pubs.acs.org/Biochemistry Published on Web 07/06/2009r2009 American Chemical SocietyBiochemistry 2009, 48, 72297239 7229DOI: 10.1021/bi900670dStructure-Dependent Charge Density as a Determinant of Antimicrobial Activityof Peptide Analogues of DefensinYang Bai,Shouping Liu,Ping Jiang,Lei Zhou,,@Jing Li,,@Charles Tang,)Chandra Verma,^Yuguang Mu,*,Roger W. Beuerman,*,,@,#and Konstantin Pervushin*,School of Biological Sciences, Nanyang Technological University, 60 Nanyang Drive, Singapore 637551,Singapore Eye ResearchInstitute, 11 Third Hospital Avenue, #06-00, Singapore 168751,)Department of Pathology, Pharmaceutical Microbiology Laboratory,Singapore General Hospital, Singapore 168751, ^Bioinformatics Institute (A-STAR), 30 Biopolis Street, #07-01 Matrix, Singapore138671,@Department of Ophthalmology, Yong Loo Lin School of Medicine, National University of Singapore, 5 Lower Kent RidgeRoad, Singapore 119074, and#School of Chemical and Biomedical Engineering, Nanyang Technological University, 62 NanyangDrive, Singapore 637459Received April 19, 2009; Revised Manuscript Received June 5, 2009ABSTRACT: Defensinsaresmall (3-5kDa) cysteine-richcationicproteinsfoundinbothvertebratesandinvertebrates constituting the front line of host innate immunity. Despite intensive research, bactericidal andcytotoxic mechanisms of defensins are still largely unknown. Moreover, we recently demonstrated that smallpeptides derived from defensins are even more potent bactericidal agents with less toxicity toward host cells.In this paper, structures of three C-terminal (R36-K45) analogues of human -defensin-3 were studied by1HNMRspectroscopy and extensive molecular dynamics simulations. Because of indications that these peptidesmight target the inner bacterial membrane, they were reconstituted in dodecylphosphocholine or dodecylpho-sphocholine/1-palmitoyl-2-oleoyl-sn-glycero-3-[phospho-rac-(1-glycerol)] mixedmicelles, andlipidbicellesmimickingthephospholipid-constitutedbilayermembraneofmammalianandbacterial cells. Theresultsshowthat the binding affinity and partitioning into the lipid phase and the ability to dimerize and accrete well-defined structures upon interactions with lipid membranes contribute to compactization of positive chargeswithin peptide oligomers. The peptide charge density, mediated by corresponding three-dimensionalstructures, wasfoundtodirectlycorrelatewiththeantimicrobial activity. Thesenovel observationsmayprovide a new rationale for the design of improved antimicrobial agents.Inrecentyears,strainsofbacteriawithresistancetooneormore conventional antibiotics have emerged at an alarming rate.An urgency to design new generation antimicrobials withimproved properties has developed. One of the promisingavenues has been natural cationic peptides, such as tachyplesin,cecropin,melitten,anddefensins(1).Inmammals,atleasttwodistinct groups of antimicrobial peptides (AMPs)1arefound,defensinsandcathelicidin-derivedpeptides.Bothhavetwodis-tinct roles, antimicrobial activity and serving as signalingmole-cules on the borderline of innate and acquired immunity (2, 3).Defensins are short cationic peptides 28-42 amino acidsin length usually containing three intramolecular disulfidebonds(4-6).Theirmodeofactioninvolveselectrostaticinter-actionswiththemembrane,leadingtopermeabilitychangesorpore formation in bacteria, fungi membrane, and viruses envel-opes (1, 7). The lowdegree of sequence similarity amongmembersof themammalian-defensinfamily, inadditiontothe highly conserved secondary and tertiary structures as well assimilaritiesofchargedensity, suggeststhattheirantimicrobialactivitymaydependonthestructureandchargedistributionrather thantheir primarysequence (8). Because theylackadistincthydrophobiccore,foldingofdefensinsismainlystabi-lizedbythepresence of threedisulfide bonds (9). Studies ofpeptideanalogues of human-defensin-3(hBD-3) (2, 10, 11)have pointed to the distribution of the positively charged aminoacidsandhydrophobicsidechainsasimportantparametersofantimicrobial activity, while the overall hydrophobicity maycorrelate with cytotoxicity to mammalian cells.Ourpreviousworkdemonstratedthatselectedlinearanalo-gues of hBD-3 (full length, G1-K45) might exhibit significantlyhigher antimicrobial activitywithlower cytotoxicitythantheThis work was supported by the Nanyang Technological UniversityURECAprogramandNational Medical ResearchCouncil ofSinga-pore (NMRC) Grants 1106/2007, 0808/2003, CPG/007/2004, andNMRC-IBG. This work was also supported by academic research fund(AcRF) Tier 2 by the MOE (Grant T206B3210RS).*To whom correspondence should be addressed. R.W.B.: SingaporeEyeResearchInstitute,11ThirdHospitalAvenue,#06-00,Singapore168751; fax, (65) 6322-4599; e-mail, rbeuer@pacific.net.sg. Y.M.:School of Biological Sciences, Nanyang Technological University,NanyangDrive60, Singapore637551; e-mail, ygmu@ntu.edu.sg. K.P.:SchoolofBiologicalSciences,NanyangTechnologicalUniversity,NanyangDrive60, Singapore637551; fax, (65) 6791-3856; e-mail,kpervushin@ntu.edu.sg; telephone, (65)-65141916.1Abbreviations: Acm, acetamidomethyl; AMP, antimicrobial pep-tide; cfu, colony forming unit; COSY, correlation spectroscopy; DMPC,1,2-dimyristoyl-sn-glycero-3-phosphocholine; DHPC, 1,2-diheptanoyl-sn-glycero-3-phosphocholine; DPC, dodecylphosphocholine; DOSY,diffusion-orderedspectroscopy; ESI-MS, electrosprayionization massspectrometry; hBD, human-defensin; hBD-3, human-defensin-3;HPLC, high-performanceliquidchromatography; HSQC, heteronuc-lear single-quantumcoherence; LPS, lipopolysaccharides; NMR, nucle-ar magnetic resonance; NOE, nuclear Overhauser effect; NOESY,nuclearOverhausereffectspectroscopy; POPG, 1-palmitoyl-2-oleoyl-sn-glycero-3-[phospho-rac-(1-glycerol)]; SERI, Singapore Eye ResearchInstitute;SGH,SingaporeGeneralHospital;rmsd,root-mean-squaredeviation; ROE, rotational frame Overhauser effect; ROESY, rota-tional frame Overhauser effect spectroscopy; TOCSY, total correlationspectroscopy; w/v, weight/volume.7230 Biochemistry, Vol. 48, No. 30, 2009 Bai et al.native protein (12). The latter effect is related to the removal ofdisulfide bridges resulting in increased structural flexibility of thelinear forms. Since peptides with varying hydrophobicity valuesare found tobe potent antimicrobials, there are apparentlyunknowndeterminantsofthethree-dimensional(3D)structurethat balance charge density and hydrophobicity. Althoughdefensins as withany other antimicrobial peptide must firstpenetrate an outer barrier, e.g., mainly lipopolysaccharides(LPS)andpeptidoglycaninbacteriaoraglycocalixlayerandmatrixproteinsinmammaliancells(13-15),themaintargetisthelipidicplasmamembrane(12, 13, 16). Asimulationofthelipid composition of bacterial membranes can be used as a targetto predict antimicrobial potency (17). Thus, understandingbacteriakillingmechanismsbyantimicrobial peptidesrequiresstudies at atomic resolution of the intrinsic structural propertiesof peptides as well as their modulation by model membranes.As we demonstrate in this study, some of the newly designed,evenshorterdefensinC-terminalanalogues(R36-K45)exhibitimproved antimicrobial activities with minimal cytotoxicity.Three C-terminal (R36-K45) analogues of hBD-3, C2, Y2,and F2, were selected so their 3D structures and interaction withmembrane lipids could be studied using NMR spectroscopy andMDsimulations(Table1). Wefoundthattheabilityofthesepeptides to dimerize both in aqueous solution and on themembrane surfaces leads to a large increase in the local positivechargedensity(18).Thisseemstobeanimportantfeatureforthese peptides to increase both the affinity and possible interfacearea interacting with lipid membranes. Moreover, as indicated byMD simulations, accretion of awell-defined structure featuringthemostcompactdensityof positivechargesleadstoamoreefficient bactericidal activity. Our studies suggest that the higherdensityof positivechargeswithinthescaffoldof well-definedstructure might predict higher antimicrobial activity.MATERIALS AND METHODSPeptide Selection and Synthesis. A series of human-defensin-3(hBD-3) C-terminal (R36-K45) analogues wereproducedusingsolidphasepeptidesynthesis(SPPS),viaeitherchemical modificationof twocysteineresiduesintheoriginalsequenceorsubstitutionofthemwithotheraminoacids. TheseriesofhBD-3C-termianl(R36-K45)analogueswasderivedand named by the two amino acids in the center of the sequences,at the two Cys positions. HPLC and ESI-MS were subsequentlyperformed for peptide purification and identification. Followingantibacterial andcytotoxicityassays, C2, F2, andY2peptideswere selected for structural analyses. The original hBD-3C-terminal(R36-K45)sequenceisRGRKCCRRKK.TheC2sequence is RGRKC(Acm)C(Acm)RRKK(Acm, acetamido-methylgroup). TheF2sequenceisRGRKFFRRKK.TheY2sequence is RGRKYYRRKK.AntimicrobialActivityAssays.ThebacteriumPseudomo-nas aeruginosa ATCC 9027 was used in this study. It was chosenbecause P. aeruginosa is known to cause corneal ulcers that aredifficult totreat andrequireanintensiveregimenof fortifiedtopicalantibiotics. Resistance to some commonly usedantibio-tics has been documented among clinical ocular isolates ofP. aeruginosa, thusaddingtoasenseofurgencyinthesearchforalternativetherapeuticagents(19). Bacteriawerenomorethan five passages removed from the original master lot obtainedfromtheAmericantypeCultureCollection(ATCC). Bacteriawere prepared with a final cell count of 105-106cfu/mL. The cfuof survivingbacteriawerecountedafter incubationwithtestpeptides at 22.5 Cfor 4 h in purified water media. Theantibacterial activity of the peptide was manifested as thelogarithmic reduction of the bacteria on exposure to the peptide.Peptides that exhibited high bacterial killing activities weresubsequently selected and tested in high-salt restricted conditions[0.9%(w/v) sodiumchloride buffer andincubationfor 6h].Bacteria unexposed to peptides were also counted parallel to thetest as a control.Cytotoxicity Assay. For the cytotoxicity assays, the primaryconcern was for the fragile epithelial cells that cover the surface ofthe eye. This is a mucosal surface, and an important use of AMPswould be for the eye or other mucosal surfaces such as the lungsandmouth, all of whichhaveanepithelial cell covering(12).Cytotoxicitywasdetermined by measuringtheamount of ATPgenerated by viable cells in culture using the CellTiter-Gloluminescent cell viability assay (Promega, Madison, WI). Synthe-sized peptides were incubated with a human conjunctival epithe-lial cellline (IOBA-NHC)as described previously(12,20). Theluminescent signal was read by a microplate reader (TecanGeniosPro, Tecan Asia, Singapore). Blank wells containing onlymediumwere included in each assay, and the reading wassubtracted fromeach sample and control well. Using theluminescenceoftheuntreatedwellsas100%, thecell viabilitywas calculated as (luminescence of the control -luminescence ofthe sample)/luminescence of the control 100%. Each conditionwas assayed in triplicate, and the experiments were repeated threetimes.Hemolysis Assay. The hemolytic activity of the peptides wasestimatedbytheamount of hemoglobinreleasedfromrabbiterythrocytes (12). Rabbit erythrocytes were chosen becauseAMP-inducedtoxicityeasilyoccursinthesebloodcellswhencomparedtohumanredbloodcells(21). Thehemolyticacti-vityof eachsamplewas calculatedas (ODsample-ODblank)/(ODpositive control-ODblank) 100%. Each condition wasassayedintriplicate, andtheexperimentswererepeatedthreetimes. Rabbit procedures were approved by the IACUCofSingHealth.NMRSamplePreparation. DPC(5.27mg) (Avanti PolarLipids) with or without 1.16 mg of POPG(Avanti Polar Lipids) wasdirectlydissolvedin0.5mLof 30 mMphosphate buffer. This resultedin30mMDPCor 10:1(molar ratio) DPC-POPGmembranemimickingsolutions. DMPC(41mg) (Avanti PolarLipids) wasTable 1: Biophysical Properties of Three hBD-3 Analogous Peptidespeptide molecular weight pI (iso-electric point) overall hydrophobicityanet () charge no. of hydrophobic residues sequencebC2 1432.79 11.57 19 7 0 RGRKCCcRRKKF2 1378.7 12.9 16 7 2 RGRKFFRRKKY2 1410.69 11.59 16.4 7 2 RGRKYYRRKKaHydrophobicityiscalculatedonthebasisoftheHopp-Woddshydrophilicityscale.Alowervaluecorrespondstoalowerhydrophilicity.bResiduesmodified from original native peptide sequence are highlighted in bold.cCysteine residues are acetamidomethyled, C(Acm).Article Biochemistry, Vol. 48, No. 30, 2009 7231dissolved in ethanol and dried under a N2stream. DHPC(11 mg)(Avanti Polar Lipids) was dissolved in 30 mMphosphatebuffer and used to solubilize DMPC. After several cyclesof freezingandthawing,themixturewasa10%(w/v)DMPC/DHPCbicellesolutionwithaqfactorof3(DMPC:DHPCmolarratioof3:1).Relativelylargebicelles(q=3)weremadetoprovidealargeplanarregioninitsdiscoidal structureand,hence, a better mimic to a membrane bilayer (22, 23). The peptidestock solutions were directly added to the lipid-containingsolutions.NMRExperiments. NMRexperiments were performedusingBruker Avance II 600and700MHz spectrometers at25 Cfor aqueous and micelle solutions and 10 Cfor the discordbicelle medium. In the latter case, the lower temperature was usedto maintain the isotropic tumbling of bicelles (23).Two-dimensional(2D)homonuclearTOCSY,NOESY,andROESYexperimentswereconductedwithmixingtimesof70,150, and 100 ms, respectively, using 500 Mpeptides at pH5. TheY2peptide was analyzedinthe complex withDPC-POPGmixed micelles and DMPC/DHPC bicelles. F2 and C2 peptideswereanalyzedincomplexwithDPC-POPGmixedmicelles.Data were collected for each experiment for 7 h. Residue specificresonanceassignment andassignment of NOESYcross-peakswere performedusing CARA(www.nmr.ch). 3Dstructures of Y2andF2 peptides inDPC-POPGmixedmicelles andY2 inDMPC/DHPC bicelles were reconstructed using CYANA2.0 (24).2D1H-13C HSQC experiments using natural abundance13Cin5mMpeptidesampleswerealsoperformed. HSQCspectraof both F2 andY2 peptides dissolved in aDPC-POPG mixedmicelle solution were recorded. A bicelle solution was not used inthis experimentsinceY2 showednosecondarystructurediffer-ences inmicelle or bicelle solutions. Urea (8 M) was addedtoeachsample subsequently for collection of reference spectra of dena-tured peptides.Aseries of one-dimensional (1D)1Hhomonuclear experimentswere conducted for each sample with increasing peptide concen-trations (10, 20, 50, and 500 M), both inphosphate buffer and in30 mM DPC solutions.DOSYexperimentswereperformedforeachpeptidesampleinphosphatebuffer, 30mMDPCmicelles, andDPC-POPG(30 mM DPC and 3 mM POPG) mixed micelles (25). The pulsedmagnetic fieldgradient strengthwas calibratedusingthe1Hsignal in a 99.9% D2O sample with the self-diffusion coefficientofHDO[(1.902 (0.002) 10-9m2/s].Aseriesof1Dspectraweremeasuredwith adiffusiondelayof150msandincreasinggradientstrength in 32stepsfrom2to 95%of Gmax.Diffusioncoefficient DS is calculated by fitting the curve of signal intensityversus variable gradient strength. The molar fraction of peptidein lipid (FL) is calculated with eq 1 (25)DS DLFLDF1 -FL 1whereDLand DFare thediffusion coefficients oflipidmicellesand free peptides in DPC solution, respectively.Molecular Dynamics Simulations. Six sets of Hamiltonianreplica exchange molecular dynamics (hREMD) simulationswere conducted for C2, F2, and Y2 peptides with eight replicaseachat atemperatureof 300K. Thedetailsof thishREMDmethod can be found in recent work (26). The novel part of themethodisthat insteadofreducingtheintermolecularinterac-tions, we increasedthe repulsive part of the vander Waalsinteractions between the peptide atoms. The dissociation-enhanced hREMD method decreases the energy barrier betweenlocal minima which are mainly stabilized by non-native contacts.Thus, suchhREMDsimulations helpedtosampleabroaderrangeofconfigurationsofpeptideoligomers.Foreachpeptidedimer,twohREMDsimulationswere performed with differentinitialconfigurations:onebegunwiththeNMR-deriveddimerstructure and the other started with the two monomers separated.For C2 and F2 peptides, simulations provided converged resultsby 100 ns, while for Y2 peptides, 150 ns simulations were needed.Twentyof eachY2-F2dimer conformer reconstructedfromNMRdata were clustered on the basis of the pairwise rmsdwithacutoff of 1 A. The center of the largest cluster was selected as arepresentative conformer of the dimer. The two tyrosine residuesof the Y2 model were changed to two cysteine residues to build amodel of the C2 dimer. The C2, F2, andY2 models wereseparately solvated with explicit water in simulation boxes. Thecomplex is formed when the minimum distance between the twomonomers is less than 0.375 nm. The complex structures fromtheY2andC2dimersimulationswereanalyzedtoextractbindingfree energies using the MM/GBSA protocol (27).Themicellestructurewasmodeledusing64DPCmolecules(M64) (28). A single DPC molecule was extracted from a 65-merDPC micelle provided by the Tieleman group (29). The moleculewas repeatedlycopied, rotated, andtranslatedtogenerate a64-merDPCmicelle. Thecompleteparametersetforbonded,nonbonded, and 1-4 interactions of the lipid atoms as well as thenonbondedinteractionsbetweenlipidandpeptideatomswasprovided by the Tieleman group and combined with the OPLS-AA force field in the Gromacs package.Two different initial setups with regard to the relative positionof the Y2 dimer withthe micelle were constructed. Inone case, theY2dimerwasplacedinthemiddleofM64,whileintheothercase, the peptide was placed in the solvated water inthe proximityof the external surface of M64. In these two simulations, distancerestraintswithaforceconstant of 3000kJmol-1nm-2wereimposed on the peptide to maintain the backbone structure of theNMR model while allowing free motion of the side chains. Therestraintswereimposedacrossintramonomerbackboneheavyatom pairs (CR, N, and C0) and across intermonomer CR atompairs. The peptide-micelle complex was solvated in a cubic boxwith dimensions of 7 nm7 nm7 nmwith 9720 explicit TIP3Pwatermoleculesand14chlorideions,thusneutralizingthenetcharge on the system.TheMDsimulationswereimplementedusingtheGromacspackage applying the OPLS-AA (all-atom)force field (30). TheTable 2: Assays of Antimicrobial Activity of Defensins Derived Peptidesagainst P. aeruginosa ATCC 9027 (log reduction)alow salt (4 h) high salt (6 h)concn (g/mL) C2 F2 H2 Y2 Y2 W2200 0.41 0.33 0.29100 0.57 0.61 0.62 4.1950 0.63 0.87 3.07 4.07 1.0625 2.39 3.31 0.4812.5 2.10 0.49 0.096.25 2.02 0.49 0.083.125 0.33aPeptides wereincubatedwithP. aeruginosaATCC9027for 4hat22.5 C. A 0.5 log reduction equals 68% of the bacteria being killed, a 1 logreduction 90%of the bacteria, a 2 log reduction 99%of the bacteria, and a 3log reduction 99.9% of the bacteria.7232 Biochemistry, Vol. 48, No. 30, 2009 Bai et al.complete parameter set for bonded, nonbonded, and 1-4 inter-actions of the lipid atoms was the Berger force field; the proteininteractions were described by the OPLS-AA force field. To treatthe difference of the 1-4 interaction scale factor in Berger (1.0)andOPLS-AA(0.5), the half- double-pair list methodwasapplied. All bonds connectedtohydrogenatoms were con-strained in length according to the LINCS protocol. The protein,DPC, water, and ions were separately coupled to an external heatbathwitharelaxationtimeof0.1ps.Electrostaticinteractionswere treated with the particle mesh Ewald method with a cutoffof 9 A,and acutoff of 14 Awas used in the calculation ofvander Waals interactions. The simulated systems were energy-minimizedusing a steepest descent algorithmfor 2000 stepsand then simulated for 100 ps with the position of all heavy atomsrestrained to allow water molecules to relax prior to a simulationrun. ThegeneralizedBorn(GB) calculations wereperformedusingthe sander module inAMBER9. The source code ofAmber was modified toallowthe use of the OPLS-AAforce field.The modified GB model developed by A. Onufriev, D. Bashford,and D. A. Case was used (31).RESULTSAntibacterial ActivityAssays. Defensin-derivedpeptidesexhibitedvarying antibacterial activities against the commonhuman pathogen P. aeruginosa ATCC9027. The results are listedin Table 2. The Y2 peptide exhibited high antibacterial activity inthe lowmicromolar range under both conditions compared to theotherpeptides. Acooperativeincrease in theactivity ofY2wasdetected atconcentrationsfrom3.2 to 6.3 g/mL (2.5-5.0 M)underaqueousconditionsandfrom12.5to25g/mL(10-20M) under salt conditions. This suggests a structural transition ofY2 in the 5-20 M concentration range, resulting in its remark-ablyincreasedactivity. Incontrast, relativelylowantibacterialactivitywasobservedforconcentrationsofC2andF2. Upto100M, nodetectabletransitionstoahigh-activitystatewereobserved in either peptide in aqueous solution. Y2 was selected asarepresentativeof thehigh-activitygroup for subsequent struc-tural studies, while in the low-activity group, C2 was selected torepresent a native peptide and the F2 peptide was selected becauseof its high degree of sequence similarity to the Y2 peptide.CytotoxicityandHemolyticActivityagainstMamma-lian Cells. The cytotoxicity of hBD-3 analogues was tested on aFIGURE 1: Cytotoxicity assay of C2 (9), F2 (4), and Y2 (b) peptidesfor human conjunctival epithelial cells. Data for human -defensin-3(hBD-3) are also given (O). Reported values are means of performedexperiments with variation bars representing the standard deviation(rmsd).Table 3: Peptide Lipophilicity Measurement (FL)FLa(%)C2 F2 Y2DPC 35.17 31.25 41.41DPC-POPG 100 100 100aFL is the molar fraction of peptides bound to the lipid phase.Table 4: Translational Diffusion Coefficients of C2, F2, and Y2 Measuredby NMRDSa(10-10m2/s)C2 F2 Y25 M 5.32 (0.17 5.5 (0.23 5.71 (0.19500 M 2.37 (0.10 2.31 (0.23 2.01 (0.03aDS is the translational diffusion coefficient.FIGURE 2: 1D 1H spectra of Y2 (A-D) and F2 (E-H) as a functionof concentrationinthepresenceof 30mMDPCat 25C. Theconcentrations of the peptides are (AandE) 10 M, (BandF) 20 M,(C and G) 50 M, and (Dand H) 500 M.1H peaks are assigned andlabeled. Unknownpeaks whichcouldnot be assignedare labeledwithstars. Gradually changing proton resonance peaks are highlighted inred.Spectraineachcolumnsharethecommonaxislabeledatthebottom (D and H).Article Biochemistry, Vol. 48, No. 30, 2009 7233humanconjunctival epithelial cell line(IOBA-NHCcells). Nodecrease in viable cell number was observed 24 h after constantexposuretoC2,F2,orY2atconcentrationsupto200 g/mL(Figure 1). Aslight increase in the viable cell number wasobserved in the presence of C2 in the concentration rangeof 12.5-200 g/mL. On the other hand, native hBD-3 at12.5 g/mL resulted in 50% epithelial cell death, corroboratingan earlier observation (12). Only 6%of the cells remainedviable after exposure to hBD-3 for 24 h at 50 g/mL. Hemolyticactivities of bothpeptides were alsomeasuredusing freshly isolatedrabbit erythrocytes. At a concentration of 200 g/mL, 13 A. The charge densityis estimated using eq 2:F Q=V Q4=3cR3g, 2whereQisthetotalchargeofapeptidedimer(Q=14eforallpeptides). The average cRg for C2 and F2 dimers is 14.9 (0.25 AFIGURE 6: Probability of dimer formationas afunctionof simulationtimefor(A)C2, (B)F2, and(C)Y2.Thesimulationsbeganwithdifferentinitial configurations: dimerstructure(blackcurves) andtwo separated monomers (red curves).FIGURE 7: Distribution functions of the radius of gyration forcharged points in C2, F2, and Y2 dimers.Table 5: Breakdown of the Average Binding Energy between Monomers inY2 and C2 Dimeric Configurations Estimated fromMDSimulationTrajectoriesenergy componentaY2 dimer C2 dimer differenceEvdw (kJ/mol) -156 (13 -84 (14 -72Eelec (kJ/mol) 337 ( 77 537 (151 -200Egb (kJ/mol) -3386 (174 -3583 (289 198Esurf (kJ/mol) -12 (1 -7 (1 -5Ep (kJ/mol) -3048 -3046 -2Enp (kJ/mol) -168 -91 -77Etotal (kJ/mol) -3216 -3137 -79aEvdwandEelecare the vander Waals andelectrostatic interactionenergies, respectively, calculated by the Gromacs package for the peptidesonly.EgbandEsurfaresolvationenergiesofpolarandnonpolarresidues,calculated by Amber 9 using the Generalized Born model. Ep and Enp arethesumsofpolarenergy(Eelec Egb)andnonpolarenergycomponents(Evdw Esurf), respectively. Etotal is the sum of Ep and Enp.Article Biochemistry, Vol. 48, No. 30, 2009 7237each, and the averagecRg of the Y2 dimer is 13.6 ( 0.28 A. Thecharge density of the C2 and F2 dimer is (1.0 (0.05) 10-3e/A3,while the charge density of the Y2 dimer is (1.3 ( 0.07) 10-3e/A3.ThechargedensityofY2is30%higherthanthatofC2orF2. Forthesakeofcomparison, thechargedensityof theC-terminus of hBD-3(R36-K45) [ProteinDataBankentry1KJ6 (33)] was calculated in a similar way: from residue 32 to 45,the total charge is 8e withacRgof 12.3 A, resulting ina F of 1.03 10-3e/A3, which is comparable to that of C2 and F2.Table 5 shows the average binding free energies estimated fromthe final 2 ns simulations of the solvated Y2 and C2 dimers. It isclear that the Y2 dimer is more stable and this originates in betterpacking, as is evident fromthe vander Waals energy term(Evdw72 kJ/mol); this arises mainly from the - stackingof the aromatic side chains of four tyrosine residues. The dimersare electrostatically unstable, with the Y2 dimer being lessdestabilized than the C2 dimer. However, the destabilizedelectrostaticsisoffset bysolvationeffectsasthechargedsidechains are well-exposed and solvated. The net polar componentof the binding energy is also more stable for the Y2 dimer but isexceedingly small [(Eelec Egb)=2 kcal/mol], and hence, thedimerization is mainly driven by packing interactions.Two long simulations of 50 ns were employed to study the Y2peptide-micellesystem.Thesimulationsindicated stronginterac-tions between the peptide and DPC headgroups driven mainly bypolar contacts between charged amino acids and polar headgroups.One snapshot obtainedat the simulationtime of 50ns demonstratedclose contacts between the side chains of positively charged residues(Arg1,Arg3,Lys4,Arg7,andArg8)andtheDPCphospha-tidylcholine groups (Figure 8). Such interactions are expected to beeven stronger for lipids with negatively charged headgroups.DISCUSSIONDefensins are small, positively charged peptides, characterizedby the presence of conserved cysteines (6, 8) that are bridged bydisulfide links. Althoughevolutionarilyconserved, the nativedisulfidebridgesarenot critical forantibacterial activity, andshort segments of hBD-1tohBD-3are potent antimicrobialagents (2, 34-43). Disulfide bonds formed by Cys residues maynot be critical for defensinantimicrobial activity but rathercontributetotheiroverall peptidestructural stabilityandche-motaxis activities (2). Similar effects can be found in other anti-microbial peptides, like tachyplesin and polyphemusins (21, 44).Analogues of natural antibacterial peptides with deletion of thedisulfide bonds have the advantage of removing chemotaxis andso may be nonimmunogenic as drug candidates.In this study, we show that a human -defensin-3 C-terminal(R36-K45) analogue peptide, Y2, with the two cysteine residuesreplacedwithtyrosines, exhibitedhigher antibacterial activityagainsttheGram-negativebacteriumP. aeruginosaandlowercytotoxicityagainstmammaliancellscomparedtotheoriginalfull-length human defensin. The underlying physicochemicalproperties distinguishing Y2andlow-activity grouppeptides(F2andC2)includepropensitiestooligomerizeandtoaccreteunique structure both in water and on the surface of lipidmembranes. Moleculardynamicssimulationsshowedahigherstability of the Y2 and F2 dimers compared to that of C2. Thisoriginatesmainly from -stackingand hydrophobic interac-tions between the Tyr and Phe aromatic side chains. Theoligomerizationstatehas beenstudiedinmanyantimicrobialpeptides.ArecentstudyofmagaininanalogueMSI-78showedthat dimerizationoftheMSI-78peptideincreasestheelectro-staticattractionforbacterial membranesandalsoreducestheFIGURE 8: Representative snapshot of the Y2-DPC configuration in the 30 ns long MD simulation run. The Y2 dimer is highlighted as tubes.DPCmolecularbonds containing oxygen,nitrogen,and carbon are colored red, blue, and green,respectively.This figure was prepared withPyMol (DeLano Scientific LLC).7238 Biochemistry, Vol. 48, No. 30, 2009 Bai et al.affinityformammaliancell membraneswhoseouterleaflet iszwitterionic(45). Theprominentstructural differencebetweenY2andF2dimersmightunderlinedifferencesintheiractivity.The Y2 dimer maintains a hairpinlike, more compact configura-tion, instriking contrast toa twisted helical, extended structure oftheF2dimer. Thecompact structureof Y2induces ahighercharge density. The strong correlation of antibacterial action andsurface charges has been well-studied, and our finding here is inline with the notion that the electrostatic interaction between thecationicpeptides andanioniclipids isveryimportant for theantibacterial function (46-49). In addition, the Tyr residues areinvolved in aromatic stacking interactions as well as in intramo-lecularhydrogenbondswithsurroundingpolargroupsthatinturn further stabilizes the compact configuration.Antimicrobial peptides represent a new therapeutic candidateforovercomingantibioticresistanceissues. Understandingbac-terialkillingmechanismsofthesepeptidesisnecessary,sothatmodifications canbe made toimprove their pharmaceuticalefficacy.Some R-helical peptides (likemagainin, magainin ana-logue MSI-78, and pardaxin) kill bacteria via formation ofbilayer-spanning pores (48, 50-54). Very differently, -defensinsmay exert their antimicrobial activity by an electrostatic charge-basedmechanism(33,55).Arecentstudy(56)hasshownthatclusteringofanioniclipidsontothebacterialmembranemightcauseaphaseboundarydefectinthemembrane.Thesedefectsarethoughttoberesponsibleforincreasedlevelsofmembraneleakage andpermeability, or alteredmembranefunctionandsignaling activity (56). Our data point to the critical role electro-static charge density plays in antibacterial activity, which poten-tiallycausessuchadefect inthe bacterialinnermembrane.Weare currently extending these studies to examine the interactionsof these peptides withbacterial membranes usingsolidstateNMR. However, we are finding that structural parameters suchas compactness and charge density may act as filters in the designof novel antibacterial peptides or small molecule mimics.ACKNOWLEDGMENTWe acknowledge Prof. James J. Chou (Department of Biolo-gical Chemistry and Molecular Pharmacology, Harvard MedicalSchool, Boston, MA) for discussion about diffusion experimentsand diffusion coefficient calculations. We also thank Assoc. Prof.ThorstenWohland(Department of Chemistry, National Uni-versity of Singapore) for sharing his experience about the choiceof lipids for antimicrobial peptide research.SUPPORTING INFORMATION AVAILABLEFour figures and two tables. This material is available free ofcharge via the Internet at http://pubs.acs.org.REFERENCES1.Hancock, R. E., Falla, T., and Brown, M. (1995) Cationic bactericidalpeptides. Adv. Microb. Physiol. 37, 135175.2.Dhople, V., Krukemeyer, A., andRamamoorthy, A. (2006) Thehuman -defensin-3, an antibacterial peptide with multiple biologicalfunctions. Biochim. Biophys. Acta 1758, 14991512.3.Durr, U. H., Sudheendra, U. S., and Ramamoorthy, A. (2006) LL-37,theonlyhumanmemberofthecathelicidinfamilyofantimicrobialpeptides. Biochim. Biophys. Acta 1758, 14081425.4.Bevins, C. L., Martin-Porter, E., and Ganz, T. (1999) Defensins andinnate host defence of the gastrointestinal tract. Gut 45, 911915.5.Diamond, G., andBevins, C. L. (1998) -Defensins: Endogenousantibioticsoftheinnatehostdefenseresponse. Clin. Immunol. Im-munopathol. 88, 221225.6.Ganz,T.,andLehrer, R.I.(1995)Defensins.Pharmacol.Ther.66,191205.7.Boman, H. G. (1995) Peptide antibiotics andtheir role ininnateimmunity. Annu. Rev. Immunol. 13, 6192.8.Raj, P. A., and Dentino, A. R. (2002) Current status of defensins andtheirroleininnateandadaptiveimmunity. FEMSMicrobiol.Lett.206, 918.9.Bauer, F., Schweimer, K., Kluver, E., Conejo-Garcia, J. R.,Forssmann,W. G., Rosch,P., Adermann, K., and Sticht,H. (2001)Structuredeterminationof humanandmurine-defensins revealsstructural conservationintheabsenceofsignificantsequencesimi-larity. Protein Sci. 10, 24702479.10.Kluver, E., Adermann, K., andSchulz, A. (2006) Synthesis andstructure-activityrelationshipof-defensins, multi-functional pep-tides of the immune system. J. Pept. Sci. 12, 243257.11.Kluver,E.,Schulz-Maronde,S.,Scheid, S.,Meyer,B.,Forssmann,W. G., and Adermann, K. (2005) Structure-activity relation of human-defensin 3: Influence of disulfide bonds and cysteine substitution onantimicrobial activity and cytotoxicity. Biochemistry 44, 98049816.12.Liu, S., Zhou, L., Li, J., Suresh, A., Verma, C., Foo, Y. H., Yap, E. P.,Tan, D. T., and Beuerman, R. W. (2008) Linear analogues ofhuman-defensin3: Conceptsfordesignofantimicrobial peptideswith reduced cytotoxicity to mammalian cells. ChemBioChem9,964973.13.Papo,N.,andShai,Y.(2003)Canwepredictbiologicalactivityofantimicrobialpeptidesfromtheirinteractionswithmodelphospho-lipid membranes?. Peptides 24, 16931703.14.Vaara,M.(1992)Agentsthatincreasethepermeabilityoftheoutermembrane. Microbiol. Rev. 56, 395411.15.Vaara, M. (1992) Theouter membraneas thepenetrationbarrieragainstmupirocininGram-negativeentericbacteria.J.Antimicrob.Chemother. 29, 221222.16.Yu, L., Ding, J. L., Ho, B., Feng, S., andWohland, T. (2008)Investigation of the Mechanisms of Antimicrobial Peptides Interact-ingwithMembranesbyFluorescenceCorrelationSpectroscop.TheOpen Chemical Physics Journal 1, 6279.17.Epand, R. F., Savage, P. B., and Epand, R. M. (2007) Bacterial lipidcompositionandtheantimicrobialefficacyofcationicsteroidcom-pounds (Ceragenins). Biochim. Biophys. Acta 1768, 25002509.18.Suresh, A., and Verma, C. (2006) Modelling study of dimerization inmammalian defensins. BMC Bioinf. 7, S17.19.Parks, Q. M., and Hobden, J. A. (2005) Polyphosphate kinase 1 andthe ocular virulence of Pseudomonas aeruginosa. Invest. Ophthalmol.Visual Sci. 46, 248251.20.Diebold, Y., Calonge, M., EnriquezdeSalamanca, A., Callejo, S.,Corrales, R. M., Saez, V.,Siemasko,K. F.,and Stern, M. E. (2003)Characterizationofaspontaneouslyimmortalizedcell line(IOBA-NHC)fromnormal humanconjunctiva. Invest. Ophthalmol. VisualSci. 44, 42634274.21.Ramamoorthy, A., Thennarasu, S., Tan, A., Gottipati, K.,Sreekumar,S.,Heyl,D.L.,An,F.Y.,andShelburne,C.E.(2006)Deletion of all cysteines in tachyplesin I abolishes hemolytic activityandretains antimicrobial activityandlipopolysaccharide selectivebinding. Biochemistry 45, 65296540.22.Whiles, J. A., Deems, R., Vold, R. R., andDennis, E. A. (2002)Bicelles instructure-functionstudies of membrane-associatedpro-teins. Bioorg. Chem. 30, 431442.23.Yamamoto,K.,Soong,R.,andRamamoorthy,A.(2009)Compre-hensive AnalysisofLipidDynamicsVariationwith LipidComposi-tionandHydrationofBicellesUsingNuclearMagneticResonance(NMR) Spectroscopy. Langmuir 25, 70107018.24.Guntert, P. (2004) Automated NMRstructure calculation withCYANA. Methods Mol. Biol. 278, 353378.25.Wang, J., Schnell, J. R., and Chou, J. J. (2004) Amantadine partitionand localization in phospholipid membrane: A solution NMR study.Biochem. Biophys. Res. Commun. 324, 212217.26.Mu, Y. (2009) Dissociation Aided and Side Chain SamplingEnhanced Hamiltonian Replica Exchange. J. Chem. Phys. 130,164107.27.Kollman, P. A., Massova, I., Reyes, C., Kuhn, B., Huo, S., Chong, L.,Lee, M., Lee, T., Duan, Y., Wang, W., Donini, O., Cieplak, P.,Srinivasan, J., Case, D. A., and Cheatham, T. E.III (2000) Calculatingstructures and free energies of complex molecules: Combiningmolecularmechanics andcontinuummodels. Acc. Chem. Res. 33,889897.28.Tieleman,D.P.,vanderSpoel,D.,andBerendsen,H.J.C.(2000)Moleculardynamicssimulationsofdodecylphosphocholine micellesat three different aggregate sizes: Micellar structure and chain relaxa-tion. J. Phys. Chem. B 104, 63806388.Article Biochemistry, Vol. 48, No. 30, 2009 723929.Tieleman, D. P., Marrink, S. J., andBerendsen, H. J. (1997) Acomputer perspective of membranes: Molecular dynamics studies oflipid bilayer systems. Biochim. Biophys. Acta 1331, 235270.30.Van Der Spoel, D., Lindahl, E., Hess, B., Groenhof, G., Mark, A. E.,andBerendsen, H. J. (2005) GROMACS: Fast, flexible, andfree.J. Comput. Chem. 26, 17011718.31.Onufriev, A., Bashford, D., and Case, D. A. (2004) Exploring proteinnative states and large-scale conformational changes with a modifiedgeneralized born model. Proteins 55, 383394.32.Morris, K. F. J., and Johnson, C. S. Jr. (1994) Mobility ordered 2D-NMR spectroscopy. J. Am. Chem. Soc. 114, 776777.33.Schibli, D. J., Hunter, H. N., Aseyev, V., Starner, T. D., Wiencek, J.M., McCray, P. B., Tack, B. F., and Vogel, H. J. (2002) The SolutionStructures of the Human -Defensins Lead to a Better UnderstandingofthePotentBactericidalActivityofHBD3againstStaphylococcusaureus. J. Biol. Chem. 277, 82798289.34.Hoover, D. M., Wu, Z. B., Tucker, K., Lu, W. Y., and Lubkowski, J.(2003) Antimicrobial characterization of human -defensin 3 deriva-tives. Antimicrob. Agents Chemother. 47, 28042809.35.Kluver, E., Adermann, K., andSchulz, A. (2006) Synthesis andstructure-activityrelationshipof-defensins, multi-functional pep-tides of the immune system. J. Pept. Sci. 12, 243257.36.Kluver, E., Schulz-Maronde, S., Scheid, S., Meyer, B., Forssmann, W.G.,andAdermann,K.(2005)Structure-activityrelationofhuman-defensin 3: Influence of disulfide bonds and cysteine substitution onantimicrobial activity and cytotoxicity. Biochemistry 44, 98049816.37.Krishnakumari,V.,Singh,S.,andNagaraj,R.(2006)Antibacterialactivities of synthetic peptides corresponding to the carboxy-terminalregion of human -defensins 1-3. Peptides 27, 26072613.38.Mandal, M., andNagaraj, R. (2002) Antibacterial activities andconformations of synthetic R-defensin HNP-1 and analogs withone, two and three disulfide bridges. J. Pept. Res. 59, 95104.39.Mandal, M., Jagannadham, M. V., and Nagaraj, R. (2002) Antibac-terial activitiesandconformationsof bovine-defensinBNBD-12and analogs: Structural and disulfide bridge requirements for activity.Peptides 23, 413418.40.Pazgier, M., Li, X., Lu, W., and Lubkowski, J. (2007) Humandefensins: Synthesis and structural properties. Curr. Pharm. Des. 13,30963118.41.Sahl, H. G., Pag, U., Bonness, S., Wagner, S., Antcheva, N., andTossi, A. (2005) Mammalian defensins: Structures and mechanism ofantibiotic activity. J. Leukocyte Biol. 77, 466475.42.Wu, Z. B., Hoover, D. M., Yang, D., Boulegue, C., Santamaria, F.,Oppenheim, J. J., Lubkowski,J., and Lu, W. Y. (2003)Engineeringdisulfide bridges to dissect antimicrobial and chemotactic activities ofhuman -defensin 3. Proc. Natl. Acad. Sci. U.S.A. 100, 88808885.43.Zou, G. Z., de Leeuw, E., Li, C., Pazgier, M., Li, C. Q., Zeng, P. Y.,Lu, W. Y., Lubkowski, J., andLu, W. Y. (2007) Towardunder-standing the cationicity of defensins: Arg and Lys versus theirnoncoded analogs. J. Biol. Chem. 282, 1965319665.44.Powers, J. P., Tan, A., Ramamoorthy, A., and Hancock, R. E. (2005)Solutionstructureandinteractionoftheantimicrobialpolyphemu-sins with lipid membranes. Biochemistry 44, 1550415513.45.Porcelli, F., Buck-Koehntop, B. A., Thennarasu, S., Ramamoorthy,A.,andVeglia,G.(2006)Structuresofthedimericandmonomericvariants of magainin antimicrobial peptides (MSI-78 and MSI-594) inmicellesandbilayers, determinedbyNMRspectroscopy. Biochem-istry 45, 57935799.46.Prenner, E. J., Lewis, R., Kondejewski, L. H., Hodges, R. S., andMcElhaney, R. N. (1999) Differential scanning calorimetric study oftheeffectoftheantimicrobialpeptidegramicidinSonthethermo-tropicphasebehaviorofphosphatidylcholine, phosphatidylethano-lamineandphosphatidylglycerollipidbilayermembranes. Biochim.Biophys. Acta 1417, 211223.47.Abraham, T., Lewis, R., Hodges, R. S., and McElhaney, R. N. (2005)Isothermal titration calorimetry studies of the binding of a rationallydesigned analogue of the antimicrobial peptide gramicidin S tophospholipid bilayer membranes. Biochemistry 44, 21032112.48.Dathe, M., andWieprecht, T. (1999) Structural featuresof helicalantimicrobial peptides: Their potential to modulate activity on modelmembranes and biological cells. Biochim. Biophys. Acta 1462, 7187.49.Shai, Y. (2002) Modeof actionof membraneactiveantimicrobialpeptides. Biopolymers 66, 236248.50.Tossi, A., Sandri, L., andGiangaspero, A. (2000) Amphipathic,R-helical antimicrobial peptides. Biopolymers 55, 430.51.Bommineni, Y. R., Dai, H. E., Gong, Y. X., Soulages, J. L.,Fernando, S. C., DeSilva, U., Prakash, O., and Zhang, G. L. (2007)Fowlicidin-3 is an R-helical cationic host defense peptide with potentantibacterial and lipopolysaccharide-neutralizingactivities.FEBS J.274, 418428.52.Hallock, K. J., Lee, D. K., and Ramamoorthy, A. (2003) MSI-78, ananalogue of the magainin antimicrobial peptides, disrupts lipidbilayerstructure via positive curvature strain. Biophys. J. 84, 30523060.53.Kolusheva, S., Lecht, S., Derazon, Y., Jelinek, R., and Lazarovici, P.(2008) Pardaxin, afishtoxinpeptideinteractionwithabiomime-tic phospholipid/polydiacetylene membrane assay. Peptides 29,16201625.54.Rapaport, D., Peled, R., Nir, S., andShai, Y. (1996) Reversiblesurfaceaggregationinporeformationbypardaxin. Biophys. J. 70,25022512.55.Sass, V., Pag, U., Tossi, A., Bierbaum, G., andSahl, H. G. (2008)Mode of action of human -defensin 3 against Staphylococcus aureusand transcriptional analysis of responses to defensin challenge. Int. J.Med. Microbiol. 298, 619633.56.Epand, R. M., Rotem, S., Mor, A., Berno, B., and Epand, R. F. (2008)Bacterialmembranesaspredictorsofantimicrobialpotency.J.Am.Chem. Soc. 130, 1434614352.57.Koradi, R., Billeter, M., andWuthrich, K. (1996) MOLMOL: Aprogramfor display and analysis of macromolecular structures.J. Mol. Graphics 14, 5155, 29-32.