Báo cáo khoa học: Investigations into the ability of an oblique a-helical template to provide the basis for design of an antimicrobial anionic amphiphilic peptide pot

Generally, a-AMPs are cationic [11,12], which facili-tates their interaction with the anionic membranes of microbial cells, and they exert their antimicrobial action by the use of nonrec

template to provide the basis for design of an

antimicrobial anionic amphiphilic peptide

Sarah R Dennison1, Leslie H G Morton2, Klaus Brandenburg3, Frederick Harris4and

David A Phoenix1

1 Faculty of Science, University of Central Lancashire, Preston, UK

2 School of Natural Resources, University of Central Lancashire, Preston, UK

3 Forschungszentrum Borstel, Leibniz-Center for Medicine and Biosciences, Borstel, Germany

4 Department of Forensic and Investigative Science, University of Central Lancashire, Preston, UK

Globally and particularly in developing countries [1],

antimicrobial drug resistance has become a major

prob-lem, resulting in a decline in the effectiveness of existing

antimicrobial agents [2] As a consequence, infections

have been rendered more expensive and harder to treat,

and epidemics have been made more difficult to

con-trol Moreover, many previously treatable infectious

diseases such as tuberculosis now have greatly

increased rates of morbidity and mortality [3] In

response, the pharmaceutical industry has investigated

a number of compounds with the potential to act as new and effective antimicrobial agents [4], ranging from photosensitizing dyes [5] to nucleosides [6] A recent focus of these investigations has been a-helical anti-microbial peptides (a-AMPs) which are components of mammalian innate immune systems [7–10]

Generally, a-AMPs are cationic [11,12], which facili-tates their interaction with the anionic membranes

of microbial cells, and they exert their antimicrobial action by the use of nonreceptor-based mechanisms of

Keywords

anionic; antimicrobial; a-helical; membrane;

peptide

Correspondence

D A Phoenix, Deans Office, Faculty of

Science, University of Central Lancashire,

Preston PR1 2HE, UK

Fax: +44 1772 892903

Tel: +44 1772 893481

E-mail: daphoenix@uclan.ac.uk

(Received 12 January 2006, revised 12 June

2006, accepted 20 June 2006)

doi:10.1111/j.1742-4658.2006.05387.x

AP1 (GEQGALAQFGEWL) was shown by theoretical analysis to be an anionic oblique-orientated a-helix former The peptide exhibited a mono-layer surface area of 1.42 nm2, implying possession of a-helical structure

at an air⁄ water interface, and Fourier transform infrared spectroscopy (FTIR) showed the peptide to be a-helical (100%) in the presence of vesi-cle mimics of Escherichia coli membranes FTIR lipid-phase transition analysis showed the peptide to induce large decreases in the fluidity of these E coli membrane mimics, and Langmuir–Blodgett trough analysis found the peptide to induce large surface pressure changes in monolayer mimics of E coli membranes (4.6 mNÆm)1) Analysis of compression iso-therms based on mixing enthalpy (DH) and the Gibbs free energy of mix-ing (DGMix) predicted that these monolayers were thermodynamically stable (DH and DGMix each negative) but were destabilized by the pres-ence of the peptide (DH and DGMix each positive) The peptide was found

to have a minimum lethal concentration of 3 mm against E coli and was seen to cause lysis of erythrocytes at 5 mm In combination, these data clearly show that AP1 functions as an anionic a-helical antimicrobial pep-tide and suggest that both its tilted peppep-tide characteristics and the com-position of its target membrane are important determinants of its efficacy

of action

Abbreviations

a-AMP, a-helical antimicrobial peptide; AP1, GEQGALAQFGEWL; FTIR, Fourier transform infrared spectroscopy; Ole 2 PtdEtn,

dioleoylphosphatidylethanolamine; Ole2PtdGro, dioleoylphosphatidylglycerol; SUV, small unilamellar vesicle.

membrane invasion [13,14] The relatively nonspecific

nature of these mechanisms renders the development

of acquired microbial resistance to a-AMPs unlikely,

although several mechanisms of inherent resistance to

these peptides have been reported [11,15,16] The most

common of these mechanisms is exhibited by both

Gram-positive and Gram-negative pathogens and

effectively involves the reduction of anionic lipid

con-centrations in the bacterial cell envelope, thereby

inhib-iting the membrane-binding ability of cationic a-AMPs

[17–19]

Most recently, theoretical studies have suggested

that many a-AMPs may destabilize bacterial

mem-branes by the use of oblique orientated a-helical

struc-ture [20], which has been experimentally demonstrated

for the amphibian a-AMPs: aurein 1.2, citropin 1.1

and caerin 1.1 [21] These a-helices have been described

in a variety of proteins and peptides, most commonly

viral protein segments, and are differentiated from

other classes of membrane-interactive a-helices in that

they possess a hydrophobicity gradient along the

a-helical long axis This structural feature causes an

a-helix to penetrate membranes at a shallow angle of

30–60, thereby disturbing membrane lipid

organiza-tion and compromising bilayer integrity [22,23]

Among the a-AMPs predicted to form

oblique-orientated a-helices [12] are a small number that are

negatively charged, such as the amphibian peptide

maximin H5 [24] It has been suggested that anionic

a-AMPs may have evolved to counter microbe

resist-ance to cationic a-AMPs, which would seem to make

these former peptides well suited for development as

novel antimicrobial agents directed against such

organ-isms [24,25]

There appears to have been little research into the

mode of membrane interaction used by anionic

a-AMPs, although photodynamic antimicrobial studies

have shown nonpeptide anionic molecules to be

effect-ive against Gram-negateffect-ive bacteria because of their

ability to penetrate the membranes of these organisms

[26,27] Pathogenic Gram-negative bacteria are

becom-ing increasbecom-ingly problematic in areas rangbecom-ing from

health care to the food industry [28–31], therefore

in this study we analysed a novel synthetic peptide,

AP1 (GEQGALAQFGEWL), as a potential anionic

a-AMP against Escherichia coli The sequence of AP1

was designed to form a membrane-interactive oblique

orientated a-helix, shown here by theoretical analysis,

and Fourier transform infrared spectroscopy (FTIR)

confirmed that the peptide was a-helical in the

pres-ence of lipid vesicles that mimicked membranes of

E coli A standard toxicity assay showed that

AP1 inactivated the organism, and the use of

Langmuir-Blodgett troughs showed that the peptide inserted strongly into lipid monolayers that mimicked

E coli membranes Compression isotherm analysis indicated that lipid monolayers mimicking E coli membranes were thermodynamically stable but were destabilized by the presence of AP1 FTIR lipid-phase transition analysis showed that the peptide induced changes in the membrane fluidity of E coli mem-branes, which were consistent with penetration of the hydrophobic core of these membranes AP1 was found

to lyse erythrocyte membranes, and, on the basis of these combined data, it is suggested that the peptide functions as an anionic membrane-interactive a-AMP These data also suggest that the antimicrobial activity

of AP1 depends on both the structural characteristics

of its tilted peptide architecture and the lipid packing

of its target membrane

Results

The influenza HA2 fusion peptide is known to form a membrane-interactive oblique orientated a-helix [32], a secondary-structural motif recently postulated to fea-ture in the action of a range of a-AMPs [20] A seg-ment of the HA2 peptide (GLFGAIAGFIENG), which is key to its structure and underlying phobicity gradient, was used as a basis for the hydro-phobicity gradient of the AP1 peptide, thereby giving these peptides 62% sequence homology The sequence

of AP1 is predicted to produce an a-helical peptide with structural features that are characteristic of both this oblique orientated a-helix and the established anionic a-AMP, maximin H5 Figure 1A shows that, when the sequences of these three peptides were ana-lysed using extended hydrophobic moment plot meth-odology, the resulting data points were proximal and, along with those of  50% of the a-AMPs studied, are candidates to form oblique orientated a-helices Figure 1B shows that, in an a-helical conformation, AP1 would possess a hydrophobic arc size of 90, and Fig 2 indicates that this value (Fig 2A) and the mean hydrophobic moment of the peptide (0.35; Fig 2B) are highly comparable to those of maximin H5 However, Fig 2A,B also show that, for both peptides, these val-ues fall in the lower quartile of those observed for the cationic a-AMPs tested

Monolayer analysis showed that increasing con-centrations of AP1 in the subphase of a Langmuir-Blodgett trough led to progressively greater interfacial surface pressures until, at 20 lm peptide, a maximal value of 11.5 mNÆm)1 was observed (Fig 3) Above this peptide concentration, surface pressures were effectively constant, which indicates that 20 lm AP1

was the minimum bulk concentration required to

sat-urate the air⁄ water interface with the peptide under

these experimental conditions These data were used to

determine the corresponding interfacial surface area per AP1 molecule (Table 1), and, for 20 lm peptide, extrapolation provides an estimate of peptide surface

of 1.40 nm2, which is comparable to that found for peptides that adopt a-helical structure [33]

When spread from chloroform on to the subphase

of a Langmuir-Blodgett trough, AP1 formed stable monolayers Under compression, these monolayers showed collapse pressures in the region of 20 mNÆm)1 (Fig 4), indicating the presence of a well-ordered monolayer [34] Compressibility moduli, Cs1, were derived from these isotherms (Table 2) and generally decreased with increasing surface pressure, indicating that the monolayer is in the protein phase [35] Figure 4 also shows that the area per AP1 molecule

1.5

1.25

1

0.75

0.5

0.25

0

<H>

A

G10 Q3

B

A7

E11 G4 Q8 W12G1 A5 F9 E2 L13 L6

Fig 1 (A) Extended hydrophobic moment plot analysis of AP1 (m),

the known anionic a-AMP, maximin H5 (d), and peptides of the

a-AMP dataset (http://www.uclan.ac.uk/biology/bru/amp_data.htm),

all as described in the text AP1, maximin H5, and  50% of the

peptides in the dataset are represented by data points that lie in

the shaded region, delineating candidacy for oblique-orientated

a-helix formation (B) Sequence of AP1 represented as a 2D axial

projection This a-helix possesses a hydrophilic face, which is rich

in glycine residues and polar residues (circled), and a hydrophobic

face formed from bulky apolar residues with a centrally placed

glu-tamate residue.

1.20 1.00 0.80 0.60 0.40 0.20

350

300

250

200

150

100

50

0

Fig 2 Box plot for the hydrophobic arc size (A) and mean

hydro-phobic moment (B) of AP1, maximin H5, which is a known anionic

a-AMP, and the a-AMP dataset, http://www.uclan.ac.uk/biology/

bru/amp_data.htm, all determined as described in the text AP1 and

maximin H5 show comparable amphiphilic properties, which

gener-ally lie in the lower quartile range of the dataset.

0 2 4 6 8 10 12

AP1 concentration ( M )

-1 )

Fig 3 AP1 surface pressure as a function of peptide concentration Increasing concentrations of AP1 were injected into a Tris ⁄ HCl buf-fer subphase (10 mM, pH 7.5) of a Langmuir-Blodgett system At each AP1 concentration, the peptide was allowed to equilibrate, and the surface pressure determined and plotted, all as described

in the text.

Table 1 Surface excess (G) and interfacial surface area per AP1 molecule (A) for various molar subphase concentrations (C) of the peptide where p is the interfacial pressure increase Values for these parameters were derived using AP1 surface pressure data from Fig 3 with G computed using Eqn (1) and A computed using Eqn (2), all as described in the text.

C (lM)

p

A (nm 2 )

corresponding to this collapse pressure was 0.33 nm2.

The extrapolated area at p¼ 0 mNÆm)1 for the

iso-therm provides a measure of the mean monolayer

surface area per AP1 molecule [36] This area was

1.42 nm2 per AP1 molecule and is comparable to the

value of 1.40 nm2 calculated above for the peptide

using Eqns (1) and (2) (Table 1) Although towards

the lower end of the expected range, this would

approximate to that predicted for AP1 if the peptide

was orientated perpendicular to the air⁄ water interface

(1.77 nm2 [37]), but may also indicate the presence of

some non-a-helical structure in AP1

FTIR conformational analysis showed that AP1

adopted predominantly b-type structures in solution

However, at lipid to peptide ratios of 50 : 1 and above,

the peptide adopted  100% a-helical structure in the

presence of lipid assemblies that mimicked membranes

of E coli (Fig 5) This conformational behaviour is

similar to that shown by most a-AMPs, which are

gen-erally non-a-helical in solution but assume a-helical

structure at the microbial interface [38–40]

A standard toxicity assay established a minimum lethal concentration of 3 mm for AP1 when directed against E coli Analysing the number of colony form-ing units over time showed that, at this concentration, the peptide took 1 h to induce 100% cell death (Fig 6)

Microbial membrane invasion is the primary killing mechanism used by most a-AMPs [41,42] FTIR con-formational analysis shows that, in the absence of AP1, small unilamellar vesicles (SUVs) mimetic of

E coli membranes underwent transition from the gel phase to liquid crystalline phase over the temperature range 20–70C with a concomitant increase in mem-brane fluidity, as indicated by the rise in wavenumber from  2851.0 cm)1 to 2852.3 cm)1 The presence of AP1 caused no apparent shift in the temperature range

of these phase transitions but, over this temperature range, induced a significant decrease in the membrane fluidity of E coli membranes (Fig 7)

AP1 also interacted with lipid monolayers that were mimetic of E coli membranes (Fig 8), inducing

Fig 4 A pressure–area isotherm for an AP1 monolayer The

pep-tide was spread from chloroform on to a Tris ⁄ HCl buffer subphase

(10 mM, pH 7.5) The variation of surface pressure with area per

peptide molecule was monitored as monolayers were compressed

and plotted, all as described in the text.

Table 2 Compressibility moduli (Cs1) of lipid monolayers at

vary-ing surface pressure (p) (all values mNÆm)1) Values of Cs1 were

computed using data from compression isotherms (Fig 9) and Eqn

(3) Monolayers were formed from either Ole 2 PtdGro, Ole 2 PtdEtn,

cardiolipin, or lipid mixtures that corresponded to membranes of

E coli, all as described in the text.

Pressure p

(mNÆm)1)

Cs1(mNÆm)1)

Cardiolipin E coli Ole 2 PtdGro Ole 2 PtdEtn

Fig 5 FTIR conformational analysis of AP1 in the presence of SUVs with lipid compositions that correspond to those of E coli membranes, all as described in the text The numbers annotating spectra indicate peak band absorbancies For each spectrum, the relative percentages of a-helical structure (1650–1660 cm)1) and b-sheet structures (1625–1640 cm)1) were computed, all as des-cribed in the text In aqueous solution, AP1 was predominantly formed from b-type structures (A), but in the presence of E coli membrane mimics (B), the peptide was 100% a-helical.

maximal changes in surface pressure of 4.6 mNÆm)1

after 6500 s This was further investigated by

thermo-dynamic analysis of compression isotherms derived

from monolayer mimics of E coli membranes in either

the absence (Fig 9A) or presence of AP1 (Fig 9B)

Cs1 were derived from these isotherms (Table 2), and

Cs1 is seen to be generally low, indicating that the

lipid monolayers analysed were in a liquid expanded

phase [35] and, thus, are more fluid and possess high

Fig 6 Time course for the viability of E coli represented as

per-centage death rate in the presence of AP1 (3 mM) At these

con-centrations, the peptide is bactericidal, achieving a 100% death

rate after 1 h The percentage death rate was determined by

com-parison with identical noninoculated control cultures, all as

des-cribed above, and error bars represent the standard error on three

replicates.

2854

2853

2852

2851

2850

0 10 20 30 40 50 60 70 80

Temperature (°C)

Fig 7 FTIR lipid-phase transition analyses of SUVs with lipid

com-positions that correspond to those of E coli membranes, all as

des-cribed in the text In the absence of AP1 (,) model membranes of

E coli underwent a transition from the gel phase to the liquid

crys-talline phase liquid over the temperature range 30–70 C with a

concomitant increase in membrane fluidity s indicated by the rise in

wavenumber from  2851.0 cm)1to 2852.3 cm)1 The presence of

AP1 caused no apparent shift in this temperature range but induced

a significant decrease in the membrane fluidity of E coli

mem-branes, which is consistent with the peptide penetrating the

hydro-phobic core of these membranes.

Fig 8 Time course of interactions between AP1 and monolayers with lipid compositions that correspond to those of E coli mem-branes, all as described in the text Monolayers were at an initial surface pressure of 30 mNÆm)1, mimetic of naturally occurring membranes, and the peptide was introduced into the subphase to give a final concentration 20 lM, all as described in the text.

Fig 9 Compression isotherms of monolayers formed from: lipid compositions that correspond to those of E coli membranes (a), Ole2PtdEtn (b), Ole2PtdGro (c) and cardiolipin (d) The variation of surface pressure with area per lipid molecule was monitored as monolayers were compressed on a Tris ⁄ HCl buffer subphase (10 mM, pH 7.5) either in the absence of AP1 (A) or containing AP1 with a final concentration of 20 lM (B), all as described above.

compressibility Table 2 also shows that AP1 induced

a general decrease in Cs1 with rising monolayer

sur-face pressure, indicating expansion of these bacterial

membrane mimics because of peptide interactions

Values for the Gibbs free energy of mixing (DGMix)

(Table 3) were derived from the compression isotherms

shown in Fig 9 It can be seen from Table 3 that

DGMixfor E coli model membranes varies with surface

pressure and according to the presence or absence of

AP1 Table 3 shows that values of DGMix for these

E coli model membranes are much lower than RT

(2444.316 JÆmol)1), indicating that deviations from

ideal mixing behaviour are small In the absence of

AP1, negative values of DGMix were observed for

E colimodel membranes (Table 3), indicating a stable

monolayer However, in the presence of AP1 (Table 3),

positive values of DGMixare observed for E coli model

membranes, indicating that, although the lipids

form-ing these monolayers are miscible, repulsive

interac-tions are established in the presence of the peptide,

thereby decreasing membrane stability These values of

DGMix become progressively more positive as surface

pressure increases, showing that, at higher surface

pressures, mutual interactions between the component

molecules of these membranes are weaker than those

occurring in monolayers formed by their pure

compo-nents [43], becoming increasingly less stable with

compression This instability may contribute to the

susceptibility of E coli model membranes to the action

of AP1

An important determinant of the susceptibility of

membranes to a-AMPs is the packing characteristics

of the individual membrane lipids [44] To evaluate the

nature of interactions between the component lipid

molecules in E coli model membranes, the interaction

parameter, a, and the mixing enthalpy, DH, were

com-puted (Table 3) It can be seen from Table 3 that, in

the absence of AP1, values for a and DH are negative

for these model membranes, but, in the presence of the

peptide, they are positive These results confirm that

E coli membranes are thermodynamically less stable

in the presence of AP1, further supporting the

sugges-tion that this instability may contribute to the

susceptibility of E coli to the antimicrobial action of the peptide

Discussion

The biological action of many pore-forming and lytic peptides involves membrane destabilization by the use

of lipid-interactive oblique-orientated a-helical struc-ture [22], and such strucstruc-ture also appears to be used

by many a-AMPs [20] It has been suggested that anionic a-AMPs and their analogues may serve as complements to their cationic counterparts in some therapeutic contexts [24,25] Here, a synthetic peptide, AP1, was prepared to observe whether anionic pep-tides with tilted peptide characteristics could be designed to act as potential anionic a-AMPs

Theoretical analysis confirmed that the peptide pos-sessed the potential to form an a-helix with a balance between amphiphilicity and hydrophobicity, and struc-tural characteristics that are associated with oblique-orientated a-helices (Figs 1 and 2) It can be seen from Fig 1B that the AP1 a-helix possesses a glycine-rich polar face, and it has previously been shown that simi-larly located glycine residues are critical for maintaining the hydrophobicity gradients associated with mem-brane-interactive oblique-orientated a-helices [45] It can also be seen from Fig 1B that the AP1 a-helix pos-sesses a wide hydrophobic face rich in bulky amino-acid residues, and, in combination with a glycine-rich polar face, these structural characteristics give a-helices an effective inverted wedge shape It has been recently shown that a number of a-AMPs, experimentally dem-onstrated to penetrate membranes in an oblique orien-tation, appear to possess this inverted wedge shape [12,21] In addition, it can be seen from Fig 1B that a glutamate residue is centrally located in the apolar face

of the AP1 a-helix, and previous studies have shown that similarly located glutamate residues are important for the antimicrobial action of other a-AMPs also predicted to form an inverted wedge shape [46]

FTIR spectroscopy showed that AP1 was completely a-helical in the presence of model membranes mimetic

of those of E coli (Fig 5), although molecular area

Table 3 Gibbs free energy of mixing (DG Mix ), interaction parameter (a) and enthalpy of mixing (DH) at varying surface pressure (p) for lipid mixtures that correspond to membranes of E coli Values for these parameters were computed either in the presence or absence of AP1 using data from compression isotherms (Fig 9) in conjunction with Eqns (4), (5) and (6), respectively, all as described above.

Pressure

p (mNÆm)1)

determinations showed that the peptide may possess

some non-a-helical structure at an air⁄ water interface

It would seem that AP1 specifically requires the

amphiphilicity associated with the environment of a

membrane or lipid interface to form such structure

Monolayer studies confirmed that AP1 was able to

par-tition into model membranes that were mimetic of

those of E coli (Fig 8), and toxicity assay showed that

AP1 was bactericidal at 3 mm (Fig 6) In combination,

these data clearly show that the peptide is able to

func-tion as an anionic a-AMP Moreover, these results

sug-gest that interaction with bacterial membranes features

in the antibacterial action of AP1, and it is well

estab-lished that perturbation of the microbial membrane is a

primary killing mechanism used by a-AMPs [41,42]

This suggestion is strongly reinforced by the

observa-tion that AP1 showed haemolytic ability, thereby

clearly confirming that the peptide is able to induce cell

bilayer disruption AP1 was found to be haemolytic at

5 mm, thereby showing a common characteristic of

a-AMPs in that higher concentrations of these peptides

are generally required for haemolytic action than for

bactericidal action [38] The minimum lethal

concentra-tion of AP1 is far in excess of those normally required

by cationic a-AMPs to inhibit target micro-organisms

(< 20 lm), but is closer to those required by some

ani-onic a-AMPs (80 lm) [12], which are known to

gener-ally exhibit lower levels of antimicrobial efficacy than

their cationic counterparts Lipid-phase transition

ana-lysis clearly suggested that the interactions of the

pep-tide with membranes of E coli induced a significant

decrease in the membrane fluidity of E coli membranes

(Fig 7), which is consistent with the peptide

penetrat-ing deeply into the membranes hydrophobic core

To investigate further the mechanism of bacterial

membrane interaction used by AP1, thermodynamic

analysis of compression isotherms for lipid monolayer

mimetics of E coli membranes were undertaken

(Table 3, Fig 9) These analyses gave negative values

for DGMix, a and DH in the absence of AP1, indicating

membrane stability, but, in the presence of AP1,

posit-ive values of for DGMix, a and DH were obtained,

sug-gesting that the monolayer had become less stable

This shows that the association of AP1 with these

model membranes had a destabilizing effect, and, when

taken with the FTIR data above, suggests that the

peptide may promote toxicity to E coli by a lytic-type

mechanism involving disturbance of lipid acyl chains

within the membrane core [47]

It is well established that the packing characteristics

of component lipids is an important factor in

determining the stability of membrane bilayers [44] It

is interesting to note that E coli membranes possess

high levels of phosphatidylethanolamine ( 85%), which is effectively shaped like an inverted wedge and

is known to have a strong preference for the nonlamel-lar H11phase [14] Thus, according to the wedge hypo-thesis of Tytler et al [48], it may be that insertion of the inverted wedge shape formed by the AP1 a-helix into membranes of E coli leads to the formation of nonbilayer structures and thereby membrane destabil-ization Such a mechanism of membrane perturbation would be consistent with the use of a lytic-type mech-anism for antimicrobial action, as predicted by the thermodynamic analyses above and the involvement of oblique-orientated a-helical structure in AP1 The higher concentrations of peptide required for haemo-lysis would indicate that the membrane composition plays an important role in activity

In summary, AP1 was found to function as an ani-onic a-AMP, indicating that it is possible to design a-AMPs by the use of an oblique-orientated a-helical template It appears from the biophysical data that the peptide uses this structure for the destabilization of membranes of Gram-negative bacteria, thereby promo-ting the inactivation of these organisms The relatively high concentration required for the minimum lethal concentration indicates though that further lessons with respect to the amino-acid composition are still to

be learnt However, as a general lesson, the data pre-sented in this study emphasize that, in development of antimicrobial compounds, both the structural charac-teristics and composition of their target membrane are important determinants of their efficacy of action

Experimental procedures

Reagents

AP1 (GEQGALAQFGEWL) was supplied by Pepsyn (Liverpool, UK), produced by solid-state synthesis, and purified by HPLC to greater than 95%, which was con-firmed by MALDI MS Buffers and solutions for

monolay-er expmonolay-eriments wmonolay-ere prepared from Milli-Q watmonolay-er Nutrient broth was purchased from Amersham Bioscience (GE Healthcare, Chalfont St Giles, UK) Dioleoylphosphatidyl-glycerol (Ole2PtdGro) and dioleoylphosphatidylethanolam-ine (Ole2PtdEtn) were purchased from Alexis Corporation (Axxora, Bingham, UK) Cardiolipin, Hepes, Tris and all other reagents were purchased from Sigma (Sigma-Aldrich, Gillingham, UK)

Primary structure analyses

The sequences of 161 known a-AMPs were obtained from Dennison et al [41] (http://www.uclan.ac.uk/biology/bru/

amp_data.htm) and along with those of maximin H5

(ILGPVLGLVSDTLDDVLGIL [24]) and AP1 were

ana-lysed according to conventional hydrophobic moment

methodology [49] Essentially, this methodology treats the

hydrophobicity of successive amino acids in a sequence, as

vectors and then sums these vectors in two dimensions,

assuming an amino-acid side chain periodicity of 100 The

resultant of this summation, the hydrophobic moment,

pro-vides a measure of a-helix amphiphilicity The analysis of

the present study used a moving window of 11 residues,

and for each sequence under investigation, the window with

the highest hydrophobic moment was identified [49] For

these windows, the mean hydrophobic moment, <lH>,

and the corresponding mean hydrophobicity, <H>, were

computed using the online program moment helix

predic-tion (http://www.doe-mbi.ucla.edu/Services/moment/) and

the normalized consensus hydrophobicity scale of Eisenberg

et al [49] These mean values were plotted on the

hydro-phobic moment plot diagram of Eisenberg et al [50], as

modified by Harris et al [23], to identify candidate

oblique-orientated a-helix-forming segments Assuming an idealized

a-helix with a residue side chain angular periodicity of

100, a 2D axial projection of the peptide was generated

[51] The angle subtended by the hydrophobic residue

distri-bution was taken as a measure of hydrophobic arc size

Preparation of lipid unilamellar vesicles

SUVs with lipid compositions designed to mimic E coli

membranes were prepared as described by Keller et al [52]

Essentially, chloroform solutions of Ole2PtdGro, Ole2

Ptd-Etn and cardiolipin in molar proportions of 1 : 13.67 : 2

[53] were dried with nitrogen gas and hydrated with Hepes

buffer (10 mm, pH 7.5) to give final total lipid

concentra-tions of 150 mm The resulting cloudy suspensions were

sonicated at 4C with a Soniprep 150 sonicator (amplitude

10 lm) until clear (30 cycles of 30 s), centrifuged (15 min,

3000 g, 4C), and the supernatant decanted for immediate

use

FTIR conformational analysis of AP1

To give final peptide concentration ranging from 3 mm

to1 mm, AP1 was solubilized in either Hepes buffer

(10 mm, pH 7.5) or suspensions of SUVs formed from

Ole2PtdGro, Ole2PtdEtn and cardiolipin as described

above These samples were spread individually on a CaF2

crystal, and the free excess water was evaporated at room

temperature The single band components of the VAP1

amide I vibrational band (predominantly C¼O stretch) was

monitored using an FTIR ‘5-DX’ spectrometer (Nicolet

Instruments, Madison, WI, USA), and, for each sample,

absorbance spectra were produced For these spectra, water

bands were subtracted, and the evaluation of peptide band

parameters (peak position, band width and intensity)

performed Curve fitting was applied to overlapping bands using a modified version of the curfit procedure written

by Dr Moffat, National Research Council, Ottawa, Ontario, Canada The band shapes of the single compo-nents are superpositions of Gaussian and Lorentzian band shapes Best fits were obtained by assuming a Gauss frac-tion of 0.55–0.6 The curfit procedure measures the peak areas of single band components, and, after statistical evaluation, determines the relative percentages of primary structure involved in secondary-structure formation, all as described by Dennison et al [54]

FTIR analysis of phospholipid phase-transition properties

To give a final peptide concentration of 3 mm, AP1 was solu-bilized in suspensions of SUVs, which were formed from Ole2PtdGro⁄ Ole2PtdEtn⁄ cardiolipin as described above As controls, corresponding lipid SUVs were prepared with no peptide present All samples were then subjected to automa-tic temperature scans with a heating rate of 3C per 5 min and within the temperature range 0–60C For every 3 C interval, 50 interferograms were accumulated, apodized, Fourier transformed, and converted into absorbance spectra [55] These spectra monitored changes in the b fi a acyl chain melting behaviour of phospholipids, with these changes determined as shifts in the peak position of the symmetric stretching vibration of the methylene groups, ms(CH2), which

is known to be a sensitive marker of lipid order The peak position of ms(CH2) lies at 2850 cm)1 in the gel phase and shifts at a lipid specific temperature Tc to 2852.0– 2852.5 cm)1in the liquid crystalline state [55]

Antimicrobial assay

Cultures of the E coli strain W3110, which had been freeze-dried in 20% (v⁄ v) glycerol and stored at )80 C, were inoculated into 10 mL nutrient broth After overnight incubation in an orbital shaker (100 r.p.m., 37C), 100-lL aliquots of these cultures were used to inoculate 100 mL nutrient broth in 250 mL flasks, which were then incubated with shaking (100 r.p.m., 37C) until growth in the mid-exponential phase was reached (A¼ 0.6; k ¼ 600 nm) Aliquots (1 mL) of bacterial samples were centrifuged, using a bench top centrifuge (15 000 g, 3 min, 22C), and the centrifuged cells washed three times in 1-mL aliquots of Tris⁄ HCl buffer (10 mm, pH 7.5) These cells were then suspended in 1 mL of this buffer containing AP1 at a final concentration of 3 mm, which corresponds to its minimum inhibitory concentration These culture⁄ peptide mixtures were incubated at 37C, and samples taken at the begin-ning of the experiment (time zero), and at 15 minute inter-vals for 1 h and then hourly interinter-vals for 7 h At each time interval, samples were surface-spread on to nutrient agar plates, which were incubated at 37C for 12 h As a

control, bacterial cultures were similarly treated but in the

absence of peptide Colony counts were expressed as colony

forming units (CFU) ml)1 The percentage reduction in

col-ony counts for each time interval was then calculated, and

the results were presented graphically against time

Monolayer technique

All experiments were conducted at 21.0 ± 1C using a

Langmuir trough measuring 5· 16 cm, which was fitted

with two moveable barriers and was supplied by NIMA

Technology (Coventry, UK) Unless otherwise stated,

monolayer studies were performed using a Tris⁄ HCl buffer

subphase (10 mm, pH 7.5), which was continuously stirred

by a magnetic bar (5 r.p.m.) Surface tension was

monit-ored by the Wilhelmy method using a Whatman’s (Ch1)

paper plate in conjunction with a microbalance, as

des-cribed by Brandenburg et al [56] Changes in monolayer

surface pressure⁄ area were recorded as graphic output on a

PC using nima software version 5.16, which interfaces with

the Langmuir-Blodgett microbalance

Peptide surface activity

The barriers of the Langmuir-Blodgett trough were

adjus-ted to their maximum separation (surface area 80 cm2), and

this position maintained AP1 was then injected into the

buffer subphase to give final concentrations of 1–30 lm,

and, at each peptide concentration, changes in surface

pres-sure at the air⁄ water interface were monitored for 1 h The

maximal values of these surface pressure changes were then

plotted as a function of the peptide’s final subphase

concen-tration (Fig 3) From these results, the surface excess, G,

was calculated by means of the Gibbs’ adsorption isotherm,

which is given by Eqn (1) [57]:

C¼  1

where R is 8.314 JÆmol)1ÆK)1, T¼ 294 K, p is the

interfa-cial pressure increase (mNÆm)1), and c is the molar

concen-tration of peptide in the subphase These values ofG were

then used to determine values of the interfacial surface area

per AP1 molecule (A) according to Eqn (2):

A¼ 1

where N is Avogadro’s number (Table 1)

The ability of AP1 to spread on an aqueous surface and

to form a stable monolayer was investigated The barriers

of the Langmuir-Blodgett trough were adjusted to their

maximum separation (surface area 80 cm2) and this

posi-tion maintained A 10-lL aliquot of AP1 in chloroform

(1 mm) was spread on to a buffer subphase and allowed to

equilibrate for 1 h The resulting peptide monolayer was

compressed using the moveable barriers of the trough to

produce a pressure⁄ area isotherm, which was converted by nima software in to an output plot of surface pressure vs monolayer surface area per AP1 molecule (Fig 4)

Peptide interactions with lipid monolayers

The ability of AP1 to penetrate lipid monolayers at con-stant area was studied Monolayers were formed by spread-ing on to a buffer subphase, chloroform solutions of Ole2PtdGro, Ole2PtdEtn and cardiolipin in molar propor-tions of 1 : 13.67 : 2 [53] The solvent was allowed to evap-orate off over 30 min and then the monolayer compressed

at a velocity of 5 cm2Æmin)1 to give a surface pressure of

30 mNÆm)1 The barriers were maintained in this position, and peptide was then injected into the subphase to achieve the desired optimum peptide concentration of 20 lm which was determined by analysis of surface activity data des-cribed in Fig 3 This subphase concentration of AP1 gave rise to a lipid to peptide ratio of approximately 100 : 1, which was used in all other monolayer studies Interactions

of the peptide with lipid monolayers were monitored as changes in monolayer surface pressure vs time

The ability of the peptide to interact with lipid monolay-ers was also investigated using compression isotherms Monolayers were formed by spreading on to a buffer sub-phase chloroform solutions of either Ole2PtdEtn, Ole2 Ptd-Gro, cardiolipin, or these lipids in molar proportions of

1 : 13.67 : 2 [53] The solvent was allowed to evaporate off over 30 min, and monolayers then compressed using a bar-rier speed of 5 cmÆmin)1 either with AP1 absent from the subphase or included in the subphase at a final peptide con-centration of 20 lm Changes in monolayer surface pressure

vs changes in area per lipid molecule of the monolayer were monitored and recorded

Thermodynamic analysis of compression isotherm data

Thermodynamic analysis of compression isotherms was used to investigate the molecular interactions and dynamic behaviour of monolayers The compressibility modulus,

Cs1, provides a measure of the compressional elasticity of

a monolayer and can be used to characterize the phase state

of the isotherm, thereby providing information about the compactness and packing of the model membrane [35] Val-ues of Cs1(Table 1) were computed according to Eqn (3):

C1s ¼ A dp

dA

 

ð3Þ where p is the surface pressure of the monolayer, and A represents the area per molecule in the monolayer

The Gibbs free energy of mixing (DGMix) quantifies the sta-bility of monolayer mixtures, thereby providing information

on interactions between the components of the monolayers Values of DGMixwere computed according to Eqn (4):

Zp

0

½A1;2 n ðx1A1þ x2A2 .þ xnAnÞdp ð4Þ

where A1,2, nis the molecular area occupied by the mixed

monolayer, A1, A2 An are the area per molecule in the

pure monolayers of component 1, 2, .n, x1, x2 .xnare the

molar fractions of the components and p is the surface

pressure These data were then recorded as the variation of

DGMix with monolayer surface pressure (Table 3)

Numer-ical data were calculated from compression isotherms using

the methodology of Simpson [58]

The interaction parameter (a) relates the interaction of

each molar fraction of components within a monolayer

with the free energy of mixing Values of a were computed

(Table 3) according to Eqn (5):

RTðXn

1X2 .Xnþ X1Xn

2 .Xnþ X1X2 .Xn ð5Þ where X are the molar fractions of the components, R is

8.314 JÆmol)1ÆK)1, and T is 294 K These data were then

used to compute values of monolayer mixing enthalpy (DH)

(Table 3) according to Eqn (6):

DH¼RTa

where Z is the packing fraction parameter and calculated

using the Quikenden and Tam model [59]

Haemolytic assay of AP1

Haemolytic assay was conducted as described by Harris &

Phoenix [60] Essentially, packed red blood cells were

washed three times in Tris-buffered sucrose (0.25 m sucrose,

10 mm Tris⁄ HCl, pH 7.5) and resuspended in the same

medium to give an initial blood cell concentration of

 0.05% (w ⁄ v) For haemolytic assay, this concentration

was adjusted such that incubation with 0.1% (v⁄ v) Triton

X-100 for 30 min produced a supernatant with A416of 1.0,

which was taken as 100% haemolysis Aliquots (1 mL) of

blood cells at assay concentration were then used to

solubi-lize various amounts of stock AP1 solution, which had been

added to a test-tube and dried under nitrogen gas The

resulting mixtures were incubated at room temperature with

gentle shaking After 30 min, the suspensions were

centri-fuged at low speed (1500 g, 15 min, 25C), and the A416of

the supernatants determined In all cases, levels of

haemoly-sis were determined as the percentage haemolyhaemoly-sis relative to

that of Triton X-100 and the results recorded Background

haemolysis was less than 1% in all cases

Acknowledgements

We thank Jo¨rg Howe, Division of Biophysics,

Fors-chunginstitute, Borstel, Germany for his assistance

with FTIR analysis We would also like to thank Dr

Frank Grunfeld (NIMA Technology, UK) for his tech-nical advice with the monolayer studies

References

1 Byarugaba DK (2004) Antimicrobial resistance in devel-oping countries and responsible risk factors Int J Anti-microb Agents 24, 105–110

2 Stu¨renburg E & Mack D (2003) Extended-spectrum b-lactamases: implications for the clinical microbiology laboratory, therapy, and infection control J Infection

47, 273–295

3 French GL (2005) Clinical impact and relevance of anti-biotic resistance Adv Drug Deliv Rev 57, 1514–1527

4 Thomson CJ, Power E, Ruebsamen-Waigmann H & Labischinski H (2004) Antibacterial research and devel-opment in the 21st Century: an industry perspective of the challenges Curr Opin Microbiol 7, 445–450

5 Phoenix DA & Harris F (2003) Phenothiazinium-based photosensitizers: antibacterials of the future? Trends Mol Med 9, 283–285

6 Rachakonda S & Cartee L (2004) Challenges in antimi-crobial drug discovery and the potential of nucleoside antibiotics Curr Med Chem 11, 775–793

7 Papagianni M (2003) Ribosomally synthesized peptides with antimicrobial properties: biosynthesis, structure, function, and applications Biotechnol Adv 21, 465–499

8 Toke O (2005) Antimicrobial peptides: new candidates

in the fight against bacterial infections Biopolymers 80, 717–735

9 Sitaram N (2006) Antimicrobial peptides with unusual amino acid compositions and unusual structures Curr Med Chem 13, 679–696

10 Tossi A, Sandri L & Giangaspero A (2000) Amphi-pathic, alpha-helical antimicrobial peptides Biopolymers

55, 4–30

11 Diamond G (2001) Natures antibiotics: the potential of antimicrobial peptides as new drugs Biologist (London)

48, 209–212

12 Dennison SR, Harris F & Phoenix DA (2003) Factors determining the efficacy of alpha-helical antimicrobial peptides Protein Pept Lett 10, 497–502

13 Epand RM & Vogel HJ (1999) Diversity of antimicro-bial peptides and their mechanisms of action Biochim Biophys Acta 1462, 11–28

14 Blondelle SE, Lohner K & Aguilar M (1999) Lipid-induced conformation and lipid-binding properties of cytolytic and antimicrobial peptides: determination and biological specificity Biochim Biophys Acta 1462, 89–108

15 Devine DA & Hancock RE (2002) Cationic peptides: distribution and mechanisms of resistance Curr Pharm Des 8, 703–714

16 Hancock RE & Diamond G (2000) The role of cationic antimicrobial peptides in innate host defences Trends Microbiol 8, 402–410