THE JOURNAL OF BIOLOGICAL CHEMISTRY
© 2003 by The American Society for Biochemistry and Molecular Biology, Inc.
Vol. 278, No. 46, Issue of November 14, pp. 45216 –45223, 2003
Printed in U.S.A.
Pore Formation by Equinatoxin II, a Eukaryotic Protein Toxin,
Occurs by Induction of Nonlamellar Lipid Structures*
Received for publication, June 5, 2003, and in revised form, August 26, 2003
Published, JBC Papers in Press, August 27, 2003, DOI 10.1074/jbc.M305916200
Gregor Anderluh‡, Mauro Dalla Serra§, Gabriella Viero§, Graziano Guella¶, Peter Maček‡,
and Gianfranco Menestrina§储
From the ‡Department of Biology, Biotechnical Faculty, University of Ljubljana, Večna pot 111, 1000 Ljubljana, Slovenia,
§ITC-CNR, Institute of Biophysics, Section at Trento, Via Sommarive 18, 38050 Povo (Trento) Italy, ¶Laboratory of
Bioorganic Chemistry, Department of Physics, University of Trento, Via Sommarive 14, 38050 Povo (Trento), Italy
Pore formation in the target cell membranes is a common mechanism used by many toxins in order to kill
cells. Among various described mechanisms, a toroidal
pore concept was described recently in the course of
action of small antimicrobial peptides. Here we provide
evidence that such mechanism may be used also by
larger toxins. Membrane-destabilizing effects of equinatoxin II, a sea anemone cytolysin, were studied by various biophysical techniques. 31P NMR showed an occurrence of an isotropic component when toxin was added
to multilamellar vesicles and heated. This component
was not observed with melittin, ␣-staphylococcal toxin,
or myoglobin. It does not originate from isolated small
lipid structures, since the size of the vesicles after the
experiment was similar to the control without toxin.
Electron microscopy shows occurrence of a honeycomb
structure, previously observed only for some particular
lipid mixtures. The analysis of FTIR spectra of the
equinatoxin II-lipid complex showed lipid disordering
that is consistent with isotropic component observed in
NMR. Finally, the cation selectivity of the toxin-induced
pores increased in the presence of negatively charged
phosphatidic acid, indicating the presence of lipids in
the conductive channel. The results are compatible with
the toroidal pore concept that might be a general mechanism of pore formation for various membrane-interacting proteins or peptides.
Proteins and peptides with the capacity to increase membrane permeability have been elaborated by a large number of
organisms and are used as toxins, effectors in immune response or apoptosis. One of the most commonly adopted mechanisms is the formation of pores in the targeted membrane as
occurs, for example, with pore-forming toxins (PFT)1 (1, 2).
* This work was supported by grants from Consiglio Nazionale delle
Ricerche, Ministero Italiano Università e Ricerca Scientifica, and Istituto Trentino di Cultura (to G. A., M. D. S., G.V., G. G., and G. M.) and
the Ministry of Education, Science, and Sport of Slovenia (to G. A. and
P. M.). The costs of publication of this article were defrayed in part by
the payment of page charges. This article must therefore be hereby
marked “advertisement” in accordance with 18 U.S.C. Section 1734
solely to indicate this fact.
储 To whom correspondence should be addressed. Tel.: 39-0461-314256; Fax: 39-0461-810-628; E-mail: [email protected].
1
The abbreviations used are: PFT, pore-forming toxin(s); DPhPC,
1,2-diphytanoyl-sn-glycerophosphocholine; EqtII, equinatoxin II; EM,
electron microscopy; FTIR spectroscopy, Fourier-transformed infrared
spectroscopy; LUV, large unilamellar vesicles; L/P, lipid/protein ratio;
MLV, multilamellar vesicles; PA, phosphatidic acid; PC, phosphatidylcholine; PLM, planar lipid membranes; POPC, 1-palmitoyl-2-oleoyl-snglycero-3-phosphocholine; QLS, quasielastic light scattering; SM,
sphingomyelin; SUV, small unilamellar vesicles.
Bacterial PFT, protein molecules of Mr ⬎ 30,000, usually follow
two strategies; they either form a channel via insertion of a de
novo generated transmembrane ␤ barrel (examples are staphylococcal ␣-toxin, the cholesterol-dependent cytolysins, and the
protective antigen of anthrax toxin), or they insert a bundle of
preexisting ␣-helices through the membrane (like colicins and
crystal ␦-endotoxins) (3, 4). Smaller molecules, like antimicrobial peptides or peptide toxins with Mr between 1000 and 5000,
have developed a wider set of mechanisms (5). In fact, besides
the ␤-barrel (e.g. protegrin) (6) and the ␣-helix bundle (e.g.
alamethicin and other peptaibols) (7), some alternative strategies were found, which directly modify the bilayer organization
of the membrane. They range from a generic destabilization
(exemplified by the carpet-like model) (8) to the formation of
specific mixed lipid-peptide structures, like the toroidal pore,
which was observed with magainin (9) and melittin (10).
Actinoporins are a peculiar class of eukaryotic PFT with
intermediate Mr, exclusively found in sea anemones. It is a
family of cysteineless proteins with Mr around 18,000 –20,000
and a preference for sphingomyelin (SM) (11). They form cation-selective pores with a diameter of ⬃2 nm on cellular and
model membranes (12–14). The three-dimensional structure of
the soluble state of one actinoporin, equinatoxin II (EqtII; from
the sea anemone Actinia equina) was recently solved by x-ray
(15) and NMR (16). The molecule is composed of a hydrophobic
␤-sandwich core, flanked on the opposite sides by two ␣-helices.
The first 30 N-terminal residues, which include one helix, are
the best candidates for pore formation. This part of the molecule is amphipathic in character, is well conserved in all actinoporins, is clearly similar to some membrane-interacting peptides like melittin and fusogenic viral peptides (17), and is the
only portion of the molecule that can change conformation
without disrupting the general fold (15). Preliminary evidence
of the involvement of the N-terminal part in the formation of
the transmembrane channel was obtained by a low resolution
cysteine-scanning mutagenesis along the molecule (18) and by
N-terminal truncations (19). More recently, a complete cysteine-scanning mutagenesis of the region encompassing residues 10 –28 clearly demonstrated that this portion adopts
␣-helical configuration in the membrane, that the ␣-helix is
longer than in the soluble form, that it lies on the water-lipid
interface in the membrane-bound state, and that it inserts into
the membrane to line the pore interior, forming an angle of
around 20° with the bilayer normal (20). The slight increase in
helicity that occurs upon lipid binding was already noted by
FTIR spectroscopy (21) and is consistent with the length of the
homologous, membrane-perturbing, helical peptides.
In our current model, initial binding is provided by a cluster
of exposed aromatic residues (which includes Trp-112 and Trp116) (18, 22, 23), followed by N-terminal helix translocation to
45216
This paper is available on line at http://www.jbc.org
Equinatoxin II Forms Toroidal Pores
the surface of the membrane and its insertion into the lipid
bilayer (20, 23). The channel entity is not stable enough to be
isolated and analyzed as a single unit. Therefore, the number of
monomers appearing in the final pore was deduced from indirect experiments. All evidence, gathered from cross-linking
(13), steady state (12), and kinetic (24) partitioning experiments, consistently indicated a tetrameric structure. Notably,
however, the pore diameter is too large to be formed by a simple
bundle of four helices. One possibility to solve this apparent
contradiction is to assume that the pore is partially lined by
membrane lipids.
In fact, perturbation of the lamellar lipid structure has been
noted with some of the helical peptides similar to the N-terminal helix of actinoporins (e.g. mastoparan (25), staphylococcal
␦-lysin (26), the HIV-1 viral fusion peptide (27), and some
synthetic peptides with an amphipathic ␣-helical character
(28)). In addition, a direct effect of EqtII on the lipid phase has
been recently observed and attributed to the separation of a
sphingomyelin-enriched phase and to vesiculation (29). Given
its similarity with melittin, we suspected that the N-terminal
helix of actinoporins could perturb the membrane by giving rise
to the formation of a toroidal protein-lipid pore. Direct functional evidence for such a structure was already provided by
the toxin-strengthening effect of lipids that favor the positive
curvature appearing at the center of the toroidal pore and by
the fact that pore formation induced the mobilization of lipid
molecules between the two leaflets of the bilayer (30).
Here we report a new investigation of EqtII effects on membrane order, providing original evidence that it can affect the
lipid phase by a peculiar mechanism, previously observed only
in some specific lipid mixtures. Such a mechanism, which is
consistent with the concept that its biological activity is exerted
via the formation of a toroidal protein-lipid pore, represents a
novel paradigm of protein-induced membrane destabilization.
EXPERIMENTAL PROCEDURES
Toxins, Proteins, and Lipids—Native EqtII was isolated from the sea
anemones as described (31). Melittin (65– 85% pure by high pressure
liquid chromatography) and myoglobin (type II, minimum 90%, from
whale muscle) were from Sigma; pure ␣-toxin from Staphylococcus
aureus was a kind gift of Dr. Hungerer (Behringwerk, Marburg,
Germany). Lipids, 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine
(POPC), 1,2-diphytanoyl-sn-glycerophosphocholine (DPhPC), phosphatidic acid (PA), egg phosphatidylcholine (PC), and brain SM, were all
obtained from Avanti Polar Lipids (Alabaster, AL).
Lipid Vesicle Preparations—PC, POPC, and cholesterol were dissolved in chloroform, SM in chloroform/ethanol (4:1, v/v). Lipid mixtures were vacuum-dried in a film at the bottom of a round flask and
further treated in a speed-vac for 1–2 h. The lipid films were resuspended at room temperature by vigorously vortexing in the appropriate
buffer, followed by six cycles of freezing and thawing in liquid nitrogen.
Such dispersions of multilamellar vesicles (MLV) were either immediately used as such or were extruded through polycarbonate membranes
with pores of diameter 400 or 600 nm to prepare large unilamellar
vesicles (LUV) of that size. Alternatively, small unilamellar vesicles
(SUV) were prepared by sonicating the MLV on ice for 30 min, followed
by gentle sedimentation to remove titanium particles released by the
sonotrode. Lipid samples with PFT or myoglobin, were prepared similarly except that the buffer used for lipid hydration contained appropriate amounts of the proteins. For actinoporins, mixture POPC/SM or
PC/SM in a molar ratio of 1:1 or 2:1 was used, since these two mixtures
are the most sensitive in functional assays for binding and permeabilization (13, 24). For the same reason, we used POPC/cholesterol in a
molar ratio of 1:1 with ␣-toxin (32).
NMR Spectra—31P NMR spectra of the different lipid and lipid/toxin
preparations were recorded at 121.4 MHz on a Varian XL-300 spectrometer. Lipid composition was POPC/SM 2:1 (mol/mol), and lipid
concentration was 30 mM in 100 mM NaCl, 30 mM Tris-HCl, 1 mM
EDTA, pH 7.0 (NMR buffer). 0.4 – 0.7 ml of the sample were placed in
5-mm Pyrex NMR tubes.
Two different protocols were used (as indicated). The phase-cycled
Hahn echo pulse sequence was used on samples that were supple-
45217
mented with 10% deuterated water to provide an internal lock frequency control (33). Recycle time was 2 s, 90° pulse width was 30 ␮s,
delay between pulses was 100 ␮s, sweep width was 38.46 kHz, acquisition time was 0.104 s, and transient number was 2000 –10,000. Alternatively, spectra were acquired by the single pulse acquisition technique (60° pulse angle and 0.4-s interpulse time) on samples without
deuterated water. 2000 – 4000 transients were used in this case. In all
measurements, 8192 complex data points were acquired and exponentially multiplied with 75 Hz prior to Fourier transformation with continuous wave 1H-decoupling during data acquisition. 31P chemical
shifts were measured relative to 0.0 ppm indicated by sonicated unilamellar vesicles. Variable temperatures were obtained with a homemade
temperature-control unit that was calibrated on the chemical shift
difference between the residual OH and CD2H proton of 99.8% deuterated methanol (temperature accuracy ⫾0.2 °C). Samples were left 10
min to equilibrate after each change of temperature
Measurement of Vesicle Size by Quasielastic Light Scattering (QLS)—
MLV, LUV, and SUV size was determined by QLS (34) at a fixed angle
(90°) and room temperature, using a laser particle sizer (Malvern
Z-sizer 3) equipped with a 5-milliwatt helium-neon laser. For this
assay, lipid samples were diluted in NMR buffer to give similar
scattering (from 25 to 250 times, depending on their size). A 64-channel
correlator was used capable of particle size estimates in the range of
5–5000 nm. Data were analyzed by the cumulant method using
Malvern Application Software. The first cumulant provided the apparent diffusion coefficient of the particles (from which the hydrodynamic
radius can be derived by the Stokes-Einstein relation), whereas the
second cumulant gave the distribution width (35).
Permeabilization of Lipid Vesicles—Permeabilization was assayed by
measuring the leakage of calcein (36). MLV were prepared as above, but
in the presence of 80 mM calcein (pH 7.0 with NaOH). The external
calcein was removed by spinning through minicolumns (Pierce) loaded
with Sephadex G50 medium preequilibrated with NMR buffer. Fluorescence was measured in a spectrofluorimeter (SPEX Fluoromax) using a 1-cm semimicroquartz cuvette with 1 ml of NMR buffer with 1 ␮M
lipids continuously stirred. Lysins were added to vesicles at a lipid/
protein (or peptide) molar ratio of 1. Excitation wavelength was set to
485 nm, and emission was set to 520 nm with both slits set to 2 nm.
Release of calcein (R), a percentage, was calculated according to Equation 1,
R ⫽ 共Fmeas ⫺ Finit兲/共Fmax ⫺ Finit兲䡠100
(Eq. 1)
where Fmeas, Finit, and Fmax are the measured, initial, and maximal
fluorescence, respectively. Fmax was obtained by the addition of Triton
X-100 to 1 mM final concentration. Spontaneous leakage was negligible
on this time scale. The experiments were run at room temperature.
Electron Microscopy (EM)—Transmission EM was performed with a
Philips CM100 microscope operating at 80 kV. A drop of the sample was
deposited on the copper grids coated with formvar film and negatively
stained with 1% phosphotungstic acid (Sigma). Samples were dried on
air before imaging. Images were taken with a BioScan Camera model
792 (Gatan) with CCD resolution of 1024 ⫻ 1024 pixels. For size
estimation, samples were imaged after the NMR experiment. The dimensions were estimated by measuring the diameter of vesicles from
the EM image. Averages for MLV and MLV with EqtII were obtained
after fitting dimensions to Gaussian distribution. Averages for MLV
with melittin were calculated from the data, since the distribution was
not Gaussian in that case.
Preparation of Vesicles for Infrared Spectroscopy—SUV were prepared as above with POPC/SM (1:1) (5 mg/ml) for EqtII and PC/cholesterol (1:1) (6 mg/ml) for ␣-toxin. EqtII was applied at a lipid/protein
ratio (L/P) of 100. Unbound toxin, estimated by the residual hemolytic
activity, was between 5 and 10% and was not removed. Samples were
applied straight to the germanium crystal. ␣-Toxin was used at L/P of
200 and incubated at 37 °C for 1 h. Free toxin (⬃35%, as determined by
hemolytic activity) was removed by repeated ultrafiltration through
polysulfone filters of 300-kDa cut-off (NMWL; Millipore Corp.) as described (37). All preparations were in 10 mM Hepes, pH 7.0.
FTIR—FTIR spectra were collected in the attenuated total reflection
configuration, as described (21), on a Bio-Rad FTS 185 spectrometer
with a deuterium triglycine sulfate detector and KBr beam splitter, at
a nominal resolution of 0.5 cm⫺1. Three kinds of spectra were taken:
toxin alone, lipid alone (SUV of either PC/SM 1:1 or PC/cholesterol 1:1),
or toxin-treated SUV. The samples were spread on a 10-reflection
germanium crystal (45° cut), flushed with D2O-saturated nitrogen, and
housed in a vertical attenuated total reflection attachment (by Specac).
Polarized spectra were collected with a rotating wire grid polarizer (fir
45218
Equinatoxin II Forms Toroidal Pores
grid; Specac), manually positioned either parallel (0°) or perpendicular
(90°) to the plane of the internal reflections.
The orientation of a structural element was calculated from the
dichroic ratio, R ⫽ A0°/A90°, where A0° and A90° are the absorption bands
of the functional group of that element in the parallel and perpendicular
configuration, respectively (38, 39). The form factor, S, was derived
from R using Equation 2 (40),
S⫽
E2x ⫺ RE2y ⫹ Ez2
1
共3cos2␪ ⫺ 1兲共E2x ⫺ RE2y ⫺ 2Ez2兲
2
Lipid structurea
SUV
LUV400
LUV600
MLV
MLV ⫹ EqtII
MLV ⫹ EqtII
MLV ⫹ EqtII
MLV ⫹ melittin
MLV ⫹ melittin
MLV ⫹ melittin
MLV ⫹ myoglobin
MLV ⫹ ␣-toxin
(Eq. 2)
where ␪ represents the angle between the long axis of the molecule
under consideration and the transition moment of the investigated
vibration; Ex, Ey, and Ez are the components of the electric field of the
evanescent wave in the three directions (the z axis being perpendicular
to the plane of the crystal), which we calculated according to Harrick
expressions for thick films (40, 41). The following order parameters
were calculated: (a) SL, for the lipid chains, using either the symmetric
or asymmetric CH2 stretching (bands centered at 2850 and 2920 cm⫺1,
respectively) and ␪ ⫽ 90°; (b) Samide I⬘, for the amide I⬘ band (integrated
between 1600 and 1700 cm⫺1) with ␪ ⫽ 0° (42); and (c) S␣, for the
␣-helix, using the Lorentzian component at 1651 ⫾ 1 cm⫺1, obtained by
deconvoluting and curve-fitting the amide I⬘ band (between 1700 and
1600 cm⫺1) as previously described (21) and using ␪ ⫽ 39° (38). To
analyze the amide I⬘ band of the lipid-bound toxin, the spectra were
previously corrected by subtracting the contribution of the lipid alone,
with a weight that eliminates the stretching band of the phospholipid
carboxyl groups at 1738 cm⫺1.
From the order parameter, we calculated the average tilt angle ␥ of
the molecular axis with respect to the z axis according to Equation
3 (40),
1
S ⫽ 共3cos2 ␥ ⫺ 1兲
2
TABLE I
Diameter of lipid particles measured by quasielastic light scattering
or electron microscopy
Dimensions were estimated at room temperature, after having exposed the vesicles to 55 °C.
L/P
QLS
EMb
␮m
␮m
200
100
50
20
10
5
25
55
0.06 ⫾ 0.01
0.37 ⫾ 0.09
0.61 ⫾ 0.19
2.6 ⫾ 0.8
2.4 ⫾ 1.3
2.7 ⫾ 0.8
2.4 ⫾ 1.7
1.7 ⫾ 0.9
1.3 ⫾ 0.8
0.05 ⫾ 0.02
4.5 ⫾ 3.0
2.6 ⫾ 1.8
1.46 ⫾ 0.02
1.48 ⫾ 0.2
0.41 ⫾ 0.3
a
Lipid composition was, in all cases, POPC/SM 2:1 (mol/mol).
The value reported is the mean ⫾ S.D. of 69, 247, or 157 vesicles for
MLV, MLV ⫹ EqtII, or MLV ⫹ melittin, respectively. In the case of
MLV with melittin, the distribution was quite broad (approximately
40% of vesicles were smaller than 250 nm). Note that, in the presence
of heterogeneous distributions, QLS provided slightly higher estimates
than EM, probably due to the stronger scattering of larger particles.
b
(Eq. 3)
By definition, S represents the fraction of molecules aligned with the
axis direction, whereas (1 ⫺ S) is the fraction remaining disordered.
The mean angle ␴ of the ␣-helix axis with respect to the direction of the
lipid chains was recalculated from Equation 4 (40),
S␣
1
⫽ 共3cos2 ␴ ⫺ 1兲
1
2
共3cos2 ␥L ⫺ 1兲
2
(Eq. 4)
where ␥L is the angle formed by the lipid chains with the z axis. Finally,
the lipid to bound toxin ratio (L/Pb) was calculated from A90° according
to Equation 5 (42),
共1 ⫺ Samide I⬘)
䡠
(1 ⫹ SL/2)
L/Pb ⫽ 0.208共nres ⫺ 1兲
冕
冕
2980
A90°共␯L兲d␯
2800
1690
(Eq. 5)
A90°共␯amide I⬘兲d␯
1600
where nres represents the total number of residues in the toxin (179 for
EqtII and 293 for ␣-toxin).
Determination of Ion Selectivity of the EqtII Channel—Electrical
properties of EqtII pores were studied using planar lipid membranes
(PLM) exactly as described (20). On a weight basis, PLM were made of
20% SM, (80 ⫺ x)% of DPhPC and x% PA, with x ranging from 0 to 20.
EqtII was added at nanomolar concentration only to the cis side, where
voltage was applied. Reversal voltages were measured in a 10-fold KCl
gradient and translated into a permeability ratio P⫹/P⫺ (where P⫹ and
P⫺ refer to cation and anion permeability, respectively) by the Goldmann-Hodgkin-Katz equation. Initially, both sides (volume of 2 ml)
were bathed by symmetrical solution of 10 mM Tris, 100 mM KCl, pH
8.0. Thereafter, the trans side was perfused with 10 volumes of 10 mM
Tris, 1 M KCl, pH 8.0.
RESULTS
Effects of EqtII on Lipid Isotropy: NMR—The effects of
equinatoxin (EqtII) on lipid membranes were first studied by
31
P NMR spectroscopy. Initially, pure lipid samples were examined in order to find the most suitable preparation. Four
different kinds of vesicles were compared (SUV, LUV400,
FIG. 1. 31P NMR spectra of the lipid preparations used in this
paper. Various kinds of vesicles composed of POPC/SM 2:1 (mol/mol)
were prepared at 30 mM lipid concentration in NMR buffer. Spectra
were taken by a Varian XL-300 spectrometer at 121 MHz with the
phase-cycled Hahn echo pulse sequence. 2000 –10,000 transients were
accumulated at either 20 °C (left traces) or 50 °C (right traces). At each
temperature, the samples were equilibrated for at least 10 min. The size
of the lipid vesicles increases from top to bottom.
LUV600, and MLV), which share the same bilayer organization
but have quite different size, as shown by QLS (Table I). Ideally, the lipid preparation should display a pure lamellar phase
signal so that any toxin-induced change of lipid organization
(e.g. from lamellar to hexagonal or isotropic phases) (43, 44)
could be detected. Only MLV approximated this behavior,
showing, even at the higher temperature, a clear predominance
of the broad asymmetric 31P NMR spectra (with low field shoulder) typical of the lamellar phase over a small isomorphic
signal (Fig. 1). All other preparations showed the occurrence of
a peak around 0 ppm, suggesting the presence of a population
of fast tumbling phospholipid molecules, which was predomi-
Equinatoxin II Forms Toroidal Pores
45219
FIG. 3. Effects of other polypeptides on 31P NMR spectra of
MLV. 31P NMR spectra of 30 mM POPC/SM 2:1 (mol/mol) MLV with
␣-toxin (two upper traces), or melittin (lower traces) at the indicated
lipid/protein, or lipid/peptide, ratio (L/P). Spectra were recorded with
the single pulse protocol at either 20 °C (left traces) or 55 °C
(right traces).
FIG. 2. Effects of EqtII on 31P NMR spectra of MLV. Temperature dependence of 31P NMR spectra of 30 mM POPC/SM 2:1 (mol/mol)
MLV either alone (traces on the left, phase-cycled Hahn echo pulse
sequence) or with the addition of EqtII in the hydration buffer at L/P of
100 (traces on the right, single pulse protocol). Spectra in each panel
were recorded consecutively, from bottom to top, after equilibrating the
sample at the indicated temperature for at least 10 min. 4000 –10,000
transients were averaged for each spectrum.
nant, at least at the higher temperature. Clearly, the amount of
isotropic phase was inversely correlated to the size of the vesicles and directly related to the temperature. For these reasons,
as in other works (25, 26, 28, 29), we used only MLV in the rest.
Spectra of MLVs with EqtII present in the buffer during the
lipid hydration are given in Fig. 2. Even at a lipid/EqtII molar
ratio 100, practically no change was observed at 20 °C. However, upon heating the mixture up to 55 °C, an isotropic peak,
in the region of 0 ppm shift, progressively appeared with temperature and toxin content. The isotropic peak was largely
reversible with temperature, and after cooling back to 20 °C
only a small amount of isotropic phase remained (not shown).
The whole cycle was repeatable. Spectra taken with either of
the two protocols were qualitatively the same.
To clarify whether the observed effects were peculiar of
EqtII, controls were performed with the same lipid system and
three different polypeptides: S. aureus ␣-toxin (a PFT that
opens pores by inserting a transmembrane ␤-barrel) (45);
melittin, a pore-forming peptide with a strong destabilizing
capacity (46); and myoglobin, a protein that does not associate
with the membrane. Neither myoglobin (not shown) nor ␣-toxin
induced any shift or isotropic peak (Fig. 3) even at high temperatures and high protein content (the lowest L/P used was 25
for myoglobin and 55 for ␣-toxin (i.e. lower than that used for
EqtII in Fig. 2)). In the case of melittin, instead, MLV were still
in lamellar phase at L/P of 20, but a mixed population appeared
at L/P of 10, and, finally, all of the signal became isomorphic at
L/P of 5, producing a sharp 0 ppm peak even at 20 °C (Fig. 3).
At these concentrations, melittin can act as a detergent and
generate small mixed micelles, or discoids, which produce the
isotropic signal in the 31P NMR spectrum (46, 47).
The changes in NMR spectra in the presence of EqtII could
have two reasons. First, an isotropic signal may arise from
physical structures, which, at variance with MLV, allow for
complete motional averaging of the chemical shift anisotropy
(e.g. smaller vesicles or micellar structures) (see Fig. 1). This
might happen if the toxin promotes solubilization of MLV into
smaller structures by a detergent-like mechanism similar to
melittin (Fig. 3). Alternatively, the presence of relatively high
toxin concentrations might directly influence and change the
lipid phase by promoting isotropic structures like rhombic,
cubic, or inverted micellar (44).
Effects of EqtII on Lipid Morphology: QLS—Toxin effects on
the size of the MLV were directly assayed by QLS (Table I).
Down to an L/P of 50 (corresponding to a w/w ratio of 2), there
was no marked change in size induced by EqtII that could
justify the appearance of the isotropic signal in the 31P NMR
spectrum. Melittin, instead, at L/P of 5 completely destabilized
the MLV, producing structures of size comparable with the
SUV. Neither of the two control proteins, the pore-forming
␣-toxin or the lipid-inert myoglobin, modified the size of
the MLV.
Electron Microscopy—The shape and size of control and toxin-treated vesicles was further checked by transmission EM
(Fig. 4). Control MLV appeared as cauliflower-like structures
with no details within the vesicles and no penetration of the
contrasting dye, confirming that they were sealed structures
(Fig. 4A). The presence of melittin during MLV formation considerably affected their size (Fig. 4B and Table I), producing
much smaller structures already evident at L/P of 10. The
effect of EqtII was, instead, markedly different and peculiar. In
fact, whereas the size of the complexes formed in the presence
of EqtII was approximately the same as that of pure MLV, a
multitude of smaller internal structures became visible inside
the MLV (Fig. 4, C and D) after they had been exposed to the
higher temperature. Obviously, the toxin affected the organization of the internal lipid leaflets, and the newly formed
structures could indeed generate the isotropic peak of the NMR
signal, while leaving unaffected the overall size estimated by
45220
Equinatoxin II Forms Toroidal Pores
FIG. 4. EM analysis of morphology
changes in toxin-treated MLV at
room temperature. A, unmodified MLV
composed of POPC/SM 2:1. B, the same
kind of MLV but with the addition of
melittin at L/P of 10. C–F, same MLV
with the addition of EqtII at L/P of 50 and
the exposure to 50 °C. Magnification increases from C to F. Calibration bars (␮m)
are reported. G, mechanism of the honeycomb formation in MLV (reproduced, with
permission, from Ref. 44). Raising the
temperature, contact points between lamellae (b) can undergo localized fusion,
with the formation of lipid particles corresponding to HII phase (c), and induce
compartmentalization (d).
QLS. No internal structures were present when the samples
had not been heated. When such structures are magnified (Fig.
4, E and F), it appears that the inner lipid leaflets are not
always closed in concentric onion-like shells but can make
abrupt bends and several changes of direction. A similar pattern has been previously observed and described as the “honeycomb structure” (Fig. 4G). It occurs in purely lipidic MLV,
when they contain a sufficient amount of lipids with the capacity to form the hexagonal phase (HII), such as, for example, PE.
In that case, the MLV, instead of being onion-like (as in Fig.
4G, a), could become compartmentalized (as in Fig. 4G, c) when
heated through the gel-to-liquid phase transition (44, 48).
Leakage Experiments—A different mode of action of EqtII
and melittin was also confirmed by permeabilization experiments in which they were added to preformed, peptide-free,
POPC/SM 2:1 (mol/mol) MLV loaded with the fluorescent
marker calcein (Fig. 5). Melittin (at L/P of 1) induced ⬃90% of
release of calcein, indicating that, even if applied externally, it
could perturb almost all of the inner layers. EqtII was, at the
same L/P, much less efficient, releasing only less than 20% of
calcein. This indicates that EqtII, even at 100 times higher
toxin/lipid ratios than those used for the NMR experiments,
interacts only with the outermost layers and cannot penetrate
deeply through the multilayered MLV structure.
Effects of EqtII on Lipid Orientation: FTIR—Polarization
FTIR spectroscopy of EqtII in lipid vesicles was used to investigate toxin effects on the orientation of the lipid chains. Oriented lipid layers were obtained by careful deposition of SUV
onto the flat surface of a germanium crystal. Vibrational spectra in the region from 3000 cm⫺1 to 2800 cm⫺1, corresponding
to the stretching of the CH2 and CH3 groups (Fig. 6), provided
the average orientation of the lipid chains with respect to the
normal to the plane of the crystal, which is also the plane of the
deposited membranes. The dichroic ratio and the order parameter were estimated, and from this the average tilt angle was
derived as in Refs. 21 and 49 (Table II). In the case of the lipid
alone, we found 29°, a value that is usual for phospholipid
membranes above the phase transition (42, 50, 51). Upon interaction with EqtII, a decrease in order was observed, leading
to an average angle of around 33° (Table II). To the contrary,
␣-toxin does not increase lipid disorder in PC/cholesterol 1:1
membranes (Table II). Such membranes, although slightly less
ordered than PC/SM, had to be used because cholesterol is
required for ␣-toxin interaction (32). Notably, cholesterol per se
does not prevent the formation of the honeycomb structure (48).
Spectra collected with perpendicularly polarized light were
also used to evaluate the amount of toxin bound to the lipid in
the SUV/toxin samples, via Equation 5. L/P values of 115 ⫾ 15
and 300 ⫾ 20 were obtained for EqtII and ␣-toxin, respectively.
FIG. 5. Permeabilizing capacity of EqtII and melittin on MLV.
The release of calcein from POPC/SM 2:1 (mol/mol) MLV is shown. The
concentration of both lipids and lysins (melittin or EqtII) was 1 ␮M. The
measurement was done in a final volume of 1 ml of NMR buffer.
The arrows indicate the addition of 1 mM Triton X-100 to obtain the
signal corresponding to the maximal release.
Compared with the initial value in the mixture, they indicated
that ⬃90 ⫾ 2% of EqtII and 67 ⫾ 5% of ␣-toxin were bound,
which was consistent with the level of hemolytic activity remaining in solution (see “Experimental Procedures”). At variance with our recent work (20), in which L/P values of ⱖ1000
were chosen in order that most of EqtII bound would be still in
the monomeric, nonaggregated form, the L/P used here (⬃100)
warrants that the largest fraction of the bound protein is in the
pore state (13).
The polarization of the ␣-helix component of the amide I⬘ band
of membrane-bound EqtII was calculated. We obtained a dichroic
ratio (R) ⫽ 2.8 (Table II), from which the average angle ␥, formed
by the helix axis with the vertical to the plane of the crystal
surface, was calculated via Equations 1 and 2 using ␪ ⫽ 39° (38).
We obtained ␥ ⫽ 36° (Table II). However, if we consider that the
lipid chains are also forming an angle around the vertical direction and we use Equation 4 to recalculate the approximate tilting
of the ␣-helices with respect to the lipid chains, we obtain an
angle of around 22°. Noting that slightly different values of ␪ for
the EqtII ␣-helix have been reported (38), we can say that an
upper bound for the tilt angle is 35° (as the lower bound of ␪ is
30°). These results are in excellent agreement with pore conductance experiments that we recently reported, indicating that the
average orientation of the amphipathic N-term ␣-helix, with respect to the pore axis, is 21° (20).
Lipid Dependence of Pore Selectivity—The possibility that
mixed lipid/peptide structures occur during pore formation was
tested by measuring the selectivity of the pores when the overall surface charge of the membrane was varied. We found that
Equinatoxin II Forms Toroidal Pores
45221
FIG. 6. Polarized infrared-attenuated total reflection spectra of EqtII in PC/SM layers. PC/SM (1:1) SUV with absorbed EqtII were
deposited onto the surface of a germanium crystal. FTIR spectra were taken with either parallel (0°, solid lines) or perpendicular polarization (90°,
dashed lines). Traces are vertically shifted to improve presentation. Indicated features are C-H stretching (a) O-D stretching (b), C⫽O stretching
(c), amide I⬘ (d), and amide II⬘ (e). In the left inset, the region from 3000 to 2800 cm⫺1 of pure lipids without EqtII is enlarged. Four absorption bands
appear that correspond to the symmetric and antisymmetric stretching of the CH2 (at 2850 cm⫺1 and 2920 cm⫺1) and the CH3 groups (at 2872 and
2956 cm⫺1). In the right inset, the amide I⬘ region of EqtII (after subtraction of the lipid contribution) is enlarged. The best curve fit with Lorentzian
components is superimposed as a thin solid line to the 90° polarized dashed trace. The Lorentzian components used are reported with thin dotted
lines, and the ␣-helix component is evidenced in boldface type.
TABLE II
Assignment and dichroic ratio of some IR bands observed in PC/SM vesicles alone and with EqtII
Wave number
cm
Vibration
Direction (␪)a
EqtII ⫹ PC/SM (1:1)
PC/SM (1:1)
Dichroic ratiob
Form factor (S)
Angle (␥)c
Dichroic ratio
Form factor (S)
Angle (␥)c
0.64
0.61
28° ⫾ 1°
30° ⫾ 1°
1.25 ⫾ 0.01
1.24 ⫾ 0.01
2.8 ⫾ 0.01
0.55
0.56
0.48
0.78e
33.5° ⫾ 1°
33° ⫾ 1°
36° ⫾ 2°
22° ⫾ 2°e
38° ⫾ 1°
39° ⫾ 1°
1.38 ⫾ 0.02
1.43 ⫾ 0.02
⫺1
2920
2850
1651
as CH2 stretchingd
s CH2 stretchingd
amide I ␣-helix
90°
90°
30°
1.18 ⫾ 0.01
1.20 ⫾ 0.01
2920
2850
as CH2 stretching
s CH2 stretching
90°
90°
1.42 ⫾ 0.02
1.44 ⫾ 0.02
␣-Toxin ⫹ PC cholesterol (1:1)
PC cholesterol (1:1)
0.41
0.40
0.45
0.40
37° ⫾ 1°
39° ⫾ 1°
a
Direction of the variation of the dipole moment associated with the vibration with respect to the direction of the main molecular axis (aliphatic
chain or ␣-helix axis).
b
Data are averages ⫾ S.D. from two series of independent experiments.
c
Average angle between the direction of the molecular axis and the perpendicular to the crystal plane (i.e. the membrane plane).
d
s, symmetric; as, antisymmetric.
e
Recalculated according to the tilt angle between the direction of the ␣-helix axis and the average direction of the aliphatic chains.
increasing contents of negatively charged lipids of the membrane (by adding PA) increased also the cation selectivity of the
EqtII pore (Fig. 7), suggesting that lipid head groups are indeed
exposed into the pore lumen.
DISCUSSION
We observed that EqtII induces the appearance of an isotropic signal in 31P NMR spectra when mixed toxin-lipid MLV are
cycled through higher temperatures (Figs. 2 and 3). Similar
results were recently reported by another group (29) and attributed to the capacity of EqtII to induce the formation of an
SM-enriched phase and to promote vesiculation. Here we used
a wider range of techniques and experimental conditions that
allowed us to propose a different model for toxin action.
QLS and EM consistently indicated that EqtII affects the
organization of the lipid lamellae within the interior of the
vesicles, rather than changing the overall size of the multilamellar vesicles (Table I and Fig. 4). It was important to distinguish these two effects, since both could lead to the isotropic
signal. EM pictures were particularly informative, since they
suggested that EqtII was able to induce the formation of a
honeycomb structure in the MLV (Fig. 4), which is consistent
with all of our experimental findings. The honeycomb structure
FIG. 7. Selectivity of EqtII channels in membranes of different
composition. After channel formation by EqtII in a PLM, a 10-fold
gradient of KCl was established, and the reversal voltage (Vrev) was
determined by finding the potential necessary to clamp the current to
zero. Vrev (which ranged from 32 to 39 mV) was converted into a permeability ratio (P⫹/P⫺) by the Goldman-Hodgkin-Katz equation. Means ⫾
S.E. of four or five experiments are presented. PLM were composed of 20%
SM, the reported amount of PA, and, for the rest, DPhPC.
45222
Equinatoxin II Forms Toroidal Pores
FIG. 8. Proposed mechanism of pore formation by EqtII in lipid membranes. a, the toxin binds to the lipid bilayer with its amphipathic
N-terminal ␣-helix still folded as it is in solution. b, the N-terminal ␣-helix unfolds and inserts into the lipid layer, remaining flat on the membrane
surface, with the hydrophobic part embedded in the membrane and the hydrophilic part in contact with the solution. c, after aggregation, the
helices insert into the membrane and promote lipid reorganization into a toroidal pore.
was originally described for MLV containing lipids with a propensity to form the hexagonal phase HII and lipids forming the
lamellar phase, heated above their main transition temperature (44, 48). In that case, the MLV can go into a state (state c
of Fig. 4G), in which small pieces of HII-like phase are formed
at the contact points. These do not represent an extended HII
phase, and in fact, in 31P NMR spectra, they do not give rise to
a typical HII signal but rather to the superposition of an isomorphic signal onto the lamellar one (44, 48). This is exactly
what we have observed with EqtII-containing MLV, even with
lipids like PC or SM, which do not have a propensity to form HII
phase (43). Actually, in the absence of EqtII, the isomorphic
phase was irrelevant even at high temperature.
When such MLV are cooled again, they go back to state b of
Fig. 4G, with a reversibility that is almost complete in the case
of EqtII-treated MLV. State b is a highly compartmentalized
lamellar structure fully compatible with what we have seen by
EM and NMR. The presence of EqtII pores on each leaflet
ensures high membrane permeability to the negative stain,
thus explaining why all internal compartments appear black in
the EM. In state b, large lamellae are still present, providing
the 31P NMR lamellar signal. The lamellae are not always
closed around the single compartments, but they go from one
spot to another, frequently changing direction. Such changes
are evident in the EM enlargement of Fig. 4F. The overall
structure is an agglomerate of compartments retaining, as a
whole, a big size even at high temperature, consistent with the
QLS indication that MLV are not cleaved into smaller structures, as with melittin. The polygonal shape of these compartments, peculiar for the honeycomb structure, is also evident
(Fig. 4, E and F).
Until now, the honeycomb structure was observed only in the
presence of lipids with the propensity to form HII phase. Therefore,
one important question is how EqtII can induce such a propensity
in lipids (PC and SM) that do not have it. A possible explanation
comes from a model of actinoporin pore formation that we have
recently introduced (30), which assumes that these toxins perturb
the lamellar organization of the lipid by favoring the formation of a
toroidal lipid pore in the bilayer (9, 52). The essential steps of this
model are illustrated in Fig. 8. Upon binding to the lipid bilayer
(Fig. 8a), the toxin unfolds its amphipathic N-term ␣-helix. This
helix would absorb into the lipid layer but remains flat on its
surface, with the hydrophobic part embedded in the membrane and
the hydrophilic part in contact with the solution (Fig. 8b). Upon
aggregation of several monomers, the helices might insert through
the membrane and simultaneously reorganize the lipid itself to
form a toroidal pore (Fig. 8c). Thereby, a hydrophilic pore lined by
protein and partially by lipid is formed. This is confirmed by the
fact that the cation selectivity of the pore increases if a negatively
charged lipid is present in the membrane (Fig. 7) that attracts
cations, increasing their local concentration. Notably, whereas the
helix in the pore is almost perpendicular to the plane of the membrane (see FTIR results in Fig. 6), it is still lying at the interface
FIG. 9. Proposed mechanism of MLV compartmentalization induced by EqtII. A, toxin pores are created at random on the different
lamellae. B, at contact points, pores may coordinate to form a tubular
channel structure extending through the lamellae. C, by acquiring
again conformation b of Fig. 8, the toxin may favor the formation a
cylindrical lipid structure identical to the elements forming the HII lipid
phase. D, a further transition of the protein to the inserted form (i.e.
form c of Fig. 8) can finally lead to the formation of compartments. Note
that the protein is always in one of the two possible bound states (b or
c of Fig. 8) and that the helix is always lying at the water lipid interface.
between lipid and water, being in the lumen of the toroidal lipid
structure (Fig. 8c).
With a higher amount of absorbed protein and with the help
of higher temperature, the toxin may lead to compartmentalization as depicted in Fig. 9. Toxin pores, present on the different lamellae, are initially randomly distributed (Fig. 9A),
but under appropriate circumstances, they may coordinate at
given spots to form a kind of local trans-lamellar aggregate
(Fig. 9B). Examples of long range coordination of toroidal pores
in MLV structures have been already reported (10). New channel-like structures, extending through the lamellar structure
from one lamella to the other, may thus form. These structures
can be regarded as loci of generation of nonlamellar structure.
In fact, simply by reversing the toxin conformational transition
that has led to the formation of the toroidal pore (step b to c of
Fig. 8 reversed to b again), it is possible to create a layer of
toxin adsorbed inside a cylindrical lipid structure (Fig. 9C),
which is essentially identical to the contact HII phase of Fig. 4G
(although depicted here in a parallel section). Such formation of
small HII-like cylinders, or tethers, may occur simultaneously
at many places in the MLV as in the proposed honeycomb
model and thus originate the observed 31P NMR isomorphic
signal. A further transition of the protein to the inserted form
(i.e. form c of Fig. 8) could easily lead to the formation of
Equinatoxin II Forms Toroidal Pores
compartments (Fig. 9D) by going through a phase similar to
the inverted micellar intermediate and the stalks (43, 53, 54).
Although Fig. 9 exemplifies the events only in a twodimensional representation, it is easy to conceive that the
sequential formation and coalescence of many of these tethers, or stalks, could eventually originate the typical threedimensional compartmentalization of the honeycomb structure (Fig. 4E). In a similar way, inverted micellar
intermediates were observed to coalesce and give rise to
extended structures like rod micellar intermediates and line
defects in pure membranes (43). A similar mechanism of
action has been proposed also for F-actin, which establishes
toroidal pores and lipid nanotubes (55), and for Bax-type
apoptotic proteins (56).
This model is fully consistent with the FTIR results shown in
Fig. 6. One implication of the toroidal toxin-lipid pore is that
the aliphatic chains of the lipids in the torus, instead of being
oriented perpendicular to the plane of the membrane, assume
all orientations between perpendicular and parallel. Our polarization experiments indicate that, upon formation of the pore, a
fraction of the lipid becomes disordered, which can be calculated in ⬃8% by the decrease of the order parameter S (Table
II). Using the determined L/P of 115, we can thus estimate that
around nine lipid molecules are disordered by each EqtII monomer. We can see that this is indeed consistent with the toroidal model. In fact, if we calculate the number of lipid molecules
that could fit into the nonoriented toroidal structure, assuming
the thickness of the membrane is 6 nm, the radius of the
aqueous pore is 1 nm (14), and the volume of the hemitorus of
disordered lipid is V ⫽ ␲a2(␲c ⫺ 4/3a) (where a is the radius of
the disk that, rotating, generates the torus and c is the radius
of rotation), we get a figure of 50 – 60. This corresponds well to
the total number of lipids disordered by the toxin in each pore,
if we remember that one pore is formed by a tetrameric aggregate of the toxin (12, 13, 57).
The alternative explanation that lipid disorder could derive from
the effect of the toxin portion protruding outside the bilayer is less
likely. In fact, ␣-toxin, which forms a ␤-barrel pore inserted perpendicularly through the lamellar membrane (45) (Fig. 3), does not
increase lipid disorder in FTIR analysis (Table II), even if the
portion protruding outside the bilayer is larger than that of the
actinoporin (seven monomers of ⬃270 amino acid residues (45)
instead of four monomers of ⬃150 residues (20)). One reason for
this difference could be that ␣-toxin uses ␤-structure to create the
hole through the membrane, whereas EqtII uses an ␣-helix (like
magainin, for which the toroidal pore was originally introduced) (9,
58). The ␣-helix of EqtII shares structural homology with melittin
and fusogenic viral peptides (17). In our conditions, melittin behaved different from EqtII and more like a detergent that produces
isomorphic structures (SUVs or bicelles). However, it has been
shown that, under appropriate conditions of lipid composition, L/P,
and temperature, also melittin can induce toroidal pores (10). In
addition, many other related ␣-helical peptides, including fusogenic
viral ones, have a similar lipid-perturbing activity (26–28, 59).
Acknowledgments—We thank Luka Malenšek for assistance with the
electron microscope. We are also indebted to Alekos Athanasiadis, who
first suggested that we investigate the effects of EqtII on the phase
morphology of lipids in vesicles.
REFERENCES
1. Menestrina, G., Dalla Serra, M., and Lazarovici, P. (eds) (2003) Pore-forming
Peptides and Protein Toxins, Taylor & Francis Group, London, UK
2. van der Goot, F. G. (ed) (2001) Pore-forming Toxins, Springer Verlag, Berlin
3. Heuck, A. P., Tweten, R. K., and Johnson, A. E. (2001) Biochemistry 40,
9065–9073
4. Lesieur, C., Vécsey-Semjén, B., Abrami, L., Fivaz, M., and van der Goot, F. G.
(1997) Mol. Membrane Biol. 14, 45– 64
5. Zasloff, M. (2002) Nature 415, 389 –395
6. Heller, W. T., Waring, A. J., Lehrer, R. I., and Huang, H. W. (1998) Biochemistry 37, 17331–17338
45223
7. Chugh, J. K., and Wallace, B. A. (2001) Biochem. Soc. Trans. 29, 565–570
8. Shai, Y., and Oren, Z. (2001) Peptides 22, 1629 –1641
9. Matsuzaki, K., Sugishita, K., Ishibe, N., Ueha, M., Nakata, S., Miyajima, K.,
and Epand, R. M. (1998) Biochemistry 37, 11856 –11863
10. Yang, L., Harroun, T. A., Weiss, T. M., Ding, L., and Huang, H. W. (2001)
Biophys. J. 81, 1475–1485
11. Anderluh, G., and Maček, P. (2002) Toxicon 40, 111–124
12. Varanda, A., and Finkelstein, A. (1980) J. Membr. Biol. 55, 203–211
13. Belmonte, G., Pederzolli, C., Maček, P., and Menestrina, G. (1993) J. Membr.
Biol. 131, 11–22
14. Tejuca, M., Dalla Serra, M., Alvarez, C., Potrich, C., and Menestrina, G. (2001)
J. Membr. Biol. 183, 125–135
15. Athanasiadis, A., Anderluh, G., Maček, P., and Turk, D. (2001) Structure 9,
341–346
16. Hinds, M. G., Zhang, W., Anderluh, G., Hansen, P. E., and Norton, R. S. (2002)
J. Mol. Biol. 315, 1219 –1229
17. Belmonte, G., Menestrina, G., Pederzolli, C., Križaj, I., Gubenšek, F., Turk, T.,
and Maček, P. (1994) Biochim. Biophys. Acta 1192, 197–204
18. Anderluh, G., Barlič, A., Podlesek, Z., Maček, P., Pungerčar, J., Gubenšek, F.,
Zecchini, M., Dalla Serra, M., and Menestrina, G. (1999) Eur. J. Biochem.
263, 128 –136
19. Anderluh, G., Pungerčar, J., Križaj, I., Štrukelj, B., Gubenšek, F., and Maček,
P. (1997) Protein Eng. 10, 751–755
20. Malovrh, P., Viero, G., Dalla Serra, M., Podlesek, Z., Lakey, J. H., Maček, P.,
Menestrina, G., and Anderluh, G. (2003) J. Biol. Chem. 278, 22678 –22685
21. Menestrina, G., Cabiaux, V., and Tejuca, M. (1999) Biochem. Biophys. Res.
Commun. 254, 174 –180
22. Malovrh, P., Barlič, A., Podlesek, Z., Menestrina, G., Maček, P., and Anderluh,
G. (2000) Biochem. J. 346, 223–232
23. Hong, Q., Gutierrez-Aguirre, I., Barlič, A., Malovrh, P., Kristan, K., Podlesek,
Z., Maček, P., Turk, D., González-Mañas, J. M., Lakey, J. H., and Anderluh,
G. (2002) J. Biol. Chem. 277, 41916 – 41924
24. Tejuca, M., Dalla Serra, M., Ferreras, M., Lanio, M. E., and Menestrina, G.
(1996) Biochemistry 35, 14947–14957
25. Hori, Y., Demura, M., Niidome, T., Aoyagi, H., and Asakura, T. (1999) FEBS
Lett. 455, 228 –232
26. Lohner, K., Staudegger, E., Prenner, E. J., Lewis, R. N. A. H., Kriechbaum, M.,
Degovics, G., and McElhaney, R. N. (1999) Biochemistry 38, 16514 –16528
27. Pereira, F. B., Valpuesta, J. M., Basañez, G., Goñi, F. M., and Nieva, J. L.
(1999) Chem. Phys. Lipids 103, 11–20
28. Liu, F., Lewis, R. N. A. H., Hodges, R. S., and McElhaney, R. N. (2001)
Biochemistry 40, 760 –768
29. Bonev, B. B., Lam, Y. H., Anderluh, G., Watts, A., Norton, R. S., and Separovic,
F. (2003) Biophys. J. 84, 2382–2392
30. Alvarez, C., Dalla Serra, M., Potrich, C., Bernhart, I., Tejuca, M., Martinez, D.,
Pazos, I. F., Lanio, M. E., and Menestrina, G. (2001) Biophys. J. 80,
2761–2774
31. Maček, P., and Lebez, D. (1988) Toxicon 26, 441– 451
32. Forti, S., and Menestrina, G. (1989) Eur. J. Biochem. 181, 767–773
33. Rance, M., and Byrd, R. A. (1983) J. Magn. Reson. 52, 221–240
34. Mayer, L. D., Hope, M. J., and Cullis, P. R. (1986) Biochim. Biophys. Acta 858,
161–168
35. Santos, N. C., and Castanho, M. A. R. B. (1996) Biophys. J. 71, 1641–1650
36. Kayalar, C., and Düzgünes, N. (1986) Biochim. Biophys. Acta 860, 51–56
37. Ferreras, M., Höper, F., Dalla Serra, M., Colin, D. A., Prévost, G., and Menestrina, G. (1998) Biochim. Biophys. Acta 1414, 108 –126
38. Tamm, L. K., and Tatulian, S. A. (1997) Q. Rev. Biophys. 30, 365– 429
39. Goormaghtigh, E., Raussens, V., and Ruysschaert, J. M. (1999) Biochim.
Biophys. Acta 1422, 105–185
40. Axelsen, P. H., Kaufman, B. K., McElhaney, R. N., and Lewis, R. N. A. H.
(1995) Biophys. J. 69, 2770 –2781
41. Harrick, N. J. (1967) Internal Reflection Spectroscopy, Harrick Scientific Corp.,
Ossining, NY
42. Tamm, L. K., and Tatulian, S. A. (1993) Biochemistry 32, 7720 –7726
43. Seddon, J. M. (1990) Biochim. Biophys. Acta 1031, 1– 69
44. Cullis, P. R., Hope, M. J., de Kruijff, B., Verkleij, A. J., and Tilcock, C. P. (1985)
in Phospholipids and Cellular Regulation (Kuo, J. F., ed) Vol. 1, pp. 1–59,
CRC Press, Inc., Boca Raton, FL
45. Song, L., Hobaugh, M. R., Shustak, C., Cheley, S., Bayley, H., and Gouaux,
J. E. (1996) Science 274, 1859 –1866
46. Dufourcq, J., Faucon, J. F., Fourche, G., Dasseux, J. L., Le Maire, M., and
Gulik-Krzywicki, T. (1986) Biochim. Biophys. Acta 859, 33– 48
47. Pott, T., Paternostre, M., and Dufourc, E. J. (1998) Eur. Biophys. J. Biophys.
Lett. 27, 237–245
48. de Kruijff, B., Verkleij, A. J., van Echteld, C. J., Gerritsen, W. J., Mombers, C.,
Noordam, P. C., and de Gier, J. (1979) Biochim. Biophys. Acta 555, 200 –209
49. Menestrina, G. (2000) in Bacterial Toxins, Methods and Protocols (Holst,
O., ed) pp. 115–132, Humana Press, Totowa, NJ
50. Hübner, W., and Mantsch, H. H. (1991) Biophys. J. 59, 1261–1272
51. Mueller, E., Giehl, A., Schwarzmann, G., Sandhoff, K., and Blume, A. (1996)
Biophys. J. 71, 1400 –1421
52. Epand, R. M. (1998) Biochim. Biophys. Acta 1376, 353–368
53. Siegel, D. P. (1984) Biophys. J. 45, 399 – 420
54. Yeagle, P. L. (1997) Curr. Top. Membr. 44, 375– 401
55. Grigoriev, P. A., Tarahovsky, Y. S., Pavlik, L. L., Udaltsov, S. N., and Moshkov,
D. A. (2000) IUBMB Life 50, 227–233
56. Basanez, G., Sharpe, J. C., Galanis, J., Brandt, T. B., Hardwick, J. M., and
Zimmerberg, J. (2002) J. Biol. Chem. 277, 49360 – 49365
57. De Los Rios, V., Mancheno, J. M., Martinez del Pozo, A., Alfonso, C., Rivas, G.,
Onaderra, M., and Gavilanes, J. G. (1999) FEBS Lett. 455, 27–30
58. Matsuzaki, K. (1998) Biochim. Biophys. Acta 1376, 391– 400
59. Bechinger, B., Kinder, R., Helmle, M., Vogt, T. C. B., Harzer, U., and Schinzel,
S. (1999) Biopolymers 51, 174 –190