Phospholipid-driven differences determine the action of the synthetic antimicrobial peptide OP-145 on Gram-positive bacterial and mammalian membrane model systems

(1)

Phospholipid-driven differences determine the action of the synthetic

antimicrobial peptide OP-145 on Gram-positive bacterial and

mammalian membrane model systems

Nermina Malanovic

a,b,

⁎

, Regina Leber

a

, Maria Schmuck

b,c

, Manfred Kriechbaum

d

, Robert A. Cordfunke

e

,

Jan W. Drijfhout

e

, Anna de Breij

f

, Peter H. Nibbering

f

, Dagmar Kolb

g,h

, Karl Lohner

a,b

a

Institute of Molecular Biosciences, Biophysics Division, University of Graz, NAWI Graz, Graz, Austria

b

BioTechMed-Graz, Austria

c

Institute of Environmental Biotechnology, Graz University of Technology, Austria

d_{Institute of Inorganic Chemistry, Graz University of Technology, Austria} e

Department of Immunohematology and Blood Transfusion, Leiden University Medical Center, Leiden, The Netherlands

f

Department of Infectious Diseases, Leiden University Medical Center, Leiden, The Netherlands

g

Institute of Cell Biology, Histology, and Embryology, Medical University of Graz, Graz, Austria

h

Core Facility Ultrastructure Analysis, Center for Medical Research, Medical University of Graz, Graz, Austria

a b s t r a c t

a r t i c l e i n f o

Article history: Received 2 April 2015

Received in revised form 10 July 2015 Accepted 20 July 2015

Available online 23 July 2015

Keywords: Antibacterial activity Cytotoxicity Membrane mimics Membrane biophysics Lipoteichoic acid Peptidoglycan

OP-145, a synthetic antimicrobial peptide developed from a screen of the human cathelicidin LL-37, displays strong antibacterial activities and is— at considerably higher concentrations — lytic to human cells. To obtain more insight into its actions, we investigated the interactions between OP-145 and liposomes composed of phosphatidylglycerol (PG) and phosphatidylcholine (PC), resembling bacterial and mammalian membranes, re-spectively. Circular dichroism analyses of OP-145 demonstrated a predominantα-helical conformation in the presence of both membrane mimics, indicating that the different membrane-perturbation mechanisms are not due to different secondary structures. Membrane thinning and formation of quasi-interdigitated lipid–peptide structures was observed in PG bilayers, while OP-145 led to disintegration of PC liposomes into disk-like micelles and bilayer sheets. Although OP-145 was capable of binding lipoteichoic acid and peptidoglycan, the presence of these bacterial cell wall components did not retain OP-145 and hence did not interfere with the activity of the peptide toward PG membranes. Furthermore, physiological Ca++_{concentrations did neither inﬂuence the} mem-brane activity of OP-145 in model systems nor the killing of Staphylococcus aureus. However, addition of OP-145 at physiological Ca++_{-concentrations to PG membranes, but not PC membranes, resulted in the formation of} elongated enrolled structures similar to cochleate-like structures. In summary, phospholipid-driven differences in incorporation of OP-145 into the lipid bilayers govern the membrane activity of the peptide on bacterial and mammalian membrane mimics.

1. Introduction

The increasing incidence of multidrug resistant (MDR) bacterial strains is becoming a signiﬁcant problem for the treatment of various in-fections. This presses the need for the development of novel antimicro-bial strategies to combat infections with MDR pathogens[1]. One strategy has focused on the characterization of naturally occurring as well as newly designed antimicrobial peptides (AMPs) as a novel class of antibiotics[2–4]. Some AMPs have already been clinically tested as a topical treatment[5]and studied for prevention of implant surface and catheter-associated infections[6,7].

AMPs act in many cases on the lipid matrix of bacterial cell membranes[8]or have multiple cellular targets[3,9]. Therefore, the likelihood of emergence of resistance to AMPs is thought

Abbreviations: AMP, antimicrobial peptide; ANTS, 8-aminonaphthalene-1,3,6-trisulfonic acid; BSA, bovine serum albumin; CD, circular dichroism; DPPC, 1,2-dipamitoyl-sn-glycero-3-phosphocholine; DPPG, 1,2-dipamitoyl-sn-glycero-3-phospho-rac-glycerol; DSC, differential scanning calorimetry; DPX, p-xylene-bis-pyridinium bromide; Fmoc, 9H-ﬂuorenylmethyloxycarbonyl; IL-8, interleukin 8; LTA, lipoteichoic acid; LUV, large unilamellar vesicle; MLVs, multilamellar vesicles; NMM, N-methylmorpholin; PBS, phosphate buffered saline; PC, phosphatidylcholine; PGN, peptidoglycan; PI, propidium iodide; POPC, 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine; POPG, 1-palmitoyl-2-oleoyl-sn-glycero-3-phospho-rac-glycerol; SAXS, small-angle X-ray scattering; WAXS, wide-angle X-ray scattering

⁎ Corresponding author at: Institute of Molecular Biosciences, Biophysics Division, University of Graz, NAWI Graz, Graz, Austria.

E-mail address:nermina.malanovic@uni-graz.at(N. Malanovic).

http://dx.doi.org/10.1016/j.bbamem.2015.07.010

Contents lists available atScienceDirect

Biochimica et Biophysica Acta

(2)

to be considerably less than that for conventional antibiotics. The mechanism of action of many AMPs relates to targeting the micro-bial cytoplasmic membrane, interacting with the lipid matrix and subsequent permeabilization of the membrane[10,11]. Their anti-microbial activity strongly depends on the nature of the peptide and the lipid composition of the membrane [10]. The plasma membrane of Gram-positive bacteria contains a high amount of negatively charged phosphatidylglycerol (PG)[12], while zwitter-ionic phosphatidylcholine (PC) is the major component of mam-malian cytoplasmic membranes[13–15]. Hence, afﬁnity to PG and PC may primarily govern the interactions between antimicrobial peptides and the membranes of their targets.

One of the best studied AMPs is human LL-37, which belongs to the cathelicidin peptide family that is conserved among diverse mammalian species[16,17]. This peptide family plays an important role in the immune system and possesses a wide spectrum of activities against bacteria, fungi, viruses and cancer cells[18]. Its molecular mode of membrane disruption strongly depends on the target membrane lipid composition[19_–23].

Modifications in the sequence of LL-37 have been introduced to improve its bactericidal properties[24,25]. In this context, we have designed OP-145, previously termed P60.4Ac[26]. In their study, Nell et al.[26]scanned the LL-37 sequence with a window size of 22 to 25mers identifying a 24mer (amino acid sequence 13–36) as the most promising segment in terms of antimicrobial activity and with similar efficacy as LL-37 in terms of LPS and LTA neutralization and lower pro-inflammatory activity. This peptide was then further modified and several amino acids were replaced in particular at the C-terminal part of the peptide to favor the formation of an ideal amphipathic helix. Fur-thermore, in order to improve the stability of the peptide against pro-teolytic degradation the N- and C-terminus of OP-145 were blocked by N-acetylation and C-amidation, respectively. This synthetic peptide has been proven to be safe and successful as a treatment for chronic otitis media in a clinical phase I/II trial[27]. One of the most frequent pathogens associated with this disease is the Gram-positive bacteria Staphylococcus aureus[28,29]including MRSA which is also a major pathogen linked to osteomyelitis[30,31]. Thus in the present study, we focused on the interactions between OP-145 and membrane model systems mimicking Gram-positive bacterial and mammalian cytoplasmic membranes in order to obtain better insight into the mechanism(s) of action of this peptide.

2. Experimental procedures 2.1. Materials

Phospholipids (N99% purity) were obtained from Avanti Polar Lipids (Alabaster, AL). ANTS (8-aminonaphthalene-1,3,6-trisulfonic acid, disodium salt), bodipy-cadaverine and DPX (p-xylene-bis-pyridinium bromide) were purchased from Molecular Probes (Eugene, OR). Choles-terol, LTA and peptidoglycan (PGN) were from Sigma-Aldrich Handels Gmbh (Vienna, Austria).

OP-145 (acetyl-IGKEFKRIVERIKRFLRELVRPLR-amide) was prepared by solid phase strategies on an automated multiple peptide synthesizer (SyroII, MultiSyntech, Witten, Germany) as described previously[32]. The purity of the peptides was N95%, as determined by UPLC–MS (Acquity, Waters, Milford, MA). Peptide's integrity was conﬁrmed using MALDI-TOF mass spectrometry (Microﬂex, Bruker, Bremen, Germany), showing the expected molecular masses. The lyophilized peptide was stored at−20 °C until use, then dissolved in H2O with 0.01% acetic acid

to a stock of 10 mg/ml, and aliquots were stored at−20 °C. 2.2. Antimicrobial activity

S. aureus JAR060131[33]was cultured to mid-logarithmic phase in tryptic soy broth and washed once with PBS. Approximately

1 × 106_{CFU/ml of PBS were incubated with OP-145 (}_ﬁnal

concentra-tions: 0–51.2 μM) for 2 h at 37 °C under shaking conditions. Thereaf-ter, the number of viable bacteria was determined by plating 10-fold serial dilutions on diagnostic sensitivity agar. Antimicrobial activity is expressed as the 50% inhibitory concentration (IC50), i.e., the

lowest peptide concentration that killed_{≥50% of bacteria.} 2.3. Lysis of human dermalﬁbroblasts

Human dermalﬁbroblasts (PromoCell GmbH, Heidelberg, Germany) were cultivated inﬁbroblast growth media-2 (PromoCell GmbH). All cells were kept in a 5% CO2atmosphere at 37 °C. Cells were cultured

to ~ 90% conﬂuence and then detached by accutase (PAA, Pasching, Austria). Enzymic activity was stopped with PBS containing 10% fetal bovine serum. Cells were washed with PBS and re-suspended in the culture medium. Approximately 1 × 105_{cells were incubated with}

0–80 μM OP-145 for 1 h at 37 °C. Five microliters of 50 μg/ml propidium iodide (PI, Invitrogen, Camarillo, CA) was added to the samples and after 5 min at room temperature in the dark fluores-cence was measured with afluorescence spectrometer (SPEX Fluoro Max-3 spectrofluorimeter, Horiba Scientific, Irvine, CA) using an ex-citation wavelength (λex) of 536 nm and an emission wavelength

(λem) of 617 nm. Cytotoxicity was calculated from the percentage

of PI-positive cells in medium alone (P0) and in the presence of

peptide (PX) according to Eq.(1):

%PI‐uptake ¼100 Pð X−P0Þ P100−P0

ð Þ ð1Þ

whereby Triton-X-100 (2.5%) was used to determine 100% PI-positive cells (P100). Data are expressed as IC50values, i.e. the lowest

peptide concentration that lyses_{≥50% of the cells.} 2.4. Hemolysis

Whole venous blood of healthy volunteers was collected in citrate tubes. Erythrocytes were isolated by centrifuging the blood for 10 min at 2000 ×g and subsequently washed twice in PBS. A 0.5% erythrocyte solution in PBS was exposed to OP-145 (ﬁnal concentrations: 1.6– 51.2μM) in the presence of 50% human plasma. After 1 h at 37 °C under static conditions, the erythrocyte suspensions were centrifuged and the optical density of the supernatants was determined at 415 nm. The percentage of hemolysis was calculated relative to the pos-itive control (2.5% Triton X-100). Data are expressed as IC50values, the

lowest peptide concentration that lyses≥50% of the cells. 2.5. Determination of secondary structure

Secondary structure of OP-145 was obtained from circular dichroism (CD) spectra measured on a Jasco J-715 spectropolarimeter (Jasco, Gross-Umstadt, Germany) using a 0.02 cm quartz cuvette equilibrated at 25 °C. Spectra were recorded at 25 °C between 180 and 260 nm with a 0.2 nm step resolution. The measurements were performed with 200μM OP-145 in the presence and absence of large unilamellar vesicles (LUVs) composed of POPG and POPC, respectively. The peptide-to-lipid molar ratio was 1:25. The background signal of 10 mM Hepes (pH 7.0) was subtracted from the respective spectra. Five spectra were averaged before secondary structure contents have been estimated by CDSSTR program on the online server DICHROWEB

[34].

2.6. Preparation of liposomes

Lipidﬁlms were prepared using an appropriate amount of phospho-lipid solved in chloroform/methanol (9:1 v/v), evaporated under a stream of nitrogen and stored in vacuum overnight. After addition of

(3)

sodium phosphate buffered saline (PBS, 20 mM NaPi, 130 mM NaCl, pH 7.0), formation of lipid vesicles was achieved by intermittent vig-orous vortexing at a temperature above the phase transition of the respective phospholipids. In case of samples containing peptide, LTA or PGN appropriate amounts had been dissolved in PBS prior to vortex mixing. The preparation of DPPG/LTA vesicles was performed under sonication. LTA concentration has been calculated with a mo-lecular weight of 6200 g/mol estimated for 24 repeating units of the polyglycerolphosphate chain[35]. Large unilamellar vesicles were obtained by several cycles of extrusion of the hydrated liposomes through a polycarbonateﬁlter (Millipore-Isopore™) of 0.1 μm pore size following a protocol described previously[36].

2.7. Differential scanning calorimetry (DSC)

DSC measurements were performed using a Microcal VP-DSC high-sensitivity differential scanning calorimeter (Microcal, Northampton, MA). Scans of 1 mg/ml lipid concentration were recorded at a constant rate of 30 °C/h and data were analyzed using Microcal's Origin software. Calorimetric enthalpies were calculated by integration of the peak areas after baseline correction and normalization to the mass of phospholipid. The phase transition temperature was deﬁned as the temperature at the peak maximum.

2.8. Fluorescence spectroscopy and leakage assay

Leakage of the aqueous content of the 8-aminonaphthalene-1,3,6-trisulfonic acid/p-xylene-bis-pyridinium bromide (ANTS/DPX)-loaded large unilamellar vesicles (LUVs) composed of POPG, POPG/LTA, POPC or POPC/cholesterol upon incubation with peptide was determined as described[36]. In the case of POPG/PGN vesicles, 0.1% PGN has been added to the 50_{μM POPG large unilamellar vesicles prior to the} incuba-tion with peptide. Fluorescence emission was recorded as a funcincuba-tion of time before and after the addition of incremental amounts of peptide ranging from 0.125 up to 8μM, corresponding to peptide-to-lipid molar ratios from 1:400 to 1:6.25 (0.25–16 mol%). The measurements were performed on a SPEX Fluoro Max-3 spectroﬂuorimeter (Jobin-Yvon, Longjumeau, France) combined with Datamax software. Percent-age of leakPercent-age was calculated from the fraction of the leakPercent-age (IF)

according to Eq.(2): IF¼ F−F0 ð Þ Fmax−F0 ð Þ ð2Þ

where F is the measuredﬂuorescence, F0is the initialﬂuorescence

with-out peptide, and Fmaxis theﬂuorescence corresponding to 100% leakage

gained by addition of 1% Triton X-100.

2.9. Small- and wide-angle X-ray scattering (SWAXS)

For SWAXS measurements samples of 50 mg/ml lipid concentration were measured on a high-ﬂux laboratory small-angle X-ray scattering camera S3-Micro (Hecus X-Ray Systems, Graz, Austria) equipped with a high-brilliance micro-beam delivery system operating at a low power of 30 W (50 kV and 0.6 mA), with point-focus optics (FOX3D). The used X-ray wavelength (λ) was 1.54 Å and the small-angle X-ray scattering (SAXS) curves were recorded with a 1D silicon strip detector operating in single-photon counting mode (Mythen-1K, Dectris, Baden, Switzerland) and a 2D hybrid pixel detector (Pilatus 100K, Dectris) in the angular range 0.1°b 2θ b 8.5° in a distance of 27.7 cm (1D) and of 29.7 cm (2D) from the center of the sample. Additionally, wide-angle X-ray scattering (WAXS) patterns were measured with a 1D-position sensitive gas detector (Hecus 50 M, Graz, Austria) within the angular range of 17°b 2θ b 27°. The incident main-beam was blocked by a pre-cisely stepper-motor controlled beam-stopper made of a tungsten blade, located in front of the detector. The calibration of the q-scale

was performed by measuring silver-behenate with a deﬁned lamellar spacing of 58.38 Å, where q, the reciprocal scattering vector, is related to the scattering angle 2_{θ, as given in Eq.}(3):

q¼4_λπ sin θð Þ: ð3Þ

Background corrected scattering data of DPPG were evaluated based on a global analysis technique[37,38], while DPPC was analyzed apply-ing the Indirect Fourier Transformation method, which evaluates the particle distance distribution function p(r) in real space by using the program GNOM[39]. Subsequent modeling of lipid_{–peptide disks was} performed as described earlier[40].

2.10. Determination of bodipy-cadaverine displacement

The displacement assay has been performed according to Zorko et al.

[41]using 1μM bodipy-cadaverine and 13 μg of LTA/ml. 2.11. Binding assay for PGN

Binding afﬁnity of OP-145 toward staphylococcal PGN has been test-ed by the method describtest-ed by Torrent et al.[42]. Bovine serum albumin (BSA) and lysozyme have been included as negative and positive con-trols, respectively.

2.12. Neutralization assay

UV-inactivated S. aureus JAR was pre-incubated with 0–3.2 μM OP-145 for 30 min at 37 °C. Five times diluted whole blood from a healthy volunteer was stimulated with S. aureus/peptide mixtures, the bacteri-um, the peptide or no stimulus. After 18 h incubation at 37 °C and 5% CO2the levels of the IL-8 in the supernatants were quantiﬁed by ELISA

(Biosource Life Technologies) according to supplier's instructions. The percentage of inhibition of IL-8 release was calculated as follows: (1− IL-8 level by S aureus JAR exposed to OP-145 / IL-8 level caused by S aureus JAR) × 100%.

2.13. Freeze fracture transmission electron microscopy

For freeze fracture electron microscopy, DPPG liposomes were pre-pared in 10 mM Hepes, 130 mM NaCl, 2.5 mM CaCl2, pH 7.0 at aﬁnal

concentration of 5 mg/ml lipid in the presence and absence of 2 mol% OP-145. Five microliters of the liposomal solution were mixed with 30% glycerol (v/v), frozen in liquid propane, and stored in liquid nitro-gen until further use. Samples were fractured in a Balzers BAF400D freeze-etching apparatus (Oerlikon Balzers Coating AG, Balzers, Liechtenstein) under vacuum with a pressure between 1.3 × 10−4and 1.3 × 10−5Pa. Replicas were produced by vacuum deposition of plati-num and carbon on the surface. The replicas were put into a sodium hy-pochlorite solution for about 3 h and stored overnight in 50% NaOH. Before mounting them on a copper grid, replicas were washed with dis-tilled water at least three times. Images were taken using a FEI Tecnai G2 20 transmission electron microscope (FEI, Eindhoven, The Netherlands) with a Gatan ultrascan 1000 CCD camera. Acceleration voltage was 120 kV.

3. Results

3.1. Activity of OP-145 toward bacterial and mammalian cells

Cytotoxic activity of OP-145 was assessed toward bacterial and human cells showing that OP-145 was highly effective against S. aureus, asN50% of bacteria were killed within 1 h at 0.8–1.6 μM. As expected, considerably higher concentrations of the peptide were required to lyse human dermal

(4)

ﬁbroblasts and human erythrocytes, i.e. IC50values amounted to≥32 μM

and≥102.4 μM, respectively.

3.2. Structural properties of OP-145

CD experiments revealed no deﬁned secondary structure of OP-145 in 10 mM Hepes. However, a predominant_{α-helical conformation was} detected for OP-145 when analyzed in the presence of lipid bilayers, POPG and POPC, respectively (Fig. 1). The fraction ofα-helix differed only slightly between POPG and POPC, suggesting that the ability of the peptide to discriminate between bacterial and mammalian mem-branes is not due to different secondary structures of OP-145.

3.3. Interaction of OP-145 with bacterial membrane mimic

The interaction of OP-145 with liposomes composed of 1,2-dipalmitoyl-phospatidylglycerol (DPPG), a bacterial membrane mimic, was assessed by differential scanning calorimetry (DSC) as well as by small and wide angle X-ray scattering (SAXS and WAXS) techniques. DSC revealed that the thermotropic phase be-havior of DPPG was characterized by two phase transitions, which can be attributed to the pre-transition from the lamellar gel (Lβ′)

to the ripple phase (Pβ′) at 34.4 °C and the main or chain-melting

transition from the Pβ′to theﬂuid phase (Lα) at 41 °C in agreement

with published data[19](Table 1). Upon cooling DPPG from the ﬂuid to the gel phase, a minor hysteresis of the main transition (Tm= 40.8 °C) was observed. In the presence of OP-145, the

heating scans showed a minor increase of the main transition perature accompanied with a strong tailing toward the high tem-perature side at low and the rise of a shoulder at high peptide concentrations (data not shown). In addition, the main transition enthalpy is increased by about 20% with increasing peptide concentration (Table 1). These observations suggest the occurrence of two phase transitions, which was more prominently seen in the cooling scans, where a phase separation into peptide-enriched and -poor lipid do-mains was clearly observed (Fig. 2A). The two phase transitions still over-lap at lower peptide concentration but started to separate above 1 mol% OP-145. No changes in the transition temperature of the peptide-poor (largely unperturbed) DPPG domain were observed, while the transition temperature of peptide-enriched (strongly affected) DPPG domains shifted to lower values (Fig. 2A,Table 1). As can be deduced from the en-thalpy of these transitions, the fraction of peptipoor DPPG domains de-creased with increasing peptide concentration, which correlated with the elevating fraction of the peptide-enriched DPPG domain.

In the gel phase (25 °C) the SWAXS patterns of pure DPPG liposomes were characteristic for positionally weakly correlated bilayers of oligolamellar vesicles in the Lβ′-phase with tilted hydrocarbon chains

(Fig. 2C/D). Applying a full-q range global_{ﬁtting model}[37,38]a lamel-lar repeat distance (d-spacing) of 107.5 Å and a phosphate–phosphate distance of 45.3 Å were calculated (Fig. 2C,Table 2). In the presence of 2 mol% OP-145, SWAXS patterns were obtained that largely resembled the features of DPPG in the presence of LL-37[19]. The quasi-Bragg peak was hardly discernible and the diffuse scattering as well as increase of lamellar d-spacing to 177.5 Å indicate loss of correlation be-tween the bilayers, which was more clearly detected in cooling scans.

Moreover, the SAXS pattern showed additional diffuse scattering around q = 0.2 Å−1(Fig. 2C). In this q-range a reasonable but not per-fect_{ﬁt could be obtained, when the X-ray curve was analyzed based on a} single bilayer structure, which is also indicative for the presence of a second coexisting structure. Most importantly, a reduced phosphate– phosphate distance (dPP) of about 40.9 Å (Table 2) was estimated,

which suggests the presence of a thinner quasi-interdigitated bilayer structure like in the case of LL-37. The wide-angle X-ray data further support this assumption showing a loss of tilt in hydrocarbon chain packing. Now a single symmetric peak at q = 1.51 Å−1(d = 4.16 Å) was dominant (Fig. 2D), characteristic for hexagonally packed hydro-carbon chains oriented perpendicular to the bilayer plane as observed for interdigitated bilayers[43]. In addition, the increased total enthalpy measured for DPPG in the presence of OP-145 is in agreement with the formation of an interdigitated structure[44,45]. Also in the physiologi-cally relevantﬂuid phase, the DPPG bilayer was noticeably thinner in the presence of OP-145 (dPP= 37.1 Å as compared to dPP= 39.7 Å of

Fig. 1. Structural properties of OP-145. A) CD spectra of 200μM OP-145 in Hepes buffer solution (black line), in the presence of 5 mM POPG (ﬁlled squares) and 5 mM POPC (open circles), respectively. B) Respective fraction of secondary structure at indicated conditions.

Table 1

Thermodynamic parameters (pretransition and main phase transition temperature (Tpre/Tm) and corresponding enthalpies (ΔHpre/ΔHm)) of DPPG bilayers in the

absence and presence of OP-145. Data are the representative results of three indepen-dent experiments with standard deviations less than 5%.

OP-145 (mol%) Heating Cooling Tpre (°C) ΔHpre (kcal/mol) Tm (°C) ΔHm (kcal/mol) Tm (°C) ΔHm (kcal/mol) 0 34.4 1.6 41.0 10.9 40.8 10.0 0.25 34.3 0.7 41.2 10.9 41.0 11.0 0.5 33.7 0.7 41.1 10.6 41.0 11.0 1 34.1 0.9 41.3 13.3 40.8 8.3 39.8a 3.7a 2 34.0 0.8 41.3 12.3 40.9 7.1 39.0a 5.0a 4 36.2 0.5 41.2 12.7 41.0 4.3 38.4a 8.3a a Peptide-enriched domains.

(5)

pure DPPG;Table 2) in accordance with observations from DPPG/LL-37 mixtures[20]. In summary, the DSC and X-ray data demonstrated that OP-145, like otherα-helical peptides such as LL-37, PGLa or mellitin

[19,20], induces a quasi-interdigitated membrane structure in the gel phase and membrane thinning in theﬂuid phase in a bacterial mem-brane model.

To test bilayer permeability of our model system, we used large unilamellar vesicles composed of POPG loaded with ANTS/DPX and in-cubated them with OP-145 at 37 °C. Release of entrappedﬂuorescence marker molecules from POPG was observed already at low concentra-tions tested and almost full leakage was achieved at 4 mol% (=2μM) OP-145. At a total concentration of 1μM OP-145 50% membrane leakage from POPG vesicles was obtained (Fig. 2B), which is in the range of the IC50(1.6μM) value for the killing of S. aureus.

3.4. Interaction of OP-145 with LTA and PGN

Next, we investigated the impact of LTA and PGN on the interaction of OP-145 with the bacterial membrane mimic, DPPG. In the presence of 1.2 mol% LTA a very broad phase transition regime with two major peaks at 41.7 °C and 47.2 °C was observed (Fig. 3A, left panel), whereby theﬁrst peak corresponds to highly enriched DPPG domains in accor-dance with previously published data[46]. In the case of DPPG/PGN (0.1 wt.%) mixtures, the main transition showed rather a dominant shoulder at the high temperature side than a second phase transition, which may be due to binding of PGN to the PG surface (Fig. 3A, right panel). In both the PGN and the LTA-containing liposomes addition of OP-145 (2 mol%) resulted in the disappearance of the second transition or the shoulder, indicating that the peptide interacts preferentially with the LTA as well as PGN enriched bilayer domains. Concomitantly, an in-crease of the fraction of DPPG undergoing the main phase transition was observed suggesting that part of the DPPG is displaced from DPPG/LTA or DPPG/PGN domains. The bilayer permeability of our model system in the presence of LTA or PGN has been tested with ANTS/DPX loaded large unilamellar vesicles composed of POPG and LTA at a molar ratio

of 9:1 to be consistent with the biological distribution of those two lipids in S. aureus membrane[47]and ANTS/DPX loaded POPG vesicles in the presence of 0.1% weight PGN. Interestingly, leakage from POPG/LTA ves-icles but not from POPG/PGN was reduced by about 25% in the presence of OP-145 indicating a slight hindrance of the penetration of OP-145 by the presence of LTA in the liposome (Fig. 3B).

The probability of OP-145 to bind to LTA was confirmed in a fluores-cence assay using bodipy-cadaverine showing that at 0.1μM OP-145 more than 80% of thefluorophore was displaced (Fig. 3C). Moreover, OP-145 can also bind to PGN as assessed in a binding assay by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE). Similar to the positive control lysozyme, which strongly binds to PGN, the major fraction of OP-145 was recovered together with the PGN insoluble frac-tion in pellet (Fig. 3C). OP-145 was not found in the pellet, when the binding assay was performed in the absence of PGN (data not shown) and only traces were found in samples with BSA, which was used as a control for molecules that do not bind to PGN.

Staphylococcal LTA and PGN are also potent stimuli of cytokine releaseand important mediators of inﬂammation[48]. Thus in addition, neutralization assays were performed showing that OP-145 inhibited dose-dependently the production of IL-8 by whole blood cells in re-sponse to UV-inactivated S. aureus. Already 0.05μM peptide was sufﬁ-cient to inhibit two-thirds of the stimulated IL-8 production and about 90% was inhibited in the presence of 1.6μM OP-145 (Fig. 3D). No effect on IL-8 production was detected, when OP-145 was incubated in PBS without stimulating the whole blood cells with UV-inactivated S. aureus (data not shown). This further suggests that the high bind-ing capacity of OP-145 to LTA and PGN plays a role in its immune-modulatory activity to neutralize the bacterial cell wall components and inhibit cytokine production. The potential of OP-145 to neutral-ize LTA has also been reported by Nell et al.[26].

3.5. Interaction of OP-145 with mammalian model system

In order to assess the effect of OP-145 on mammalian membrane model system, we used multilamellar vesicles (MLVs) composed of DPPC. The thermotropic phase behavior of DPPC was characterized by two phase transitions being attributed to the pre-transition from the lamellar gel (Lβ′) to the ripple phase (Pβ′) at 35.1 °C and the main or

chain-melting transition from the Pβ′to theﬂuid phase (Lα) at 41.7 °C

(Fig. 4A,Table 3). Upon incubation with OP-145, the thermotropic behavior of DPPC was strongly altered. The enthalpy of the pre- and main transition of DPPC decreased, whereas their transition tempera-tures and half-width were unaffected (Fig. 4A,Table 3). Above 2 mol% OP-145 a broad low cooperative transition resulting from peptide-enriched DPPC domains underlying the sharp main transition evolved, which increased in enthalpy upon higher peptide concentrations.

Fig. 2. Effect of OP-145 on bacterial membrane mimics. A) Concentration dependent thermotropic behavior of DPPG vesicles in the presence of OP-145 as observed in DSC cooling scans. B) Leakage of POPG LUVs at indicated OP-145 concentration. C) SAXS and D) WAXS of DPPG in the presence of 2 mol% OP-145 (gray, bottom) at 25 °C upon cooling. The corresponding proﬁles of pure DPPG at 25 °C are shown in black. Solid lines in SAXS diagram represents the full q-range ﬁts of the experimental curves shown as circles.

Table 2

Lamellar spacing (d) and phosphate–phosphate distance (dPP) of DPPG in the absence and

presence of 2 mol% OP-145. Data are the representative results of three independent experiments with standard deviations less than 5%.

T (°C)

Pure ±2 mol% OP-145 dPP(Å) d (Å) dPP(Å) d (Å) DPPG 25 (gel phase) 45.3 100.9 43.1 100.4 50 (ﬂuid phase) 39.7 116.4 37.1 110.9 25 (gel phase)a _44.1 _107.5 _40.9 _177.5 a

(6)

At the highest peptide concentration used (4 mol% OP-145), the ther-mogram revealed a minor fraction of peptide unaffected DPPC bilayers. Based on the estimation of the enthalpy of the sharp and broad compo-nent, we concluded that ~ 10% DPPC MLVs were still present. Both the low enthalpy and cooperativity of the underlying transition indicated the presence of small lipid–peptide aggregates which suggests a detergent-like action of OP-145 leading to disintegration of the multilamellar PC liposomes as also reported for LL-37[19].

Permeability of PC bilayers in the presence of OP-145 was measured using POPC vesicles. Whereas addition of 2 mol% of OP-145 to the bacte-rial mimetic system POPG induced 50% leakage, this value was not reached for POPC vesicles even at the highest concentration of OP-145 tested (Fig. 4B). Moreover, the leakage of POPC is attenuated in the pres-ence of 25 mol% cholesterol.

SAXS data revealed the typical pattern of multilamellar DPPC vesicles with characteristic Bragg peaks up to the 5th order in the gel phase. In good agreement with published values[20,49], a lamellar repeat distance of 64.6 Å as well as a phosphate–phosphate distance of 43.4 Å was calcu-lated for DPPC vesicles in the gel phase (Fig. 4C/D). In the gel phase in the presence of 2 mol% OP-145 the SAXS pattern was characterized by the loss of Bragg peaks and occurrence of a strong diffuse scattering as well as by a broad weak peak around q = 1.47 Å−1(d = 4.27 Å) in the wide-angle region, the latter indicating a loss of hydrocarbon chain pack-ing order. These observations suggested that, alike LL-37, the peptide leads to disintegration of the multilamellar DPPC arrangement and that the diffuse scattering arises from small lipid–peptide particles. Indeed, data evaluation and modeling of the SAXS patterns of the gel phase yielded disk-like particles with an estimated maximum diameter of about 500 Å and a thickness representing the lipid-bilayer of about 60 Å (Fig. 5). The lipid-core itself is modeled by a disk with dimensions

480 × 60 Å. Additionally to this lipid core, an outer peptide layer of a width of 10 Å with an electron density approximately matching that of the phospholipid headgroup was added radially, thus resulting in a disk with the total dimensions of 500 × 60 Å.

Data evolution and modeling of the experimental SAXS-curve in the ﬂuid phase (50 °C) was performed using a two-dimensional bilayer model with an electron density across the bilayer perpendicular to its plane and a bilayer thickness of 45.5 Å, a phosphate–phosphate distance of 37.8 Å and a hydrocarbon chain length of 15 Å (Fig. 5A). The modeling revealed that the lateral dimensions of the bilayer are beyond resolution of this method and thus can be considered as inﬁnitely large as com-pared to the bilayer thickness suggesting that the disk-like micelles fuse into large bilayer sheets (Fig. 5A). Minor amounts of multilamellar DPPC aggregates are still present as can be deduced from the small Bragg peak at position q = 0.092 Å−1in agreement with DSC data.

3.6. Effect of calcium on the membrane activity of OP-145

As small amounts of Ca2+_{-ions can drastically raise the phase}

transi-tion of anionic membrane lipids[50]we determined the effect of phys-iological calcium concentrations (Ca++_{/lipid = 1/0.5) in the presence}

and absence of OP-145 on both the bacterial and mammalian mimetic system. Results revealed that the main transition of DPPG shifted dras-tically to higher temperatures by about 15 °C (Tm= 55.7 °C) (Fig. 6B),

while Ca++_{did not signi}_{ﬁcantly affect the thermotropic behavior of}

DPPC (Fig. 6A). The thermograms for DPPC in the presence of 2 mol% OP-145 were practically identical with and without Ca++(seeFigs. 4A and6A) indicating that the divalent cation does not affect the interac-tion of the peptide with zwitterionic lipid.

Fig. 3. Effect of OP-145 on bacterial cell wall components, LTA and PGN. A) Thermotropic behavior of DPPG/LTA vesicles (left panel) and DPPG/PGN vesicles (right panel) in the presence of 2 mol% OP-145 (II). The corresponding DSC thermograms of pure DPPG with and without LTA (left panel) or PGN (right panel) are shown in black/gray (I). B) Leakage of POPG/LTA LUVs (black bars) and POPG/PGN (white bars). Dashed lines indicate leakage of pure POPG at respective OP-145 concentration. C) Displacement of Bodipy-cadaverine from LTA in the presence of indicated OP-145 concentration (left panel) and PGN binding of OP-145 as analyzed by gel electrophoresis (right panel). Lysozyme was used as a positive control and BSA as a negative control. Lysozyme or OP-145 bound to PGN is detected in the pellet, while unbound protein remains in the supernatant. D) IL-8 production by whole blood cells in response to 2.5 × 106_{CFU/ml of}

(7)

However, a different behavior was observed in DPPG liposomes, where an exothermic component at higher temperature upon incubation with OP-145 was detected. Such behavior was observed with negatively charged phospholipids at Ca++:lipid molar ratios higher than 1:2 and was attributed to the formation of cylindrical structures like cochleate cyl-inders[50,51]. We used Ca++_{concentrations below this molar ratio and}

thus in the pure lipid system we observed only an increase in transition temperature resulting from charge neutralization of the phospholipid, as a consequence of decreased intramolecular repulsion[52]. Electron microscopy conﬁrmed that pure DPPG in the presence of Ca++

still

forms oligolamellar vesicles, while upon addition of OP-145 rather elon-gated enrolled structures similar to cochleate cylinders were found (Fig. 6C). Although OP-145 induced completely different morphological changes of DPPG membranes in the presence of calcium, this had no in ﬂu-ence on its bactericidal activity toward S. aureus, i.e. its antimicrobial ac-tivity was not impaired up to 10 mM CaCl2(data not shown).

4. Discussion

A general mechanism of bacterial killing by a number of cationic AMPs is thought to be due to targeting anionic phospholipids on the membrane surface, which leads to membrane perturbation, eventually rupture and cell death. S. aureus mainly contains negatively charged PG in the plasma membrane[12], while zwitterionic PC is the major component of the mammalian outer membrane[13,14]. In this context, the main_{ﬁnding of the present study is that the nature of the interaction} of OP-145 with Gram-positive bacterial and mammalian model mem-branes is governed by the phospholipids. OP-145 induces membrane thinning in bacterial and a detergent-like action on mammalian mem-brane mimics. Secondary structure of the peptide may not be a determi-nant in the discrimination between the bacterial and mammalian model membranes as CD measurements revealed no signiﬁcant differences in α-helical content of OP-145 in bacterial and mammalian membrane mimics, which also has been suggested earlier for LL-37[53].

Fig. 4. Effect of OP-145 on mammalian membrane mimics. A) Concentration dependent thermotropic behavior of DPPC vesicles in the presence of OP-145 as observed in heating scans. Enlarged DSC data for 2 mol% OP-145 is shown in inset. B) Leakage of LUVs composed of POPC (ﬁlled squares) and POPC containing 25 mol% cholesterol (open squares) at indicated OP-145 concentration. C) SAXS and D) WAXS of DPPC in the presence of 2 mol% OP-145 (in gray) at 25 °C. The corresponding proﬁles for pure DPPC at 25 °C are shown in black.

Table 3

Thermodynamic parameters (transition temperature (Tm), and enthalpy (ΔHm)) of

the main transition of DPPC in the absence and presence of OP-145. Data are the representative results of three independent experiments with standard deviations less than 5%. OP-145 (mol%) Tm (°C) ΔHm (kcal/mol) 0 41.7 8.0 0.25 41.7 7.5 0.5 41.5 7.9 1 41.7a 6.9 2 41.5a 5.3 4 41.6a _2.4 a

(8)

Notifying that DPPG and DPPC bilayers share the same basic fea-tures (cylindrical shape, chain tilt and thermodynamic behavior) and that the only marked difference is the charge of the head group of the phosholipids, the different mode of action of OP-145 might be provoked by the different penetration depth of the peptide in the two membrane models. Thus, electrostatic interactions between the cationic peptide and the anionic lipid head groups inhibit deeper insertion into the acyl chain region of PG mem-branes[54]. Hence, the observed thinning of the PG bilayer, i.e. for-mation of a quasi-interdigitated structure in the gel-phase, results from the larger“void” created below the peptide attracting the opposing lipid monolayer, as described in the“free” volume (void) model of membrane perturbation by antimicrobial peptides[10]. In contrast, the deeper pene-tration of OP-145 in neutral DPPC bilayers leads to disintegration and for-mation of disk-like micelles. Micellization of PC bilayers has also been reported for other toxic peptides, such asδ-lysin or melittin, and deter-gents[40,55–57]. Support for this model comes also from earlier experi-ments on the lipid chain-length dependence on the induction of a quasi-interdigitated structure by antimicrobial peptides such as LL-37, PGLa and melittin, which demonstrated that reducing the chain length in PG bilayers hindered and vice versa increasing the chain length in PC bilayers promoted the fraction of this peculiar structure[19]. For exam-ple, formation of disk-like micelles by LL-37 was not observed for the longer-chain distearoyl-PC and diarachidoyl-PC. In contrast, a coexis-tence of interdigitated and non-interdigitated lipid domains was observed.

As discussed by Freire et al.[58]it is a pertinent question whether bacterial cell wall components serve as electrostatic barriers capturing AMPs and in turn preventing their interaction with the cytoplasmic membrane. They conclude that in the end this is a matter of concentra-tions of AMPs and membrane components as well as of afﬁnity of AMPs toward the different membrane components. Indeed, our data clearly indicate that OP-145 can also bind to the bacterial cell wall components LTA and PGN. Leakage experiments demonstrated that PGN did not impair the capacity of OP-145 to permeabilize PG membranes, which was slightly reduced by the presence of LTA. Thus, OP-145 is still capable to reach the PG lipid bilayer so that membrane permeabilization is not hindered by these molecules.

Given the importance of divalent cations in membrane physiology and activity of antimicrobial peptides[59,60], we also studied the effect of OP-145 on the two model systems in the presence of physiological Ca++_{concentrations. Whereas no signi}_{ﬁcant changes for zwitterionic}

phospholipids were observed, Ca++ _{had a major impact on the}

Fig. 5. Modeling of the DPPC membranes in the presence of OP-145. A) SAXS pattern of DPPC in the presence of 2 mol% OP-145 at 25 °C (gel phase, open circles) and 50 °C (fluid phase, open squares) with corresponding two-dimensional (infinitely) bilayer model withfitted axial electron density (black line), resulting in a bilayer thickness of 45.5 Å, dPPof 37.8 Å and chain length of 15 Å. The scattering curves (50° and 25°,

respec-tively, are shifted vertically for better clarity). B) Distance-distribution function p(r) of DPPC membranes in the presence of 2 mol% OP-145 (open circles) at 25 °C upon cooling. The calculated p(r)-function of a model consisting of a bilayer-disk model with dimen-sions of 60 × 480 Å, with an axial electron-density step proﬁle as shown in the insert, and a surrounding 10 Å circular ring (electron-density 10% lower than the headgroups representing the peptide) is shown as a gray line.

Fig. 6. Effect of Ca++

on OP-145 interaction with model membranes. Thermotropic behav-ior of DPPC (A) and DPPG (B) vesicles in the presence of 2.5 mM CaCl2and 2 mol% OP-145

(II). The corresponding DSC thermograms of pure lipids with and without CaCl2are shown

in black/gray (I). C) Electron micrographs of freeze fracture replicas of 50μM DPPG in the presence of 2.5 mM CaCl2with 2 mol% OP-145 or no peptide at 25 °C (gel phase) and 65 °C

(9)

morphology and phase transition properties of the anionic phospho-lipids. Owing to the strong electrostatic repulsion at low ionic strength, anionic lipids destabilize bilayers by allowing the hydrophilic surface to acquire a high curvature, thus leading to smaller particles. These effects on anionic phospholipids were diminished at higher salt concentra-tions, resulting in formation of large sealed vesicular structures[61]. In general, cations may directly protonate the phospholipids or compete with the phospholipid polar group for water leading to an increase in phase transition temperature[52]. All these features (dehydration, membrane curvature) enhance the formation of elongated multilayered cylindrical structures[50], further referred to as cochleates. At physio-logical Ca++ concentrations (2.5 mM CaCl2), we observed such a

cochleate formation only after addition of OP-145 to DPPG. Presumably, OP-145 facilitates the bridging of anionic DPPG molecules in the pres-ence of Ca++_{as the positive charges of the amphipathic OP-145 are}

ex-posed in a close distance along one side of the_{α-helix. It is highly} plausible that the peptide may preferentially interact along the helix with phospholipid head groups of the planar bilayer that is enabled to wrap around the incorporated peptide molecules in cylinders, which are characterized rather by an elliptical than circular cross section as compared to the“spirally twisted snail's shell” shown for calcium induced PG[62]and phosphatidylserine cochleates[51]. Epand et al.[63]reported on several antimicrobial cationic oligo-acyl-lysyl peptide mimics that can form cochleates also in the ab-sence of divalent cations and that their assembly was favored by in-creased hydrophobicity and disfavored by extending the distance between the cationic groups. Cochleate structures found in the pres-ence of oligo-acyl-lysyl peptide mimics are also not as tightly rolled up as the Ca++_{-bridged cochleates of PG}_[64]_{. As described above}

under our experimental conditions these structures were only ob-served in the presence of Ca++_{. In this respect it should be noted}

that addition of low mM concentrations of Ca++_{can lead to}

de-creased activity of AMPs[65]. Although our experimental approach does not allow making conclusions in terms of displacement of Ca++ by OP-145, calorimetry and electron microscopy clearly showed that in the presence of physiological Ca++_{concentrations}

the peptide interferes with membrane structure in accordance with the antimicrobial activity observed even at elevated Ca++levels (up to 10 mM).

OP-145 exhibits potent antimicrobial activity against S. aureus with LC99.9%values of 1.6μM (LC99.9%= lowest peptide concentration that

resulted inN99.9% killing of S. aureus), while disruption of the respective model membrane DPPG by OP-145 was induced at about 2-fold higher total peptide concentration (full leakage at 4μM). The correlation between vesicle permeabilization and antimicrobial activity has been al-ready discussed in detail[11,66–69]. Recently, several biophysical tech-niques have been translated to studies using live bacteria, which allow a correlation of data gained from liposome-based and cell-based studies

[70]. In this context, Roversi et al.[71]showed an extremely high cover-age of both leaﬂets of the outer and inner Escherichia coli membranes by PMAP-23, a cationic amphipathic helix from the cathelicidin fam-ily. Bacterial killing started at a molar ratio of bound peptide per lipid of about 1:30 and all bacteria were killed at a molar ratio of 1:4, cor-responding closely to those estimated by Castanho and co-workers

[70]for other peptides, based on the partition constants derived from binding studies on model membranes. This emphasizes that the critical parameter deﬁning the correlation between model sys-tems and bacterial killing is not the total peptide concentration in the assay but the membrane-bound peptide-to-lipid ratio.

The impact of membrane partitioning of peptides can also be deduced from our experiments using PC to mimic mammalian mem-branes. OP-145 showed a detergent-like mode of action by nearly complete disruption of PC bilayer as deduced from X-ray and DSC experiments, but membrane permeabilization did not exceed 30% at the highest total peptide concentration used (8μM). The different extent of membrane perturbation can be explained by the different

experimental protocols used for DSC/X-ray scattering and leakage stud-ies. In the former case OP-145 is a component of the hydration buffer, i.e. is already present during the process of lipid hydration, which facilitates direct interaction of the peptide with fatty acyl chains of the phospho-lipids favoring partitioning of the amphipathic peptide into the bilayer, while in the latter case the peptide interacts with the preformed lipid vesicle and has to overcome the interfacial barrier. Thus, only at aﬁrst glance, membrane selectivity observed in leakage experiments (almost 100% for POPG vs. about 12% for POPC at a total peptide concentration of 2μM) seems to be contradictive to the mode of action, detergent-like disruption for PC bilayers as compared to bilayer thinning for DPPG, but rather reﬂects different membrane partitioning of OP-145. More-over, the presence of cholesterol in POPC liposomes further reduced membrane permeabilization. This is in agreement with earlier reports that this essential mammalian membrane component attenuates the binding of AMPs[72_–74]including LL-37[75]concomitantly reducing toxicity of AMPs, which would be in accordance with observations that OP-145 was found to be safe in pig guinea, rabbits and rats, and primary skin and eye irritation/corrosion study, ototoxicity as well as after intravenous administration[26].

In summary, the present results demonstrate that the nature of the interaction of OP-145 with model membranes (bacterial and mammali-an cell membrmammali-anes) is directed by the phospholipids. OP-145 induces membrane thinning in bacterial and a detergent-like action on mamma-lian models, thus sharing the same mechanistic features as its naturally occurring parent peptide LL-37. The same mechanistic features were also reported for other amphipathicα-helical peptides such as the frog skin peptide PGLa, the bee venom toxin melittin or LL-37[19]

from which the synthetic peptide OP-145 was developed by a screening approach. Furthermore, the bacterial cell wall components, PGN and LTA seemingly do not retain OP-145 and hence do not hinder OP-145 to interact and permeabilize the cytoplasmic membrane.

Transparency document

TheTransparency documentassociated with this article can be found, in the online version.

Acknowledgments

The research leading to these results has received funding from the European Community's Seventh Framework Program (FP7/2007– 2013) under grant agreement n° 278890 (BALI Consortium).

The authors want to thank Dagmar Zweytick and Georg Pabst, University of Graz, for their helpful discussion in data analysis. References

[1] R.E. Hancock, Collateral damage, Nat. Biotechnol. 32 (1) (Jan 2014) 66–68.

[2] K. Lohner, Development of Novel Antimicrobial Agents: Emerging Strategies, Horizon Scientiﬁc Press, 2001.

[3] M. Zasloff, Antimicrobial peptides of multicellular organisms, Nature 415 (6870) (Jan 24 2002) 389–395.

[4] R.E. Hancock, A. Patrzykat, Clinical development of cationic antimicrobial peptides: from natural to novel antibiotics, Curr. Drug Targets Infect. Disord. 2 (1) (Mar 2002) 79–83.

[5] E. Andres, Cationic antimicrobial peptides in clinical development, with special focus on thanatin and heliomicin, Eur. J. Clin. Microbiol. Infect. Dis. 31 (6) (Jun 2012) 881–888.

[6] M.N. Melo, D. Dugourd, M.A. Castanho, Omiganan pentahydrochloride in the front line of clinical applications of antimicrobial peptides, Recent Pat. Antiinfect. Drug Discov. 1 (2) (Jun 2006) 201–207.

[7] M. Kazemzadeh-Narbat, B.F. Lai, C. Ding, J.N. Kizhakkedathu, R.E. Hancock, R. Wang, Multilayered coating on titanium for controlled release of antimicrobial peptides for the prevention of implant-associated infections, Biomaterials 34 (24) (Aug 2013) 5969–5977.

[8] K. Lohner, S.E. Blondelle, Molecular mechanisms of membrane perturbation by anti-microbial peptides and the use of biophysical studies in the design of novel peptide antibiotics, Comb. Chem. High Throughput Screen. 8 (3) (May 2005) 241–256.

[9] J.D. Hale, R.E. Hancock, Alternative mechanisms of action of cationic antimicrobial peptides on bacteria, Expert Rev. Anti-Infect. Ther. 5 (6) (Dec 2007) 951–959.

(10)

[10]K. Lohner, New strategies for novel antibiotics: peptides targeting bacterial cell membranes, Gen. Physiol. Biophys. 28 (2) (Jun 2009) 105–116.

[11]W.C. Wimley, Describing the mechanism of antimicrobial peptide action with the interfacial activity model, ACS Chem. Biol. 5 (10) (Oct 15 2010) 905–917.

[12] W.M. O'Leary, WSG, Gram-positive bacteria, in: C. Ratledge, S.G. Wilkinson (Eds.),Microbial Lipids, vol. 1, Academic Press, New York, NY 1988, pp. 117–201.

[13] Z. Sojcic, H. Toplak, R. Zuehlke, U.E. Honegger, R. Buhlmann, U.N. Wiesmann, Cultured human skinﬁbroblasts modify their plasma membrane lipid composition andﬂuidity according to growth temperature suggesting homeoviscous adaptation at hypothermic (30 °C) but not at hyperthermic (40 °C) temperatures, Biochim. Biophys. Acta 1104 (1) (1992 Feb 17) 31–37.

[14]J.T. Dodge, G.B. Phillips, Composition of phospholipids and of phospholipid fatty acids and aldehydes in human red cells, J. Lipid Res. 8 (6) (Nov 1967) 667–675.

[15]J.E. Rothman, J. Lenard, Membrane asymmetry, Science 195 (4280) (Feb 25 1977) 743–753.

[16] M. Zanetti, R. Gennaro, D. Romeo, Cathelicidins: a novel protein family with a com-mon proregion and a variable C-terminal antimicrobial domain, FEBS Lett. 374 (1) (Oct 23 1995) 1–5.

[17]J.M. Kahlenberg, M.J. Kaplan, Little peptide, big effects: the role of LL-37 in inﬂam-mation and autoimmune disease, J. Immunol. 191 (10) (Nov 15 2013) 4895–4901.

[18] J.H. Wong, X.J. Ye, T.B. Ng, Cathelicidins: peptides with antimicrobial, immunomod-ulatory, anti-inﬂammatory, angiogenic, anticancer and procancer activities, Curr. Protein Pept. Sci. 14 (6) (Sep 2013) 504–514.

[19] E. Sevcsik, G. Pabst, A. Jilek, K. Lohner, How lipids inﬂuence the mode of action of membrane-active peptides, Biochim. Biophys. Acta 1768 (10) (Jun 23 2007) 2586–2595.

[20] E. Sevcsik, G. Pabst, W. Richter, S. Danner, H. Amenitsch, K. Lohner, Interaction of LL-37 with model membrane systems of different complexity: inﬂuence of the lipid matrix, Biophys. J. 94 (12) (Jun 2008) 4688–4699.

[21] U.H. Durr, U.S. Sudheendra, A. Ramamoorthy, LL-37, the only human member of the cathelicidin family of antimicrobial peptides, Biochim. Biophys. Acta 1758 (9) (Sep 2006) 1408–1425.

[22] K.A. Henzler Wildman, D.K. Lee, A. Ramamoorthy, Mechanism of lipid bilayer disruption by the human antimicrobial peptide, LL-37, Biochemistry 42 (21) (Jun 3 2003) 6545–6558.

[23] K.A. Henzler-Wildman, G.V. Martinez, M.F. Brown, A. Ramamoorthy, Perturbation of the hydrophobic core of lipid bilayers by the human antimicrobial peptide LL-37, Biochemistry 43 (26) (Jul 6 2004) 8459–8469.

[24]C. Nagant, B. Pitts, K. Nazmi, M. Vandenbranden, J.G. Bolscher, P.S. Stewart, et al., Identiﬁcation of peptides derived from the human antimicrobial peptide LL-37 active against bioﬁlms formed by Pseudomonas aeruginosa using a library of truncat-ed fragments, Antimicrob. Agents Chemother. 56 (11) (Nov 2012) 5698–5708.

[25]G. Wang, B. Mishra, R.F. Epand, R.M. Epand, High-quality 3D structures shine light on antibacterial, anti-bioﬁlm and antiviral activities of human cathelicidin LL-37 and its fragments, Biochim. Biophys. Acta 1838 (9) (Jan 23 2014) 2160–2172.

[26] M.J. Nell, G.S. Tjabringa, A.R. Wafelman, R. Verrijk, P.S. Hiemstra, J.W. Drijfhout, et al., Development of novel LL-37 derived antimicrobial peptides with LPS and LTA neu-tralizing and antimicrobial activities for therapeutic application, Peptides 27 (4) (Apr 2006) 649–660.

[27] F. Peek, M.J. Nell, R. Brand, T. Jansen-Werkhoven, E. Van Hoogdalem, J. Frijns, Double-blind placebo-controlled study of the novel peptide drug P60.4Ac in cronic middle ear infection, ICAAC, 2009. L1–L337.

[28] S. Saini, N. Gupta, Aparna, Seema, O.P. Sachdeva, Bacteriological study of paediatric and adult chronic suppurative otitis media, Indian J. Pathol. Microbiol. 48 (3) (Jul 2005) 413–416.

[29]M. Verhoeff, E.L. van der Veen, M.M. Rovers, E.A. Sanders, A.G. Schilder, Chronic suppurative otitis media: a review, Int. J. Pediatr. Otorhinolaryngol. 70 (1) (Jan 2006) 1–12.

[30] M. Street, H. Crawford, Pediatric humeral osteomyelitis, J. Pediatr. Orthop. 6 (Nov 2014).

[31] O. Lesens, F. Desbiez, C. Theis, T. Ferry, M. Bensalem, H. Laurichesse, et al., Staphylo-coccus aureus-related diabetic osteomyelitis: medical or surgical management? A French and Spanish retrospective cohort, Int. J. Low. Extrem. Wounds (2014 Dec 16),http://dx.doi.org/10.1177/1534734614559931.

[32] H.S. Hiemstra, G. Duinkerken, W.E. Benckhuijsen, R. Amons, R.R. de Vries, B.O. Roep, et al., The identiﬁcation of CD4+ T cell epitopes with dedicated synthetic peptide libraries, Proc. Natl. Acad. Sci. U. S. A. 94 (19) (Sep 16 1997) 10313–10318.

[33] T.F. Moriarty, L. Debefve, L. Boure, D. Campoccia, U. Schlegel, R.G. Richards, Inﬂuence of material and microtopography on the development of local infection in vivo: ex-perimental investigation in rabbits, Int. J. Artif. Organs 32 (9) (Sep 2009) 663–670.

[34]L. Whitmore, B.A. Wallace, DICHROWEB, an online server for protein secondary structure analyses from circular dichroism spectroscopic data, Nucleic Acids Res. 32 (Jul 1 2004) W668–W673 (Web Server issue).

[35] T. Gutberlet, K. Milde, H. Bradaczek, H. Haas, H. Mohwald, Miscibility of lipoteichoic acid in dipalmitoylphosphatidylcholine studied by monoﬁlm investigations and ﬂuorescence microscopy, Chem. Phys. Lipids 69 (2) (2005) 151–159.

[36] D. Zweytick, G. Deutsch, J. Andra, S.E. Blondelle, E. Vollmer, R. Jerala, et al., Studies on lactoferricin-derived Escherichia coli membrane-active peptides reveal differences in the mechanism of N-acylated versus nonacylated peptides, J. Biol. Chem. 286 (24) (Jun 17 2011) 21266–21276.

[37] G. Pabst, M. Rappolt, H. Amenitsch, P. Laggner, Structural information from multilamellar liposomes at full hydration: full q-rangeﬁtting with high quality X-ray data, Phys. Rev. E Stat. Phys. Plasmas Fluids Relat. Interdiscip. Topics 62 (3 Pt B) (2000 Sep) 4000–4009.

[38]G. Pabst, Global properties of biomimetic membranes: perspectives on molecular features, Biophys. Rev. Lett. 1 (2006) 57–84.

[39]D.I. Svergun, Determination of the regularization parameter in indirect-transform methods using perceptual criteria, J. Appl. Crystallogr. 25 (Part 4) (1992) 495–503.

[40] K. Lohner, E. Staudegger, E.J. Prenner, R.N. Lewis, M. Kriechbaum, G. Degovics, et al., Effect of staphylococcal delta-lysin on the thermotropic phase behavior and vesicle morphology of dimyristoylphosphatidylcholine lipid bilayer model membranes. Dif-ferential scanning calorimetric, 31P nuclear magnetic resonance and Fourier trans-form infrared spectroscopic, and X-ray diffraction studies, Biochemistry 38 (50) (Dec 14 1999) 16514–16528.

[41]M. Zorko, R. Jerala, Alexidine and chlorhexidine bind to lipopolysaccharide and lipoteichoic acid and prevent cell activation by antibiotics, J. Antimicrob. Chemother. 62 (4) (Oct 2008) 730–737.

[42] M. Torrent, S. Navarro, M. Moussaoui, M.V. Nogues, E. Boix, Eosinophil cationic pro-tein high-afﬁnity binding to bacteria-wall lipopolysaccharides and peptidoglycans, Biochemistry 47 (11) (Mar 18 2008) 3544–3555.

[43] P. Laggner, K. Lohner, G. Degovics, K. Muller, A. Schuster, Structure and thermody-namics of the dihexadecylphosphatidylcholine–water system, Chem. Phys. Lipids 44 (1) (Jun 1987) 31–60.

[44]K. Lohner, A. Schuster, G. Degovics, K. Muller, P. Laggner, Thermal phase be-haviour and structure of hydrated mixtures between dipalmitoyl- and dihexadecylphosphatidylcholine, Chem. Phys. Lipids 44 (1) (Jun 1987) 61–70.

[45] J.M. Boggs, G. Rangaraj, Phase transitions and fatty acid spin label behavior in inter-digitated lipid phases induced by glycerol and polymyxin, Biochim. Biophys. Acta 816 (2) (Jun 27 1985) 221–233.

[46] T. Gutberlet, J. Frank, H. Bradaczek, W. Fischer, Effect of lipoteichoic acid on thermo-tropic membrane properties, J. Bacteriol. 179 (9) (May 1997) 2879–2883.

[47]H.U. Koch, R. Haas, W. Fischer, The role of lipoteichoic acid biosynthesis in mem-brane lipid metabolism of growing Staphylococcus aureus, Eur. J. Biochem. 138 (2) (Jan 16 1984) 357–363.

[48] S. Morath, A. Stadelmaier, A. Geyer, R.R. Schmidt, T. Hartung, Synthetic lipoteichoic acid from Staphylococcus aureus is a potent stimulus of cytokine release, J. Exp. Med. 195 (12) (Jun 17 2002) 1635–1640.

[49] J.F. Nagle, S. Tristram-Nagle, Structure of lipid bilayers, Biochim. Biophys. Acta 1469 (3) (Nov 10 2000) 159–195.

[50] D. Papahadjopoulos, Effects of bivalent cations and proteins on thermotropic prop-erties of phospholipid membranes, J. Colloid Interface Sci. 58 (1977) 459–470.

[51] D. Papahadjopoulos, W.J. Vail, K. Jacobson, G. Poste, Cochleate lipid cylinders: forma-tion by fusion of unilamellar lipid vesicles, Biochim. Biophys. Acta 394 (3) (Jul 3 1975) 483–491.

[52] H. Hauser, Effect of inorganic cations on phase transitions, Chem. Phys. Lipids 57 (2–3) (Mar 1991) 309–325.

[53] Z. Oren, J.C. Lerman, G.H. Gudmundsson, B. Agerberth, Y. Shai, Structure and organi-zation of the human antimicrobial peptide LL-37 in phospholipid membranes: rele-vance to the molecular basis for its non-cell-selective activity, Biochem. J. 341 (Pt 3) (Aug 1 1999) 501–513.

[54] M. Dathe, J. Meyer, M. Beyermann, B. Maul, C. Hoischen, M. Bienert, General aspects of peptide selectivity towards lipid bilayers and cell membranes studied by variation of the structural parameters of amphipathic helical model peptides, Biochim. Biophys. Acta 1558 (2) (Feb 1 2002) 171–186.

[55]B. Bechinger, K. Lohner, Detergent-like actions of linear amphipathic cationic anti-microbial peptides, Biochim. Biophys. Acta 1758 (9) (Sep 2006) 1529–1539.

[56]E.J. Dufourc, J.-F. Faucon, G. Fourche, J. Dufourcq, T. Gulik-Krzywicki, M. le Maire, Reversible disc-to-vesicle transition of melittin–DPPC complexes triggered by the phospholipid acyl chain melting, FEBS 201 (2) (2005) 205–209.

[57] T.M. Bayerl, G.D. Werner, E. Sackmann, Solubilization of DMPC and DPPC vesicles by detergents below their critical micellization concentration: high-sensitivity differ-ential scanning calorimetry, Fourier transform infrared spectroscopy and freeze-fracture electron microscopy reveal two interaction sites of detergents in vesicles, Biochim. Biophys. Acta 984 (2) (Sep 4 1989) 214–224.

[58]J.M. Freire, D. Gaspar, A.S. Veiga, M.A. Castanho, Shifting gear in antimicrobial and anticancer peptides biophysical studies: from vesicles to cells, J. Pept. Sci. 21 (3) (Mar 2015) 178–185.

[59]J. Turner, Y. Cho, N.N. Dinh, A.J. Waring, R.I. Lehrer, Activities of LL-37, a cathelin-associated antimicrobial peptide of human neutrophils, Antimicrob. Agents Chemother. 42 (9) (Sep 1998) 2206–2214.

[60] S.G. Dashper, N.M. O'Brien-Simpson, K.J. Cross, R.A. Paolini, B. Hoffmann, D.V. Catmull, et al., Divalent metal cations increase the activity of the antimicrobial peptide kappacin, Antimicrob. Agents Chemother. 49 (6) (Jun 2005) 2322–2328.

[61] R.M. Epand, S.W. Hui, Effect of electrostatic repulsion on the morphology and ther-motropic transitions of anionic phospholipids, FEBS Lett. 209 (2) (Dec 15 1986) 257–260.

[62] A.J. Verkleij, B. de Kruyff, P.H. Ververgaert, J.F. Tocanne, L.L. van Deenen, The inﬂu-ence of pH, Ca2+

and protein on the thermotropic behaviour of the negatively charged phospholipid, phosphatidylglycerol, Biochim. Biophys. Acta 339 (3) (Mar 29 1974) 432–437.

[63]R.F. Epand, H. Sarig, D. Ohana, B. Papahadjopoulos-Sternberg, A. Mor, R.M. Epand, Physical properties affecting cochleate formation and morphology using antimicro-bial oligo-acyl-lysyl peptide mimetics and mixtures mimicking the composition of bacterial membranes in the absence of divalent cations, J. Phys. Chem. B 115 (10) (Mar 17 2011) 2287–2293.

[64]R.M. Epand, R.F. Epand, C.J. Arnusch, B. Papahadjopoulos-Sternberg, G. Wang, Y. Shai, Lipid clustering by three homologous arginine-rich antimicrobial peptides is insensitive to amino acid arrangement and induced secondary structure, Biochim. Biophys. Acta 1798 (6) (Jun 2010) 1272–1280.

[65] R.E. Hancock, H.G. Sahl, Antimicrobial and host-defense peptides as new anti-infective therapeutic strategies, Nat. Biotechnol. 24 (12) (Dec 2006) 1551–1557.

[66] W.C. Wimley, K. Hristova, Antimicrobial peptides: successes, challenges and unan-swered questions, J. Membr. Biol. 239 (1–2) (Jan 2011) 27–34.

(11)

[67] M.N. Melo, M.A. Castanho, The mechanism of action of antimicrobial peptides: lipid vesicles vs bacteria, Front. Immunol. 3 (2012) 236,http://dx.doi.org/10.3389/ ﬁmmu.2012.00236.

[68] J.P. Powers, R.E. Hancock, The relationship between peptide structure and antibacte-rial activity, Peptides 24 (11) (Nov 2003) 1681–1691.

[69]N. Papo, Y. Shai, Can we predict biological activity of antimicrobial peptides from their interactions with model phospholipid membranes? Peptides 24 (11) (Nov 2003) 1693–1703.

[70] M.N. Melo, R. Ferre, M.A. Castanho, Antimicrobial peptides: linking partition, activity and high membrane-bound concentrations, Nat. Rev. Microbiol. 7 (3) (Mar 2009) 245–250.

[71] D. Roversi, V. Luca, S. Aureli, Y. Park, M.L. Mangoni, L. Stella, How many antimicrobial peptide molecules kill a bacterium? The case of PMAP-23, ACS Chem. Biol. 9 (9) (Sep 19 2014) 2003–2007.

[72] E.J. Prenner, R.N. Lewis, M. Jelokhani-Niaraki, R.S. Hodges, R.N. McElhaney, Choles-terol attenuates the interaction of the antimicrobial peptide gramicidin S with phos-pholipid bilayer membranes, Biochim. Biophys. Acta 1510 (1–2) (Feb 9 2001) 83–92.

[73] J.R. Brender, A.J. McHenry, A. Ramamoorthy, Does cholesterol play a role in the bac-terial selectivity of antimicrobial peptides? Front. Immunol. 3 (2012) 195,http://dx. doi.org/10.3389/ﬁmmu.2012.00195.

[74]D.K. Lee, A. Bhunia, S.A. Kotler, A. Ramamoorthy, Detergent-type membrane frag-mentation by MSI-78, MSI-367, MSI-594, and MSI-843 antimicrobial peptides and inhibition by cholesterol: a solid-state nuclear magnetic resonance study, Biochem-istry 54 (10) (Mar 17 2015) 1897–1907.

[75] R. Sood, Y. Domanov, M. Pietiainen, V.P. Kontinen, P.K. Kinnunen, Binding of LL-37 to model biomembranes: insight into target vs host cell recognition, Biochim. Biophys. Acta 1778 (4) (Apr 2008) 983–996.