Lipid Phase Coexistence Favors Membrane Insertion of Equinatoxin-II, a Pore-forming Toxin from Actinia equina *

Equinatoxin-II is a eukaryotic pore-forming toxin be-longing to the family of actinoporins. Its interaction with model membranes is largely modulated by the presence of sphingomyelin. We have used large unilamellar vesicles and lipid monolayers to gain further information about this interaction. The coexistence of gel and liquid-crystal lipid phases in sphingomyelin/phosphatidylcholine mixtures and the coexistence of liquid-ordered and liquid-disordered lipid phases in phosphatidylcholine/choles-terol or sphingomyelin/phosphatidylcholine/cholesterol mixtures favor membrane insertion of equinatoxin-II. Phosphatidylcholine vesicles are not permeabilized by equinatoxin-II.

Equinatoxin II (Eqt-II) 1 is a member of the actinoporins, a group of sea anemone cytolysins (1). It is a 179-amino acid residue protein with a molecular mass of 19.8 kDa and an isoelectric point of 10.5 (2). Its three-dimensional structure has been solved by x-ray crystallography and NMR (3,4). Eqt-II forms cation-selective pores with a diameter of ϳ2 nm in cell and model membranes (5)(6)(7). The mechanism of pore formation is a multistep process consisting of (i) membrane binding of the water-soluble monomer, (ii) oligomerization on the membrane surface, and (iii) pore formation (1,(5)(6)(7)(8)(9)(10)(11). This mechanism is common to other actinoporins like sticholysin-II from Stichodactyla helianthus (12,13). Membrane insertion of Eqt-II and sticholysins is favored by the presence of sphingomyelin within the target membrane (6, 8, 14 -16). The recent finding of a phosphocholine binding site in the three-dimensional structure of sticholysin-II (13) supports the role of sphingomyelin as a specific receptor for actinoporins, as other authors have suggested (17,18). However, the presence of sphingomyelin is not strictly necessary for the lytic activity of these toxins, which are also active in phosphatidylcholine/cholesterol mixtures (14,16). Therefore, other factors are likely to govern their mechanism of action.
Mixtures of sphingomyelin, phosphatidylcholine, and cholesterol are characteristic of the so-called rafts, microdomains in which the concentration of membrane components (lipids or proteins) and their physicochemical properties are different from the surrounding environment. The increasing amount of information pointing to the existence of lipid domains in cell and model membranes and their implication in many crucial biological processes has been extensively reviewed (19 -26). One important characteristic of rafts is their resistance to detergent solubilization (27)(28)(29)(30). This property is associated with the fact that lipids in rafts exist in the liquid-ordered (L o ) phase, where their acyl chains are extended and ordered as in the gel phase but possess lateral and rotational mobilities characteristic of the liquid-disordered (L d ) phase (31,32). In monolayers, bilayers, and animal cell membranes, L o and L d fluid phases are immiscible (33). Other examples of lipid phase coexistence are known, the most common one being probably the coexistence of gel and fluid phases in certain lipid mixtures or in pure lipid bilayers near the gel-fluid transition temperature (34).
In the present work, we have analyzed a variety of parameters that determine the formation of distinct lipid phases (lipid composition, temperature, presence of different sterols, and enzymatic activity of phospholipase C). A strong correlation was found between the coexistence of lipid phases and the pore-forming activity of Eqt-II. Epifluorescence microscopy imaging of supported lipid monolayers revealed the preferential localization of this eukaryotic toxin at the interface between lipid phases.
Eqt-II Purification-Eqt-II was purified from the liquid exuded by Actinia equina specimens freshly collected in the Bay of Biscay. We followed the purification protocol described in Ref. 2. The purified protein was concentrated to ϳ10 mg/ml with an Amicon 8050 concentrator (Danvers, MA) ultrafiltration unit equipped with a regenerated nitrocellulose filter (Millipore Corp., Bedford, MA) with a molecular mass cut-off of 10 kDa. Aliquots were stored at Ϫ20°C, and once thawed they were not refrozen. Protein concentration was estimated spectrophotometrically using a molar extinction coefficient at 280 nm of 3.61 ϫ 10 4 M Ϫ1 cm Ϫ1 (35).
Labeling of Equinatoxin II with Texas Red TM -To a 190 M solution of Eqt-II in distilled water, 130 l of 1 M NaHCO 3 were added to raise the pH to 8.3. The labeling reaction was started by the addition of 200 l of TR stock solution (5 mg/ml) (final Eqt-II/TR molar ratio of 5:1). The mixture was incubated for 60 min at room temperature with constant stirring and protected from light. To inactivate any remaining free dye, 47 l of hydroxylamine were added to the mixture, and the solution was stirred for an additional 30 min at room temperature. To purify the labeled protein, the mixture was loaded on a Sephadex G-15 column and eluted with 10 mM Hepes, 200 mM NaCl, pH 7.5. 500-l fractions were collected, and absorption spectra from 250 to 650 nm were measured. Protein concentration and the degree of labeling were determined as indicated by the manufacturer of the protein labeling kit.
Leakage of Liposomal Contents-The appropriate lipids were mixed in organic solvent, evaporated thoroughly, and resuspended in 10 mM Hepes 200 mM NaCl, pH 7.5, containing 25 mM ANTS and 90 mM DPX. Large unilamellar vesicles (LUV) were prepared by the extrusion method (36), using polycarbonate filters with a pore size of 0.1 m (Nucleopore, Pleasanton, CA). Nonencapsulated fluorescent probes were separated from the vesicle suspension through a Sephadex G-75 gel filtration column (Amersham Biosciences). Solution osmolarities were checked with an Osmomat 030 instrument (Gonotec, Berlin, Germany). Phospholipid concentration was measured according to Bartlett (37).
The leakage of encapsulated solutes was assayed as described by Ellens et al. (38). The probe-loaded liposomes (final lipid concentration ϭ 0.1 mM) were treated with the appropriate amounts of Eqt-II in a fluorometer cuvette at 25°C with constant stirring. Changes in fluorescence intensity were recorded in a PerkinElmer Life Sciences LS-50 spectrofluorometer (Beaconsfield, UK) with excitation and emission wavelengths set at 350 and 510 nm, respectively. An interference filter with a nominal cut-off value of 470 nm was placed in the emission light path to minimize the contribution of the light scattered by the vesicles to the fluorescence signal. The percentage of leakage was calculated after the complete release of the fluorescent probe by the addition of the nonionic detergent Triton X-100 (final concentration ϭ 0.1% w/v). When PLC was used, the assay was carried out under optimal conditions for its activity; buffer was 10 mM Hepes, 200 mM NaCl, 10 mM CaCl 2 , pH 7.5, and the experiment was carried out at 37.6°C with constant stirring. Concentrations were 0.1 mM, 0.3 M, and 1.5 units/ml for lipid, Eqt-II, and PLC, respectively. To stop the enzyme reaction, o-phenantroline was added at a final concentration of 6 mM.
Surface Pressure Measurements-Surface pressure measurements were carried out with a MicroTrough-S system from Kibron (Helsinki, Finland) at 25°C with constant stirring. The aqueous phase consisted of 1.1 ml of 10 mM Hepes, 200 mM NaCl, pH 7.5. The lipid, dissolved in chloroform/methanol (2:1), was gently spread over the surface. The desired initial surface pressure was attained by changing the amount of lipid applied to the air-water interface. After 10 min to allow for solvent evaporation, the protein was injected through a hole connected to the subphase. The final protein concentration in the Langmuir trough was 1 M. The increment in surface pressure versus time was recorded until a stable signal was obtained.
Supported Phospholipid Monolayers-Monolayers were formed by spreading chloroform/methanol (3:1, v/v) solutions (1 mM) of the phos-pholipid mixture on top of a buffered subphase (10 mM Hepes, 200 mM NaCl, pH 7.5) in a thermostated Langmuir-Blodgett ribbon trough (NIMA Technologies, Coventry, UK). To allow the observation by epifluorescence microscopy, 1% (mol/mol) of NBD-PC was included. After 10 min to allow for solvent evaporation, monolayers were compressed at 25 cm 2 /min to an initial surface pressure of 20 mN/m. After 10 min for equilibration, TR-labeled Eqt-II from a 45 M buffered stock solution was injected into the subphase. Insertion was followed by monitoring the increase in surface pressure. The surface pressure stabilized at 25 mN/m, and at this point, the monolayer was transferred onto glass coverslips at a velocity of 5 mm/min. The ribbon trough was provided with a feedback mechanism that kept the surface pressure constant by compressing the monolayer, thereby compensating the loss of material that took place during the transfer.
Epifluorescence microscopy observation of the planar supported monolayers was carried out with a Zeiss Axioplan II fluorescence microscope (Carl Zeiss, Jena, Germany). Images from NBD-labeled phospholipid and TR-labeled protein were recorded separately from the same sample by switching fluorescence filters to select the proper emission wavelength range. The experiment was carried out at 25°C.

RESULTS
Interaction of Eqt-II with SM-PC Mixtures-In most cases, the interaction of Eqt-II and other actinoporins with model membranes requires the presence of SM in the target membrane (6,15). To gain more information on the interaction of Eqt-II with model membranes, we prepared LUV composed of SM and PC in different proportions. Protein-vesicle interaction was monitored through the release of fluorescent dyes that had been entrapped in the vesicles. In Fig. 1A, we observe a strong dependence of the release of encapsulated ANTS/DPX on the SM content of the vesicles. For SM molar fractions between 0.3 and 0.7, the percentages of leakage ranged from 44 to 54%, and maximum leakage was obtained when the mixture was approximately equimolar. When one of the two phospholipids predominated, the release was reduced to ϳ25%, and for LUV made of 100% PC or 100% SM, the release was close to 8%.
Next, we prepared lipid monolayers with PC/SM mixtures to determine whether this behavior could be observed in other model membranes. The initial surface pressure ( 0 ) was set at 20 mN/m, and we measured the increase in surface pressure (⌬) after injection of 1 M Eqt-II into the aqueous subphase (Fig. 1A). The insertion followed the same pattern observed in LUV: 1) maximum values for ⌬ (between 12 and 14 mN/m) were observed when the molar fractions of PC and SM were similar; 2) when one of the lipids predominated, ⌬ was reduced to 8 -9 mN/m, and 3) for 100% PC and 100% SM monolayers, ⌬ was practically the same (5.3 and 5.5 mN/m, respectively).
We also measured the "critical pressure" ( c ) for different PC/SM mixtures (i.e. the initial surface pressure ( 0 ) above which no ⌬ is observed after injection of Eqt-II into the subphase). In all of the lipid compositions tested, the higher the 0 , the smaller the ⌬, because tighter lipid packing prevented protein insertion (Fig. 1B). The critical pressures were calculated by linear fitting of the experimental ⌬ versus 0 (initial surface pressure) values and extrapolation to ⌬ ϭ 0. Fig. 1C shows the c values obtained for different SM/PC mixtures. Again, the highest c values (around 36 mN/m) were observed in mixtures approximately equimolar and decreased when one of the lipids was predominant. This means that at intermediate SM molar fractions there are more binding sites available for Eqt-II, and more protein molecules are able to penetrate the monolayer. The mixtures producing maximum penetration were also associated with the maximum values of proteininduced leakage from LUV (Fig. 1A). The gel to liquid-crystal phase transition temperatures (T m ) for PC and SM are Ϫ5°C (39) and 38°C (40), respectively. A detailed phase diagram of egg PC and bovine brain SM has been published (39). According to those data, at our experimental temperature (25°C), pure egg PC exists in the fluid lamellar phase. The addition of SM gives rise to a (PC ϩ SM) fluid lamellar phase plus a pure SM gel phase. Phase separation is clear above 33 mol % SM, and above 90 mol % SM only the gel phase occurs. Our data ( Fig. 1) show maximum protein insertion and maximum bilayer permeabilization at SM mol % between 30 and 70 (i.e. in the phase diagram region where phase separation predominates). In addition, binary phase diagrams at 23°C of palmitoyloleoyl phosphatidylcholine and palmitoyl sphingomyelin mixtures also showed gel/fluid coexistence at palmitoyl sphingomyelin proportions between 30 and 70 mol % (41). Thus, it seems that the coexistence of gel and fluid lipid phases favors the degree of protein insertion and the extent of vesicle permeabilization.
The Effect of Temperature-We studied the effect of temperature on the EqT-II-induced release of fluorescent solutes en-capsulated in PC/SM (1:1) LUV (Fig. 2). The highest percentages of leakage were observed between 11 and 25°C. At temperatures higher than 25°C, the toxin activity started to decrease, a trend that was even more pronounced above 30°C. Above 40°C, the leakage was reduced to 13%. One possible explanation for this behavior would be a potential thermal destabilization of the protein. In a control experiment, we measured the fluorescence of 7.9 M ANS in the presence of 2.6 M Eqt-II as a function of temperature (data not shown). Between 16 and 54°C, the ANS fluorescence remained low and constant, an indication that in this temperature interval the folding of the protein was compact (42). Therefore, the decrease in EqT-II activity shown in Fig. 2 was not due to protein denaturation. However, a partial phase diagram for mixtures of SM and PC of natural origin (43) reveals that, above 32°C, only the fluid phase exists. The egg PC/bovine brain SM phase diagram (39) also suggests that at 37°C complete miscibility of these two lipids occurs. In addition, the phase diagram for the equimolar mixture of palmitoyl sphingomyelin/palmitoyloleoyl phosphatidylcholine shows that above 35°C only the liquiddisordered phase exists (41). Thus, in our opinion, the temperature-dependent decrease in protein activity results from changes in the membrane structure associated with an increased lipid miscibility and the disappearance of coexisting lipid phases.
Interaction of Eqt-II with PC/Cholesterol Mixtures-Actinoporins can also permeabilize PC-cholesterol membranes (14,16). In Fig. 3A, we represent the kinetics of Eqt-IIinduced leakage of ANTS/DPX encapsulated in LUV made of PC-cholesterol (70:30). At mammalian physiological temperatures, there was almost no leakage. However, at lower temperatures, the release increased, and at 4°C it was 22%. Fixing the temperature at 4°C, the extent of leakage increased with the cholesterol content of the model membrane (Fig. 3B). Thus, the effect of cholesterol is temperature-and concentration-dependent.
In lipid monolayers formed at an initial pressure of 20 mN/m, the increase in surface pressure after injecting 1 M Eqt-II into the subphase depends linearly on the amount of cholesterol (Fig. 4A). We also observed a small increment in the critical pressures (Fig. 4B).
The effects of cholesterol are not as conspicuous as those of SM but might be related with a similar feature (i.e. the formation of different lipid phases within the model membrane). At low temperatures, cholesterol is likely to interact with the phospholipid to form an L o lipid phase in coexistence with the bulk fluid phase (44 -46), thereby favoring the lytic action of Eqt-II.
Interaction of Eqt-II with SM-PC/Cholesterol Mixtures-The L o phase has intermediate physical properties between the gel and the liquid-crystalline phases, and it has been related to the sphin- Final release values were 75, 59, and 45% for the liposomes containing ergosterol, cholesterol, and cholestenone, respectively. Therefore, the permeabilizing activity of Eqt-II seems to correlate with the ability of the different sterols to create lipid domains within the model membrane. This dependence was observed over a wide range of lipid/protein ratios for the three lipid compositions (data not shown).
The insertion of Eqt-II into lipid monolayers was also dependent on the ability of the lipid mixture to form domains (Fig. 7B). Combined Action of Phospholipase C and Equinatoxin-II-The lack of a measurable permeabilizing activity does not necessarily mean that the protein is not interacting with the membrane. For instance, although PC LUV are refractory to Eqt-II-induced permeabilization, the toxin partitions into PC LUV (14). Toxin binding to the membrane is probably just one step in the permeabilizing process. Moreover, the interaction of Eqt-II with PC monolayers at initial surface pressures below 25 mN/m gives rise to an increase in the surface pressure (Fig. 4B).
In order to explore conditions that could render PC vesicles susceptible to Eqt-II-induced permeabilization, we studied its activity in the presence of PLC from Bacillus cereus. PLC is a phosphodiesterase that hydrolyzes PC and generates diacylglycerol (DAG) and phosphorylcholine. When PLC or Eqt-II was added to PC LUV, there was no protein-induced leakage of ANTS/DPX (Fig. 8A, lower traces). However, when the vesicles were preincubated during 5 min with Eqt-II, the subsequent addition of PLC gave rise to the release of the fluorophore after a lag period of 2 min (Fig. 8A, upper trace). This leakage was dependent on PLC activity, since the addition of o-phenantroline, a specific inhibitor of PLC (52), abolished this effect (Fig.  8B). In a control experiment, we observed that LUV of PC-DAG (9:1) were not permeabilized by Eqt-II (data not shown), thus indicating that the homogeneously distributed DAG does not render the vesicles susceptible to the lytic activity of the toxin. It has been proposed that it is the localized generation of DAG-rich domains within the outer leaflet of the membrane what promotes vesicle aggregation and fusion without leakage of encapsulated solutes (53)(54)(55). The time interval comprised between the addition of o-phenantroline and the arrest of leakage is probably due to the slow access of the inhibitor (56). 2 The time length for o-phenantroline to produce maximum inhibition (i.e. tens of seconds) is several orders of magnitude longer than the time required for DAG to diffuse out of the asymmet-  (57,58). Lateral diffusion is even faster, with "hopping" or "jumping" events occurring with a frequency of about 10 Ϫ7 s (59). Therefore, the inhibition of PLC activity results in an arrest of Eqt-II insertion, because DAG diffusion overcomes DAG generation and interdomain interfaces are blurred out.
Fluorescence Microscopy of Transferred Monolayers-Finally, we tried to visualize the localization of Eqt-II after insertion into monolayers known to contain coexisting domains. For this purpose, we labeled the toxin with the fluorescent marker TR. Labeling of the toxin did not affect its hemolytic activity or its ability to insert into monolayers (data not shown). We built a monolayer composed of SM/PC/cholesterol (50:15:35), which also included 1% NBD-PC, a fluorescence-labeled lipid that is excluded from ordered phases and accumulates into disordered regions of the film. The initial surface pressure was 20 mN/m, and the toxin was injected directly into the subphase (10 mM Hepes, 200 mM NaCl, pH 7.5). Protein insertion originated an increase in the surface pressure, which stabilized at 25 mN/m. At this point, the monolayer was transferred onto a glass support while keeping constant the surface pressure. The transferred monolayer was placed under the fluorescence microscope, and the selection of the fluorescence filter permitted the visualization of either the NBD-PC or the TR-labeled toxin in the same preparation. Fig. 9, A-C, shows the distribution of NBD-PC in the film. Two phases coexist; the dark area corresponds to an ordered phase from which the fluorescent probe is excluded, and the bright area corresponds to a disordered phase where NBD-PC accumulates. Fig. 9, D-F, also shows the topological distribution of the fluorescently labeled Eqt-II. Although a fraction of the protein fluorescence can be observed in the form of dispersed bright spots located in both lipid phases, equinatoxin-II binds preferentially at their interface. DISCUSSION At 25°C, when the gel and liquid-crystal phases coexist in the SM/PC mixtures (39,41), the lytic activity of Eqt-II shows a marked dependence on the SM/PC ratio in the model membrane; it is maximum when the mixture is approximately equimolar and decreases when one of the two components predominates (Fig. 1, A and B). On the other hand, in LUV made of SM/PC (1:1), the Eqt-II-induced release of ANTS/DPX shows a strong dependence on temperature. The extent of permeabilization is markedly reduced above 25°C (Fig. 2). This is not the result of protein denaturation, because its hydrophobic core remains inaccessible to ANS in the temperature interval where changes in activity are recorded. 3 Above 40°C, when the two lipids are totally miscible (39,43), the percentage of leakage is practically the same as that obtained with either 100% PC or 100% SM. Although specific interactions between actinoporins and SM (6, 15) cannot be excluded, it is the coexistence of lipid phases that seems to modulate the interaction of the toxin with the membrane.
The addition of cholesterol to PC model membranes enhances the lytic activity of Eqt-II (Figs. 3 and 4). The same effect has been described for sticholysin-II, a related actinoporin from S. helianthus (16). The phase diagrams of mixtures of different phosphatidylcholines with cholesterol obtained by a variety of experimental approaches indicate the existence of two immiscible fluid phases (33, 41, 44, 46, 60 -62). At 37°C, Eqt-II is unable to permeabilize LUV made of PC/cholesterol (70:30), probably because the lipids are organized in one uniform L d phase. At 4°C, which is close to the transition temperature of PC (39), the mixtures containing 15-30% cholesterol fall into the phase diagram region where two immiscible lipid phases coexist (44), and the toxin resumes its lytic activity.
The incorporation of cholesterol to PC/SM mixtures gives rise to lipid compositions typically associated with membrane rafts (for a review, see Ref. 63). A ternary phase diagram has recently been published for palmitoyloleoyl phosphatidylcholine/palmitoyl sphingomyelin/cholesterol mixtures at different temperatures (41). In our experimental conditions, the SM/PC/ cholesterol (50:15:35) mixture clearly lies in the liquid-ordered/ liquid-disordered coexistence region mixture and is particularly sensitive to the action of the toxin as evidenced by the large increment observed in the critical pressure permitting insertion of the protein, which increased from 36.8 mN/m (in the PC/SM equimolar mixture) to 46.8 mN/m. The specific interactions of cholesterol with saturated phospholipids might give rise to condensed complexes (47,64,65) in which lipids undergo an area contraction that permits the accommodation of more protein molecules into the monolayer. A close relationship between condensed complexes, liquid-ordered phases, and rafts has been established (66). The use of sterols distinct from cholesterol also highlights the correlation between phase coexistence and lytic activity of Eqt-II (Fig. 7). Whereas the presence of ergosterol increases the extent of permeabilization in LUV and the critical pressure in monolayers, cholestenone induces opposite effects. Ergosterol is a fungal sterol that promotes tight packing of saturated phospholipids and domain formation (49,50), whereas cholestenone does not interact with 3 I. Gutiérrez-Aguirre, unpublished observation. sphingomyelin, does not induce domain formation (49,51), and restores detergent solubilization (30).
The combined action of PLC and Eqt-II further illustrates the relevance of lipid phase coexistence for Eqt-II activity (Fig.  8). PC LUV are not permeabilized by Eqt-II or PLC alone. However, the PLC-induced local accumulation of DAG creates conditions for the permeabilization of the otherwise insensitive PC LUV. Hønger et al. (67) have found a direct correlation between the total area occupied by the interfaces between gel and fluid lipid domains and the activity of phospholipase A 2 . Moreover, Nielsen et al. (68), using atomic force microscopy, have shown that the activity of phospholipase A 2 on 1,2-dipalmitoyl-sn-glycero-3-phosphocholine monolayers originates 3-5-Å deep depressions in the membrane that are interpreted as areas where the product of its catalytic activity (i.e. lyso-PC) concentrates. The edges between the intact bilayer and the product-enriched domains must facilitate the accessibility of phospholipase A 2 (and presumably also of PLC) to the region of the lipid molecule that undergoes its catalytic activity and, therefore, may play a role in triggering the activity burst (54).
The strong correlation between Eqt-II activity and the coex-istence of lipid phases could be the result of the accumulation of the protein at the interface between immiscible lipid phases. In monolayers, this specific localization is favored when the protein does not show a defined preference for any of the different phases (69). The free energy per length associated to the boundaries between liquid phases is referred to as line tension, and it has been proposed that if such an interface exists in cell membranes, it is likely to be decorated with specific proteins and/or lipids (33). Our epifluorescence experiment with TR-labeled Eqt-II in monolayers confirms that the protein shows a certain preference to bind at the boundaries between ordered and disordered regions (Fig. 9). If coexistence of immiscible phases is comparable in monolayer and bilayer systems, as has been recently suggested (62), it seems that the accumulation of the protein at the boundaries between gel-fluid or liquid-ordered/ liquid-disordered lipid phases could precede membrane permeabilization. Lipid packing defects and differences in membrane thickness occurring at these interfaces (70 -73) might facilitate the interaction with the protein. Most likely, this interaction is governed by a structural motif that binds phosphocholine and is conserved among actinoporins (13).
FIG. 9. Epifluorescence microscopy images of Eqt-II inserted into a SM/PC/Chol monolayer. The initial surface pressure of the SM/PC/Chol/NBD-PC (50:14:35:1) monolayer was initially set at 20 mN/m. After the injection of the TR-labeled protein into the subphase, the surface pressure stabilized at 25 mN/m. At this point, the monolayer was transferred onto a glass support at 5 mm/min while keeping the surface pressure constant. The experiment was carried out at 25°C. The subphase was 10 mM Hepes, 200 mM NaCl, pH 7.5. A-C, different frames of the monolayer viewed through an NBD filter (emission at 520 nm). D-F, the same frames viewed through a TR filter (emission at 590 nm). Scale bars, 50 m.
A number of receptors for pore-forming toxins are components of lipid rafts and therefore permit the accumulation of toxins in two dimensions and provide a mechanism that facilitates the oligomerization of the toxin prior to pore formation (for a review, see Ref. 74). Association with the interfaces between domains is an even more efficient concentration strategy because it confines the toxin to a linear space where oligomerization and pore formation can take place at very low protein bulk concentrations. Moreover, lipid molecules at interfaces might be intrinsically more disordered, perhaps offering less resistance to protein insertion. In the case of equinatoxin-II, the insertion is limited to its N-terminal ␣ helix (9,10). This process can be compared with the insertion of a wedge into a fracture line and would expose momentarily the hydrophobic cores of immiscible lipid phases. However, protein insertion gives rise to a concomitant increase in the surface pressure of the outer monolayer of the membrane (9), which may push adjacent lipid molecules to fill the open gap. Therefore, the insertion of Eqt-II could be associated to the redistribution of lipid molecules around it and might result in the formation of a lipidic pore whose walls might be delimited by the hydrophilic face of its amphipathic N-terminal ␣-helix and the polar head groups of the phospholipids (10). Such a toroidal pore structure has been postulated for actinoporins (10 -13), antimicrobial peptides (75,76), and apoptotic proteins (77,78).