Protein-solvent interactions.

The central importance of solvent interactions in stabilization of specific protein structure has long been recognized. Decades ago, Tanford and Kirkwood treated in detail the interaction of charges with solvent, and showed how desolvation/burial of charges upon protein folding was an important factor in stability 1. The influence of their model, with further elaborations, can still be seen in much subsequent work on protein electrostatics and implicit solvent models. A little later, Kauzmann provided a seminal insight into the second major 'theme' in protein-solvent interaction: The hydrophobic effect and how burial of hydrophobic amino acid side chains could stabilize proteins and play a role in determining their structure2. The view that hydrophobicity is the major contributor to protein stability is widely held 3, although current studies of solvation recognize the importance of other types of solvent-protein interaction, including van der Waals, polar, charged, ionic and hydrogen bonding interactions. The view of solvation as a stabilizing force was further expanded to include the possibility that solvent interactions play a role in specifying structure and function; that water is in effect the ‘21st amino acid’. The field of experimental and theoretical studies, even for this rather specialized topic, is now too vast to be covered in any single review. We have selected four topic for discussion in this review: Peptide-water interactions, New experimental probes of protein hydration, New solvent models for long protein-solvent simulations, and Thermal hysteresis proteins. The selection was guided by the theme of this issue: Protein folding. The rationale for discussing peptides is that the effect of solvent on conformations/nascent folding is best understood in these systems. Peptides have long been used as more experimentally and computationally tractable test systems for studying protein-solvent interactions and developing simulation methods. Thus, much of what is known about the specific and quantitative effects of solvent on proteins is derived from peptide studies. This applies particularly to protein folding. It also goes without saying that many peptides are not simply smaller versions of proteins, but have their own biological importance. New experiments that directly access the hydration of proteins, especially the interior of folded and around unfold proteins, have major implications for how the field views the role of water removal during folding. X-ray crystallography has long been used to analyze water around proteins, and with the routine production of stunningly high resolution structures a wealth of water structure has been revealed. Most X-ray structures, however, are now solved at cryogenic temperatures. This fact, combined with the presence of the crystal lattice makes the relevance of this water structure to biological temperatures and the solution phase problematic at best, and beyond the scope of this review. These concerns and other theoretical issues related to crystallographically observed water have been recently reviewed elsewhere at length. 4–6 In contrast to the rather 'static' picture of protein hydration obtained from crystallography, a 'revised', more dynamic view of protein hydration has resulted from advances in several specific types of spectroscopy, discussed below. Protein folding and unfolding occur on the microsecond to second timescale. Thus very long timescale molecular dynamics (MD) simulations of these events are required. At this point in time, this means one must use implicit solvent models for solvent to be realistically and tractably included. We thus focus on two such implicit models which are currently practical in MD simulations: The Generalized Born (GB) and Poisson Boltzmann (PB) implicit solvent. These two models have changed the way many simulations of proteins in water are done. They have also done much to extend the time scale of routine simulations into the multi-nanosecond to sub-microsecond scale while retaining an accurate treatment of solvent effects. Both these models have been used to study protein electrostatics in a wide variety of applications aside from MD, but this section is not aimed as a general review of GB/PB implicit solvent models. The final topic, thermal hysteresis proteins (THP's), was included because these proteins are unique in their ability to recognize and selectively bind solvent water in its solid phase (i.e. as ice). Study of these proteins is an active area of research. Although the mechanism of thermal hysteresis is not fully understood, study of the unique aspects of THP hydration is leading to qualitatively new information about protein solvation.