Confinement can induce unusual behavior in the properties of matter. Using molecular dynamics simulations, we show here that water confined to carbon nanotubes of a critical size under ambient conditions (1 bar, 300 K) can undergo a transition into a state having icelike mobility with an amount of hydrogen bonding similar to that in liquid water. The onset of this behavior occurs rapidly, raising the possibility that confinement inside nanotubes, and perhaps even buckyballs, can provide an environment in which the dynamics of phase changes may be studied directly by simulation. Moreover, because of a variety of evidence suggesting that water ordering may modulate proton conductance via a "proton wire" hydrogen bonding network, the ability to modulate water ordering with geometry suggests a possible mechanism for a switchable nanoscale semiconductor.

In this paper we report on the molecular dynamics simulation of a fluid phase hydrated dimyristoylphosphatidylcholine bilayer. The initial configuration of the lipid was the x-ray crystal structure. A distinctive feature of this simulation is that, upon heating the system, the fluid phase emerged from parameters, initial conditions, and boundary conditions determined independently of the collective properties of the fluid phase. The initial conditions did not include chain disorder characteristic of the fluid phase. The partial charges on the lipids were determined by ab initio self-consistent field calculations and required no adjustment to produce a fluid phase. The boundary conditions were constant pressure and temperature. Thus the membrane was not explicitly required to assume an area/phospholipid molecule thought to be characteristic of the fluid phase, as is the case in constant volume simulations. Normal to the membrane plane, the pressure was 1 atmosphere, corresponding to the normal laboratory situation. Parallel to the membrane plane a negative pressure of -100 atmospheres was applied, derived from the measured surface tension of a monolayer at an air-water interface. The measured features of the computed membrane are generally in close agreement with experiment. Our results confirm the concept that, for appropriately matched temperature and surface pressure, a monolayer is a close approximation to one-half of a bilayer. Our results suggest that the surface area per phospholipid molecule for fluid phosphatidylcholine bilayer membranes is smaller than has generally been assumed in computational studies at constant volume. Our results confirm that the basis of the measured dipole potential is primarily water orientations and also suggest the presence of potential barriers for the movement of positive charges across the water-headgroup interfacial region of the phospholipid.

We introduce a new force field (43A1-S3) for simulation of membranes by the Gromacs simulation package. Construction of the force fields is by standard methods of electronic structure computations for bond parameters and charge distribution and specific volumes and heats of vaporization for small-molecule components of the larger lipid molecules for van der Waals parameters. Some parameters from the earlier 43A1 force field are found to be correct in the context of these calculations, while others are modified. The validity of the force fields is demonstrated by correct replication of X-ray form factors and NMR order parameters over a wide range of membrane compositions in semi-isotropic NTP 1 atm simulations. 43-A1-S3 compares favorably with other force fields used in conjunction with the Gromacs simulation package with respect to the breadth of phenomena that it accurately reproduces.

We present analysis of new configurational bias Monte Carlo and molecular dynamics simulation data for bilayers of dipalmitoyl phosphatidyl choline and cholesterol for dipalmitoyl phosphatidyl choline:cholesterol ratios of 24:1, 47:3, 11.5:1, 8:1, 7:1, 4:1, 3:1, 2:1, and 1:1, using long molecular dynamics runs and interspersed configurational bias Monte Carlo to ensure equilibration and enhance sampling. In all cases with cholesterol concentrations above 12.5% the area per molecule of the heterogeneous membrane varied linearly with cholesterol fraction. By extrapolation to pure cholesterol, we find the cross-sectional area of cholesterol in these mixtures is approximately 22.3 A(2). From the slope of the area/molecule relationship, we also find that the phospholipid in these mixtures is in a liquid ordered state with an average cross-sectional area per lipid of 50.7 A(2), slightly above the molecular area of a pure phospholipid gel. For lower concentrations of cholesterol, the molecular area rises above the straight line, indicating the "melting" of at least some of the phospholipid into a fluid state. Analysis of the lateral distribution of cholesterol molecules in the leaflets reveals peaks in radial distributions of cholesterols at multiples of approximately 5 A. These peaks grow in size as the simulation progresses, suggesting a tendency for small subunits of one lipid plus one cholesterol, hydrogen bonded together, to act as one composite particle, and perhaps to aggregate with other composites. Our results are consistent with experimentally observed effects of cholesterol, including the condensation effect of cholesterol in phospholipid monolayers and the tendency of cholesterol-rich domains to form in cholesterol-lipid bilayers. We are continuing to analyze this tendency on longer timescales and for larger bilayer patches.

In this paper we calculate surface conformation and deformation free energy associated with the incorporation of gramicidin channels into phospholipid bilayer membranes. Two types of membranes are considered. One is a relatively thin solvent-free membrane. The other is a thicker solvent-containing membrane. We follow the approach used for the thin membrane case by Huang (1986) in that we use smectic liquid crystal theory to evaluate the free energy associated with distorting the membrane to other than a flat configuration. Our approach is different from Huang, however, in two ways. One is that we include a term for surface tension, which Huang did not. The second is that one of our four boundary conditions for solving the fourth-order differential equation describing the free energy of the surface is different from Huang's. The details of the difference are described in the text. Our results confirm that for thin membranes Huang's neglect of surface tension is appropriate. However, the precise geometrical form that we calculate for the surface of the thin membrane in the region of the gramicidin channel is somewhat different from his. For thicker membranes that have to deform to a greater extent to accommodate the channel, we find that the contribution of surface tension to the total energy in the deformed surface is significant. Computed results for the shape of the deformed surface, the total energy in the deformed surface, and the contributions of different components to the total energy, are presented for the two types of membranes considered. These results may be significant for understanding the mechanisms of dimer formation and breakup, and the access resistance for ions entering gramicidin channels.

We have carried out an atomic-level molecular dynamics simulation of a system of nanoscopic size containing a domain of 18:0 sphingomyelin and cholesterol embedded in a fully hydrated dioleylposphatidylcholine (DOPC) bilayer. To analyze the interaction between the domain and the surrounding phospholipid, we calculate order parameters and area per molecule as a function of molecule type and proximity to the domain. We propose an algorithm based on Voronoi tessellation for the calculation of the area per molecule of various constituents in this ternary mixture. The calculated areas per sphingomyelin and cholesterol are in agreement with previous simulations. The simulation reveals that the presence of the liquid-ordered domain changes the packing properties of DOPC bilayer at a distance as large as approximately 8 nm. We calculate electron density profiles and also calculate the difference in the thickness between the domain and the surrounding DOPC bilayer. The calculated difference in thickness is consistent with data obtained in atomic force microscopy experiments.

It is known from experimental studies that lipid bilayers composed of unsaturated phospholipids, sphingomyelin, and cholesterol contain microdomains rich in sphingomyelin and cholesterol. These domains are similar to "rafts" isolated from cell membranes, although the latter are much smaller in lateral size. Such domain formation can be a result of very specific and subtle lipid-lipid interactions. To identify and study these interactions, we have performed two molecular dynamics simulations, of 200-ns duration, of dioleylphosphatidylcholine (DOPC), sphingomyelin (SM), and cholesterol (Chol) systems, a 1:1:1 mixture of DOPC/SM/Chol, and a 1:1 mixture of DOPC/SM. The simulations show initial stages of the onset of spontaneous phase-separated domains in the systems. On the simulation timescale cholesterol favors a position at the interface between the ordered SM region and the disordered DOPC region in the ternary system and accelerates the process of domain formation. We find that the smooth alpha-face of Chol preferentially packs next to SM molecules. Based on a comparative analysis of interaction energies, we find that Chol molecules do not show a preference for SM or DOPC. We conclude that Chol molecules assist in the process of domain formation and the process is driven by entropic factors rather than differences in interaction energies.

No abstract