Energetics of Pore Formation Induced by Membrane Active Peptides
†
Ming-Tao Lee,
‡
Fang-Yu Chen,
‡
and Huey W. Huang*
,§
Department of Physics, National Central UniVersity, Chung-Li 32054, Taiwan, and Department of Physics & Astronomy,
Rice UniVersity, Houston, Texas 77251
ReceiVed December 1, 2003; ReVised Manuscript ReceiVed January 15, 2004
ABSTRACT: Antimicrobial peptides are known to form pores in cell membranes. We study this process in
model bilayers of various lipid compositions. We use two of the best-studied peptides, alamethicin and
melittin, to represent peptides making two types of pores, that is, barrel-stave pores and toroidal pores. In
both cases, the key control variable is the concentration of the bound peptides in the lipid bilayers (expressed
in the peptide-lipid molar ratio, P/L). The method of oriented circular dichroism (OCD) was used to
monitor the peptide orientation in bilayers as a function of P/L. The same samples were scanned by
X-ray diffraction to measure the bilayer thickness. In all cases, the bilayer thickness decreases linearly
with P/L and then levels off after P/L exceeds a lipid-dependent critical value, (P/L)*. OCD spectra showed
that the helical peptides are oriented parallel to the bilayers as long as P/L < (P/L)*, but as P/L increases
over (P/L)*, an increasing fraction of peptides changed orientation to become perpendicular to the bilayer.
We analyzed the data by assuming an internal membrane tension associated with the membrane thinning.
The free energy containing this tension term leads to a relation explaining the P/L-dependence observed
in the OCD and X-ray diffraction measurements. We extracted the experimental parameters from this
thermodynamic relation. We believe that they are the quantities that characterize the peptide-lipid
interactions related to the mechanism of pore formation. We discuss the meaning of these parameters and
compare their values for different lipids and for the two different types of pores. These experimental
parameters are useful for further molecular analysis and are excellent targets for molecular dynamic
simulation studies.
Membrane active peptides, including antimicrobials and
toxins, are known to induce transmembrane pores. The first
peptide discovered to do so is alamethicin (1, 2). At first,
alamethicin was thought to induce pores (which were
detected by ion conduction) only by a transmembrane electric
potential (see review in ref 3). However, numerous experi-
ments (4-8) indicated that alamethicin could insert into
bilayers in the absence of an external field (see review in
ref 9). Although it was believed that alamethicin insertion
would create pores, a direct correlation with ion conduction
was difficult to establish. Later, with the combination of
oriented circular dichroism (10, 11) and neutron diffraction
(12, 13), we showed the direct correlation between ala-
methicin insertion (without voltage) and transmembrane
pores. Two other extensively studied peptides, bee venom
toxin melittin (14) and frog peptide magainin (15), also
exhibited similar behaviors. Pores were evidently formed by
both melittin (16-18) and magainin (19, 20) because they
caused leakage of fluorescent dyes from lipid vesicles. In
the last 15 years, a great variety of antimicrobial peptides
have been shown to induce transmembrane pores in bacterial
cells as well as in lipid vesicles (see reviews in refs 21 and
22-24). Understanding the mechanism of pore formation
induced by peptides will provide insights into the functions
of antimicrobial peptides, which are essential components
of the innate immune system, and facilitate the development
of new anti-infective therapeutics. Pore formation is also
potentially useful for gene and drug deliveries (24).
To understand how peptides induce pore formation,
consider the simpler case of pores in pure lipid bilayers,
which have been extensively studied both experimentally and
theoretically (25-31). In pure lipid bilayers, pores are always
produced under tension. The initiation of a pore is a dynamic
process that is difficult to analyze, because it often involves
nucleation defects (28, 30, 32). However, once a pore is
formed, its essential mechanics is well understood. A pore
in a pure lipid bilayer is governed by the energy E
R
, which
is defined as the energy difference between a bilayer with a
circular pore of radius R and a bilayer without a pore (25,
26):
The first term represents the free energy cost of creating the
rim or the edge of the pore; γ is the line tension, or the energy
cost per unit length of the edge. The second term represents
the (negative) work done by the membrane tension σ to create
a pore of area πR
2
. The driving force for pore opening is
the membrane tension, while that for closure is the line
tension. For given γ and σ, E
R
is maximum at R ) γ/σ.
†
This work was supported by NIH Grants GM55203 and RR14812
and by the Robert A. Welch Foundation (to H.W.H) and by National
Science Council (Taiwan) through Contract NSC92-2112-M-008-013
(to F.-Y.C.).
* To whom correspondence should be addressed. Tel: 713 3484899.
Fax: 713 3484150. E-mail: hwhuang@rice.edu.
‡
National Central University.
§
Rice University.
E
R
) γ2πR - σπR
2
(1)
3590 Biochemistry 2004, 43, 3590-3599
10.1021/bi036153r CCC: $27.50 © 2004 American Chemical Society
Published on Web 03/04/2004
Thus a pore in a pure lipid bilayer is unstable: a pore of
radius smaller than γ/σ tends to close, whereas a pore of
radius larger than γ/σ tends to expand indefinitely. This
general behavior of pores in pure lipid bilayers suggests that
the pore-inducing peptides must (1) create a stress in the
membrane equivalent to a membrane tension to open pores
and (2) also stabilize the pores once produced.
In previous publications (33, 34), we have presented
experimental evidence and thermodynamic arguments for the
tension effect. Here we will provide a qualitative argument
for the stability of the peptide-induced pores. The main
purpose of this paper is to present experimental parameters
of lipid-peptide interactions that underline the cause and
the effect of the peptide-induced tension. We will discuss
the meaning of these parameters and show that the interac-
tions depend on the size of lipid headgroup and the chain
cross section. We will also show how the interactions are
different between toroidal pores and barrel-stave pores (35).
These experimental parameters provide excellent targets for
molecular dynamics simulations.
Among the known antimicrobials, only alamethicin and
its analogues have been shown to form barrel-stave pores.
Melittin, magainin, protegrin, and perhaps most cationic
antimicrobial peptides form toroidal pores (35). In this paper,
alamethicin and melittin are used to study these two types
of pores. Alamethicin and melittin are the most studied
peptides and hence have the most complete experimental
data, including that of single crystals.
MATERIALS AND METHODS
Materials. 1,2-Diphytanoyl-sn-glycero-3-phosphocholine
(DPhPC),
1
1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine
(POPC), 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC),
1,2-dioleoyl-sn-glycero-3-phosphoethanolamine (DOPE), and
1,2-dierucoyl-sn-glycero-3-phosphocholine (DiC22:1PC) were
purchased from Avanti Polar Lipids (Alabaster, AL). Ala-
methicin and melittin were purchased from Sigma-Aldrich
Chemical Co. (St. Louis, MO). Sigma alamethicin is a
mixture of components, principally alamethicin I (85% by
high-performance liquid chromatography) and alamethicin
II (12%), which differ by one amino acid (36). The same
material has been used in all previous studies (9, 11, 33, 34,
37, 38). Two grades of melittin were used, the sequencing
grade (product no. M-1407) and the grade of purity 93%
HPLC (product no. M-2272). Both gave the same results in
this study. Yang et al. (35) also found no difference between
Sigma melittin and pure synthetic melittin in this type of
study as long as there was no added Ca
2+
in the sample.
Poly(ethylene glycol) (PEG20000) was purchased from Merk
Co. (Hohenfrunn, Germany). All materials were used as
delivered.
Sample Preparation. Two experimental methods were used
in this study. One was oriented circular dichroism (OCD;
10, 11) for the measurement of peptide orientation in lipid
bilayers. Another was lamellar X-ray diffraction (LXD) for
the measurement of membrane thickness. The samples used
in both methods were in the form of oriented multilayers, a
stack of parallel lipid bilayers on a solid substrate. The
preparation of such oriented samples followed the method
described in the previous study (33). Briefly, lipid and
peptide of chosen peptide-to-lipid molar ratio (P/L) were
codissolved in a solvent of 1:1 (v/v) methanol and chloro-
form. The lipid concentration was about 1 mg per 20 µLof
solvent. The solution of appropriate amount was spread onto
a cleaned quartz surface,10 µL or less of solution (depending
on the P/L) onto a 14-mm diameter area for OCD or 100
µL of solution onto a 20-mm square area for LXD. When
the solvent dried, the sample was vacuumed to remove the
remaining solvent residues and then slowly hydrated with
water vapor until it appeared transparent. A good sample
was visually smooth and showed at least 5 orders of Bragg
diffraction by LXD.
OCD Measurement. The procedure of OCD measurement
has been described in Chen et al. (33). All sample temper-
atures were set at 30 °C. A water solution of poly(ethylene
glycol) (PEG20000) was inside a sealed sample chamber to
control the relative humidity, which in turn set the degree
of hydration of the sample. The concentration of PEG
solution used in this study was 4.75 g of PEG20000 in 10.00
g of water, which gave a vapor pressure equivalent to 98%
relative humidity (RH) at 30 °C. The hydration equilibrium
of the sample was ensured by an agreement of at least three
OCD spectra measured over a period of 6 h. OCD was
measured with a Jasco J-810 spectropolarimeter with light
incident normal to the sample surface (11). The background
OCD spectra of pure lipid bilayers (i.e., without peptides)
were measured separately and were removed from the spectra
of the corresponding samples containing peptides.
The reason we chose 98% RH (rather than 100% RH) for
this experiment was that for both OCD and LXD measure-
ments, the sample substrate was oriented vertically. At levels
of humidity higher than 98% RH, the membranes deposited
on one substrate would flow. This is not to say that it is
impossible to make measurements at 100% RH. An oriented
membrane sample could be covered with another substrate
to prevent the sample flow, as we have done previously for
OCD (11, 33) and for LXD (37, 39). However, it would take
a long equilibrating time to change the hydration level of a
covered (i.e., two-substrate) sample, and hydration changes
are necessary in an X-ray experiment for the purpose of phase
determination. Our previous experiments mentioned above
have shown that the dependence of the peptide transition on
hydration is gradual. There is no qualitative difference
between the orientation transitions measured at 98% RH and
at 100% RH (33).
The OCD studies of four peptide/lipid systems, Ala/
(DOPC/PE 2:1), Mel/DPhPC, Mel/DiC22:1PC, and Mel/
POPC, are reported here. The results of two other systems,
Ala/DPhPC and Mel/DOPC, were reported previously (33,
34)
LXD Measurement. The sample chamber for LXD was
the same as that used in our previous studies (40, 41), except
that the relative humidity was controlled by a series of PEG
solutions enclosed inside the chamber. This was to ensure
that the hydration levels of the samples were the same in
the OCD and LXD measurements. The temperature was set
1
Abbreviations: DPhPC, 1,2-diphytanoyl-sn-glycero-3-phospho-
choline; POPC, 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine; DOPC,
1,2-dioleoyl-sn-glycero-3-phosphocholine; DOPE, 1,2-dioleoyl-sn-glyc-
ero-3-phosphoethanolamine; DiC22:1PC, 1,2-dierucoyl-sn-glycero-3-
phosphocholine; Ala, alamethicin; Mel, melittin; P/L, the molar ratio
of the bound peptide to lipid; (P/L)*, the threshold (or critical) peptide
concentration for pore formation; OCD, oriented circular dichroism;
LXD, lamellar X-ray diffraction.
Energetics of Pore Formation Biochemistry, Vol. 43, No. 12, 2004 3591
at 30 °C, the same temperature for OCD measurements. In
addition to the measurement at 98% RH, a series of
measurement were made at lower levels of humidity for the
purpose of phase determination. Precise RH reading for these
lower levels of humidity was not necessary because the
swelling method for phase determination uses the lamellar
repeat spacing as the variable.
LXD was measured with Cu KR radiation generated at
30 kV/30 mA by θ-2θ scan from θ ) 0.5-7.5° with a step
size ∆θ ) 0.01° at 1 s per step. The equilibrium of the
sample at each humidity setting was ensured by an agreement
of at least three consecutive diffraction patterns the average
of which was subsequently analyzed. Only samples that
produced at least five discernible diffraction peaks were
accepted. Each peptide-lipid combination was measured
with at least two separately prepared samples. Each sample
was measured twice at least 10 h apart to check the
reproducibility.
The procedure for data reduction was described in many
of our papers (37, 39-42). Briefly, the procedure started
with background removal and absorption and diffraction
volume corrections. Then the integrated peak intensities were
corrected for the polarization and the Lorentz factors. The
magnitude of the diffraction amplitude was the square root
of the integrated intensity. The phases were determined by
the swelling method (43). With their phases determined, the
diffraction amplitudes were Fourier transformed to obtain
the transbilayer electron density profiles. The profiles were
not normalized to the absolute scale, but they gave the correct
peak-to-peak distances, since the latter are independent of
normalization (37).
LXD measurements of two peptide/lipid systems, Mel/
DiC22:1PC and Mel/POPC, are reported here. Four other
systems, Ala/DPhPC, Ala/(DOPC/PE 2:1), Mel/DPhPC, and
Mel/DOPC, were published previously (33, 34).
RESULTS
Fraction of Peptide Molecules Oriented Normal to the
Bilayer as a Function of P/L. The OCD spectra of ala-
methicin and melittin have been extensively discussed in
previous publications (33, 34). We used the same method
here to analyze the new results. Figure 1 shows the raw data
of OCD measurements for Ala/(DOPC/PE 2:1), Mel/DPhPC,
Mel/DiC22:1PC, and Mel/POPC systems, each for a series
of P/L. Lipids with unsaturated bonds exhibited high UV
absorption at wavelengths below 200 nm, which made the
OCD spectra in that region extremely noisy, and therefore,
they are not shown in Figure 1.
Alamethicin and melittin both form helices when bound
to lipid bilayers. Each has been found to be oriented either
parallel or perpendicular to the plane of the bilayer, depend-
ing on the sample condition. In previous experiments, we
have carefully obtained the mutually normalized OCD spectra
FIGURE 1: OCD spectra of alamethicin in the DOPC/DOPE 2:1 mixture bilayers, melittin in DPhPC bilayers, melittin in DiC22:1PC
bilayers, and melittin in POPC bilayers at 30 °C and 98% RH. Appropriate lipid background, that is, the OCD of the same amount of lipid,
was removed from each spectrum. The CD amplitudes are in an arbitrary unit. The I and S spectra of alamethicin and melittin were
reproduced from data of refs 33 and 34, respectively. They are the spectra of the helical peptide oriented perpendicular and parallel to the
bilayer, respectively. I and S were relatively normalized to each other. High UV absorption by unsaturated lipids made the spectra below
∼200 nm unacceptably noisy. Nevertheless the somewhat incomplete spectra are adequate for spectra fitting to determine the fraction of
the peptide molecules in the I (or S) state.
3592 Biochemistry, Vol. 43, No. 12, 2004 Lee et al.
for these two orientations by using one sample at different
temperatures or RHs, and denoted them as I and S spectra,
respectively, for perpendicular and parallel (to the plane of
the bilayer) orientations. These standard spectra were repro-
duced in each panel in Figure 1. Each newly measured OCD
spectrum was fitted, after the lipid background removal, with
a linear combination of I and S, aI + bS, and then replotted
with the original amplitude multiplied by a factor 1/(a + b)
in Figure 1. The fraction of the peptide molecules in the I
state (perpendicular to the bilayer) is denoted as φ ) a/(a +
b). φ is plotted as a function of 1/(P/L) for the four peptide/
lipid systems in Figure 2.
Membrane Thickness as a Function of P/L. Diffraction
patterns of Mel/DiC22:1PC and Mel/POPC systems are
shown in Figure 3 for a series of P/L at the highest RH
measured. The complete data include two other sets of
diffraction patterns measured at lower RH for the purpose
of phasing. Each pattern has at least 5 Bragg orders. No peak
broadening with order was observed, indicating that the
undulation fluctuations were negligible at hydration levels
below 98% RH (44). After the data reduction (see Materials
and Methods), each sample has three sets of diffraction
amplitudes at three different repeat spacings. For the purpose
of phase determination, the amplitudes were relatively
normalized according to the Blaurock method (43) and
plotted against the scattering momentum, q. Four examples
are shown in Figure 4, where the phases were chosen
according to the swelling principle (45, 46).
With the phases determined, the diffraction amplitudes
were Fourier transformed to the transbilayer electron density
profiles shown in Figure 5. We then plotted the peak-to-
peak (PtP) spacing against P/L for each sample (Figure 6).
The error bars in Figure 6 represented the ranges of
reproducibility of four measurements, two measurements for
each of two separately prepared samples. The hydrocarbon
FIGURE 2: Fractions of the peptide molecules in the I state (helices
perpendicular to the bilayer), φ, are plotted as a function of the
inverse of P/L. For each peptide/lipid system, φ is zero at low values
of P/L. However, φ increases linearly with -1/(P/L)asP/L exceeds
a certain threshold value, in agreement with the prediction given
by eq 5. The intercept of the linear fit at the high P/L region with
the baseline φ ) 0 gives the threshold peptide concentration, (P/
L)*. The slope gives the value β according to eq 5.
FIGURE 3: X-ray diffraction patterns of melittin/DiC22:1PC and
melittin/POPC systems for a series of P/L at the highest hydration
level measured. The patterns are displaced for clarity. The steps at
2θ ≈ 4° were the results of using an X-ray attenuator that reduced
the count rates for the first two diffraction peaks to not saturate
the photon counter. Note that each pattern has at least 5 Bragg
orders.
FIGURE 4: Phasing diagrams for the X-ray diffraction of pure
DiC22:1PC and DiC22:1PC containing melittin at P/L ) 1/100 (top)
and of pure POPC and POPC containing melittin at P/L ) 1/100
(bottom). The abscissa is the X-ray momentum transfer q ) 4π
sin θ/λ, where λ is the X-ray wavelength, 1.54 Å. The phases were
chosen according to the swelling method (43, 45, 46).
Energetics of Pore Formation Biochemistry, Vol. 43, No. 12, 2004 3593
thickness, h, of the bilayer was estimated by subtracting twice
the length of the glycerol region (from the phosphate to first
methylene of the hydrocarbon chain), that is, ∼10 Å, from
PtP (47-51).
EXPERIMENTAL PARAMETERS
Table 1 was compiled from the OCD and LXD data above.
In addition, we also included the parameters calculated from
the published OCD data for Ala/DPhPC and Mel/DOPC and
the published LXD data for Ala/DPhPC, Ala/(DOPC/PE 2:1),
Mel/DPhPC, and Mel/DOPC (33, 34). The bilayer stretch
moduli, K
A
’s, were measured by Rawicz et al. (52) for more
than 10 lipids using the vesicle aspiration method. Surpris-
ingly the K
A
’s are all about 240 pN/nm within experimental
errors. Hence we use this average value for our discussion.
The hydrocarbon thickness h of each pure lipid bilayer was
calculated from its PtP by subtracting twice the length of
the glycerol region (10 Å) as mentioned in the Results
section. The lipid area cross section, A
L
, was calculated from
the hydrocarbon volume (48) of each lipid divided by its h
value.
1. Area Expansion per Peptide, A
P
. The idea that peptide
binding creates membrane tension was derived from the
observation that every antimicrobial peptide that we inves-
tigated caused membrane thinning. The peptides that we have
investigated include alamethicin (37), magainin (42), pro-
tegrin (53), melittin (34), and many of their analogues. (One
exception is θ-defensin the thinning effect of which is 1 order
of magnitude smaller compared with the aforementioned
peptides (54). Interestingly, its mechanism is also different
(55-57)). We assumed that the membrane thinning is caused
by the peptides stretching the membrane area, which is the
direct result of the peptide molecules being embedded in the
headgroup region (see the cartoon in Figure 7). Other
experiments supporting the peptide embedment in the head-
group region include solid-state NMR (58, 59), Raman (60),
fluorescence (61), differential scanning calorimetry (DSC,
62), and titration calorimetry (63, 64). A membrane area
expansion caused by melittin adsorption has also been
observed by vesicle aspiration (65) at constant vesicle
volume, while no permeation through the membrane oc-
curred. We have argued, on the basis of the theory of
membrane elasticity (66), that the peptide molecules embed-
ded in the headgroup region are dispersed rather than
aggregated. This was indeed supported by evidence from
fluorescence energy transfer (67-69), NMR (58), and
electron paramagnetic resonance (EPR, 70) studies. Thus the
fractional increase of the monolayer area due to peptide
binding is ∆A/A ) ∆A
L
/A
L
) (A
P
/A
L
)(P/L), where P/L is
the bound peptide-to-lipid molar ratio, A
L
the area cross
section per lipid, and A
P
the area increase caused by one
FIGURE 5: Electron density profiles of melittin/DiC22:1PC systems
for a series of P/L from 0 to 1/15 (top) and of melittin/POPC
systems for a series of P/L from 0 to 1/25 (bottom). The profiles
are not normalized and are displaced for clarity. The short vertical
bars indicate the positions of the peaks from which the peak-to-
peak distances, PtP, were measured.
FIGURE 6: Peak-to-peak distance (PtP) versus P/L for melittin in
DiC22:1PC (top) and for melittin in POPC (bottom). The error bars
represent the ranges of reproducibility by four measurements. The
arrows indicate (P/L)* determined by the OCD measurement (see
Figure 2). PtP decreases linearly with P/L below (P/L)* and is
constant above (P/L)* within the experimental errors.
3594 Biochemistry, Vol. 43, No. 12, 2004 Lee et al.
