The antibacterial activity of the amphipathic a-helical peptides varied with their size, 15 residues beingthe optimal length, independent of the membrane hydrophobic core thickness and t
Trang 1The antibiotic activity of cationic linear amphipathic peptides:
lessons from the action of leucine/lysine copolymers on bacteria
Laure Be´ven1, Sabine Castano2, Jean Dufourcq2, A˚ke Wieslander3and Henri Wro´blewski1
1
UMR CNRS 6026, Universite´ de Rennes 1, France;2Centre de Recherche Paul Pascal, CNRS, Pessac, France;
3
Department of Biochemistry and Biophysics, Stockholm University, Sweden
Peptides composed of leucyl and lysyl residues (LK
pep-tides) with different compositions and sequences were
compared for their antibacterial activities usingcell wall-less
bacteria of the class Mollicutes (acholeplasmas,
mycoplas-mas and spiroplasmycoplas-mas) as targets The antibacterial activity
of the amphipathic a-helical peptides varied with their size,
15 residues beingthe optimal length, independent of the
membrane hydrophobic core thickness and the amount of
cholesterol The 15-residue ideally amphipathic a helix with
a +5 positive net charge (KLLKLLLKLLLKLLK) had the
strongest antibacterial activity, similar to that of melittin In
contrast, scrambled peptides devoid of amphipathy and the
less hydrophobic b-sheeted peptides [(LK)nK], even those
15-residue long, were far less potent than the helical ones
Furthermore, the growth inhibitory activity of the peptides
was correlated with their ability to abolish membrane
potential These data are fully consistent with a
predominantly flat orientation of LK peptides at the lipid/ water interface and strongly supports that these peptides and probably the linear polycationic amphipathic defence pep-tides act on bacterial membranes in four main steps accordingto the carpet model: (a) interfacial partitioning with accumulation of monomers on the target membrane (limitingstep); (b) peptide structural changes (conformation, aggregation, and orientation) induced by interactions with the lipid bilayer (as already shown with liposomes and erythrocytes); (c) plasma membrane permeabilization/ depolarization via a detergent-like effect; and (d) rapid bacterial cell death if the extent of depolarization is main-tained above a critical threshold
Keywords: amphipathic peptides; antibacterial activity; bacterial cell death; membrane depolarization; mollicutes
The amphipathic a helix concept helps in understandingthe
behaviour of very different classes of proteins and peptides,
especially those actingon membranes [1–4] This concept,
which has been useful in the field of peptidic cytotoxins to
develop new analogues, and to better understand their
mode of action [5–7], is still the basis for the rational design
of new antimicrobial compounds, i.e analogues or chimeras
of natural products [8], or radically new molecules [9,10]
The need to understand their mode of action and improve
their efficacy and/or selectivity towards microorganisms led
to the synthesis of new active peptides, made possible largely
by the progress in solid state synthesis As a result, a large
wealth of information was obtained on many natural
peptides endowed with cytotoxic (includingantimicrobial)
activity However, answers are still missing regarding: (a) the
requirements for optimized activity and selectivity; and (b) the mechanisms of action, especially in bacterial cells In
an effort of rationalization, a minimalistic approach was initiated by the pioneeringwork of De Grado and Lear [11] usingresidue substitutions or designingsimplified sequences [2,12–14] Usingthe very minimal requirement of an amphipathic structure able to associate properly with either hydrophilic or hydrophobic side chains, several families of peptides were designed including those composed of only leucines and lysines (LK peptides) [11,14–17] Ideally secon-dary amphipathic structures (helices or b sheets) can thus be obtained by playingonly with the composition (i.e the L/K molar ratio) and charge periodicity [11,18–21] The lytic activities of these peptides vary accordingto the L/K ratio [22,23] and a specific sequence is not required to get an adequate polar/apolar topology and a strong membranolytic activity, matchingthose of the stronger natural toxins [20,24]
In homologous series of such LK peptides, the observed lytic activities on zwitterionic liposomes and erythrocytes are similar havingan optimum at a length of 15 residues, i.e there is a delicate balance between hydrophobicity and charge repulsion (two antagonistic forces) [14,22,25] Due to the lipid affinity of such peptides, there is an increase of lytic activity with chain length Moreover, in contrast with earlier studies [26], neither a threshold in length nor a matching between peptide length and membrane thickness was observed [21,24,25] This strongly supports a mechanism
Correspondence to H Wro´blewski, UMR CNRS 6026,
Universite´ de Rennes 1, Campus de Beaulieu,
35042 Rennes Cedex, France Fax: + 33 2 23 23 50 52,
E-mail: wroblews@univ-rennes1.fr
Abbreviations: CFU, colony forming unit; Dns, dansyl; MIC, minimal
inhibitory concentration; MDC, minimal deformingconcentration;
DpH, transmembrane pH gradient (DpH ¼ pH in ) pH out );
DY, membrane electrical potential (DY ¼ Y in –Y out ).
(Received 12 December 2002, revised 19 March 2003,
accepted 24 March 2003)
Trang 2of action based upon the invasion of the outer membrane
leaflet by the peptides and their insertion in a flat orientation
to form rafts or carpets [21,24,25,27,28]
In this work, we studied the antibacterial activities of a
homologous series of 8–22-residue LK peptides having
different sequences, hydrophobicities and secondary
struc-tures Our goal was to assess whether or not the rules drawn
from the studies with model membranes and erythrocytes
[21,25] are also valid for bacteria Mollicutes were chosen as
targets because these bacteria are devoid of both a cell wall
and an outer membrane [29] makingthem sensitive to many
natural membrane-active peptides Peptide activities were
monitored at different physiological levels: growth
inhibi-tion, plasma membrane depolarizainhibi-tion, and cell shape
modification [30] Furthermore, to get a better insight into
the mechanism of action, we compared the activities of a
series of peptides differingin length on Acholeplasma
laidlawiicells whose membrane thickness can be tuned via
the lipid diet [31,32]
Materials and methods
Chemicals
Ultrapure mellitin was from Sigma Dansylated (Dns)
peptides were synthesized by Fournier Pharma (Heidelberg,
Germany) and all the other peptides were from Neosystem
(Strasbourg, France) [24,25] The purity of the peptides was
97% as estimated by HPLC and their mass was consistent
with that expected from the sequence They were stored as dry powders at )20 C and dissolved just before use in methanol to give 1 or 10 mMstock solutions The peptides were named accordingto their length and K-residue periodicity (2.0, 3.0 or 3.6 indicatingwhether they were designed as amphipathic b strands, 3.10helices or a helices, respectively; Table 1) A scrambled 15-residue peptide [scr-LK15(W14)] was used as a nonamphipathic control peptide,
in which the K residues are distributed to achieve a hydrophobic moment close to zero [25] This peptide has free N and C termini, and leucine 14 is substituted by a tryptophan residue
HPLC Retention times of the peptides were measured by reversed-phase HPLC on a C18 Purospher RP-18 end-capped semipreparative column (125· 4 mm, 5-lm particle size) in conjunction with a Waters Millenium HPLC system, as described previously [25]
Antimicrobial assays The mollicutes Acholeplasma laidlawii A-EF22, Myco-plasma gallisepticum S6, M mycoides ssp mycoides SC KH3J, Spiroplasma citri R8A2, S floricola BNR1, and
S melliferum BC3 were cultured as described previously [30] The minimal inhibitory concentrations (MICs) were determined in 96-well microtitre plates by growing the
Table 1 Listing of the peptides used in this work Dansylation of peptides is indicated by Dns and D L, length of the peptide (residuesÆmol)1);
S, membrane-bound peptide conformation (note that although peptides LK8(3.6), LK9(3.6) and DnsLK9(3.6) were designed to form ideally amphipathic a helices, they adopt an extended conformation when interactingwith lipid bilayers; similarly, peptide LK16(W15)(3.0) which was designed to form an ideally amphipathic 3 10 helix is in fact a-helical when interactingwith spiroplasma lipids; C, Charge Lysine periodicity (2.0, 3.0
or 3.6) is indicated in parentheses to help peptide identification in the text.
Alpha-amphi series
3 10 -amphi
Beta-amphi series
Scrambled
Natural
Trang 3bacteria in the presence of twofold serial dilutions of
peptide The startingcell concentration in each well was 106
colony-formingunits (CFU)ÆmL)1 All assays were
per-formed in triplicate Bactericidal activities were assessed by
spreadingon agar plates cells treated for 2 h with different
peptide concentrations The minimal lethal concentration
was defined as the lowest peptide concentration capable of
killing99% of the cells in a suspension containing
106CFUÆmL)1 In addition, to determine the influence of
the hydrophobic core thickness of the cell membrane on
peptide antibiotic activity, A laidlawii A-EF22 was adapted
to grow in a lipid-free medium supplemented with
appro-priate fatty acids, as described previously [32,33]
Light microscopy
All of the experiments were performed on spiroplasma cell
suspensions containing1010CFUÆml)1 (A600¼ 1.0) in
50 mMsodium phosphate buffer pH 7.0, 50 mM D-glucose
and 549 mM D-sorbitol Dark-field optics were used to
analyse the effects of the peptides on spiroplasma motility
and cell morphology, as described previously [30,33–35]
Microphotographs were taken using Kodak T-Max 35-mm
film (ISO 400 or 3200)
Membrane potential measurement
Alterations of membrane potential by peptides in A
laid-lawiiand S melliferum were probed spectrofluorometrically
usingthe fluorescent dye
3,3¢-dipropyl-2,2¢-thiadicarbocya-nine iodide [36] A detailed description of the experimental
conditions and of the calibration method of fluorescence
signal vs membrane potential is given in previous reports
[33,34]
Determination of intracellular pH The intracellular pH of A laidlawii and S melliferum cells was determined by spectrofluorometry usingthe internally conjugated fluorescent probe 5(6-)-carboxyfluorescein suc-cinimidyl ester [37] as described previously [33]
Protein and cholesterol determination Protein was determined with the bicinchoninic acid method [38] usingBSA as standard Peptide concentrations were estimated from absorbance measurements using
e340¼ 4640M )1Æcm)1 for dansylated peptides and
e280¼ 5600M )1Æcm)1for tryptophan-containingpeptides Total cholesterol was determined in the chloroform/meth-anol membrane lipid fraction with the Sigma Diagnostics Cholesterol reagent kit
Results
Antibacterial activity of the peptides as a function
of length and structure The first step of our work was to screen the growth inhibition activities of 15 LK peptides towards six different species of mollicutes Melittin was used as a reference because it is a well known cytotoxic peptide which is lethal for mollicutes [30,33] The diversity of the LK peptides used
in this study (Table 1) allowed us to assess the importance
of both peptide length and structure for antibacterial activity
Table 2 shows that the susceptibility of mollicutes to melittin and the LK peptides was independent of the amount of cholesterol in the membranes of these bacteria
Table 2 Antibacterial activities of LK peptides and melittin against several species of mollicutes.
Cholesterol (%) a
MICs (l M )
Alb (2.1)
Mg (7.2)
Mm (3.5)
Sc (25.2)
Sf (16.8)
Sm (22.2) Peptides
a
The percentage (by mass) of cholesterol in the whole membrane lipid fraction of the different bacteria (% chol.) is indicated below the abbreviation of their name b Bacterial targets: A laidlawii (Al), M gallisepticum (Mg), M mycoides ssp mycoides SC (Mm), S citri (Sc),
S floricola (Sf), and S melliferum (Sm) R, No activity at concentrations £ 100 l
Trang 4The most potent of the LK peptides was, overall,
LK15(3.6), i.e the peptide designed to adopt an ideal
amphipathic a-helical conformation (KLLKLLLKLLLK
LLK) With MICs ranging from 1.56 to 6.25 lM, this
molecule exhibited a growth inhibition activity similar to
that of melittin (MIC, 0.39–12.5 lM) In this series, the
12-residue peptide was much less potent but, interestingly,
dansylation increased its activity whilst the same
modifi-cation decreased the activity of the 15-residue peptide
Shorter peptides (8 or 9 residues) displayinga b
confor-mation in the membrane-bound state [24] were poorly
effective or harmless, even at concentrations up to 100 lM
Longer peptides [DnsLK19(3.6) and DnsLK22(3.6)] were
also less active, but the relationship between length and
activity was less clear than in the case of the shorter
peptides In addition, the L14W substitution in LK15(3.6)
had no effect on activity Collectively, these observations
indicate that for ideally amphipathic a helices composed of
only leucine and lysine, a sequence of 15 residues
constitutes the optimal length to inhibit the growth of
mollicutes
Table 2 also shows that with MICs ranging from 0.78 to
12.5 lM, LK16(W15)(3.0), the peptide designed to form
ideally amphipathic 3.10 helices but which, in fact, is
a-helical in membranes (data not shown), was as effective
as LK15(3.6) (L/K¼ 2) In comparison, peptides of the
beta-amphi series designed to form ideally amphipathic b
strands, were significantly less effective The MICs of
Dns-LK15(2.0) (L/K¼ 1) were similar to those of DnsLK9(3.6)
So, here again a decrease of the length in the same series
below 15 residues resulted in a loss of activity Finally,
scramblingthe sequence of LK15(W14)(3.6) [peptide
scr-LK15(W14)] strongly decreased its activity, notably
against M mycoides ssp mycoides SC and the three
spiroplasmas This second set of data pinpoints the
import-ance of peptide charge topology, showing this time that
provided they have the optimum length (i.e 15 residues),
ideally amphipathic helices are significantly more potent in
inhibiting the growth of mollicutes than irregular sequences
with the same composition and than ideally amphipathic
b sheets
It should also be noted that platingpeptide-treated
mollicutes on agar broth revealed that the most active
peptides (MICs < 50 lM) were bactericidal as no surviving
cells could be detected by this method after a 2 h peptide
treatment
Spiroplasma cell deformation by the peptides
In contrast with other mollicutes (e.g acholeplasmas and
mycoplasmas), spiroplasmas exhibit a helical shape and
motility which are altered by membranotropic peptides, a
phenomenon easy to observe by dark-field light microscopy
[30,33–35] In the second step of this work, we have thus
assessed, using S melliferum as a target, the cell deforming
activity of LK15(3.6) compared with melittin Upon
treatment with the latter, the cells lost their motility and
helical shape Deformation of 100% of the cells could be
achieved within seconds with 1 lM melittin, which
ham-pered a reliable microphotographic recording However, as
the effects were both concentration- and time-dependent, it
was possible to find conditions compatible with the
technique Hence, 100% of the cells (1010CFUÆmL)1) were deformed with 0.1 lM melittin within 5 min whilst only 50% were deformed within the same time with 0.01 lM melittin At this latter concentration, a limit of 55–60% was reached after 10 min (Fig 1) In the same conditions, LK15(W14)(3.6) was less efficient than melittin at the lowest concentration (20% cells deformed after 10 min) but equally active at the highest one
As the spiroplasma cell deformation test proved to be relevant and reliable to investigate the membranotropic activity of antibacterial peptides, this technique was used to assess the importance of the structure of the LK peptides Table 3 [columns MDC (minimal deformingconcentra-tion)50and MDC100] reveals that, consistent with the results
of growth inhibition tests, the activity of melittin was matched by LK15(3.6), LK15(W14)(3.6), and to a lesser extent by LK16(W15)(3.0) In contrast, scr-LK15(W14) exhibited a deformingactivity about two orders of magni-tude weaker and the Dns(LK)n-series peptides were harm-less independently of their length
Hence, the secondary structure and amphiphilicity of model peptides composed of L and K residues were also critical parameters in the spiroplasma cell deformation test
Fig 1 Time-course of S melliferum cell deformation by melittin and LK15(3.6) Spiroplasma cells (1010CFUÆmL)1) energ ized with 50 m M
D -glucose in 50 m M sodium phosphate buffer (pH 7.0, 32 C) con-taining549 m M D -sorbitol as osmoprotectant were treated with different peptide concentrations and pictures were recorded with a light microscope equipped with dark-field optics The cells were considered deformed as soon as they lost their helicity Points on the curves are the average of three determinations (SD £ 4.5%).
Trang 5The 15-residue ideally amphipathic helix LK15(3.6) proved
again to be the optimal structure in contrast with b-sheeted
structures or a scrambled sequence
Membrane depolarization by the peptides
We have previously shown that there is a correlation
between the ability of peptides to inhibit the growth of
mollicutes, to abolish the membrane potential, and to
deform spiroplasma cells [30,33–35] The next step of this
work was to investigate the depolarizing activity of the LK
peptides Fig 2 shows that the relative efficacy of LK15(3.6)
compared with melittin was the same in A laidlawii and
S melliferum with respect to the extent of membrane
depolarization and the delay necessary to reach a
steady-state level of depolarization Actually, 0.1 lM melittin
totally depolarized the A laidlawii membrane within
5 min and that of S melliferum by 41% Under the same
conditions (109CFUÆmL)1), 0.1 lMLK15(3.6) depolarized
the A laidlawii membrane by 64% and that of S melliferum
by 16%
S melliferum beingmore robust than A laidlawii was
then used to compare the depolarizingactivities of 15- and
16-residue LK peptides havingdifferent structures Table 3
(DDY columns) shows that the depolarization of the
S melliferum membrane mirrored the spiroplasma cell
deformation phenomenon described above Indeed, here
again melittin, LK15(3.6), LK15(W14)(3.6), and LK16
(W15)(3.0) proved to be more effective than scr-LK15(W14)
whilst the three Dns(LK)n peptides produced a negligible
effect at best
Abolition of the transmembrane pH gradient
by the peptides
After havinganalysed the effects of melittin and LK15(3.6)
on the membrane potential of A laidlawii and S
melli-ferum, we have investigated the ability of these peptides to
alter (RT/F)· DpH, i.e the second component of the
protonmotive force The transmembrane pH gradient
(DpH¼ pHin) pHout) was thus measured at 37C for
A laidlawiiand 32C for S melliferum vs extracellular pH (pHout) (Fig 3) When energized with 50 mM D-glucose,
A laidlawiiand S melliferum cells (109CFUÆmL)1) gener-ated in slightly buffered solutions a D
to be stable for at least 20 min This transmembrane gradient increased linearly from 0.23 to 1.31 in A laidlawii (Fig 3A) and from 0 to 1.34 in S melliferum (Fig 3B) when pHoutdecreased from 7.5 to 5.0 Immediately after the addition of 0.1 lMmelittin or LK15(3.6), the DpH increased transiently for about 2 min and then dropped within 2 min
to reach a steady value, always lower than that observed in the absence of peptide As expected from DY measurements (see above), 0.1 lMLK15(3.6) diminished the DpH in both mollicutes almost as efficiently as melittin The linear relationship of DpH vs pHoutindicates that the activity of both peptides was independent of the extracellular pH within the explored range (5.0–7.5) which largely covers the conditions found by bacteria either in their animal hosts or
in culture media
Table 3 Effect of melittin and LK peptides on Spiroplasma melliferum
cell shape and membrane potential MDC 50 and MDC 100 are the
min-imal concentrations (l M ) required for the deformation of 50% and
100% of the cells, respectively, upon 5 min of action No stands for no
observed effect DDY is the percentage of membrane depolarization
induced by 0.1 and 1 l M peptide concentrations (unperturbed
poten-tial: 68 ± 5 mV, inside negative; SD, 4%) upon 10 min of action.
Peptides MDC 50 MDC 100
DDY 0.1 l M 1 l M
LK15(W14)(3.6) 0.05 0.1 17 65
LK16(W15)(3.6) 0.1 0.2 16 65
Fig 2 Time-course of A laidlawii and S melliferum plasma mem-branes depolarization by melittin and LK15(3.6) The cells (109 CFUÆmL)1) were energized with 50 m M D -glucose in 5 m M Hepes buffer (pH 7.0) containingeither 150 m M NaCl (A laidlawii) or
128 m M NaCl (S melliferum) Measurements were performed at 37 C for A laidlawii and 32 C for S melliferum The arrows indicate the time at which the peptides were injected into the cell suspensions The curves are the means of three determinations (SD £ 4%) A laidlawii (- -) and S melliferum (—).
Trang 6Influence of membrane thickness on the antibacterial
activity of the peptides
The possibility of modifyingthe A laidlawii membrane
lipid bilayer via the fatty acids incorporated into the
growth medium [29,31,32] was exploited to study the
influence of membrane thickness on the antibacterial
activity of LiKj a-helical amphipathic peptides (i¼ 2j)
A laidlawii was thus grown under conditions allowing
23, 25, 26.3 or 28-A˚ membrane hydrophobic core
thick-ness to be obtained (Table 4) As previously shown [33],
the MIC of honey bee melittin increased from 0.78 to
3.12 lM when increasingthe hydrophobic thickness from
23 to 28 A˚ However, even with the thickest cell
membrane, the activity of melittin was still very high
With activities similar to those of mellitin, LK15(3.6) and
Dns-LK15(3.6) were the most efficient of the LK
peptides for the four types of membranes, i.e
independ-ent of the membrane thickness Their inhibitory activities
decayed as peptide length was either decreased or
increased but, similar to the results obtained in the
growth inhibition experiments (Table 2), the loss of
activity due to peptide lengthening was less sharp than
that due to peptide shortening It should also be stressed
that the 25-A˚ hydrophobic core membrane which
contained exclusively 16 : 1c fatty acyl chains and no
cholesterol was overall more sensitive to the peptides
than the three other membranes
Fig 3 Effect of melittin and LK15(3.6) on the transmembrane pH gradient of A laidlawii (A) and S melliferum (B) DpH ¼ pH in ) pH out Measurements were performed at 37 C for A laidlawii and 32 C for S melliferum Each point on the curves is the mean of three independent determinations (SD £ 3%).
Table 4 Influence of Acholeplasma laidlawii membrane thickness on the antibacterial activity of melittin and LK peptides of the alpha-amphi series Data are expressed as l M MIC R, No growth inhibition for concentrations £ 100 l M A laidlawii (strain A-EF22) was grown in lipid-free medium supplemented with: (a) 75 l M tetradecanoic acid (14 : 0) + 75 l M Dcis-9-tetradecanoic acid (14 : 1c); (b) 150 l M Dcis-9-hexadecanoic acid (16 : 1c); (c) 150 l M Dcis-9-octadecanoic acid (18 : 1c); and (d) 150 l M Dcis-9-octadecanoic acid (18 : 1c) + 20 l M
cholesterol These conditions give the following compositions in membrane lipids: (a) 14 : 0 + 14 : 1, molar ratio, 50/50; (b) 100%
16 : 1c; (c) 100% 18 : 1c; and (d) 18 : 1c + cholesterol, molar ratio, 75/25 The average thickness of the A laidlawii A-EF22 membrane hydrophobic core, in these conditions, was previously determined to be
23 A˚, 25 A˚, 26.3 A˚ and 28 A˚ [32].
Peptide
Membrane thickness
23 A˚ 25 A˚ 26.3 A˚ 28 A˚
DnsLK9(3.6) 3.12 0.78 12.5 50
LK15(3.6) 0.39 0.78 1.56 0.78 DnsLK15(3.6) 0.78 0.78 3.12 1.56
DnsLK22(3.6) 6.25 6.25 12.5 12.5 Melittin 0.78 0.78 1.56 3.12
Trang 7Previous studies based on the action of minimalist LK
peptides on model membranes and erythrocytes showed
that the critical parameters governing activity are peptide
length, total hydrophobicity, amphipathy, and secondary
structure, which collectively determine peptide membrane
affinity [21,24,25] However, due to the complexity of
bacterial cells compared to liposomes and erythrocytes, it
was necessary to check whether these rules hold also for the
antimicrobial activity of these peptides In this work, we
have taken advantage of the fact that bacteria of the class
Mollicutesare devoid of an outer membrane and of a cell
wall, to avoid possible interferences of these structures in the
interactions between peptides and the bacterial plasma
membrane
Most of the LK peptides studied here exhibited an
antibacterial activity in agreement with previous studies on
closely related compounds [15,23] The activity varied with
peptide length, the optimum occurring for the 15-residue
ideally amphipathic a-helical structure [LK15(3.6)] the
activity of which was similar to that of melittin, a bee
venom peptide known as one of the most efficient natural
peptides in killingbacteria [8,39–41]
Lengthening peptides of the alpha-amphi series (generic
composition LiKjwith i¼ 2j) over 15 residues did not result
in a parallel increase in activity This was previously found
for their efficacy to induce leakage in lipid vesicles and
erythrocytes, and was shown to be due to the
self-association of such a helices in solution [25] However,
unlike haemolysis the bactericidal activity dropped more
severely from 18- to 21-residue length before a slight
increase for the 22-residue peptide (Fig 4) Thus, two
competingprocesses probably occur in bacteria and
eryth-rocytes as in the case of artificial membranes [25,42,43]: (a) a
progressive increase of the membrane affinity with peptide
length; and (b) a drop of activity due to decreased free
energy in solution upon oligomer formation The opposite
effects produced by dansylatingshort or longpeptides are
consistent with this interpretation Indeed, for short
pep-tides the dansyl group promotes activity by increasing
hydrophobicity, and thus membrane affinity, whilst for
peptides longer than 15 residues, already too hydrophobic
to remain in the monomeric state in buffer, dansylation
favours oligomerization at the expense of lipid affinity The
total lack of activity of LK18(3.6) and LK21(3.6) compared
to LK22(3.6) does not contradict this interpretation because
these peptides have no K at the N terminus but, instead,
several L residues increasinghydrophobicity and thus
oligomerization tendency
The beta-amphi series peptides [i.e (LK)nK peptides] are
much less hydrophobic and have a larger charge repulsion
They are thus monomeric in water and have a lower affinity
for lipids than the alpha-amphi series peptides (i.e LiKj
peptides with i¼ 2j) of same length [21] This explains why
these peptides are less active against bacteria (as observed in
this work) and also less haemolytic [21,25]
In addition to total charge and hydrophobicity, the
topology of the distribution of L and K residues proved to
be important: the more the peptides were amphipathic, the
more they were active against mollicutes This is clearly
shown for L10K5 peptides by comparingthe MICs of the
ideally amphipathic peptide L10K5(3.6) and of the scrambled one [scr-LK15(W14)], although LK16W15(3.0) which folds into a nonamphipathic a helix, keeps also quite
a high activity These secondary amphipathic peptides proved to be more active than previously studied primary amphipathic ones in which the positive charges were clustered at the C termini of the molecules (see e.g [33]) Hence, contrary to haemolysis, the antibacterial activity seems to be more sensitive to peptide aggregation and less sensitive to amphipathy
LK peptides have different secondary structures when bound to lipids: the LiKjpeptides (with i¼ 2j and n > 12) are a-helical [24] whilst the shorter and/or alternated ones [(KL)nK] fold into antiparallel b sheets [25] It is also noteworthy that the scrambled peptide scr-LK15(W14) is b-sheeted when bound to dimyristoyl phosphatidyl choline [24] but a-helical in the presence of spiroplasma lipids [44] This might be due to the fact that dimyristoyl phospha-tidylcholine is zwitterionic whilst the spiroplasma mem-brane contains anionic lipids The beta-amphi series peptides were less active against mollicutes than their more hydrophobic ideally amphipathic a-helical homologues However, both a-helical and b-sheeted peptides acted on
Fig 4 Comparison of the effects of alpha-amphi series peptide length on antibacterial and haemolytic activities Antibacterial activities (black curve) are expressed as MIC)1 The data were normalized in such a way that the highest activity, corresponding to a MIC of 0.78 l M , was given a value of 100 Haemolytic activities (grey curve) were taken from [25] and expressed as the inverse of the concentration inducing 50% of lysis (LC150).
Trang 8different species of mollicutes with the same rankingwithin
their respective series: A laidlawii> M
gallisepti-cum> S citri S floricola S melliferum M
myco-mycoides ssp mycoides SC, while for melittin S citri and
S floricolawere much more sensitive Such a rankingseems
therefore more relevant to peculiarities of these bacteria
whose lipid composition varies accordingto the species,
than to properties of the peptides Despite the fact they are
less efficient than their a-helical homologues, peptides of the
beta-amphi series are also intrinsically capable of killing
mollicutes This suggests that within this series, longer
peptides should prove more efficient in growth inhibition
tests than those used in this work
Measurements of DY and DpH in A laidlawii and
S melliferum revealed that the antibacterial activities of
the peptides were correlated with their ability to depolarize
the plasma membrane (Tables 2 and 3, and Figs 2 and 3) In
the case of S melliferum, the loss of cell motility and helicity
induced by the action of LK peptides was also correlated
with membrane depolarization as previously observed with
several natural antibacterial peptides [30] However, MICs
were about one order of magnitude higher than depolarizing
concentrations This difference is probably due to
lipopro-teins present in the culture medium used for growth
inhibition assays As serum components compete with
membranes for the bindingof membrane-active peptides
[45–47], they should indeed increase their apparent MICs
Our data indicate that the bactericidal activity of LK
peptides towards mollicutes is due to their ability to
permeabilize the plasma membrane; this raises the question
of the molecular mechanism governing their action In the
case of hydrophobic peptides such as alamethicin,
experi-mental data indicate that permeabilization occurs through
the formation of ion-conductingtransmembrane channels
in accordance with the barrel-stave model [35,48]
How-ever, such a mechanism hardly fits LK peptides, even the
a-helical ones, because of their polycationic nature and
charge periodicity (+1 per a-helix turn) Indeed, the
transfer of five or more positive charges per molecule from
water into the hydrophobic core of the lipid bilayer is
energetically extremely unfavourable unless they are
pro-perly counterbalanced by a set of negative charges in register
with them Hence, if transient transmembrane bundles of
LK peptide helices were to exist, such bundles would be very
unstable because of K+/K+electrostatic repulsions [49,50]
It should also be stressed that the a helix LK15(3.6) is
anyway too short to span the membrane bilayer
hydropho-bic core, even in the case of the thinnest A laidlawii
membrane (see Table 4) A 15-residue helix would be
22.5 A˚ long, i.e very close to the thickness of the membrane
hydrophobic core (23 A˚), but polarity of the N and C
termini should hamper their localization within an apolar
environment In contrast, with a length of 50 A˚, the
b strand DnsLK15(2.0) is too longfor a correct
transmem-brane fit In fact, PM-IRRAS spectra show unambiguously
that both a and b ideally amphipathic LK peptides are
layingflat on the interface between water and lipids
includingthose of S melliferum [44] In the same conditions,
scr-LK15(W14) and LK16W15(3.0) exhibited a mainly
a-helical folding, without amphipathy, and a slightly tilted
orientation with respect to the lipid/water interface plane
[44] Such a flat orientation should thus be considered the
most stable one for LK peptides, even if other orientations are possible (see below) As suggested by the carpet model, aggregation on the membrane surface should enhance peptide dynamic reorientations and the subsequent forma-tion of transient transmembrane pores [51]
Amongthe different mechanisms proposed to explain peptide action on membranes, the interfacial models such as
rafts or carpets [28,52,53] thus seem to be more relevant to polycationic amphipathic molecules than the barrel-stave model This view is strongly supported by the data of Table 4 and Fig 5 showing that the activity of the LK peptides is essentially independent of membrane thickness Indeed, the helical 15-residue peptides are the most active ones independent of the mollicute species and, for the same peptide, independent of the membrane thickness, whilst the formation of transmembrane bundles of helices would require longer molecules for a better match between membrane thickness and peptide length Hence, the anti-bacterial action of these peptides comprises four main steps: step 1, interfacial partitioningand exofacial accumulation of
Fig 5 Graphical illustration of the effects on antibacterial activity of the length of alpha-amphi series peptides vs A laidlawii membrane thick-ness Antibacterial activities are expressed as MIC)1 The data were normalized in such a way that the highest activity, corresponding to a MIC of 0.78 l , was given a value of 100.
Trang 9monomers on the target membrane (limiting step); step 2,
peptide structural changes (conformation, aggregation, and
orientation) induced by interactions with the lipid bilayer, as
indicated by previous studies with liposomes and
erythro-cytes [17,25]; step 3, plasma membrane permeabilization/
depolarization via detergent-like effects; step 4, rapid
bacterial cell death if the extent of depolarization is
maintained above a critical threshold At step 3, bound
peptides can modifiy the membrane curvature stress above a
certain peptide/lipid ratio which should also contribute to
their toxicity (see for example [54]) Steps 3 and 4
(permeabilization and physiological consequences,
respect-ively) are identical to those induced by ion-channel forming
peptides such as alamethicin although the molecular
mechanisms of permeabilization (step 3) differ Whilst for
alamethicin and analogues there is strong evidence that
membrane permeabilization is due to the formation of
dynamic barrel-staves [35,48], carpet-formingpeptides
behave rather like detergents disrupting the lipid bilayer
when a threshold concentration of peptide monomer is
reached; at this stage, transient transmembrane pores might
be formed [53] Some cationic peptides such as magainin are
also capable of formingtransient toroidal pores composed
of dynamic, peptid/lipid supramolecular complexes [55]
Whatever the permeabilization mechanism, a sudden
membrane depolarization should prevent bacteria from
settingup appropriate countermeasures and lead to a rapid
cell death if the peptide concentration is maintained above a
critical threshold In the case of bacteria like mollicutes
which are devoid of a cell wall (murein), cell death can occur
still faster upon the action of the most efficient peptides
since in this case massive entry of water into the cytoplasm
can lead to cell burst
Mechanisms other than membrane permeabilization have
also been proposed to explain the antimicrobial activity of
some cationic peptides In some cases, the inhibition of
intracellular proteins might indeed be responsible for
bacterial cell death (see for example [56,57] for brief
discussions) We believe that these different views are not
contradictory but rather reflect, even for a same peptide,
differences in the context of its action, the variables (notably
the properties of the target cell) being probably too many to
allow for a single and general mechanism of action
Acknowledgements
We are pleased to thank Dr K Bu¨tner, now at Neosystem, for kindly
providingpeptides and W Ne´ri for peptide purification.
This work was supported by the GDR CNRS 790 (Peptides et
Prote´ines Membranotropes) and the Ministe`re de lEnseignement
Supe´rieur et de la Recherche’ (ACC Physico-Chimie des Membranes
Biologiques and the Programme de Recherche Fondamentale en
Microbiologie et Maladies Infectieuses et Parasitaires).
References
1 Segrest, J.P (1977) Amphipathic helixes and plasma lipoproteins:
thermodynamic and geometric considerations Chem Phys Lipids
18, 7–22.
2 De Grado, W.F., Ke´zdy, F.J & Kaiser, E.T (1981) Design,
synthesis and characterization of a cytotoxic peptide with
melittin-like activity J Am Chem Soc 103, 679–681.
3 Eisenberg, D (1984) Three-dimensional structure of membrane and surface proteins Annu Rev Biochem 53, 595–623.
4 Segrest, J.P., de Loof, H., Dohlman, J.G., Brouillette, C.G & Anantharamaiah, G.M (1990) Amphipathic helix motif: classes and properties Proteins 8, 103–117.
5 De Grado, W.F., Musso, G.F., Lieber, M., Kaiser, E.T & Ke´zdy, F.J (1982) Kinetics and mechanism of hemolysis induced by melittin and by a synthetic melittin analogue Biophys J 37, 329–338.
6 Cornut, I., Thiaudie`re, E & Dufourcq, J (1993) The amphipathic helix in cytotoxic peptides In The Amphipathic Helix (Epand R.M., ed.), pp 173–219 CRC Press, Boca Raton.
7 Saberwal, G & Nagaraj, R (1994) Cell-lytic and antibacterial peptides that act by perturbingthe barrier function of membranes: facets of their conformational features, structure-function corre-lations and membrane-perturbingabilities Biochim Biophys Acta
1197, 109–131.
8 Merrifield, R.B., Boman, H.G., Andreu, D., Li, Z.Q., Fink, J & Wade, D (1990) The cecropins: an exemple of the use of peptide synthesis to study a biochemical problem In Peptides (Giralt, E & Andreu, D., eds), p.3 ESCOM, Leiden.
9 De Grado, W.F (1999) De novo design of antibacterial beta-peptides J Am Chem Soc 121, 122000–122201.
10 Porter, E.A., Wang, X., Lee, H.-S., Weisblum, B & Gellman, S.H (2000) Non-haemolytic beta-amino-acid oligomers Nature 404, 565.
11 De Grado, W.F & Lear, J.D (1985) Induction of peptide conformation at a polar/water interfaces A study with model peptides of defined hydrophobic periodicity J Am Chem Soc.
107, 7684–7689.
12 Lee, S., Mihara, H., Aoyagi, H., Kato, T., Izumiya, N & Yamasaki, N (1986) Relationship between antimicrobial activity and amphiphilic property of basic model peptides Biochim Bio-phys Acta 862, 211–219.
13 Alouf, J.E., Dufourcq, J., Siffert, O., Thiaudie`re, E & Geoffroy,
C (1989) Interaction of staphylococcal delta-toxin and synthetic analogues with erythrocytes and phospholipid vesicles Biological and physical properties of the amphipathic peptides Eur J Bio-chem 183, 381–390.
14 Dathe, M., Schumann, M., Wieprecht, T., Winkler, A., Beyer-mann, M., Krause, E., Matsuzaki, K., Murase, O & Bienert,
M (1996) Peptide helicity and membrane surface charge mod-ulate the balance of electrostatic and hydrophobic interactions with lipid bilayers and biological membranes Biochemistry 35, 12612–12622.
15 Blondelle, S.E & Houghten, R.A (1992) Design of model amphipathic peptides havingpotent antimicrobial activities Biochemistry 31, 12688–12694.
16 Reynaud, J.A., Grivet, J.P., Sy, D & Trudelle, Y (1993) Inter-actions of basic amphiphilic peptides with dimyristoylphos-phatidylcholine small unilamellar vesicles: optical, NMR, and electron microscopy studies and conformational calculations Biochemistry 32, 4997–5008.
17 Cornut, I., Bu¨ttner, K., Dasseux, J.L & Dufourcq, J (1994) The amphipathic a-helix concept Application to the de novo design of ideally amphipathic Leu, Lys peptides with hemolytic activity higher than that of melittin FEBS Lett 349, 29–33.
18 Brack, A & Spach, G (1981) Multiconformational synthetic polypeptides J Am Chem Soc 103, 6319–6323.
19 Blondelle, S.E., Ostrech, J.M., Houghten, R.A & Pe´rez-Paya, E (1995) Induced conformational states of amphipathic peptides in aqueous/lipid environments Biophys J 68, 351–359.
20 Cornut, I., Bu¨ttner, K., Be´ven, L., Duclohier, H & Dufourcq, J (1994) Minimal requirements for the design of highly cytotoxic alpha-helical amphipathic peptides In Peptides 1994 – Proceedings
Trang 10of the 23 European Peptide Symposium (Maia, H.L.S., ed.), pp.
666–667 ESCOM, Leiden.
21 Castano, S., Desbat, B & Dufourcq, J (2000) Ideally amphipathic
b-sheeted peptides at interfaces: structure, orientation, affinities for
lipids and hemolytic activity of (KL) m K peptides Biochim
Bio-phys Acta 1463, 65–80.
22 Kiyota, T., Lee, S & Sugihara, G.V (1996) Design and
synthesis of amphiphilic alpha-helical model peptides with
systematically varied hydrophobic-hydrophilic balance and their
interaction with lipid- and bio-membranes Biochemistry 35,
13196–13204.
23 Dathe, M., Wieprecht, T., Nikolenko, H., Handel, L., Maloy,
M.L., McDonald, D.L., Beyermann, M & Bienert, M (1997)
Hydrophobicity, hydrophobic moment and angle subtended by
charged residues modulate antibacterial and haemolytic activity of
amphipathic helical peptides FEBS Lett 403, 208–212.
24 Castano, S., Desbat, B., Laguerre, M & Dufourcq, J (1999)
Structure, orientation and affinity for interfaces and lipids of
ideally amphipathic lytic L i K j (i ¼ 2j) peptides Biochim Biophys.
Acta 1416, 176–194.
25 Castano, S., Cornut, I., Bu¨ttner, K., Dasseux, J.L & Dufourcq, J.
(1999) The amphipathic helix concept: length effects on ideally
L i K j (i ¼ 2j) peptides to acquire optimal hemolytic activity
Bio-chim Biophys Acta 1416, 161–175.
26 McLean, L.R., Haganan, K.A., Owen, T.J & Krstenansky, J.L.
(1991) Minimal peptide length for interaction of amphipathic
alpha-helical peptides with phosphatidylcholine liposomes
Bio-chemistry 30, 31–37.
27 Bechinger, B (1997) Structure and functions of channel-forming
peptides: magainins, cecropins, melittin and alamethicin.
J Membr Biol 156, 197–211.
28 Shai, Y (1999) Mechanism of binding, insertion and
destabiliza-tion of phospholipid bilayer membranes by alpha-helical
anti-microbial and cell non-selective membrane-lytic peptides Biochim.
Biophys Acta 1462, 55–70.
29 Razin, S (1992) Peculiar properties of mycoplasmas: the smallest
self-replicatingprocaryotes FEMS Microbiol Lett 100, 423–
432.36.
30 Be´ven, L & Wro´blewski, H (1997) Effect of natural amphipathic
peptides on viability, membrane potential, cell shape, and motility
of mollicutes Res Microbiol 148, 163–175.
31 Silvius, J.R & McElhaney, R.N (1978) Lipid compositional
manipulation in Acholeplasma laidlawii B Effect of exogenous
fatty acids on fatty acid composition and cell growth when
endogenous fatty acid production is inhibited Can J Biochem.
56, 462–469.
32 Thurmond, R.L., Niemi, A.R., Lindblom, G., Wieslander, A˚ &
Rilfors, L (1994) Membrane thickness and molecular orderingin
Acholeplasma laidlawii strain A studied by 2H NMR
spectro-scopy Biochemistry 33, 13178–13188.
33 Be´ven, L., Chaloin, L., Vidal, P., Heitz, F & Wro´blewski, H.
(1997) Effects on mollicutes (wall-less bacteria) of synthetic
pep-tides comprisinga signal peptide or a membrane fusion peptide,
and a nuclear localization sequence (NLS): a comparison with
melittin Biochim Biophys Acta 1329, 357–369.
34 Be´ven, L., Duval, D., Rebuffat, S., Riddell, F.G., Bodo, B &
Wro´blewski, H (1998) Membrane permeabilisation and
anti-mycoplasmic activity of the 18-residue peptaibols, trichorzins PA.
Biochim Biophys Acta 1372, 78–90.
35 Be´ven, L., Helluin, O., Molle, G., Duclohier, H & Wro´blewski, H.
(1999) Correlation between anti-bacterial activity and pore sizes
of two classes of voltage-dependent channel-forming peptides.
Biochim Biophys Acta 1421, 53–63.
36 Waggoner, A.S (1979) Dye indicators of membrane potential.
Ann Rev Biophys Bioeng 8, 47–68.
37 Chitarra, L.G., Breeuwer, P., Van den Bulk, R.W & Abee, T (2000) Rapid fluorescence assessment of intracellular pH as a viability indicator of Clavibacter michiganensis subsp Michiga-nensis J Appl Microbiol 88, 809–816.
38 Smith, P.K., Krohn, R.I., Hermanson, G.T., Mallia, A.K., Gartner, F.H., Provenzano, M.D., Fujimuto, E.K., Goeke, N.M.
& Klenk, D.C (1985) Measurement of protein usingbicinchoninic acid Anal Biochem 150, 76–85.
39 Wade, D., Andreu, D., Mitchell, S.A., Silveira, A.M., Boman, A., Boman, H.G & Merrifield, R.B (1992) Antibacterial peptides designed as analogs or hybrids of cecropins and melittin Int J Pept Protein Res 40, 429–436.
40 Juvvadi, P., Vunnam, S., Merrifield, E.L., Boman, H.G & Mer-rifield, R.B (1996) Hydrophobic effects on antibacterial and channel-formingproperties of cecropin A-melittin hybrids J Pept Sci 2, 223–232.
41 Dathe, M & Wieprecht, T (1999) Structural features of helical antimicrobial peptides: their potential to modulate activity on model membranes and biological cells Biochim Biophys Acta
1462, 71–87.
42 Oren, Z & Shai, Y (2000) Cyclization of a cytolytic amphipathic alpha-helical peptide and its diastereomer: effect on structure, interaction with model membranes, and biological function Biochemistry 39, 6103–6114.
43 Feder, R., Dagan, A & Mor, A (2000) Structure-activity relationship study of antimicrobial dermaseptin S4 showingthe consequences of peptide oligomerization on selective cytotoxicity.
J Biol Chem 275, 4230–4238.
44 Castano, S., Desbat, B., Wro´blewski, H & Dufourcq, J (2002) Hemolytic and antibacterial activities of LK peptides of various topologies: a monolayer and PM-IRRAS approach In Mem-branes and Proteins (Templer, R.H & Leatherbarrow, R., eds), pp 191–198 Proceedings of the 1st Annual Conference of Biophysical Chemistry, London, 2001.
45 Wang, Y., Agerberth, B., Lothgren, A., Almstedt, A & Johansson, J (1998) Apolipoprotein A-I binds and inhibits the human antibacterial/cytotoxic peptide LL-37 J Biol Chem 293, 33115–33118.
46 Sorensen, O., Bratt, T., Johnsen, A.H., Madsen, M.T & Borre-gaard, N (1999) The human antibacterial cathelicidin, hCAP-18
is bound to lipoproteins in plasma J Biol Chem 294, 22445– 22451.
47 Nir-Paz, R., Pre´vost, M.-C., Nicolas, P., Blanchard, A & Wro´blewski, H (2002) Susceptibility of Mycoplasma fermentans and Mycoplasma hyorhinis to membrane-active peptides and enrofloxacin in human tissue cell cultures Antimicrob Agents Chemother 46, 1218–1225.
48 Duclohier, H & Wro´blewski, H (2001) Voltage-dependent pore formation and antimicrobial activity by alamethicin and ana-logues J Membr Biol 184, 1–12.
49 Vog t, B., Ducarme, P., Schinzel, S., Brasseur, R & Beching er, B (2000) The topology of lysine-containing amphipathic peptides in bilayers by circular dichroism, solid-state NMR, and molecular modeling Biophys J 79, 2644–2656.
50 Escrive, C & Laguerre, M (2001) Molecular dynamics simula-tions of the insertion of two ideally amphipathic lytic peptides LK(15) and LK(9) in a 1,2-dimyristoylphosphatidyl-choline monolayer Biochim Biophys Acta 1513, 63–74.
51 Raghunathan, G., Seetharamulu, P., Brooks, B.R & Guy, H.R (1990) Models of d-hemolysin membrane channels and crystal structures Proteins 8, 213–225.
52 Bechinger, B (1999) The structure, dynamics and orientation of antimicrobial peptides in membranes by multidimensional solid-state NMR spectroscopy Biochim Biophys Acta 1462, 157–183.