solution was injected intothe trough, and the surface pressure was monitored untilequilibrium was attained. The addition of peptide solution intothe dish was continued until the surface pressure ceased fromincreasing, at which point the critical micelle concentration(cmc) is achieved.To determine critical pressure of insertion (CPI) a lipid

monolayer was made by spreading lipid solution (1 or 0.1 mg/mL solutions of DPPC and DPPG in chloroform or chloroform/methanol (3:1, v/v) mixture) at the air-water interface of thedish until a constant surface pressure was obtained. A smallamount of peptide solution (close to the cmc of a peptide) wasinjected into the subphase beneath the monolayer, and thesurface pressure was monitored until equilibrium. The processwas repeated with different initial surface pressures in order toobtain the CPI of the peptide into the lipid monolayer.Carboxyfluorescein Leakage Assay. Six different model cell

membranes were used: DPPC, DPPG, mammalian model(DOPC/DOPE/Chol 2/1/1.5 mol %),19 Escherichia coli model(E. Coli, DOPE/DOPG 80/20 mol %), Staphylococcus aureusmodel (S. aureus, DOPG/CL 55/45 mol %), and Bacillus subtilismodel (B. subtilis, DOPE/DOPG/CL 12/84/4 mol %).20 Largeunilamellar vesicles were prepared by dissolving the appropriateamount of lipid in chloroform, drying the solvent under a streamof nitrogen, and keeping the sample under vacuum for at least 24 hto ensure complete solvent removal. The lipid filmswere hydratedfor 30 min (at 55 �C for DPPC and DPPG, and at 30 �C formammalian and bacterial model membranes) in leakage buffer(10 mM Tris-HCl, 150 mMNaCl, 1 mM EDTA pH 7.45 in 18.2MΩ Milli-Q water) containing 70 mM carboxyfluorescein toobtain the final lipid concentration of 1 mg/mL. Lipid suspen-sions were then vigorously stirred. Five freeze/thaw cycles wereperformed to maximize carboxyfluorescein encapsulation. Thelipid suspension was further extruded through 200 nm polycar-bonate membrane (Nuclepore Track-Etch membrane,Whatman) 31 times (at 55 �C for DPPC and DPPG, and at30 �C for bacterial model membranes). Free carboxyfluoresceinwas separated from the encapsulated with a Sephadex G-50 sizeexclusion columnusing leakage buffer for equilibrium and elusion.Leakage experiments were carried out using 2 mL of carboxy-

fluorescein-containing vesicles diluted 10 times with the leakagebuffer on a Varian Cary Eclipse spectrofluorimeter. Measure-ments were carried out with excitation and emission wavelengthsdetermined for each experiment (490-500 nm and 518-521nm, excitation and emission, respectively). Excitation and emis-sion slits were 2.5 nm, integration time was 1.5 s, and photo-multiplier tube voltage was 540 and 480 V for saturated lipids(DPPC and DPPG) and model membranes, respectively. Thebaseline fluorescence (F0) was monitored for 30 s before theaddition of the peptide. After the peptide was added, thefluorescence signal intensity was monitored for approximately10 min or until no further changes occurred. The final fluores-cence signal intensity (F) was then measured. To determine themaximum fluorescence signal (FM) corresponding to completedisruption of the vesicles, 20 mL of 10% triton X-100 (Bioshop)were added to the mixture at the end of the experiment andfluorescence intensity increase was monitored for 5 min. Theleakage fraction was calculated as % leakage = [(F - F0) �100%]/(FM - F0).The concentration of lipid phosphorus was measured by

phosphate assay.21 Vesicles with carboxyfluorescein (200 μL)



were mixed with 0.45 mL of H2SO4 and heated with a blockheater for 25min at 200-215 �C. After cooling for 5min, 150 μLof 30%H2O2 was added to each test tube followed by heating foran additional 30 min. Deionized water (3.9 mL) was added aftercooling to room temperature followed by 0.5 mL of 2.5%ammonium molybdate and 0.5 mL of 10% ascorbic acid. Eachtest tube was vigorously stirred after each addition. The solutionwas incubated in a water bath at 100 �C for 7 min. Absorption ofthe signal at 820 nm was measured. The phosphate standardsolutions were prepared using sodium phosphate monobasic upto 0.65 mM in 18.2 MΩ Milli-Q water.

’RESULTS AND DISCUSSION

Single and Multiple Amino Acid Substitutions ChangeSurface/Membrane Activity of Anoplin. Point mutation ofnatural antimicrobial peptides is a widely used technique toimprove their potential therapeutic properties.1-4 This approachhelps identify key amino acids in the primary sequence re-placement of which can increase bactericidal activity and/ordecrease cytotoxic effects. Upon these modifications certainphysicochemical properties of peptides change, such as second-ary structure, charge, hydrophobicity, and hydrophobic moment,leading to alteration of biological activity. To sufficiently exhibitantimicrobial activity and to transport efficiently, peptides needto be soluble in aqueous phase, which requires low hydropho-bicity. On the other hand, to interact with the hydrophobic lipidtails to disturb the membrane structure and permeability, highhydrophobicity is needed.22 Therefore, hydrophobicity andamphipathicity are often found as factors determining hemolyticand antibacterial activity of the peptides.1-3

Having only 10 amino acids, anoplin is one of the shortestnatural antimicrobial peptides.15 Recently, detailed structure/function studies of 37 analogues of this peptide helped identifyseveral potent mutants with improved antibacterial properties.16

Among these, Ano8K (replacement of threonine with lysine atposition 8) and Ano1K5 V8K (replacement of glycine withlysine, arginine with valine, and threonine with lysine at positions1, 5, and 8, respectively) retained low or no hemolytic activity

while having increased lytic activity toward both gram-positiveand gram-negative bacteria (Table 1).16 Additional positivecharge and amphipathicity change were suggested to be themain factors for increased selectivity of Ano8K and Ano1K5 V8Ktoward bacteria, respectively. Amphipathicity of a peptide isdetermined by the value of mean hydrophobic moment.1-3 Ifa peptide adopts R-helical conformation, as, for example, wassuggested for anoplin, helical wheel or helical net diagrams canprovide straightforward examination of the peptide’s amphi-pathicity (Figure 1). Calculations of hydrophobic moment showthat Ano8K has the highest amphipathicity followed by Ano1K5V8K and finally wild type anoplin assuming the peptides fold into100% helix (Table 1). The former assumption is, however, nevercorrect, and for short peptides like anoplin the highest R-helicalcontent can only be 50-60% due to the terminal effects.6,7

Therefore, in this work several physicochemical techniques wereused in order to determine relative helicities and amphipathicitiesof anoplin derivatives and correlate those with peptides biologicalactivities.Monolayers at the air-water interface are commonly used to

mimic cell membranes while controlling parameters such asmolecule lateral packing and curvature.22,23 At the interface,antimicrobial peptides adopt a conformation similar to that in abiological membrane,22,24 thus becoming amphipathic. At criticalmicelle concentration, the air/water interface is so completelysaturated with surfactant, or peptide in this case, that micellesform upon any further peptide addition.24 The lower the cmc, themore surface-active the peptide is, thus reflecting peptideamphipathicity.24 By measuring the surface tension as a functionof peptide concentration, we were able to determine the satura-tion surface pressure for anoplin and its two derivatives(Figure 2). The values obtained are 18.1, 19.4, and 22.2 mN/m, for Ano8K, anoplin, and Ano1K5 V8K, respectively. Thesequence correlates with overall hydrophobicity of the peptidesrather than amphipathicity (Table 1). The concentration atwhich the saturation occurs (cmc) was found to be about 1.2μg/mL for all three peptides, irrespective of the calculatedhydrophobic moment. Thus, according to cmc data there is noconsiderable difference between the peptides in terms of

Table 1. Properties of Anoplin and Derivatives

MIC/μg/mLa CPIc/mN/m

peptide sequence Qa ÆHæa μHa S. aureus E. Coli EC50/μg/mL πb/mN/m DPPC DPPG

anoplin GLLKRIKLL þ3 -0.113 0.366 13 26 - 19.4 28.5 54.0

Ano8K GLLKRIKKLL þ4 -0.205 0.455 5 10 - 18.1 29.5 68.5

Ano1K5 V8K KLLKVIKKLL þ4 -0.101 0.438 9 6 231 22.2 27.5 85.0aCharge Q, mean hydrophobicity ÆHæ, hydrophobic moment μH, minimum inhibitory concentration (MIC), and median efficacious concentration(EC50) are from ref 22. b Presure π at critical micellar concentration (cmc). cCritial pressure of insertion (CPI).

Figure 1. Helical wheel diagram of anoplin, Ano8K, and Ano1K5 V8K. Eisenberg consensus hydrophobicity scale was used. (Eisenberg, D.; Weiss, R.M.; Terwilliger, T.C.; Wilcox, W. Faraday Symp. Chem. Soc. 1982, 17, 109-120).



amphipathicity. This result disagrees with the calculations(Table 1) and may partly reflect the fact that the peptides donot completely fold at the interface.Peptide affinity to a cell membrane can be determined by

measuring the critical pressure of insertion.24 If a monolayer ofphospholipid is prepared at the air/water interface at a certainsurface pressure and a peptide is injected beneath the monolayer,lipid film surface pressure would increase as peptide moleculesbind to and insert into the lipid monolayer at constant surfacearea.24 The degree of pressure increase depends on the lipid type,initial film surface pressure, peptide surface activity, and con-centration used. Critical pressure of insertion—essentially thelipid monolayer surface pressure that prevents the peptide frominserting into the monolayer—can be determined by extrapolat-ing the initial pressure (πi) as a function of pressure increase(Δπ), where pressure increase is equal to zero.24 All threepeptides were found to possess similar affinity to zwitterionicphospholipid monolayer DPPC (Figure 3A). CPI values forAno8K, anoplin, and Ano1K5 V8K were determined to bebetween 27.5 and 29.5 mN/m with little difference betweenthe peptides (Table 1). Since the lateral packing of biologicalmembrane is believed to be in the range of 25-35mN/m,22-25 itis likely that all three peptides have limited effect on zwitterionicmembrane and thus should have little toxic effect on mammaliancells. This result is consistent with the results of biologicalassays.15,16

When a monolayer of anionic DPPG is used, CPI and there-fore affinity of all three peptide are considerably higher(Figure 3B). We determined the CPI for Ano1K5 V8K, Ano8K,and anoplin to be 85.0, 68.5 and 54.0 mN/m, respectively. Thesevalues are substantially greater than with DPPC indicating theimportance of electrostatic interaction for all three peptides intheir recognition and binding to cell membranes. This is alsoconsistent with the fact that the wild type anoplin has loweroverall charge and also the lowest CPI in DPPG. At the sametime, since CPI for Ano1K5 V8K is higher than for Ano8Kdespite the same charge, overall hydrophobicity and amphipathi-city of the peptides are likely to have contributed. The datasuggest that at the interface Ano1K5 V8K is more amphipathic.This level of amphipathicity can be achieved if the peptide adoptsa more helical conformation than Ano8K.To determine how the kinetics of peptide binding and

association with the cell membrane changes with peptide phys-icochemical properties, a time course of the surface pressure

increase was monitored after peptide injection into the subphaseunder the phospholipid monolayers with surface pressure kept inthe range of biological membrane pressure (Figure 4). Figure 4data clearly show the difference between zwitterionic and anionicmembranes. All three peptides inserted into DPPC monolayersin a substantially smaller degree than into DPPG monolayers.There was little change of the surface pressure observed forDPPC monolayers. In the case of inserting into DPPG mono-layers, both Ano8K and Ano1K5 V8K increased the initial filmpressure by approximately 10 mN/m whereas the anoplinincrease was ∼8 mN/m. There was no visible difference in thetime of insertion, and already after∼200 s the process of peptideinsertion was completed for all three peptides.UVRR Spectroscopy Probes Conformational Difference of

Anoplin Derivatives. There is a certain correlation between theconformation and amphipathicity of antimicrobial peptides, andconformational studies can often help deduce the mechanism ofaction or toxicity.1-3 However, for short peptides such studiesbecome more challenging.1-3 For example, CD spectroscopyresults demonstrated that anoplin adopts unordered structure inaqueous buffer and a favors R-helical conformation in a mem-brane-mimicking environment.9,15,16,26 However, results fromvarious studies were conflicting. The percentage of R-helicalcontent was found to be between 24 and 77%.9,15,16,26 Thediscrepancy in the calculated secondary structure is likely due topoor sensitivity of CD spectroscopy to the conformation of shortpeptides.5 Besides, CD spectroscopy was unable to detect anydifference in the secondary structure of anoplin, Ano8K, andAno1K5 V8K.16 Therefore, UV resonance Raman spectroscopy

Figure 3. Insertion of anoplin (circles and solid line), Ano8K (squaresand dotted line), and Ano1K5 V8K (triangles and dashed line) intoDPPC (A) and DPPG (B) monolayers. Extrapolation at zero change inpressure (Δπ = 0) gives the critical pressure of insertion.

Figure 2. Peptide-induced variations of the surface pressure as afunction of peptide concentration in the subphase for anoplin(circles), Ano8K (squares), and Ano1K5 V8K (triangles).



was used as a more sensitive method for discriminating betweendifferent conformations of short peptides.UV resonance Raman spectroscopy is a powerful technique to

study the structure/function relationships of proteins. Threeamide bands, amide I, II, and III, present in the Raman spectraof proteins are used to extract molecular structural infor-mation.17,27-29 The position and relative intensities of thesebands are particularily sensitive to the peptide’s conformation.Conformational information can be extracted by performingspectral deconvolution using pure secondary structure referencespectra (for example, R-helix, β-sheet, etc.) to identify thequantity of that particular secondary structure.30-34 This ap-proach is useful for long peptides/proteins, where contributionfrom the terminal residues is negligible. Alternatively, a peak-fitting procedure can be performed where several individualbands are placed into a complex amide band region.17,18 Suchan approach is more appropriate for short peptides like anoplin.Raman spectroscopic studies of antimicrobial peptides are

usually concerned with the structural changes in aqueous versusmembrane-mimicking environment,5,8,35 such as trifluoroe-thanol,36 in order to elucidate the conformational change anddynamics of peptide/membrane binding. Trifluoroethanol aidspeptide folding by creating a hydrophobic environment andremoving solvent hydrogen bonding.36 In this work, TFE wasused to determine the change that anoplin UVRR spectraundergo when the peptides fold. Yet, due to interference fromTFE Raman bands, the amide III region could not be reliablyused to determine the peptide secondary structure change(Figures 5-7). Therefore, in the present work the amide I band

was used to extract the structural information of anoplin and itsderivatives. Analysis was facilitated by the lack of aromatic aminoacids in the primary structure of anoplin derivatives.The amide I region of a peptide Raman spectrum can

occasionally be properly fit with three peaks located at 1650-1656, 1664-1670, and 1680 cm-1 corresponding to R-helix, β-sheet, and unordered structure, respectively.37,38 However, wecould not satisfactorily model the anoplin amide I region usingonly three peaks. Instead, five peaks were selected in the amide Iregion, similar to a study of a number of short peptides with KLrepeating sequences.18 Figures 5-7 show results of peak fitting,and Table 2 shows the optimized locations of the bands for allthree derivatives. The consistency of the locations of the peaksacross three different peptides in two different solvents supportsthe choice of this fitting method. In comparison to the 1615,1635, 1651, 1664, and 1679 cm-1 peaks found by Guiffo-Sohet al.,18 the average peak locations for all three peptides weredetermined to be 1612, 1631, 1650, 1668, and 1687 cm-1

(Table 2). With the exception of the first (1612 cm-1), thepeaks were assigned to R- helix, β- sheet, β-strand, and un-ordered structural components, respectively.

Figure 5. Peak fitting of UV resonance Raman spectra of anoplin in PBS(A) and 50% TFE (B). Original spectrum (thick) is shown on top of thefitted trace (dashed) with individual fitted peaks (thin) and baseline(flat) below. TFE contributions are marked by asterisks.

Figure 6. Peak fitting of UV resonance Raman spectra of Ano8K in PBS(A) and 50% TFE (B). Original spectrum (thick) is shown on top of thefitted trace (dashed) with individual fitted peaks (thin) and baseline(flat) below. TFE contributions are marked by asterisks.

Figure 4. Relative increase of the pressure of DPPC (A) andDPPG (B)monolayers after inserting anoplin (solid line), Ano8K (dotted line), andAno1K5 V8K (dashed line).



Visually, anoplin and its two derivatives show a clear shoulderappearing to the left of the main amide I band in TFE(Figures 5-7). These results are very similar to previous experi-ments concerning anoplin, deamidated anoplin, and fully D-ami-no acid substituted anoplin.35 Peak fitting was performed to get abetter idea of what secondary structure the anoplin derivativesare adopting. Table 3 summarizes these results reported as eachband’s percentage area contribution to the total amide I area (notthe total percent of a particular secondary structure).According to the results of peak fitting, the proportion of the

amide I peak corresponding to the R-helical conformation ofanoplin and Ano8k in PBS buffer is equal to∼7% of total amide Iarea, whereas for Ano1K5 V8K this number increases to 10%(Table 3). The R-helical content of each anoplin rises in 50%TFE. The increase is 28% for wild type anoplin and ∼60% forboth substituted derivatives (Table 2). The final R-helical area asa percentage of total amide I area after folding in 50% TFE is 9%,11%, and 17% for native anoplin, Ano8K, and Ano1K5 V8K,respectively (Table 3).The substantial increase in the R-helical content in 50% TFE

supports other literature findings that anoplin adopts a mostly R-helical secondary structure when folded.15,39 It appears thatalthough both anoplin derivatives have a higher but approxi-mately equal tendency to adopt R-helical structure in TFEcompared to native anoplin, they do not have a similar final R-helical content. In fact, as native anoplin gets one and then threesubstitutions, the TFE R-helix content steadily increases(Table 3). A likely explanation for this trend is the preferenceof certain residues for specific secondary structures.40-42 It iswell-known and generally accepted that lysines prefer to adoptR-helical structures. Therefore, by substituting a lysine in place of athreonine (favors a β-sheet structure)40 for Ano8K and replacingan additional glycine (favors a disordered structure)40 forAno1K5 V8K, it is only natural that the substitutions wouldincrease the peptide’s preference for R-helical structure. It alsoappears that the two lysine substitutions in Ano1K5 V8K wereinfluential enough to even increase the R-helix content in PBS(Table 3).The β-sheet content was approximately equal for wild type

anoplin and Ano8K at 6% and 7% of total amide I area, whileAno1K5 V8K had higher β-sheet content at 10% of total amide Iarea (Table 3). The β-sheet content of wild type anoplin and

Ano1K5V8K remained the same in 50%TFE but rose for Ano8Kby 67% (Table 3). This increase for Ano8K brought its β-sheetcontent up to the same amount as Ano1K5 V8K (10%)compared to anoplin in TFE (7%, Table 2). The higher β-sheetcontent in Ano1K5 V8K is likely due to the addition of a β-sheetfavoring valine in place of an arginine, which has no secondarystructure preference.40 The lack of change in β-sheet content fornative anoplin and Ano1K5 V8K in 50% TFE compared to PBSalso supports evidence that anoplin adopts an R-helical structurewhen folded.The unordered structure content of wild type anoplin in PBS

was 73% of total amide I area and decreased to 69% and 61% inAno8K and Ano1K5 V8K, respectively (Table 3). The unorderedstructure content of each peptide dropped in 50% TFE. Thedecrease was approximately the same for wild type anoplin andAno1K5 V8K at 12% and 11%, respectively, but it was greater forAno8K at 17% (Table 3). The final level of unordered structureafter folding in 50% TFE gradually decreases from anoplin toAno8K and Ano1K5 V8K similar to the unordered structurelevels in PBS (Table 3). This content is inversely related to theR-helix content of each peptide and is explained by the sameconcept. The newly substituted residues favorR-helices, and thusas the amount of R-helical content increased, the unorderedstructure content decreased. The trend also continued once thepeptides were in the membrane-mimicking TFE solvent.Whereas the R-helix content increased in TFE, the unorderedstructure content decreased.The amide II region is not often used for peptide secondary

structure determination due to its lower sensitivity and lowanalyzability. Still, some work has been done to determine amide

Table 2. Assignment of the UVRR Amide Bands of AnoplinDerivatives in PBS and 50% TFE

anoplin Ano8K Ano1K5 V8K

Raman band PBS TFE PBS TFE PBS TFE assignmenta

amide I 1609b 1616 1608 1613 1611 1616

1631 1631 1628 1628 1632 1635 β-sheet

1653 1650 1645 1651 1652 1651 R-helix1665 1668 1665 1668 1672 1667 β-strand

1686 1688 1687 1688 1687 1688 unordered

amide II 1522 1521 1524 1529 1523 1529 R-helix1540 1540 1544 1542 1541 1543 R-helix1560 1557 1561 1555 1561 1553 unordered

1582 1567 1578 1563 1583 1562 β-sheeta From Guiffo-Soh et al.2 bPosition of peak maximum measured in cm-1.Precision of the measurements was (5 cm-1 as estimated from thespectrometer slit width.

Table 3. Relative Area of the Peaks Corresponding to r-Helix, β-Sheet, and Unordered Structure in Anoplin, Ano8K,and Ano1K5V8K as Percentages of Total Amide I Area in PBSand 50% TFE

R-helix β-sheet unordered

PBS TFE PBS TFE PBS TFE

anoplin 7% 9% 6% 6% 73% 65%

Ano8K 7% 11% 7% 10% 69% 59%

Ano1K5 V8K 11% 17% 10% 10% 61% 55%

Figure 7. Peak fitting of UV resonance Raman spectra of Ano1K5 V8Kin PBS (A) and 50%TFE (B). Original spectrum (thick) is shown on topof the fitted trace (dashed) with individual fitted peaks (thin) andbaseline (flat) below. TFE contributions are marked by asterisks.



II band shifts due to secondary structure.17,43 In this work, theanoplin amide II region appeared to undergo a change in 50%TFE compared to PBS buffer, and attempts were made to fit theregion to support amide I results. Four bands were used to fit theamide II region based on the 1511, 1550, 1554, and 1564 cm-1

bands (corresponding to R-helix, R-helix, β-sheet, and unor-dered structures, respectively) found by JiJi et al.17 The locationsof the peaks for each sample are shown in Table 2. Similar toamide I peaks, the amide II peaks of all peptides showedconsistency among themselves and compared to the results ofJiJi et al.17 Although results of amide II fitting appeared promisingbased on peak locations, the quality of results declined oncequantitative and comparative analysis was attempted. Consistentpeak-fitting results could not be produced to support or opposeamide I findings. No clear patterns or trends were found in amideII fitting results. Lack of success was deemed to be due tosensitivity of amide II band to hydrogen bonding rather than topeptide secondary structure43 and difficulty in fitting of the amideII region.Overall UVRR spectroscopy was certainly effective in discri-

minating the conformation of anoplin and its two derivatives.The data indicate that Ano8K and Ano1K5 V8K are more helicalin membrane-mimicking environment than is the wild typeanoplin. As a result, amphipathicities of these peptides are likely

greater than that of the original anoplin and do not follow thesame trend as the calculated average hydrophobic moment(Table 1). The increased amphipathicity in addition to theincreased charge contributes to the higher antimicrobial activityof the two derivatives. Since the trisubstituted anoplin has thehighest helicity among the three peptides, it is likely the mostamphipathic, which explains the higher hemotoxicity of thepeptide (Table 1).Membrane Lytic Activity of Anoplin Derivatives. Finally,

the ability of all three anoplin derivatives to rupture model cellmembranes was determined using dye leakage experiments.When encapsulated inside the lipid vesicles at high concentra-tion, carboxyfluorescein fluorophore self-quenches.44 Upon ad-dition, the peptides bind and rupture this model cell membrane,releasing the entrapped fluorophore and thus leading to increasein fluorescence intensity. Membrane lytic activity is measuredrelative to the activity of nonionic surfactant triton X-100, whichcompletely ruptures vesicles. Six different model cell membraneswere tested in this work: single saturated phospholipid (DPPCand DPPG) and four model mixtures with unsaturated phos-pholipids mimicking mammalian, gram-negative bacteria E. Coli,and gram-positive bacteria S. aureus and B. subtilis.As we observed in the monolayer experiments, anoplin has

much higher affinity toward anionic DPPG than zwitterionic

Figure 8. Carboxyfluorescein fluorescence signal increase after the addition of Anoplin (circles and solid line), Ano8K (squares and dotted line), andAn1K5 V8K (triangles and dashed line) into DPPC (A), DPPG (B), mammalian model (C), E. Colimodel (D), S. aureusmodel (E), and B. subtilismodel (F).



DPPC monolayer due to electrostatic attraction (Figure 3). Inthat respect, the two derivatives Ano8K and Ano1K5 V8K haveeven greater binding selectivity due to the additional positivecharge. The selectivity does not, however, simply extrapolate tomembrane-rupturing activity thus showing nonadditivity ofelectrostatic and hydrophobic interactions.1-3

All three anoplin derivatives demonstrated similar poor activ-ity toward zwitterionic DPPC and mammalian model mem-branes (Figure 8A, C), consistent with poor binding of thepeptides to neutral membranes (Figure 3A). Upon addition ofpure buffer to mammalian model membranes, a small fluores-cence intensity increase (<5%) was detected, possibly due toosmotic shock.44 Yet, upon the addition of the peptides intomammalian model membrane, a small decrease of fluorescenceintensity was detected (Figure 8C). It has previously been shownthat model membranes composed of several different types oflipids with different transition temperatures contain more mem-brane defects and allow spontaneous diffusion of carboxy-fluorescein.45 Therefore, the decrease in fluorescence signalintensity observed for the mammalian model could be attributedto the binding of anionic fluorophore to the cationic peptideswith the formation of a signal quenching fluorophore-peptidecomplex.46-48

All three anoplin derivatives were found to induce similarleakage from DPPG vesicles (Figure 8B). We detected a smallpercentage of released dye at low peptide to lipid ratio. As therelative concentration of the peptide increased, up to 25% of thedye was released. When bacterial model cell membranes wereused, however, we noticed a difference in peptide activity, withAno1K5 V8K being the most active, followed by Ano8K and thewild type anoplin (Figure 8D-F). For both gram-positive modelsAno1K5 V8K induced almost complete dye leakage at higherpeptide/lipid ratio, indicating total vesicle disruption (Figure 8E,F). Ano8K and anoplin were found to be considerably less active,especially against the B. subtilis model. Both Ano1K5 V8K andAno8K had lytic activity higher than anoplin against the E. Colimodel (Figure 8D).Overall the data indicate that the two derivatives of anoplin

have on average higher lytic activity than wild type anoplinagainst all bacterial model cell membranes used. Although thehigher lytic activity of the derivatives correlates with greatercharge, resulting in greater initial interaction and peptide/lipidbinding, differences in the activity between the two derivativesAno1K5 V8K and Ano8K point to a more complex mechanismof membrane disruption. Degrees of helicity and amphiphathi-city were previously reported as being important parameters inpeptide antimicrobial activity.3,49 According to UVRR spec-troscopy, Ano1K5 V8K has the greater percent of helicalcontent in TFE, followed by Ano8K and anoplin. Therefore,it is reasonable to say that Ano1K5 V8K, with the greatesthelicity, is the most amphipathic among the three peptides in acell membrane environment thus leading to higher antimicro-bial activity of the peptide, as well as higher hemolytic activity.1

Another explanation of the higher membrane lytic activity ofthe trisubstituted peptide against bacterial model membranescomes from the fact that it likely has a narrower polar face thananoplin and Ano8K (Figure 1) and, as a result, it destabilizes thebilayer to a greater degree.1 It has to be mentioned, however,that the whole concept of polar angle might not be appropriatefor such short peptides due to relatively low helicity of thepeptides.

’CONCLUSION

In the present work we applied several biophysical techniquesto compare physicochemical properties of anoplin and its twoderivatives. Using UV resonance Raman spectroscopy, we wereable to detect differences between the conformations of thesepeptides, showing the power and uniqueness of the method inthe conformational studies of short antimicrobial peptides. Bothderivatives have a higher tendency to fold into an R-helix inmembrane- mimicking environment, with the trisubstitutedAno1K5 V8K still folding in the most defined R-helix. As a resultthe two derivatives are likely more amphipathic at the cellmembrane. The high amphipathic level, in addition to an extrapositive charge, leads to higher antimicrobial activity and in thecase of Ano1K5 V8K to slightly higher hemolytic activity.

’AUTHOR INFORMATION

Corresponding Author*Phone (613)-520-2600 x6043; fax (613)-520-3749; [email protected].

’ACKNOWLEDGMENT

Financial support was provided by ERA, NSERC, and CFI.

’REFERENCES

(1) Toke, O. Biopolymers 2005, 80, 717–735.(2) Sathe, M.; Wieprecht, T. Biochim. Biophys. Acta 1999,

1462, 71–87.(3) Yeaman, M. R.; Yount, N. Y. Pharm. Rev. 2003, 5, 28–56.(4) Chen, Y.; Mant, C. T.; Farmer, S. W.; Hancock, R. E. W.; Vasil,

M. L.; Hodges, R. S. J. Biol. Chem. 2005, 280, 12316–12329.(5) Ozdemir, A.; Lednev, I. K.; Asher, S. A. Biochemistry 2002,

41, 1893–1896.(6) Wallace, J.; Daman, O. A.; Harris, F.; Phoenix, D. A. Theor. Biol.

Med. Modell. 2004, 1, 5.(7) Phoenix, D. A.; Harris, F.; Daman, O. A.; Wallace, J. Curr. Protein

Peptide Sci. 2002, 3, 201–221.(8) Ianoul, A.; Mikhonin, A.; Lednev, I. K.; Asher, S. A. J. Phys. Chem.

A. 2002, 106, 3621–3624.(9) Konno, K.; Rangel, M.; Oliveira, J. S.; Cabrera, M. P. S.; Fontana,

R.; Hirata, I. Y.; Hide, I.; Nakata, Y.; Mori, K.; Kawano, M.; Fuchino, H.;Sekita, S.; Neto, J. R. Peptides 2007, 28, 2320–2327.

(10) Fontana, R.; Mendes, M. A.; de Souza, B. M.; Konno, K.; C�esar,L. M. M.; Malaspina, O.; Palma, M. S. Peptides 2004, 25, 919–928.

(11) Gao, B.; Sherman, P.; Luo, L.; Bowie, J.; Zhu, S. FASEB J. 2009,23, 1230–1245.

(12) Prates, M. V.; Sforca, M. L.; Regis, W. C. B.; Leite, J. R. S. A.;Silva, L. P.; Pertinhez, T. A.; Araujo, A. L. T.; Azevedo, R. B.; Spisni, A.;Bloch, C., Jr. J. Biol. Chem. 2004, 279, 13018–13026.

(13) Jin, L. L.; Li, Q.; Song, S. S.; Feng, K.; Zhang, D. B.;Wang, Q. Y.;Chen, Y. H. Comp. Biochem. Physiol., Part B: Biochem. Mol. Biol. 2009,154, 174–178.

(14) Abbassi, F.; Lequin, O.; Piesse, C.; Goasdoue, N.; Foulon, T.;Nicolas, P.; Ladram, A. J. Biol. Chem. 2010, 285, 16880–16892.

(15) Konno, K.; Hisada, M.; Fontana, R.; Lorenzi, C. C.; Naoki, H.;Itagaki, Y.;Miwa, A.; Kawai, N.; Nakata, Y.; Yasuhara, T.; RuggieroNeto,J.; de Azevedo, W. F., Jr.; Palma, M. S.; Nakajima, T. Biochim. Biophys.Acta—Prot. Struct. Mol. Enzymol. 2001, 1550, 70–80.

(16) Ifrah, D.; Doisy, X.; Ryge, T. S.; Hansen, P. R. J. Peptide Sci.2005, 11, 113–121.

(17) JiJi, R. D.; Balakrishnan, G.; Hu, Y.; Spiro, T. G. Biochemistry2006, 45, 34–41.

(18) Guiffo-Soh, G.; Hernandez, B.; Coic, Y.; Boukhalfa-Heniche,F.; Fadda, G.; Ghomi, M. J. Phys. Chem. B 2008, 112, 1282–1289.



(19) Smirnov, A. I. J. Magn. Reson. 2008, 190, 154–159.(20) Epand, R. F.; Savage, P. B.; Epand, R. M. Biochim. Biophys. Acta

—Biomembranes 2007, 1768, 2500–2509.(21) Chen, P. S.; Toribara, T. Y., Jr.; Warner, H. Anal. Chem. 1956,

28, 1756–1758.(22) Maget-Dana, R. Biochim. Biophys. Acta—Biomembranes 1999,

1462, 109–140.(23) Brockman, H. Curr. Opin. Struct. Biol. 1999, 9, 438–443.(24) Eiriksdottir, E.; Konate, K.; Langel, U.; Divita, G.; Deshayes, S.

Biochim. Biophys. Acta—Biomembranes 2010, 1798, 1119–1128.(25) Ambroggio, E. E.; Separovic, F.; Bowie, J.; Fidelio, G. D.

Biochim. Biophys. Acta—Biomembranes 2004, 1664, 31–37.(26) Cabrera, M. P. D. S; Arcisio-Miranda, M.; de Costa, L. C.; de

Souza, B. M.; Costa, S. T. B.; Palma, M. S.; Neto, J. R.; Procopio, J. Arch.Biochem. Biophys. 2009, 486, 1–11.(27) Ianoul, A.; Boyden, M. N.; Asher, S. A. J. Am. Chem. Soc. 2001,

123, 7433–7434.(28) Chi, Z.; Chen, X. G.; Holtz, J. S.; Asher, S. A. Biochemistry 1998,

37, 2854–2864.(29) Huang, C.-Y.; Balakrishnan, G.; Spiro, T. G. J. Raman Spectrosc.

2006, 37, 277–282.(30) Xu, M.; Shashilov, V. A.; Ermolenkov, V. V.; Fredriksen, L.;

Zagorevski, D.; Lednev, I. K. Protein Sci. 2007, 16, 815–832.(31) Xu, M.; Ermolenkov, V. V.; Uversky, V. N.; Lednev, I. K.

J. Biophoton. 2008, 1, 215–229.(32) Williams, R. W.; Dunker, K. J. Mol. Biol. 1981, 152, 783–813.(33) Williams, R. W. J. Mol. Biol. 1983, 166, 581–603.(34) Williams, R. W. Methods Enzymol. 1986, 130, 311–331.(35) Pripotnev, S.; Won, A.; Ianoul, A. J. Raman Spectrosc. 2010,

41, 1355–1359.(36) Buck, M. Q. Rev. Biophys. 1998, 31, 297–355.(37) Apetri, M. M.; Maiti, N. C.; Zagorski, M. G.; Carey, P. R.;

Anderson, V. E. J. Mol. Biol. 2005, 355, 63–71.(38) Maiti, N. C.; Apetri, M. M.; Zagorski, M. G.; Carey, P. R.;

Anderson, V. E. J. Am. Chem. Soc. 2004, 126, 2399–2408.(39) Cabrera, M. P. D. S; Arcisio-Miranda, M.; Costa, S. T. B.;

Konno, K.; Ruggiero, J. R.; Procopio, J.; Neto, J. R. J. Peptide Sci. 2008,14, 661–669.(40) Chou, P. Y.; Fasman, G. D. Biochemistry 1974, 13, 222–245.(41) Chakrabartty, A.; Baldwin, J. A.; Baldwin, R. L. Nature 1991,

351, 586–588.(42) Li, S.-C.; Deber, C. M. Nat. Struct. Biol. 1994, 1, 368–373.(43) Myshakina, N. S.; Ahmed, Z.; Asher, S. A. J. Phys. Chem. B 2008,

112, 11873–11877.(44) Schwarz, G.; Robert, C. H. Biophys. J. 1990, 58, 577–583.(45) Hays, L.; Crowe, J. H.; Wolkers, W.; Rudenko, S. Cryobiology

2001, 42, 88–102.(46) Magzoub, M.; Ogle-cka, K.; Pramanik, A.; Eriksson, L. E. G.;

Gr€aslund, A. Biochim. Biophys. Acta—Biomembranes 2005,1716, 126–136.(47) Ringstad, L.; Kacprzyk, L.; Schmidtchen, A.; Malmsten, M.

Biochim. Biophys. Acta—Biomembranes 2007, 1768, 715–727.(48) Van den Bogaart, G.; Guzm�an, J. V.; Mika, J. T.; Poolman, B.

J. Biol. Chem. 2008, 283, 33854–33857.(49) Dathe, M.; Wieprecht, T. Biochim. Biophys. Acta 1999,

1462, 71–87.

effect of point mutations on the secondary structure and membrane interaction of antimicrobial...

Documents