After the equilibration stage for all systems, external electric fields of magnitude E = 0.5 V.nm–1 and 1.0 V.nm–1 were applied in the direction perpendicular to the membrane. Fig. 1 depicts configurations taken from the simulations of model membranes subject to both TM voltages. In all cases, we observe the first of water fingers penetrating the hydrophobic core of the bilayer. As later confirmed by the analysis of the trajectories of all systems, and in agreement with Tieleman's observations (Tieleman, 2004), it appears that these fingers penetrate the bilayer hydrophobic core from either side of the bilayer, regardless of the direction of the applied field. These fingers expand toward the opposite interface or join other water fingers to ultimately form water wires that extend from one interface to the other of the bilayer hydrophobic core (Fig. 1 b). At a later stage, polar lipid headgroups migrate from the membrane-water interface to the interior of the bilayer, forming within hydrophilic pores that surround and stabilize the water columns as reported in the study by Tieleman (2004). These structures of the nonregular shapes of water channels are very different from the putative "cylindrical lipid pores" that are often postulated. This feature is also clear from previous MD simulations of membrane electroporation (Tieleman, 2004) and from MD simulations of permeation of membranes subject to mechanical stress (Leontiadou et al., 2004). Another noticeable fact brought by simulations is that despite the fact that the large water pores, after penetration of the lipid headgroup, are lined by "hydrophilic polar heads", a large fraction of the surface accessible to the solvent is in contact with lipid acyl chains (cf. Fig. 1 e).
Two bare fully hydrated bilayers containing, respectively, 64 and 256 lipid molecules were considered in this study. Both systems underwent the same process, underlining the fact that that there is no size effect on the observed perturbations. The overall characteristics of the lipid bilayer—A, the average area occupied per DMPC molecule, and d, the system's repeat distance—appear not to be much perturbed as long as large water pores are not formed (cf. Fig. 2). As expected, a noticeable expansion of the membrane is observed subsequent to the large water pores formation. Note that A and d are somewhat coupled since water migrating from the bulk phase (decrease of d) penetrates the lipid core region (increase of A).
The kinetics of the membrane penetration of water fingers and of lipid headgroups is highly dependent, as expected, on the intensity of the applied field. Whereas at 0.5 V.nm–1 the first water fingers develop within a nanosecond (others in the case of large systems may appear later in the run), the same process is much faster (200 ps) for the 1.0 V.nm–1 run. Migration of the headgroups toward the interior of the bilayer takes longer (4 ns and 1 ns, respectively for the bilayer subject to 0.5 V.nm–1 and 1 V.nm–1). The latter may be traced back on Fig. 2, as the effect is a drastic expansion of the lipid patch.
After the fast reorientation of the water molecules dipole under action of the electric field, one notices a slower reorientation of the lipid headgroup dipoles. The reorientation of the phosphatidylcholine dipole during application of electric fields has been observed by electron spin resonance and NMR experiments (Stulen, 1981) and recently by infrared (Le Saux et al., 2001; Miller, 2002), which, in agreement with our data, shows that the conformation of the headgroups and more generally the properties of the interface are greatly affected by electroporation (Robello and Gliozzi, 1989).
To investigate the resealing process, the TM voltage was switched off after the membrane breakdown. Within a few nanoseconds, the lipid bilayers underwent a complete reconstitution. All water molecules forming the channel, as well as the lipid molecules in which headgroups were initially imbedded in the hydrophobic core, were expelled toward the lipid-water interface, and A and d nearly recovered their equilibrium resting state values (Fig. 2). Interestingly enough, the complete resealing of the bilayers for all systems studied occurs for all samples within 2–4 ns. This appears to be the upper limit for resealing of the largest pores formed in this study, i.e., those stabilized by the lipid headgroups, in absence of ions. One expects indeed that for a salt solution, these pores could be stabilized by the interactions between ions and the polar headgroups of the lipids. The lower bound for coalescence time in our study is recorded for the single-file water wires that span the membrane (hydrophobic pores). Those vanish by complete repartitioning at the interface within 250–500 ps depending on the shape of the wires.
The evolution of the intrinsic overall transmembrane electrostatic potential, as electroporation and resealing of the bilayer take place are reported in Fig. 3. Estimates from the MD trajectories show that the increase of to 3 and 6 V across the membrane is quasi-immediate after the application of the external field, as is the drop back to 0 V when the field is switched off. As indicated above, the voltage difference across the system is found, as in the investigation by Tieleman (2004), to satisfy where Lz is the size of the simulation box in the direction perpendicular to the applied field. The changes in the potential difference recorded for the system around the electroporation phenomena (large hole formation) is probably meaningless since such a quantity should depend on the concentration of holes in the system. It is noteworthy that the changes in the potential due to the rearrangement of lipid headgroups, which is taking place at a much longer timescale, as well as the change due to the formation of water wires is very small compared to the large potential drop corresponding to polarization of the water, in agreement with Tieleman's observations (private communication).
Similar responses to the field were witnessed for the peptide nanotube/bilayer system (Fig. 4). However, no pore formation in the vicinity of the peptide channel was observed. The lipid molecules located nearby the peptide are known to strongly hydrogen bond to the peptide as previously reported (Tarek et al., 2003). Such interactions play a significant role in stabilizing those specific lipids. We can therefore speculate that at high membrane protein concentrations, the applied TM voltage necessary to break these hydrogen bonds should increase the electroporation threshold, which corroborates the experiments of Troiano et al. (1999) reporting on membrane/gramicidin stability.
The DNA/lipid system simulation was undertaken starting from a well-equilibrated 12-basepair 5'-cgcgaattcgcg-3' ecor1 DNA duplex placed near a model POPC bilayer. We followed the perturbation of the system under a 1.0 V.nm–1 transverse electric field during 2 ns. During the MD trajectory, several pores formed in the bilayer, and the DNA duplex, the structure of which was hardly modified, diffused toward the interior of the membrane (Fig. 5). Once the DNA migrates to the bilayer core using the water pores beneath as a conduit, it comes in contact with lipid headgroups lining along the boundaries of the pore. At this stage, the interactions between the DNA and the membrane gave rise to a stable DNA/membrane complex as inferred from mediated gene delivery studies (Golzio et al., 2002).
We also considered a second starting configuration of the system where the DNA was displaced laterally. The results were quite different, as the electroporation of the membrane does not produce any water column just beneath the DNA. In this case translocation of the plasmid was not observed. The above results tend to indicate that local electroporation of the bilayer is a requisite to transmembrane transfer of species.