Research Article

Received: 27 March 2014 Revised: 25 June 2014 Accepted: 11 July 2014 Published online in Wiley Online Library: 14 August 2014

(wileyonlinelibrary.com) DOI 10.1002/psc.2681

J. Pept. Sci. 2014; 20: 8

Interaction of a novel antimicrobialpeptide isolated from the venom ofsolitary bee Colletes daviesanus withphospholipid vesicles and Escherichiacoli cells

Sabína Čujová,a,b Lucie Bednárová,a Jiřina Slaninová,a Jakub Strakac

and Václav Čeřovskýa*

The peptide named codesane (COD), consisting of 18 amino acid residues and isolated from the venom of wild bee Colletesdaviesanus (Hymenoptera : Colletidae), falls into the category of cationic α-helical amphipathic antimicrobial peptides. In ourinvestigations, synthetic COD exhibited antimicrobial activity against Gram-positive and Gram-negative bacteria and Candidaalbicans but also noticeable hemolytic activity. COD and its analogs (collectively referred to as CODs) were studied for themechanism of their action. The interaction of CODs with liposomes led to significant leakage of calcein entrapped in bacterialmembrane-mimicking large unilamellar vesicles made preferentially from anionic phospholipids while no calcein leakage wasobserved from zwitterionic liposomes mimicking membranes of erythrocytes. The preference of CODs for anionic phospholipidswas also established by the blue shift in the tryptophan emission spectra maximawhen the interactions of tryptophan-containingCOD analogs with liposomes were examined. Those results were in agreement with the antimicrobial and hemolytic activities ofCODs. Moreover, we found that the studied peptides permeated both the outer and inner cytoplasmic membranes of Escherichiacoli. This was determined by measuring changes in the fluorescence of probe N-phenyl-1-naphthylamine and detectingcytoplasmic β-galactosidase released during the interaction of peptides with E. coli cells. Transmission electron microscopyrevealed that treatment of E. coli with one of the COD analogs caused leakage of bacterial content mainly from the septalareas of the cells. Copyright © 2014 European Peptide Society and John Wiley & Sons, Ltd.

Additional supporting information may be found in the online version of this article at the publisher’s web site.

Keywords: antimicrobial peptides; wild-bee venom; CD spectroscopy; large unilamellar vesicles; membrane permeabilization; electronmicroscopy

* Correspondence to: Václav Čeřovský, Institute of Organic Chemistry and Biochem-istry, Academy of Sciences of the Czech Republic, Flemingovo nám. 2, 16610Prague 6, Czech Republic. E-mail: cerovsky@uochb.cas.cz

a Institute of Organic Chemistry and Biochemistry, Academy of Sciences of theCzech Republic, Flemingovo nám. 2, 16610 Prague 6, Czech Republic

b Faculty of Science, Department of Biochemistry, Charles University in Prague,Hlavova 8, 12843 Prague 2, Czech Republic

c Faculty of Science, Department of Zoology, Charles University in Prague, Viničná7, 12843 Prague 2, Czech Republic

885

Introduction

Cationic antimicrobial peptides (AMPs) isolated from insectscomprise the most abundant group among those listed in theantimicrobial peptide database [1]. These are usually classified onthe basis of their sequences and structural features into severalcategories. The linear peptides, which can form α-helical structureand do not contain cysteine residues, constitute the most studiedgroup with respect to their mechanism of action. These peptidesconsist of 10–50 amino acid residues, have a net positive chargeof +2 to +10 because of considerable content of basic amino acidsresidues, and have a substantial proportion of hydrophobicresidues. Upon interaction with the cell membrane or in amembrane-mimicking environment, they adopt highly amphi-pathic conformations with hydrophobic and hydrophilic residuesseparated into distinct regions on the molecular surface [2–4]. Thehydrophilic cationic regions interact electrostatically with theanionic components of themicrobial cell envelope and cytoplasmicmembrane, which is followed by the insertion of the peptide intothe membrane accompanied by the interactions of hydrophobicpeptide’s regions with lipid bilayer of the membrane. This finally

85–895

causes disruption of the membrane’s structure through diversemechanisms, which have a lethal impact on bacteria [2–7]. Somepeptides may penetrate the cell membrane and target key intracel-lular components or interfere with bacterial metabolic pathways [6].An intrinsic distinction in composition between bacterial andeukaryotic membranes provides a basis for the preference ofcationic AMPs to act upon a specific membrane. The prevalentnet negative charge of bacterial membranes due to the composi-tion of their phospholipids plays a major role in the attraction of

Copyright © 2014 European Peptide Society and John Wiley & Sons, Ltd.

ČUJOVÁ ET AL.

886

cationic AMPs, while membranes of eukaryotic cells enriched inzwitterionic phospholipids and cholesterol (CH) are indifferentto the AMPs [4,7].

Studies of AMP interactions with membranes on molecular levelare important for understanding processes of membrane disrup-tion and permeabilization. Liposomes with compositions reflectingthose of bacterial or eukaryotic membranes constitute a simple butvaluable tool for learning the principles of peptide-membraneinteractions and may help to explain the mechanism of bacterialsensitivity – and of eukaryotic cells’ insensitivity – to AMPs [8,9].

Numerous AMPs belonging to the class of α-helical peptideshave been isolated from the venom of stinging hymenopteranssuch as honeybees, wasps, ants, and bumblebees [10]. In our labo-ratory, we have identified several novel peptides in this category inthe venom of wild bees [11–13]. These are composed of 12–18amino acid residues and show significant antibacterial and antifun-gal activities, accompanied by varying levels of toxicity to eukary-otic cells that are expressed as hemolytic activity in relation to redblood cells.

In this study, we describe the isolation and structural charac-terization of novel AMP designated as codesane (COD), whichwe have identified in the venom of the wild bee Colletes daviesanus(a common summer species nesting solitary and collecting pollenfrom Asteraceae flowers). The study further examines the interac-tion of COD and its analogs with artificial membranes, namelyliposomes constructed from various phospholipids.

For the experiments, we prepared negatively charged liposomes[large unilamellar vesicles (LUVs)] composed of phosphatidylethanol-amine (PE) and phosphatidylglycerol (PG) as a model of Escherichiacolimembranes, while liposomes composed of uncharged phos-phatidylcholine (PC) and CH served as model membranes oferythrocytes. We also examined CODs’ permeabilization of theouter and inner membranes of E. coli and investigated the effectof one selected analog on the morphology of E. coli by transmis-sion electron microscopy.

Figure 1. RP-HPLC profile of Colletes daviesanus venom extract at 220nm.An elution gradient of solvents from 5% to 70% acetonitrile/water/0.1%TFA was applied for 60min at flow rate 1ml/min.

Materials and Methods

Materials

Fmoc-protected amino acids and Rink Amide MBHA resin werepurchased from IRIS Biotech GmbH (Marktredwitz, Germany).Phospholipids 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine (PE),1,2-dioleoyl-sn-glycero-3-[phospho-rac-1-glycerol] (PG), and 1,2-dioleoyl-sn-glycero-3-phosphocholine (PC) were purchasedfrom Avanti Polar Lipids (Alabaster, AL, USA). CH, calcein, N-phenyl-1-naphthylamine (NPN), 2-nitrophenyl β-D-galactopyranoside (ONPG),and acrylamide were supplied from Sigma-Aldrich. All otherreagents, solvents, buffers, and HPLC-grade acetonitrile were ofthe highest purity available from commercial sources. As testorganisms, we used the following: Bacillus subtilis 168, kindlyprovided by Prof. Hiroshi Yoshikawa (Princeton University); E. coliB andMicrococcus luteus No. CCM 144, from the Czech Collectionof Microorganisms, Brno, Czech Republic; Staphylococcus aureusand Pseudomonas aeruginosa, obtained as multiresistant clinicalisolates; and Candida albicans (F7-39/IDE99), from the mycolog-ical collection of the Faculty of Medicine, Palacky University,Olomouc, Czech Republic. Tetracycline, polymyxin, Luria-Bertani(LB) broth, and LB agar were obtained from Sigma-Aldrich,Prague, Czech Republic. Blood samples for determination ofhemolytic activity were obtained from healthy volunteers.

wileyonlinelibrary.com/journal/jpepsci Copyright © 2014 European Pe

Sample Preparation and Peptide Purification

Bee specimens of C. daviesanus were collected in the area of the cityof Kladno in the Czech Republic andwere then kept frozen at�20 °C.The venom reservoirs of five individuals were dissected under amicroscope, and their contents were extracted using a mixtureof acetonitrile and water (1 : 1) containing 0.5% TFA (40 μl). Theextract was centrifuged, and the supernatant was fractionatedby RP-HPLC as described elsewhere [11–13]. The selectedfractions (peaks detected at 220 nm, Figure 1) were manuallycollected, the solvent evaporated in a speed-vac, and thematerial tested for antimicrobial activity against M. luteus by dropdiffusion assay. The active fraction was further analyzed by massspectrometry (MS) and subjected to Edman degradation.

Mass Spectrometry

Mass analyses of the peptides were performed in positive ionizationmode using a Micromass Q-TOF micro mass spectrometer (WatersCorporation, Milford, MA, USA) equipped with an electrospray ionsource in the 150–2000Da range. A mixture of acetonitrile andwater (1 : 1) containing 0.1% formic acid was continuously deliveredto the ion source at a 0.1ml/min flow rate. Samples dissolved in themobile phase were introduced using a 2μl loop. The capillaryvoltage, cone voltage, desolvation temperature, and source tem-perature were 3.5 kV, 20V, 150 °C, and 90 °C, respectively.

Peptide Sequencing by Edman Degradation

The N-terminal amino acid sequence was determined on theProcise Protein Sequencing System (PE Applied Biosystems, 491Protein Sequencer, Foster City, CA, USA) using the manufacturer’spulse-liquid Edman degradation chemistry cycles.

Peptide Synthesis

COD and its analogs were synthesizedmanually using a solid-phasemethod according to the Nα-Fmoc chemistry protocol in 5mlpolypropylene syringes with a Teflon filter at the bottom on RinkAmide MBHA resin (110mg). Fmoc-protected amino acids (4 eq,0.28mmol) were coupled using N,N -diisopropylcarbodiimide(7 eq, 0.49mmol) and 1-hydroxybenzotriazole (5 eq, 0.35mmol) ascoupling reagents in N,N-dimethylformamide (DMF, 0.6mL). Thefree amino group conversion during the coupling was monitoredusing 1% bromophenol blue indicator in DMF. Deprotectionsof the α-amino group were performed with 20% piperidine inDMF. Peptides were fully deprotected and cleaved from the resinusing a 2ml mixture of TFA/thioanisole/H2O/1,2-ethanedithiol/

ptide Society and John Wiley & Sons, Ltd. J. Pept. Sci. 2014; 20: 885–895

INTERACTION OF ANTIMICROBIAL PEPTIDE WITH VESICLES

triisopropylsilane (90 : 3 : 2.5 : 2.5 : 2) for 3.5h, and the peptides werethen precipitated with tert-butyl methyl ether. Crude peptides werefurther purified by preparative HPLC using a Vydac C-18 column(10×250mm; 5μm) at a 3ml/min flow rate using the solventgradient ranging from 5% to 70% acetonitrile/water/0.1% TFA over60min. The purity and identity of peptide products were checkedby analytical HPLC and electrospray ionization MS (ESI-MS), respec-tively (Table 1).

Antimicrobial Assay

Qualitative estimations of antimicrobial properties were madeusing the drop diffusion test on Petri dishes by double-layertechnique, as described elsewhere [11]. The antimicrobial activityof CODs against three Gram-positive (M. luteus, B. subtilis, and S.aureus) and two Gram-negative (E. coli and P. aeruginosa) bacterialstrains and against C. albicans was expressed as the minimuminhibitory concentration (MIC). MICs were established by observingbacterial growth in multi-well plates [11]. Mid-exponential phasebacteria were added to individual wells containing solutions ofthe peptides at different concentrations in LB broth (0.5% NaCl).The final volume in the wells was 0.2ml, and the final peptideconcentration was in the range of 0.5 to 100μM. The plates wereincubated at 37 °C for 20h while being continuously shaken in aBioscreen C instrument (Oy Growth Curves AB Ltd., Helsinki,Finland). Absorbance was measured at 540nm every 15min, andeach peptide was tested at least three times in duplicates.Routinely, 1 × 104 to 5×104 colony forming units of bacteria perwell were used to determine activity. Tetracycline in concentrationrange 0.5–50μM was tested as a standard.

Hemolytic Activity

Hemolytic activity is expressed as the peptide concentrationrequired for lysis of 50% of human erythrocytes in the assay (LC50values). The peptides in different concentrations ranging from 2.5to 200μM were incubated with 4% suspension of human red bloodcells in physiological solution at a final volume of 0.2ml for 1 h at37 °C. The samples were then centrifuged at 250 g for 5min, andthe absorbance of the supernatant was measured at 540 nm usinga Tecan Infinite M200 PRO reader (Tecan Austria GmbH, Austria).Red blood cells incubated in physiological solution were used asblank control for zero hemolysis and cells treated by 0.2% TritonX-100 served as control for 100% hemolysis. Each peptide wastested in duplicate in at least two independent experiments.

Table 1. Primary structures, MS data, and RP-HPLC retention times (tR)of COD and its analogs

Peptide Sequence Monoisotopic molecularmass (Da)

tR (min)

Calculated Found

COD GMASLLAKVLPHVVKLIK 1915.22 1915.5 36.08

COD-1 GMAKLLAKVLPHVVKLIK 1956.28 1956.5 33.22

COD-2 GMASKLAKVLPHVVKLIK 1930.23 1930.4 28.16

COD-3 GMASLLAKVLPKVVKLIK 1906.25 1906.5 37.14

COD-4 GMASLWAKVLPHVVKLIK 1988.21 1988.4 33.58

COD-5 GMASLwAKVLPHVVKLIK 1988.21 1988.4 31.02

Bold letters show amino acid residues replaced. D-amino acid isshown as a bold lower case letter.

J. Pept. Sci. 2014; 20: 885–895 Copyright © 2014 European Peptide Society a

Circular Dichroism

Circular dichroism (CD) spectra were collected by Jasco 815spectrometer (Tokyo, Japan) in spectral range 185–300nm overtwo computer controlled scans using a 0.1 cm quartz cell at roomtemperature with following setup: 0.5nm step resolution, 1 nmbandwidth, 20 nm/min scanning speed with 8 s response time, or5 nm/min scanning speed with 32 s response time. Peptides weremeasured in water and in the presence of sodium dodecyl sulfate(SDS) (constant peptide concentration 100μM and SDS concentra-tion range 0.016–16mM), in 50mM phosphate buffer, pH7.4 andin the presence of LUVs composed of PE/PG (7 : 3, mol/mol), PC/CH (9 : 1, mol/mol), and PC/CH (6 : 4, mol/mol) (constant peptideconcentration 50μM, LUVs concentration range 0.015–3mM). Afterbaseline correction, the final spectra were expressed as a molarellipticity θ (deg. cm2/dmol) per residue. The α-helix fraction (fH)was calculated assuming a two-state model [14,15] and using theonline CD analysis program Dichroweb (http://dichroweb.cryst.bbk.ac.uk) [16].

Preparation of LUVs

Appropriate amounts of phospholipid stock solutions were mixedin chloroform (1ml) to obtain the desired compositions of LUVs:PE/PG (7 : 3, mol/mol), PC/CH (9 : 1, mol/mol), and PC/CH (6 : 4,mol/mol). The chloroform was then removed by a stream ofnitrogen, resulting in deposition of phospholipids as a film on thewall of the test tube. Residual solvent was removed by vacuumdrying overnight. LUVs were prepared by vortexing the driedphospholipid film in 50mM Na2HPO4.2H2O containing 150mM NaCland 1mM EDTA, pH7.4 (1ml). The suspensions were furtherprocessed with five cycles of freezing and thawing and subsequentlyextruded 31 times through a 100nm pore size polycarbonatemembrane (Nuclepore Track-EtchedMembranes, Avanti Polar Lipids)using an Avanti Mini-Extruder (Avanti Polar Lipids). Calcein-entrapped vesicles were prepared similarly as in the preceding textusing the same buffer, which contained 70mM calcein. The non-entrapped calcein was removed by a mini-column centrifugationmethod in a plastic syringe (3×5 cm, plugged with filter pad) filledwith hydrated Sephadex G-50 coarse gel [17,18]. Vesicle suspension(400μl) was added dropwise on top of the gel in the syringe, andthe calcein-entrapped vesicles were eluted by centrifugation at2000 rpm for 10min. The non-entrapped calcein remained boundto the gel, while vesicles passed through the gel and were collected.

887

Calcein Leakage

Leakage of calcein from the LUVs was monitored by measuringfluorescence intensity at an excitation wavelength of 490 nm andemission wavelength of 520 nm on a Tecan Infinite M200 PROreader (Tecan Austria GmbH, Austria). Particular mixtures of CODsand LUVs in a 96-well microtiter plate contained 50μl of the givenpeptide in the final concentrations 0.25, 0.5, 0.75, 1.0, 1.5, 2.5, 5, and10μM and 50μl of calcein-entrapped liposomes (30μM). Maximumleakage of calcein from LUVs was achieved by 0.2% Triton X-100in buffer (50μl). The percentage of calcein leakage caused by CODswas calculated using the equation 100× (F� F0)/(Ft� F0), where F isthe fluorescence intensity gained by CODs treatment, and F0 and Ftare the intensities in the absence of CODs and in the presence ofTriton X-100, respectively [19]. Each peptide was tested in at leastfour independent experiments.

nd John Wiley & Sons, Ltd. wileyonlinelibrary.com/journal/jpepsci

ČUJOVÁ ET AL.

888

Intrinsic Tryptophan Fluorescence

The effect of model membranes on the fluorescence spectra of thetryptophan-containing codesane analogs (COD-4 and COD-5) wasmonitored in 96-well plates on the Tecan Infinite M200 PRO reader.LUVs in given concentrations up to 600μM in 50mM Na2HPO4.2H2Ocontaining 150mMNaCl and 1mM EDTA, pH7.4 (240μl) were addedto the wells of the microtiter plate. This was followed by addingCODs in the same buffer (60μl) to achieve 10μM final peptide con-centration. The mixtures were then incubated at 25 °C for 15min.The fluorescence of tryptophan residue was excited at 280nm,and the emission was recorded at wavelengths from305 to 400nm.

Quenching of Trp Fluorescence by Acrylamide

Quenching of the tryptophan fluorescence of COD-4 and COD-5was effected using acrylamide as the quencher in 96-well plateson the Tecan Infinite M200 PRO reader. Concentrations of thesubstances in the phosphate buffer (as in the preceding text in eachwell) were as follows: peptide (10μM), LUVs (300μM), and acrylam-ide (0, 60, 120, 240, or 300mM). The total volume in each well was300μl. Mixtures were incubated at 25 °C for 15min. The fluores-cence of tryptophan was excited at 295nm (instead of 280nm) toreduce the absorbance of acrylamide, as described elsewhere[20,21]. Data were analyzed according to the Stern–Volmer equa-tion: F0 /F=1+ Ksv(Q), where F0 and F are fluorescence intensitiesat 350nm of the peptide in the absence and in the presence ofacrylamide, respectively. Ksv represents the Stern–Volmer quenchingconstant, and Q represents the acrylamide concentration [22].

Outer Membrane Permeabilization Assay

The outer membrane permeation activity of CODs was evaluated bymeasuring the NPN fluorescence probe uptake in 96-well plates onthe Tecan Infinite 200 spectrometer. E. coli were grown in LB brothto mid-exponential phase, then washed in 5mM sodium HEPESbuffer (pH=7.2), and re-suspended in the same buffer to obtain finaloptical density of 0.4 at 600nm. The peptides in this buffer (90μl)were then mixed with the cell suspension of E. coli (100μl) andNPN (10μl). Final peptide concentration in the wells was 0.25, 0.5,0.75, 1.0, 1.5, 2.5, 5, and 10μM, and final NPN concentration was10μM. Changes in fluorescence were recorded for 15min using exci-tation and emission wavelengths of 350 and 420nm, respectively.

Inner Membrane Permeabilization Assay

Determination of inner membrane permeabilization by CODs wascarried out by measuring the activity of β-galactosidase releasedfrom E. coli cells [23,24]. Cells of E. coli that had been grown in thepresence of 2% lactose to mid-exponential phase were washedtwice in phosphate buffer (10mM sodium phosphate, 100mM NaCl,pH=7.4) and re-suspended in that buffer to the same opticaldensity as in the preceding text. Bacteria suspension (100μl) wasmixed with ONPG (10μl) and peptides (90μl) so that the final ONPGconcentration was 1.5mM, and the final peptide concentration ineach well was 1, 1.5, 2.5, or 5μM. The plates were incubated at37 °C for 90min while being continuously shaken in a Bioscreen Cinstrument (Oy Growth Curves AB Ltd., Helsinki, Finland). Hydrolysisof ONPG was followed spectrophotometrically at 420 nm. To distin-guish between the release of β-galactosidase from the cells andONPG uptake into cells, the control experiment was carried out asfollows: bacteria suspension in phosphate buffer was treated withCOD-1 (5μM) in the absence of ONPG at 37 °C for 90min, and then,

wileyonlinelibrary.com/journal/jpepsci Copyright © 2014 European Pe

themixture was centrifuged. The presence of β-galactosidase in thesupernatant was measured after addition of ONPG based on its hy-drolysis as in the preceding text. For the comparison, the bacteriasuspension was treated with COD-1 in the presence of ONPG underthe same conditions.

Transmission Electron Microscopy

The effect of COD-1 peptide on the structure of Gram-negativeE. coli bacteria was studied by negative staining method. Bacterialcells were treated with peptide at two different concentrations(30 and 100μM) for 60min. Untreated E. coli bacteria were used asa control. Peptide-treated bacterial cells and untreated control cellswere adsorbed on parlodion-carbon-coated copper grids for5–10min. After short washing, the samples were negatively stainedusing 0.25% phosphotungstic acid (pH7.3) with 0.01% bovine se-rum albumin in dH2O for 30–40 s and then air dried. The specimenswere then examined using a JEOL JEM 1011 transmission electronmicroscope (JEOL Ltd., Tokyo, Japan) operating at 60 kV.

Result and Discussion

Peptide Isolation and Primary Structure Determination

Fractionation of the extract from five bee venom reservoirs byRP-HPLC resulted in several intense peaks detected at 220 nm(Figure 1). Only one (peak at tR=35.7) of the 16 collected fractionsexhibited antimicrobial activity againstM. luteus. Its molecular massmeasured by internally calibrated ESI-Q-TOF MS was manuallycalculated from the m/z values of multiple (2x, 3x, and 4x) chargedmolecular ions found in the mass spectrum (Figure S1), resulting in amonoisotopic molecular mass of 1914.9. This is in agreement withthe calculated value of 1915.22 based on the sequence determinedby Edmandegradation. The Edmandegradation (see Supporting Infor-mation) gave the entire sequence as follows: GMASLLAKVLPHVVKLIK.The deconvoluted mass of this peptide designated codesane (COD)signifies that the C-terminus is amidated.

Peptide Synthesis and Design of Analogs

Codesanes were prepared by standard Fmoc chemistry yielding onaverage 120mg of crude products. This material was furtherpurified by preparative RP-HPLC to provide peptides of analyticalHPLC purity ranging from 96% to 99%. The retention times andmolecular masses of the synthetic peptides obtained byMS analysesconfirming their identities are shown in Table 1 (see also Figure S2for mass spectra).As suggested by Schiffer–Edmundsonwheel projection (Figure 2)

and by CDmeasurement (Figure 3), COD is predisposed to adopt an‘ideal’ amphipathic α-helical secondary structure in the presence ofmembrane-mimicking SDS. COD has 11 hydrophobic residuessituated on one side of the helix and the remaining seven hydro-philic residues, including three basic lysines, on the opposite side.The proline residue in position 11 may impose a kink in the helicalstructure. The cationicity of COD (net charge +4) in combinationwith hydrophobic residues present in the sequence creates its anti-microbial properties. In order to study the interactions of COD withartificial membranes (liposomes and membranes of E. coli cells), weprepared analogs in which we (i) replaced Ser4, Leu5, and His12with Lys, thus resulting in three analogs COD-1, COD-2, andCOD-3, respectively, each having one additional net positivecharge and (ii) replaced Leu6 situated in the center of the

ptide Society and John Wiley & Sons, Ltd. J. Pept. Sci. 2014; 20: 885–895

Figure 2. Schiffer–Edmundson wheel projection of COD. Hydrophobicamino acid residues are in red, and hydrophilic residues in blue (lysines)and black (glycine, serines).

INTERACTION OF ANTIMICROBIAL PEPTIDE WITH VESICLES

hydrophobic segment with Trp or D-Trp in order to use the trypto-phan fluorescence of COD-4 and COD-5 (Figure 2).

Biological Activity

COD showed high antimicrobial activity against Gram-positivebacteria and moderate activity against Gram-negative E. coli.However, its activity against pathogenic P. aeruginosa and yeastC. albicanswas low. The toxicity to human red blood cells determinedas LC50=105μM in a hemolytic test was notably high, clearlyreflecting its high amphipathic and hydrophobic character (Table 2).Based on these results and on our empirical data [12,13], we have

synthesized a set of analogs in which selected amino residues in the

Figure 3. UVcircular dichroism spectra of COD in the presence of SDS (A) and PE :Pratio, respectively. Because of the opaque sample at higher large unilamellar vesicl

Table 2. Antimicrobial and hemolytic activity of COD and its analogs

Antimicrobial activit

Peptide M.l. B.s. S.a.

COD 1.7 2.5 3.7

COD-1 0.8 1.3 2.7

COD-2 1.4 3.1 35

COD-3 1.2 3.4 3.5

COD-4 1.7 4.7 6

COD-5 2.8 2.3 28.3

MIC, minimum inhibitory concentration; LC50, peptide concentration lysing 50%Staphylococcus aureus; E.c., Escherichia coli; P.a., Pseudomonas aeruginosa; C.a.

J. Pept. Sci. 2014; 20: 885–895 Copyright © 2014 European Peptide Society a

sequence were replaced by L-Lys in order to enhance the cationicityof the peptide and promote the increase of its antimicrobial activityagainst some bacteria while decreasing its hemolysis. Note that theincrement of peptide net positive charge does not always improvethe biological activity of α-helical AMPs as described, e.g. in thestudy of temporin analogs [25].

We expected that the incorporation of L-Trp or D-Trp into thesequence at the same position would not affect some structuralparameters of the peptide such as cationicity, hydrophobicity, andamphipathicity. On the other hand, we anticipated an impact onpeptide α-helicity, especially in the case of incorporating D-Trp, whichwould thus influence both antimicrobial and hemolytic activities.

The increment of peptide net positive charge due to the singlereplacement of Ser4 and His12 with Lys within the hydrophilicsegment of α-helix (Figure 2) resulted in a decrease of hemolyticactivity in the COD-1 and COD-3 analogs, but this had no substan-tial effect on antimicrobial activity. Only the COD-1 analog showedenhanced potency against Gram-negative P. aeruginosa, and bothanalogs exhibited twofold higher activity against E. coli andC. albicans. The substitution of Lys in the perimeter of the hydro-phobic segment (Leu5) rendered a parallel decrease in both the anti-microbial and hemolytic activities of COD-2. Replacing Leu6 situatedin the middle of the hydrophobic segment of the α-helix with L-Trpproduced the analog COD-4 having hydrophobicity and amphi-pathicity comparable with those of COD. It retained antimicrobialas well as hemolytic activities roughly comparable with those ofthe natural peptide. The substitution made in the same position byD-Trp resulted in the COD-5 analog, which showed a remarkable dropin activity against pathogenic S. aureus and P. aeruginosa in parallelwith a substantial decrease in hemolytic activity.

G (7 : 3) (B) at the various SDS/peptide (S/P) and lipid/peptide (L/P) concentrationes concentrations the spectra were measured from 195nm in the case of B.

y MIC (μM) Hemolysis

E.c. P.a. C.a. LC50 (μM)

3.7 31.7 25 105

1.5 17.5 13.1 150

2.9 100 56.3 >200

1.8 47.5 12.1 >200

2.3 53.3 8.7 162.5

5.2 >100 20.8 >> 200

of human erythrocytes.M.l.,Micrococcus luteus; B.s., Bacillus subtilis; S.a.,, Candida albicans.

nd John Wiley & Sons, Ltd. wileyonlinelibrary.com/journal/jpepsci

889

ČUJOVÁ ET AL.

890

Structural Characterization of CODs by Circular Dichroism

The CD spectra of COD and its analogs measured in watershowed a predominantly unordered structure (negative Cottoneffect around 197 nm) for all peptides (Figures 3 and S3). Thesmall differences in spectral intensity among analogs reflectthe particular amino acids substitutions in their sequences. Thelowest intensity was for COD-5, in which D-Trp is substituted forLeu6 (Figure S3). The CD spectra underwent considerablechanges upon adding increasing concentrations of SDS. Thespectra demonstrate a predominantly α-helical structure atconcentrations of SDS> 2mM, as indicated by the appearanceof typical bands at 207 and 222 nm (Figures 3 and S3).

In the presence of low SDS concentrations (<2mM), COD andsome of its analogs (COD-1, COD-2, and COD-5) are mostly in aβ-sheet structure with a low proportion of unordered structure.The substitutions of His12 by Lys (COD-3) and Leu6 by L-Trp (COD-4)modified the structural change dynamics such that the helicalstructure of these two analogs was more pronounced even in thepresence of 0.16mM SDS, with a noticeable difference of COD-4helical content (36%) versus its COD-5 enantiomer (23%) (Table 3).

The helical content (57%) observed for COD-3 at high SDSconcentrations is slightly superior to that achieved for thenatural peptide (51–55%). Relatively low helical content (41%and 37–44%) at high SDS concentration was observed for the an-alogs COD-4 and COD-5 containing L-Trp and D-Trp, respectively(Table 3). The shoulder around 230 nm observed in the CDspectrum of COD-4 in 8 and 16mM SDS (Figure S3) is probablydue to the exciton couplet between Phe11 and His12 [26]. Inthe cases of the other studied analogs, this spectral bandoriginating because of interaction between the side chains ofthese aromatic amino acids is overlapped by the more intensenegative spectral band at 224 nm. Its presence together withits intensity reflects the portion of α-helical conformation.

The α-helical contents of all measured peptides in the presenceof negatively charged PE/PG (7 : 3) LUVs were noticeably lower thanthosemeasured in the presence of SDS (Table 3 and Figure S3). TheCD spectra typical for α-helical content with two characteristicminima at 208 and 224nm were observed for all peptides in theconcentration ratio lipid/peptide= 60. Natural peptide (COD) andTrp-containing analogs (COD-4 and COD-5) showed those charac-teristic spectra already at the ratio lipid/peptide= 20 (Table 3 andFigure S4). On the other hand, the spectra of all studied peptidemeasured in the presence of neutral PC/CH (9 : 1) LUVs showed very

Table 3. Calculated α-helical fractions (fh) in percentages of COD and its an(7 : 3) at the peptide concentration 100 and 50μM, respectively

Peptide α-Hel

[SDS]/[peptide]

0a 0. 16 1.6 20 40 80

COD 16 18 27 44 47 51

COD-1 16 17 28 52 52 49

COD-2 15 16 25 50 47 50

COD-3 17 19 34 55 56 57

COD-4 16 17 36 43 43 41

COD-5 15 16 23 39 41 44

aIn water.bIn 50mM phosphate buffer, pH7.4.

wileyonlinelibrary.com/journal/jpepsci Copyright © 2014 European Pe

weak tendency to form α-helical structure, while in the case of PC/CH (6 : 4) LUVs, no formation of helical structure was observed(Figures S5 and S6). The negative minimum in the spectra at198nmcharacteristic for all peptide in the absence of liposomeswereshifted to 204nm in the presence of LUVs at higher [lipide]/[peptide]ratio and was accompanied by decrease of spectral intensity indicat-ing probable formation of β-aggregates (Figures S5 and S6).

Calcein Leakage

The leakage of fluorescent dye calcein from LUVs was a measure ofmembrane-permeabilizing activity of codesanes. Cationic AMPs usu-ally exhibit membrane-permeabilizing activity preferentially againstnegatively charged liposomes, whereas their activity against neutralliposomes made from zwitterionic phospholipids is restricted [19].This activity is generally believed to reflect the ability of peptidesto permeabilize or disrupt bacterial and erythrocyte cell membranes.LUVs composed of the mixture of zwitterionic PE and anionic PGin a 7 : 3 molar ratio were used as a model of bacterial cell mem-brane. LUVs used as the model of mammalian membranes wereconstructed from zwitterionic PC and CH in the molar ratio 6 : 4or 9 : 1. The CH component was intended to stabilize the lipidbilayer, thereby reducing membrane perturbation and thuscalcein leakage. The composition of liposomes consisting of40% CH is close to that typical of erythrocytes where the CH/phospholipid molar ratio is 0.84 [27]. Liposomes containing10% CH are frequently used in experiments as a general modelfor the membranes of mammalian cells [28].As shown in Figure 4a, COD-1 and COD-3 already caused 100%

calcein leakage from negatively charged LUVs composed of PE :PG (7 : 3) at a very low concentration (2.5 μM), whereas COD andCOD-4 reached this maximum at 5 μM concentration. On theother hand, COD-2 and COD-5 caused calcein leakage only athigh concentrations. These results roughly correspond to theMIC values of the analogs (Table 2), which indicate COD-1 tobe the most active peptide of the series and COD-2 and COD-5to be the weakest ones.Calcein leakages caused by CODs from neutral LUVs made from

PC and CH were markedly lower than those from negativelycharged liposomes (Figure 4b and c). The percentage of leakagewas clearly dependent on the content of CH in the liposome. Whenthe liposomes contained 40% CH, the leakage caused by COD-2and COD-5 was negligible (Figure 4b). The most hemolytic

alogs determined by CD spectroscopy in the presence of SDS and PE : PG

ical fraction (fh)

[PE : PG (7 : 3)]/[peptide]

160 0b 0.3 6 20 60

55 10 10 13 25 34

54 10 10 12 16 36

48 10 10 12 13 32

57 10 10 13 14 44

41 11 11 14 38 46

37 10 10 13 25 23

ptide Society and John Wiley & Sons, Ltd. J. Pept. Sci. 2014; 20: 885–895

Figure 4. Calcein leakage from large unilamellar vesicles of various phos-pholipids composition as a function of peptides concentration. (A) PE/PG7 : 3, (B) PC/CH 6 : 4, (C) PC/CH 9 : 1, ( ) COD, ( ) COD-1, ( ) COD-2,( ) COD-3, ( ) COD-4, and ( ) COD-5.

Figure 5. Blue shift of the tryptophan fluorescence maximum of COD-4 and COtrations. (A) COD-4, (B) COD-5, ( ) PE/PG 7 : 3, ( ) PC/CH 6 : 4, and ( ) PC/CH

Figure 6. The uptake of NPN by E. coli cells as a measure of its outer mem-brane permeation at various concentrations of codesanes. ( ) COD, ( )COD-1, ( ) COD-2, ( ) COD-3, ( ) COD-4, ( ) COD-5, and ( ) poly-myxin as a control.

INTERACTION OF ANTIMICROBIAL PEPTIDE WITH VESICLES

J. Pept. Sci. 2014; 20: 885–895 Copyright © 2014 European Peptide Society a

peptide of the series, the natural COD, caused only 31% leakageat 10 μM concentration (Figure 4b). On the other hand, theliposomes comprised of 10% CH were considerably moresusceptible to the action of CODs (Figure 4c). The disruption ofLUVs by COD at 10 μM peptide concentration resulted in 100%leakage of calcein. The order of CODs propensity to disruptneutral LUV mimicking membranes of eukaryotic cells wasCOD>COD-4 = COD-1>COD-3>COD-5>COD-2. This more orless reflects the trend in their hemolytic activities, being thehighest for COD and the lowest for COD-5.

In general, the results of this study correlate also with the struc-tural parameters of codesanes. For example, the increment of thenet positive charge of the analogs COD-1 and COD-3 had an appar-ent effect on the calcein leakage of negatively charged liposomes.Further, the peptide with the lowest hydrophobicity (COD-2, witha Leu–to–Lys substitution) and the peptide with the lowest propen-sity to form α-helix in the presence of SDS (COD-5, Leu–to–D-Trpsubstitution) exhibited the weakest potency to permeabilize LUV,independent of their composition. These observations may be at-tributed to hydrophobic interactions of AMP with the lipidic bilayerof the vesicle membrane and to the implication of AMP α-helicityfor permeabilization activity and for their hemolytic activity. Onthe other hand, the fact that none of the studied peptides adoptedhelical structure in the presence of neutral LUVs as shown by theirCD spectra (Figures S5 and S6) suggests that the interaction ofthese peptides with the hydrophobic core of the bilayer is not sodecisively dependent on their α-helicity.

D-5 in the presence of various liposomes as a function of liposomes concen-9:1. Concentration of peptides was 10μM.

nd John Wiley & Sons, Ltd. wileyonlinelibrary.com/journal/jpepsci

891

Table 4. Tryptophan fluorescence emission maxima (λmax) of two CODanalogs and Stern–Volmer (Ksv) constants obtained in the acrylamidequenching experiments in buffer and in liposomes of variouscompositions

COD-4 COD-5

λmax Ksv λmax Ksv

(nm) (M�1) (nm) (M�1)

Buffer 346 13.2 346 19.1

PE : PG 7 : 3 323 1.6 324 1.7

PC : CH 9 : 1 333 4.2 343 5.5

PC : CH 6 : 4 336 3.7 342.5 5.1

ČUJOVÁ ET AL.

892

Intrinsic Tryptophan Fluorescence

For this study, we prepared COD-4 and COD-5 analogs in which theLeu at position 6 was replacedwith L-Trp or D-Trp, respectively. Eachsubstitution had a different effect on the biological activity of theanalogs. While COD-4 exhibited antimicrobial as well as hemolyticactivity roughly comparable with those of the natural peptide, theCOD-5 analog showed a considerable decrease in both antimicro-bial and hemolytic activities (Table 2). Tryptophan fluorescence isvery sensitive to the local environment, and therefore, its measure-ment provides a tool for monitoring the membrane localization ofpeptides containing Trp. The transition of peptides from a polarto a non-polar environment within the membrane is typicallyaccompanied by the shifting of its emission spectra maximum tothe shorter wavelengths – an effect known as blue shift [29].

In the absence of LUVs, both of these two COD analogs reachedtheir maximum fluorescence at 346 nm, which is typical for Trp fluo-rescence in an aqueous solution. When we measured the fluores-cence emission spectra in the presence of an increasingconcentration of LUVs containing anionic phospholipids PE : PG(7:3), we observed a significant shift of the spectral maximum from346nm to shorter wavelengths up to 323nm (illustrated inFigure 5). That shift was more prominent with COD-4 comparedwith COD-5, which accords with its more than twofold greater anti-microbial activity against E. coli and other bacteria.

In contrast, we observed a noticeably shorter shift of the spectralmaxima upon interactions of CODswith zwitterionic LUVs consistingof PC and CH (Figure 5). In particular, increasing the concentration ofLUVs induced only insignificant changes in the fluorescence spectraof COD-5 (Figure 5b). That was as anticipated in view of its minimalhemolytic activity (Table 2). On the other hand, the gradual decreaseof the Trp fluorescence maximum in the case of COD-4 (Figure 5a)from 346 to 331nm and 329.5 nm in the presence of increasingconcentrations of LUVs consisting of PC : CH (6 : 4) and PC : CH(9 : 1), respectively, was in accordance with its moderate hemolyticactivity. COD-5 showed no difference between the interactions withthe liposomes containing either 10% or 40% CH (Figure 5b), butCOD-4 somewhat more easily penetrated the membrane of LUVscomprised of 10% CH (Figure 5a).

Quenching of Tryptophan Fluorescence by Acrylamide

We have used acrylamide as a water-soluble quencher oftryptophan fluorescence to study the tryptophan exposure tothe aqueous environment in the absence and presence of lipidvesicles and to determine the insertion of COD-4 and COD-5 intothe membrane. Tryptophan fluorescence intensity changes wereexpressed as Stern–Volmer constants (Ksv) (Table 4). The low Ksvvalues can be interpreted as limited accessibility of the quencherto the tryptophan side chain. The Trp residues buried inside thelipid bilayer of LUVs will be protected from interaction with thesoluble fluorescence quencher while non-buried Trp will bequenched by acrylamide.

In the presence of anionic vesicles, the quenching of the fluores-cencewas considerably reduced in comparisonwith the quenchingin the buffer solution, and it was identical for both studied CODs.This suggests that the N-terminal part of CODs with Trp is burieddeeply into the membrane, and this accords with the expectedstrong interaction of CODs with negatively charged LUV. Anincrease in quenching (higher Ksv values), which corresponds to lessprotected Trp residue, was observed for both analogs in the pres-ence of zwitterionic LUVs. A difference in fluorescence quenching

wileyonlinelibrary.com/journal/jpepsci Copyright © 2014 European Pe

between CODs observed for all studied LUVs andwith the Ksv valueslower for COD-4 reflects its higher antimicrobial as well as hemolyticactivity compared with its D-enantiomeric analog (Table 4).

Outer Membrane Permeabilization of E. coli

The ability of CODs to disrupt the integrity of the outer membraneof E. coli was determined using an NPN hydrophobic fluorescentprobe, which typically provides a very weak fluorescence signal inan aqueous environment. Upon diffusion into a hydrophobicenvironment (e.g., a damaged membrane), however, the NPNfluorescence increases significantly.As illustrated in Figure 6, COD and all its analogs caused compa-

rable increase in fluorescence (i.e., comparable outer membranepermeation of the NPN probe into the membrane). The increasewas dose-dependent and linear at low peptide concentration rangefrom 0.25 to 1.5μM. We observed maximum intensity fluorescenceof NPN immediately after adding peptide in a time-dependentmanner (data not shown). The same effect was observed withpolymyxin B, a peptide antibiotic known for its outer membranedestabilization activity [30].

Inner Membrane Permeabilization of E. coli

Disruption of the bacterial inner membrane by COD and itsanalogs was tested by measuring the activity of β-galactosidase,an enzyme which leaks from cytoplasm and catalyses hydrolysisof o-nitrophenyl-β-D-galactoside (ONPG) into o-nitrophenol. Thiswas detected by light absorption at 420 nm. The obtainedresponse curves (Figure 7) showed leakage of β-galactosidase(measured as o-nitrophenol absorption) with steady-state kinet-ics for all tested peptides at 5 μM concentrations after approxi-mately 60min (Figure 7d). That indicates a substantially lowerrate of inner membrane permeabilization compared with thatof the outer membrane, similarly as published for other shortα-helical amphipathic peptides [23]. The results also show differ-ent permeation efficiency for each peptide depending on itsstructure and antimicrobial activity, which is more obvious atlower concentrations of tested peptides (<5 μM). COD-1 is theanalog with the highest antimicrobial activity against Gram-negative bacteria and proved to permeate most efficiently, whileCOD-2 and COD-5, with the lowest antimicrobial activity, perme-ated the least effectively.In the control experiment in which the bacteria were pre-treated

with COD-1 in the absence of ONPG, we observed considerable

ptide Society and John Wiley & Sons, Ltd. J. Pept. Sci. 2014; 20: 885–895

Figure 7. Time course of the release of cytoplasmic β-galactosidase as a measure of its innermembrane permeation at various concentrations of codesanes.(A) 1μM, (B) 1.5μM, (C) 2.5μM, (D) 5μM; ( ) COD, ( ) COD-1, ( ) COD-2, ( ) COD-3, ( ) COD-4, ( ) COD-5, and (+) no peptide.

INTERACTION OF ANTIMICROBIAL PEPTIDE WITH VESICLES

release of β-galactosidase into the buffer, which was equal to therelease of the enzyme observed in the simultaneous experimentcarried out in the presence of ONPG (data not shown).The effect of CODs on membrane integrity was concentration-

dependent as already shown in the literature [24,31]. Some analogs

Figure 8. Electron micrographs of negatively stained E. coli cells treated by diCOD-1, and (E,F) 100μM COD-1. Scale bar 200nm. Bacterial cells are shown in d

J. Pept. Sci. 2014; 20: 885–895 Copyright © 2014 European Peptide Society a

were capable of permeating the bacterial cell membrane even atconcentrations (1μM) lower than their MICs (Figure 7a), similarlyas in the case of outer membrane permeabilization. These resultssupport the previous observation [32] that some lytic peptidescan induce membrane perturbation at low peptide concentrations,

fferent concentrations of COD-1 for 60min. (A) untreated cells, (B–D) 30μM

ifferent stages of disruption.

nd John Wiley & Sons, Ltd. wileyonlinelibrary.com/journal/jpepsci

893

ČUJOVÁ ET AL.

894

whereas at high concentrations, they cause significant damage tothe membrane leading to cell death.

Transmission Electron Microscopy

To examine the influence of CODs on the morphology of E. coli cells,we treated the cells with the analog COD-1 and then visualized thesecells by negative staining. Untreated control bacteria revealed nativemorphology represented by an electron-dense character and well-preserved envelope structure. However, only rare flagella wereobserved (Figure 8a). Themajority (about 60%) of bacteria treatedwithCOD-1 at a concentration of 30μM lost their electron density, which ledto electron transparency (Figure 8c), but their envelopes were stillpreserved (Figure 8b and c). Only 30% of bacterial cells revealed smallholes in their outer membranes. Some of those holes were largeenough to allow leakage of the inner bacterial content (Figure 8d).

Treatment of E. coli cells with COD-1 at a concentration of 100μM

led to electron transparency for most of the bacterial cells, andinternal structures became clearly visible. More than 70% of themshowed damage to their bacterial envelopes, and particularly inthe septal area, accompanied by a huge leakage of inner bacterialcontent, even as the membrane of their polar parts seemed to bestill preserved (Figure 8e). Some of these cells were seen to becompletely destroyed (Figure 8f).

Conclusion

The venom of wild bees represents a rich source of various α-helicalamphipathic antimicrobial peptides. The new peptide known ascodesane (COD) and described in this work shows high antimicrobialactivity against Gram-positive bacteria and the Gram-negative bacte-rium E. coli and moderate activity against pathogenic P. aeruginosa,but also significant hemolytic activity, which reflects its highly hydro-phobic and amphipathic character. Modification of the peptide’sstructural parameters by replacing certain amino acid residues inthe sequencewith Lys, Trp, and D-Trp resulted in analogswith alteredbiological activities. These analogs were suitable for studying theirinteractions with model membranes. Using the CD spectroscopy,calcein leakage assay, and tryptophan fluorescence studies, wehave established that codesanes (CODs) showed a preferenceto permeabilize membranes composed of anionic phospho-lipids, which mimic membranes of bacterial cells over those com-posed of neutral lipids and CH mimicking membranes of eukaryoticcells. Moreover, our results reflect the impact of such of CODs’ struc-tural parameters as net charge, hydrophobicity, and α-helicity ontheir biological activities. At low concentrations, CODs permeabilizedthe outer and inner membranes of E. coli cells but the latter at asubstantially slower rate. Damage to E. coli cells treated with CODanalogs was clearly seen by transmission electron microscopy.

Acknowledgements

This work was supported by the Grant Agency of the Charles Uni-versity no. 645012 and by research project RVO 61388963 of the In-stitute of Organic Chemistry and Biochemistry, Academy ofSciences of the Czech Republic. We thank Gale A. Kirking at EnglishEditorial Services, s.r.o., for her assistance with the English language.

Conflicts of Interest

The authors declare that they have no conflicts of interest.

wileyonlinelibrary.com/journal/jpepsci Copyright © 2014 European Pe

References1 Wang G, Li X, Wang Z. APD2: the updated antimicrobial peptide

database and its application in peptide design. Nucleic Acids Res.2009; 37: D933–D937.

2 Epand RM, Vogel HJ. Diversity of antimicrobial peptides and theirmechanisms of action. Biochim. Biophys. Acta 1999; 1462: 11–28.

3 Tossi A, Sandri L, Giangaspero A. Amphipathic, α-helical antimicrobialpeptides. Biopolymers (Pept. Sci.) 2000; 55: 4–30.

4 Huang Y, Huang J, Chen Y. Alpha-helical cationic antimicrobial peptides:relationships of structure and function. Protein Cell 2010; 1: 143–152.

5 Yeaman MR, Yount NY. Mechanisms of antimicrobial peptide action andresistance. Pharmacol. Rev. 2003; 55: 27–55.

6 Nguyen LT, Haney EF, Vogel HJ. The expanding scope of antimicrobialpeptide structures and their modes of action. Trends Biotech. 2011;29: 464–471.

7 Teixeira V, Feio MJ, Bastos M. Role of lipids in the interaction ofantimicrobial peptides with membranes. Prog. Lipid Res. 2012; 51: 149–177.

8 Lohner K, Blondele SS. Molecular mechanisms of membrane perturbationby antimicrobial peptides and the use of biophysical studies in the designof novel peptide antibiotics. Comb. Chem. High. T Scr. 2005; 8: 241–256.

9 Shailesh S, Neelam S, Sandeep K, Gupta GD. Liposomes: a review.J. Pharm. Res. 2009; 2: 1163–1167.

10 Kuhn-Nentwig L. Antimicrobial and cytolitic peptides of venomousarthropods. Cell. Mol. Life Sci. 2003; 60: 2651–2668.

11 Čeřovský V, Hovorka O, Cvačka J, Voburka Z, Bednárová L, Borovičková L,Slaninová J, Fučík V. Melectin: a novel antimicrobial peptide from thevenom of the cleptoparasitic bee Melecta albifrons. ChemBioChem2008; 9: 2815–2821.

12 Monincová L, BuděšínskýM, Slaninová J, HovorkaO, Cvačka J, Voburka Z,Fučík V, Borovičková L, Bednárová L, Straka J, Čeřovský V. Novelantimicrobial peptides from the venom of the eusocial bee Halictussexcinctus (Hymenoptera: Halictidae) and their analogs. Amino Acids2010; 39: 763–775.

13 Čujová S, Slaninová J, Monincová L, Fučík V, Bednárová L, Štokrová J,Čeřovský V. Panurgines, novel antimicrobial peptides from the venomof communal bee Panurgus calcaratus (Hymenoptera: Andrenidae).Amino Acids 2013; 45: 143–157.

14 Backlund B-M, Wikander G, Peeters T, Graslund A. Induction ofsecondary structure in the peptide hormone motilin by interactionwith phospholipid vesicles. Biochim. Biophys. Acta 1994; 1190: 337–344.

15 Rohl CA, Baldwin RL. Deciphering rules of helix stability in peptides.Methods Enzymol. 1998; 295: 1–26.

16 Whitmore L, Wallace BA. Protein secondary structure analyses fromcircular dichroism spectroscopy: methods and reference databases.Biopolymers 2007; 89: 392–400.

17 Fry DW, White JC, Goldman ID. Rapid separation of low molecularweight solutes from liposomes without dilution. Anal. Biochem.1978; 90: 809–815.

18 DatheM, Meyer J, Beyermann M, Maul B, Hoischen C, Bienert M. Generalaspects of peptide selectivity towards lipid bilayers and cell membranesstudied by variation of the structural parameters of amphipathic helicalmodel peptides. Biochim. Biophys. Acta 2002; 1558: 171–186.

19 Chou H-T, Wen H-W, Kuo T-Y, Lin C-C, Chen W-J. Interaction of cationicantimicrobial peptides with phospholipid vesicles and theirantibacterial activity. Peptides 2010; 31: 1811–1820.

20 Park NG, Silphaduang U, Moon HS, Seo JK, Corrales J, Noga EJ.Structure� activity relationships of piscidin 4, a piscine antimicrobialpeptide. Biochemistry 2011; 50: 3288–3299.

21 Zhu WL, Lan H, Park I-S, Kim JI, Jin HZ, Hahm K-S, Shin SY. Design andmechanism of action of a novel bacteria-selective antimicrobialpeptide from the cell-penetrating peptide Pep-1. Biochem. Bioph. Res.Commun. 2006; 349: 769–774.

22 Ma Q-Q, Shan A-S, Dong N, Gu Y, Sun W-Y, Hu W-N, Feng X-J. Cellselectivity and interaction with model membranes of Val/Arg-richpeptides. J. Pept. Sci. 2011; 17: 520–526.

23 Wang K, Yan J, Dang W, Liu X, Chen R, Zhang J, Zhang B, Zhang W,Ming K, Yan W, Yang Z, Xie J, Wang R. Membrane active antimicrobialactivity and molecular dynamics study of a novel cationic antimicrobialpeptide polybia-MPI, from the venom of Polybia paulista. Peptides2013; 39: 80–88.

24 Abbassi F, Raja Z, Oury B, Gazanion E, Piesse C, Sereno D, Nicolas P,Foulon T, Ladram A. Antibacterial and leishmanicidal activities oftemporin-SHd, a 17-residue long membrane-damaging peptide.Biochimie 2013; 95: 388–399.

ptide Society and John Wiley & Sons, Ltd. J. Pept. Sci. 2014; 20: 885–895

INTERACTION OF ANTIMICROBIAL PEPTIDE WITH VESICLES

25 Mangoni ML, Carotenuto A, Auriemma L, Saviello MR, Campiglia P,Gomez-Monterrey I, Malfi S, Marcellini L, Barra D, Novellino E, Grieco P.Structure-activity relationship, conformational and biological studies oftemporin L analogues. J. Med. Chem. 2011; 54: 1298–1307.

26 Berova N, Woody R, Nakanishi K, Polavarapu P. ComprehensiveChirooptical Spectroscopy: Applications in Stereochemical Analysis ofSynthetic Compounds, Natural Products, and Biomolecules, Wiley:Hoboken,2012.

27 Verly RM, Rodrigues MA, Daghastanli KRP, Denadai AML, Cuccovia IM,Bloch CJ, Frézard F, Santoro MM, Piló-Veloso D, Bemquerer MP. Effectof cholesterol on the interaction of the amphibian antimicrobialpeptide DD K with liposomes. Peptides 2008; 29: 15–24.

28 Gonçalves S, Teixeira A, Abade J, Neves de Medeiros L, Kurtenbach E,Santos NC. Evaluation of the membrane lipid selectivity of the peadefensin Psd1. Biochim. Biophys. Acta 1818; 2012: 1420–1426.

29 Lakowicz JR. Principles of Fluorescence Spectroscopy, SpringerScience+Business Media, LLC: New York,2006.

J. Pept. Sci. 2014; 20: 885–895 Copyright © 2014 European Peptide Society a

30 Ibrahim HR, Sugimoto Y, Aoki T. Ovotransferrin antimicrobial peptide(OTAP-92) kills bacteria through a membrane damage mechanism.Biochim. Biophys. Acta 2000; 1523: 196–205.

31 Marcellini L, Borro M, Gentile G, Rinaldi AC, Stella L, Aimola P, Barra D,Mangoni ML. Esculentin-1b(1-18) – a membrane-active antimicrobialpeptide that synergizes with antibiotics and modifies the expressionlevel of a limited number of proteins in Escherichia coli. FEBS J. 2009;276: 5647–5664.

32 Bechinger B, Lohner K. Detergent-like actions of linear amphipathiccationic antimicrobial peptides. Biochim. Biophys. Acta 2006; 1758:1529–1539.

Supporting Information

Additional supporting information may be found in the onlineversion of this article at the publisher’s web site.

nd John Wiley & Sons, Ltd. wileyonlinelibrary.com/journal/jpepsci

895