INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION APPENDIX REFERENCES

External electric fields of short duration across intact cells and vesicular systems leads to transient membrane pores through which influx/efflux of impermeable molecules is believed to occur (Zimmermann, 1986; Neumann et al., 1989; Chang et al., 1992; Potter, 1988). These events are generally referred to as electroporation or electropermeabilization and have led to a number of other related processes such as electrofusion for the production of somatic cell hybrids (Al-Atabee et al., 1990) and electroinsertion for the specific incorporation of proteins in cellular membranes (Mouneimne et al., 1991) as well as recent clinical applications in targeted transdermal drug delivery and tumor treatment (Prausnitz et al., 1993; Heller et al., 1996). The induction of pores on the membrane surface of some given number, size(s), and lifetime(s) is one common feature underlying all these processes and ultimately governs the relative success of the various methods used. Consequently, significant focus has been directed toward a basic understanding of the underlying mechanisms involved.

Field-induced membrane pores have been reported on a number of cell types under a variety of experimental conditions, e.g., red blood cells and ghosts (Chang and Reese, 1990; Sowers and Lieber, 1986; Sowers, 1987), mammalian and plant cells (Tekle et al., 1990, 1991, 1994; Gabriel and Teissie, 1997; Mehrle et al., 1985, 1989), sea urchin eggs (Hibino et al., 1991, 1993; Kinosita et al., 1988), and bacterial cells (Miller et al., 1988). Irrespective of the cell type examined to date, by far the most consistent finding has been the eminent rupture of the membrane provided a critical transmembrane potential, _c (~0.3-1 V), develops across the membrane, where _c refers to the minimum amplitude required for macroscopically observable electroporation. For non-porated membranes, the magnitude of the sub-critical transmembrane potential, _i, due to an external electric field, E, is given by (Neumann and Rosenheck, 1973)

(1)

where b is the outer radii of the cell membrane and  is the angle between the field direction and any point on the membrane. The maximum potential drop at the poles of the membrane facing the electrodes (i.e., cos() = ±1 at  = , 0) and the cosine dependence have generally been confirmed by a number of studies employing potential-sensitive dyes (Ehrenberg et al., 1987).

By contrast, the density of these pores as well as their size and distribution on the perforated membrane surface is less well understood, and their structure is even far less known. One clear exception to the latter is an earlier study by Chang and Reese (1990) where a rapid freezing electron microscopy was used to show volcano-shaped pores in erythrocyte membranes exposed to an intense electric pulse. These pore structures were visually detectable ~3 ms after the pulse and resealed to their initial state in several seconds with no observable residue. The spatial disposition of the pores relative to the electroporating pulse was, however, not resolved. Nevertheless, based on these early findings and further modeling studies (Weaver and Barnett, 1992; Kakorin et al., 1996), it is currently thought the initial electroporation event involves the rearrangement of clusters of lipids into hydrophobic and hydrophilic pore structures with minimum pore size in the order of a few nanometers. Estimates of the number of pores per cell, derived from conductance and transport studies, have ranged from just a few (<10) (Saulis et al., 1991), ~700 (Sowers and Lieber, 1986) pores in ghost cells to up to ~10⁵ in mouse B cells (Neumann et al., 1998). Percolation of these nanometer-sized primary pores into larger openings have also been envisaged (Sugar and Neumann, 1984). Much of the insight into how and where these pores are distributed on the membrane surface has largely been inferred from transport studies involving influx/efflux of probe molecules (Dimitrov and Sowers, 1990; Tekle et al., 1990; Mehrle et al., 1989; Gabriel and Teissie, 1998) as well as from optical signals of membrane-bound potential-sensitive dyes (Hibino et al., 1991, 1993; Kinosita et al., 1988). Some features of the experimentally observed permeabilization patterns have recently been shown in a series of modeling studies of asymmetric electroporation (DeBruin and Krassowska, 1999a,b). However, although influx/efflux and sieving experiments are powerful methods in probing membrane pore dynamics, it is crucial to note these approaches may not necessarily indicate the type of pores created or how they may have distributed at the membrane level, as the overall transport process could depend on the nature of the probe molecules used, the mode of transport, and the kinetics of pore closure, all of which, or in some combination, could lead to potential misinterpretations. This possible anomaly, for instance, was originally anticipated by Saulis (1993) in a modeling study of asymmetric electroporation. Our earlier study (Tekle et al., 1994) had also shown selective transport of various fluorescent markers on one side of the membrane despite the fact that both hemispheres of cell membrane were permeabilized. This earlier finding implied structurally different pores may have formed at the opposite poles of the membrane, although no direct experimental evidence was available in support of the hypothesis. It is thus clear that current views on the structure, dynamics, and density of field-induced membrane pores are limited, and further studies with a particular focus to events occurring at the membrane level would be useful. In this respect, vesicles offer a relatively less complex membrane structure and have been used as model systems in a few earlier studies (Teissie and Tsong, 1981; Zhelev and Needham, 1993), including more recent ones (Tonsing et al., 1997; Kakorin et al., 1998; Kakorin and Neumann, 1998). Small unilamellar vesicles (SUVs, diameter ~100 nm) were employed in these studies, and no spatial information was thus obtained.

Here, giant vesicles (diameter >10 µm) of 1,2-di-oleoyl-sn-glycero-3-phosphocholine (DOPC) were used to monitor the induction and distribution of pores at the membrane level. A combination of membrane staining, electron microscopy, and Ca²⁺ influx measurements were employed to study the overall dynamics. The results revealed single visible pores (on the order of ~7 µm) are predominantly formed at the cathode-facing hemisphere and many small pores at the anode-facing hemisphere. Analysis of the fluorescence pattern of the stained membrane as well as electron micrographs of negatively stained vesicles shows pore formation is accompanied by loss of lipid from the membrane surface. Despite the difference in the pore size distribution, however, the overall porous area at both vesicle hemispheres was found to be comparable. The pores are resealed within a few hundred milliseconds after the pulse. These findings are compared and discussed with currently held views on asymmetric electroporation.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION APPENDIX REFERENCES

Reagents

DOPC in chloroform was purchased from Avanti Polar Lipids (Alabaster, AL). 1-(3-sulfonatopropyl)-4-[B][2-(di-n-butylamino)-6-naphthyl vinyl] pyridinium betaine (Di-4-ANEPPS) and Fluo-3 penta-ammonium salt were from Molecular Probes (Eugene, OR). CaCl₂ was from Fisher (St Louis, MO).

Preparation of vesicles

Vesicles were prepared following the procedure of Angelova et al. (1992) using low-amplitude AC fields and modified to allow exchange of hydrating liquid. A chamber was constructed using two conducting indium-tin-oxide-coated microscope cover glasses (Delta Technologies, Stillwater, MN), separated with a 1-mm-thick Teflon spacer and was fitted with a simple flow system to allow exchange of hydrating liquids. A 2-µl sample from a 20 mg/ml stock solution of DOPC in chloroform was placed on the upper glass electrode and dried extensively under a stream of nitrogen. The chamber was then reassembled and the lipid hydrated using either deionized water alone or a solution containing the desired dyes, i.e., di-4-ANEPPS or Fluo-3. The initial hydration was done by directly injecting 200 µl of the solution near the dried lipid with a micropipette. An AC field of ~70 V/cm, frequency 5 Hz, was then applied across the chamber from a waveform generator (WaveTek 183, San Diego, CA), and the formation of the vesicles was monitored under the microscope (Zeiss, Axiovert 100). Within ~1 h, several layers of large vesicles, diameter ~25 µm, were formed on the surface of the hydrated lipid. Inner layers contained progressively smaller vesicles that grew with time up to ~2 h. Before the removal of the vesicles, the bathing solution was replaced with dye-free medium to remove unincorporated dye. Membrane labeling with di-4-ANEPPS was primarily done after the vesicles have formed by bathing the vesicles in a solution containing the dye, unless stated otherwise. The larger vesicles on the outer layer were finally collected by gentle agitation of the bathing solution with a micropipette. Typical yield of vesicles of diameter b, 20 µm < b < 30 µm, was 200-300 per 100 µl of solution. These vesicle preparations were stable for several days at 4°C. In all the experiments reported here, the outside concentrations of the reagents used were, 5 µM di-4-ANEPPS, 25 µM Fluo-3, and 2.5 mM CaCl₂.

Electroporation and image acquisition equipment

The electroporation and image acquisition system is similar to that described previously (Tekle et al., 1991, 1994). Electric pulses of desired amplitude and a pulse width were supplied from a high-power pulse generator (model 360, Velonex, Santa Clara, CA) and applied across vesicles placed in an electroporation chamber. The chamber was constructed with two polished stainless steel electrodes, fixed on a 75- × 25-mm microscope glass slide. The inter-electrode separation was 1 mm. The amplitude and pulse widths for each experiment were monitored across the chamber on a digital oscilloscope (54502A, Hewlett-Packard, Palo Alto, CA) using a 100:1 probe. The chamber is mounted on an inverted fluorescence microscope (Zeiss, Axiovert-100), and the excitation and emission wavelengths were set with appropriate bandpass and cutoff filters at 520 and 610 nm for red emission and 480 and 520 for green emission, respectively. Fluorescence images were acquired with an intensified camera (SIT-68, DAGE-MTI, Michigan City, IN) and recorded onto a VHS tape (Panasonic model AG-7350). The intensity of the fluorescence emissions was also monitored on an analog scope (Hitachi, V-212) to protect the camera from excessive light exposure as well as to monitor against saturation of the measured signals. Images were later processed using the National Institutes of Health Image 1.62 software (National Institutes of Health, Bethesda, MD) on a PowerMac 9500 computer. Because the electroporation events were fast and often complete within only a few video frames, it was necessary to identify t = 0 by synchronizing the video signal with the rising edge of the electroporating pulse to eliminate timing uncertainty. A useful pulse/video synchronization circuit was built for this purpose (circuit diagram available upon request). The trigger signal is the odd/even field signal from an LM 1881 sync separator (National Semiconductor, Santa Clara, CA), which is driven by the composite video signal from the video camera. When this flip-flop is triggered, it provides an output to the pulse generator to start the experiment in synchrony with the video signal. The circuit is further designed to stamp a bright vertical bar near the upper left corner of the video image and vanishes at the instant of the output trigger. This triggering system accurately locates the video frame in which the electric pulse is applied so that the time of electric pulse application is included on the video frame recording of the experiment. Successive video frames could then be time resolved at 33 ms per frame.

Electron microscopy

Vesicle preparations were negatively stained following a similar procedure described by Moscho et al. (1996). Grids coated with a parlodion/carbon substrate were treated with poly-D-lysine (1 mg/ml), rinsed, air dried, and glow discharged just before use to make them hydrophilic. Vesicles were prepared as described earlier here, and a drop of vesicle suspension was placed on Parafilm. The grid was inverted on the drop for ~2 min. Excess liquid was removed by blotting onto a filter paper at the periphery of the grid. The grid was then negatively stained with 3 drops of 1% aqueous uranyl acetate. Excess staining solution was similarly removed using filter paper. Vesicle images were then obtained using a Philips-410 transmission electron microscope at different magnifications.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION APPENDIX REFERENCES

Our primary aim in this study was to monitor the induction and distribution of field-induced vesicle pores at the membrane level. For this purpose, a widely used potential-sensitive dye, di-4-ANEPPS, was used for its excellent membrane partitioning properties and photo-stability. It should be noted, however, that this dye has been used to monitor fast changes (microseconds time scale) in membrane potential due to externally applied fields (Lojewska et al., 1989), but these optical responses are well below the time resolution (milliseconds time scale) of our equipment here. In the present study, the dye is used as a membrane stain to visualize the rupture events and as a marker for lipid components lost from the vesicle membrane. This stated use is justified given our experimental design and setup.

Fig. 1 shows a typical series of video frames depicting the electroporation events before and after the application of a 1.1-kV/cm (Fig. 1, A-C) or 0.920kV/cm in (Fig. 1, D-F) field and a pulse width of 700 µs. For all frames, the electrodes are positioned as shown in Fig. 1 A. The membrane of the vesicle shown in Fig. 1, A-C, was labeled after the vesicle was formed. For the vesicle shown in Fig. 1, D-F, the dye was present during vesicle formation so that both the inner and outer leaflets of the membrane were labeled to avoid any asymmetric charge distribution. Fig. 1 A is the image before field application, and Fig. 1 B shows the first frame (33 ms) after the pulse. It shows that a large pore with a diameter ~6.2 µm was formed at the negative electrode-facing hemisphere, whereas a rare occurrence of a possible visible lipid aggregate loss was seen at the hemisphere facing the anode. In the next frame (66 ms, Fig. 1 C), the membrane pore was resealed, partially leaving some of the lipid on the outer surface. In Fig. 1, D-F, the time sequence was the same as in Fig. 1, A-C, and the pattern of permeabilization is representative of the successful experiments observed. Here again we found a large pore, diameter ~7.5 µm (Fig. 1 E), on the side facing the cathode, which subsequently resealed in the next frame, (66 ms, Fig. 1 F). These data clearly show that symmetric pores were not created in these vesicle systems. Further experiments and analysis revealed that both sides of the membrane are permeabilized, and a partial loss of membrane lipid occurs as a consequence of pore formation. The asymmetry of electroporation is reflected here in the size distribution of pores at the vesicle hemispheres facing the electrodes. Evidence supporting these events is detailed below.

View larger version (45K): [in this window] [in a new window]   FIGURE 1   A select series of video frames showing the electroporation events before and after electric field application. E = 1.1 kV/cm in (A-C), and E = 0.920 kV/cm in (D-F). The pulse width was 700 µs for both cases. The position of the electrodes is as indicated in A for all frames. In A-C, the membrane was labeled after the vesicle had formed. In D-F, the dye was present during vesicle formation so that both the inner and outer leaflets of the membrane are labeled to avoid any asymmetric charge distribution. (A and D) Images before field application; (B and E) The first frames (33 ms) after the pulse; (C-F) Images at 66 ms after the pulse. The frame images are pseudo colored to enhance visualization of the membrane pores and dye distribution after the field. Scale bar, 10 µm

Based on a number of image data as those shown in Fig. 1, the vesicle dimensions (diameters) were carefully measured before (see for example, Fig. 1, A and D) and immediately after (Fig. 1, C and F) the electric field. Vesicle dimensions were measured using contour templates from the vesicle images and are the averages of two readings taken perpendicular to each other. These measurements revealed that a significant reduction occurs in the size of the vesicles after the field and clearly suggest appreciable amount of lipid must have been lost as a result of pore formation. Surprisingly, however, we find that the loss of membrane lipid from the observed pores at the cathode-facing hemispheres (Fig. 1, B and E) could not account for the extent to which the vesicle size is reduced when measured after the pulse (Fig. 1, C and F). This discrepancy is elaborated below using the image data in Fig. 1, D-F, as an example.

Figure. 2 shows a geometric representation of the image data in Fig. 1 E, where r_p is the pore radius and the shaded area, S = 2r²(1  cos ), represents the surface loss due to pore formation at the cathode-facing hemisphere. If the reduction in vesicle size (Fig. 1 F) is solely due to lipid loss at the cathode side, then

(2)

where r₁ and r₂ are the radii of the vesicles before (Fig. 1 D) and after (Fig. 1 F) the field, respectively, and  is given by

(3)

The observed pore in Fig. 1 E should thus equal r_p, where the pore radius r_p (Fig. 2) is given by

(4)

The vesicle radii r₁ (Fig. 1 D) and r₂ (Fig. 1 F) were measured and found to be 13.4 and 12.7 µm, respectively. Using Eqs. 3 and 4, we find r_p = 8.1 µm. However, the pore radius (Fig. 1 E) is only ~3.8 µm, about half the value of the computed pore radius, r_p. Table 1 shows further comparisons of the pore radii, r_p, calculated based on measured vesicle sizes before and after the electric pulse with the pore radii observed, r_obs, at the cathode-facing hemisphere for different vesicle preparations. It is interesting to note that the observed pore radii are consistently about half of that of r_p. In terms of total lipid lost from the vesicle membrane, these data suggest at least two possibilities: either 1) additional lipid components must have been lost at other sites on the membrane surface or 2) the pore size at the cathode-facing hemisphere (Fig. 1 E) could have been larger at earlier times than can be detected with the current equipment (note that images captured in Fig. 1, B and E, are ~32 ms after the termination of the external pulse.). The latter possibility seems highly unlikely as it would imply the vesicle would have to lose almost half of its surface on the hemisphere facing the cathode to accommodate a pore radius of ~8.1 µm. More likely, many smaller pores are formed at the anode-facing hemisphere resulting in loss of lipid undetectable by the microscope setup.

View larger version (12K): [in this window] [in a new window]   FIGURE 2   Geometrical representation of a vesicle. The shaded area represents the membrane pore as seen in Fig. 1, B and E. x = r(1  cos). The area of the shaded region is 2rx, where rp is the pore radius. The electric field is as shown and is directed from left to right.

Lipid loss at the anode-facing hemisphere is in part implicated based upon the asymmetric distribution of the membrane-staining dye. Before field application, for example, the dye distribution is fairly homogeneous as shown in Fig. 1, A and D, but a marked asymmetry ensues in the images after the field with increased fluorescence at the anode-facing hemisphere (see Fig. 1, B, C, E, and F). These distributions remain for hundreds of seconds after the pulse and could not have been due to potential redistribution around the vesicle as the external field is absent in these time scales. A plausible cause for the uneven dye signal could arise from redistribution of the lipid bilayer itself as a consequence of pore formation such that small lipid aggregates failing to reintegrate to the original bilayer remain attached to the parent vesicle surface. This possibility is supported by electron micrograph images of porated vesicles and explains the persistent enhanced fluorescence long after the pulse has been terminated. The data further point at the site origin of the missing lipid responsible for the observed reduction in the vesicle size shown earlier.

Fig. 3 shows negatively stained electron micrographs of vesicles before (Fig. 3 A) and immediately (~2 min) after (Fig. 2, B and C) electroporation. Fig. 2, A', B', and C' are magnified (zoomed) views of the membrane in Fig. 2, A, B, and C, around the area indicated by the arrows. Clearly, major rearrangement and undulation occurs on the membrane surface as shown in Fig. 2, B' and C'. All the images are from the same vesicle preparation. The field and pulse width were 5.7 kV/cm and 700 µs, respectively. The comparisons in the electron microscope data were gathered from a subpopulation of vesicles with diameters ~1.5-5 µm. Many of the giant vesicles (diameter 5 µm) were unavailable and possibly disintegrated into the lipid aggregates and other undefined structures seen under the electron microscope. This was observed in both the control and pulsed samples and is due to the blot drying involved in preparing the negatively stained grid samples (see Materials and Methods).

View larger version (112K): [in this window] [in a new window]   FIGURE 3   Electron micrographs of vesicles stained with 1% uranyl acetate before (A) and after (B, and C) electric field application. E = 5.7 kV/cm; pulse width = 700 µs. A', B', and C' are magnified views of the area indicated by the arrow. Electron micrograph magnifications: ×24,000 (A), ×8700 (B), and ×4400 (C).

The results shown in Fig. 1 and Fig. 3 establish lipid loss at both hemispheres and that a clear pore is observed at the cathode-facing hemisphere. Whether pores were formed and subsequently resealed at the anode-facing hemisphere was finally assessed by monitoring Ca²⁺ influx from a medium containing 2.5mM CaCl₂ into vesicles loaded with Ca²⁺ indicator, Fluo-3. Intact vesicles are impermeable to Ca²⁺ present in the suspending medium. Fig. 4 shows both the rate of Ca²⁺ influx and the spatial permeabilization pattern after an electroporating field of 500 V/cm, 700-µs duration. As shown in the inset of Fig. 4, Ca²⁺ influx occurs at both the anode- and cathode-facing sites indicated by the enhanced fluorescence of Ca²⁺-Fluo-3 complex. The rate of Ca²⁺ influx is shown in the intensity versus time plot of Fig. 4. The data points were obtained by integrating the fluorescence intensity inside the vesicle from successive video frames at the indicated times and shows a rapid rise followed by a plateau within ~100-200 ms after the pulse. We should mention here that the plateau was neither caused by saturation of the indicator dye nor due to saturation of the camera's detection limit. The observed time course could thus indicate the rate of pore closure of the perforated membrane. The solid line in Fig. 4 is a fit to the data assuming a simple diffusion of Ca²⁺ ions through a time-dependent pore of radius r = r(t) (see Appendix).

View larger version (16K): [in this window] [in a new window]   FIGURE 4   Resealing of field-induced membrane pores. The vesicle loaded with Fluo-3 was porated at 500 V/cm, 700-µs pulse duration, in the presence of 2.5 mM CaCl2. The fluorescence intensity from successive video frames as that shown in the inset was integrated and plotted. The solid line is a two-parameter fit based on Eq. 8. The time of resealing from the fit was ~164 ms.

Taken together, results presented here show that the pore distribution at the membrane level is asymmetric with large pores at the cathode- and small but numerous pores at the anode-facing hemisphere. The creation of these pores is accompanied with loss of lipid from the membrane surface. We should mention here that the data shown represent results from the majority of the cases we have looked at. In general, unilamellarity of the vesicles was not checked independently but was assessed only by the sharpness of the membrane stain. Heavily multi-layered vesicles, including some with patchy stained surfaces, were easily identifiable, as were those containing other smaller vesicles within them, and were not used for the experiments. In some cases, vesicles simply disintegrated upon field application. However, this does not appear to correlate with the strength of the applied field used as many others were found to sustain much higher field strengths. In general, experiments were carried out at ~1-1.5 V over what we estimated to be a critical permeabilization potential, _c, of 648 mV. Fig. 5 shows a linear plot of Eq. 1, whose slope gives _c. The data points in the figure were determined from the minimum field strength experimentally required to observe Ca²⁺ influx into Fluo-3-loaded vesicles. The electric field strength was incremented at 50 V/cm.

View larger version (8K): [in this window] [in a new window]   FIGURE 5   Determination of the critical permeabilization potential, c. The critical potential is defined as the minimum potential required for observable pore formation. This value was determined from three vesicle sizes based on the transport of Ca2+ ions into Fluo-3-loaded vesicles. The electric field was incremented at 50 V/cm. The field strength is plotted against the inverse of the vesicle radius and considering cos() = 1 (see Eq. 1). The slope of the fitted line gives c = 648 mV.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION APPENDIX REFERENCES

The effect of an externally applied electric field on the integrity of cell membranes in many cell types and vesicles, as well as planar lipid bilayer systems, has been studied by means of a wide variety of methods, including conductometric, optical, and rapid freezing techniques. The most consistent finding in all systems investigated has been the rupture of membranes provided sufficient field strength is applied. Elegant studies using sea urchin eggs (Hibino et al., 1993) and hemispherical bilayer vesicles (Lojewska et al., 1989) have shown that the temporal and spatial variations of the membrane potential due to external fields behaves as predicted by theory. However, how and where the membrane ruptures to challenges of an excess potential have produced differing results and interpretations.

Induction of membrane pores at some critical potential implies that both hemispheres of the vesicle (cell) facing the electrodes be permeabilized. To date, an apparent symmetric permeabilization has been shown only when bipolar pulses were used (Tekle et al., 1991). The asymmetry often observed in many transport studies is believed to originate due to a resting transmembrane potential in cells (negative inside), thereby making the hemisphere facing the anode more susceptible to rupture. However, this same hypothesis was used to explain the susceptibility of the cathode-facing hemisphere, because poration on one side can conceivably double the potential on the opposite side. Teruel and Meyer (1997) proposed unequal distribution of charged phospholipid components at the membrane level to explain asymmetric Ca²⁺ entry in depolarized cells, and more recently ionic concentration gradients have been suggested to influence the creation of asymmetric pore populations (DeBruin and Krassowska, 1999a). Despite these interesting propositions, the asymmetry in electroporation still remains even under conditions where large fields are applied that could easily overwhelm the relatively small transmembrane potentials across cell membranes (Tekle et al., 1990) or even under conditions where, as shown in the present study, neither a resting membrane potential, uneven lipid distribution, nor ionic species exist in the vesicle preparations used. This last point suggests other additional factors are probably involved in determining the perforation pattern.

It is likely the pore distribution observed on the time scale of our detection system results from a complex set of lipid rearrangements that continues after the external field is turned off. One possible initial source of asymmetry may, however, reside on how the external electric field interacts with the intrinsic dipole potentials in the inner and outer layer of the membrane bilayer (Loew, 1993). The dipole potential is the potential difference between the center of the bilayer and the membrane/water interface, and its magnitude has been reported to be of the order of ~300-450 mV (Gross et al., 1994; Reyes et al., 1983). The electric field generated from these dipoles is directed from the center of the bilayer toward the surface of the inner and outer lipid layers. Considering that there would be more lipids on the outer leaflet than on the inner leaflet due to packing constraints, it is possible to imagine a net field could exist directed toward the vesicle surface. If this mechanism is operative, it could result in a larger potential drop at the cathode-facing hemisphere compared with that of the anode and may play a part in the induction of the large pores observed at the negative electrode. Although this is currently speculative, it should be possible to test this hypothesis by using factors that could modulate the dipole potential. Additionally, faster detection systems would also offer capturing events at earlier times and shade light to the evolution of the pore structures observed here.

The processes leading to the induction of pores and the resulting structure and distribution found here may in fact be common to other previously studied cellular systems for a number of reasons. As shown in Fig. 5, the magnitude of the critical potential required for permeabilization (~0.6 V) measured by Ca²⁺ influx, is similar to that found in many cellular systems, typically in the range of 0.5-1.0 V. Secondly, influx studies of a number of fluorescent indicator dyes of varying charge and size has previously been shown to be selective (Tekle et al. 1994; Gabriel and Teissie, 1997). These selective transport patterns are consistent with the pore distributions found here at the membrane level. In other systems where asymmetry is observed, it is important to note that it may not necessarily indicate the state of pore population at the membrane level. Transport through a single large pore versus that which takes place through many small pores spread over a wide region may lead to the mistaken conclusion that one side is more porated than the other. As is shown in the analysis of the lipid loss due to pore formation in the present system, the total porous area at the two hemispheres could be quite similar despite the fact that the size of the pores is markedly different (Fig. 1 and Table 1). It is, however, possible that such equivalence may not exist at times earlier than we can detect with our system.

The other important finding in this study is the demonstration that lipid loss is associated with pore formation (Figs. 1 and 3 and Table 1). This has not been observed in cellular systems and may be due to the presence of cytoskeletal and other protein components that may serve as anchors to stabilize the lipid domains of the membrane. The electron micrographs of Fig. 3 clearly show that massive restructuring of the membrane occurs due to electroporation. Although some lipid components are lost, other small cluster of aggregates remain attached to the parent vesicle membrane. We should mention here that analogous loss of lipid components due to external electric field has been implicated in an earlier study of electric birefringence of a three-component, sodium bis(2-ethylhexyl)sulfosuccinate/isooctane/H₂O, reverse micelle system (Tekle and Schelly, 1994).

In summary, we have presented data on the induction and distribution of membrane pores in DOPC vesicles showing asymmetric distribution of pores at the membrane level with large single pores at the cathode- and many small pores at the anode-facing hemisphere, with resultant partial loss of the membrane lipid. The porous areas at the two hemispheres were, however, comparable, and the pores were shown to reseal within a few hundred milliseconds after the pulse.