CONFORMATIONAL SAMPLING OF HYDROCARBON AND LIPID CHAINS IN AN ORIENTING POTENTIAL

A Distribution Biased Monte Carlo (DBMC) sampling procedure is develop ed for the efficient generation of chain conformations in the oriented environment of lipid membranes and other liquid crystalline systems. Conformations of the sn-l chain of dipalmitoyl phosphatidylcholine (DP PC) were generated by independently sampling torsion angles from conti nuous distributions in an orienting potential based on a Marcelja mean field; depending on the chain position, the convergence in the deuter ium order parameters (S-CD) was 100 to 3000 times more efficient with DBMC than with Brownian dynamics. Optimization using joint distributio n and torsional potentials of mean force yielded a further threefold i ncrease in sampling efficiency. Overall chain tilt was included using Euler angle rotations and a separate field strength for the anchor. A segmental DBMC procedure was used to generate a set of complete DPPC c onformations with well-converged conformationally averaged S-CD consis tent with experimental values. These conformations show considerable f lexibility, not only in the hydrocarbon tails, but additionally in bot h the glycerol and head-group portions of the lipid. An appendix compa res DBMC with a number of other Monte Carlo and stochastic dynamics al gorithms using the example of a bistable oscillator, and illustrates t he tuning of parameters for optimal convergence. (C) 1994 by John Wile y and Sons, Inc.