Polymorphism of the bilayer membranes in the ordered phase and the molecular origin of the lipid pretransition and rippled lamellae.

The lamellar-to-undulated-lamellar phase transition (L beta'----P beta', pretransition) in lipid bilayers is shown to be a phenomenon which exists only if lipid polar headgroups are sufficiently hydrated and if the interchain packing is sufficiently weak. The minimal lipid hydrophilicity and the critical amount of the lipid-bound water can be related to the lipid chain-melting transition temperature; the latter must not exceed some maximal, chainlength-dependent value if the pretransition is to exist. The minimally required amount of the lipid-bound water itself is essentially chainlength independent, however, and unaffected by the method of hydration variation: physical dehydration, hydrational competition between the lipid molecules and the substances dissolved in the aqueous subphase, or decreasing lipid headgroup polarity all affect the pretransition temperature similarly on the appropriate scale. Simple, phenomenological expressions for the evaluation of the bilayer subtransition, pretransition and chain-melting phase transition temperature as a function of the lipid chainlength are presented. They show that, even in excess water, bilayers will tend to undulate only as long as each of the two identical lipid chains will contain between 12 +/- 1 and 22 +/- 1 carbon atoms, the P beta-phase region for the less polar lipids being as a rule narrower. To get a theoretical means for quantitatively studying the effects of the lipid hydration on the bilayer pretransition, an interaction-balance method is proposed for describing undulated membranes at the molecular level. This is based on comparing the free-energy gain from the increased headgroup hydration with the free-energy loss caused by the reduced chain-chain attraction upon ripple formation. A rationale is thus found for scaling the pretransition temperature in terms of the hydration-induced chain-melting phase transition shift or of the lipid surface hydrophilicity. Within the framework of such a model the recently reported (de)hydration dependence of the bilayer-undulation period is reproduced with reasonable accuracy. Furthermore, it is estimated that at least 12 +/- 2 water molecules must be associated with each lipid head for the bilayer undulation to be feasible. The closer the system is to this boundary condition the longer is the repeat-distance for the surface undulations and the less stable is the undulated bilayer phase.