Section 23.4.1.  Introduction

The unique properties of water and its role in nature have preoccupied the minds of scientists and philosophers for centuries. However, only relatively recently have the tools become available to study the specific roles that water molecules play with respect to protein structure and function. When the first crystal structure of a protein was obtained by X-ray diffraction (Kendrew, 1963), the focus was on the arrangement of the amino-acid residues into secondary and tertiary structure. Although the presence of water molecules associated with the protein was noticed, little attention was given to their structure and possible functional role. The structure of the protein itself was a great novelty, and its features were eagerly analysed. For many years, the crucial role of water molecules in maintaining both the structural integrity and the functional viability of proteins was not completely obvious, although in the 1950s Kauzmann argued correctly that water plays an important role in maintaining protein structure (Kauzmann, 1959). In the late 1970s and early 1980s, reviews began to appear focusing on the properties of water relevant to interaction with proteins (Edsall & McKenzie, 1978) and the location and role of water molecules on protein surfaces (Blake et al., 1983; Edsall & McKenzie, 1983). As high-resolution structures became more easily attainable and refinement techniques improved, the importance of water molecules became increasingly apparent, and solvent structure now occupies a front seat in the realm of structural biology. There is a strong sense in the scientific community that water molecules play an integral role in many aspects of protein structure and function, and great effort is now being focused on understanding solvent effects in precise atomic detail.

In principle, water molecules can contribute both enthalpically and entropically to any process in which they are involved. The main contribution of water as solvent in the protein-folding process, for example, is entropic, driving the collapse of hydrophobic residues into the core of the protein. As is currently understood, the general shape of globular proteins is attained by this effect, with the specific structural features guided by the hydrogen bonds that define secondary structure (Hendsch & Tidor, 1994; Hendsch et al., 1996). Although the solvent contribution to the protein-folding process is beyond the scope of this chapter, it is nevertheless deserving of brief mention. It is important to understand protein three-dimensional structures as having evolved in bulk water, a fact which is largely invisible to the current methods in structural biology. The size, geometry, planarity and orientational flexibility of water molecules give them structural and functional importance. All folded globular proteins have evolved with tens to hundreds of binding sites specific for water molecules, a situation contrary to that of larger ligands which usually bind in a single or small number of specific sites found in a given protein or family of proteins. In this respect, water is unique and its ubiquitous appearance is not a direct consequence of its chemical properties alone, but has an evolutionary origin. Proteins evolved in an aqueous milieu, where over time some water molecules were specifically incorporated as integral parts of the protein architecture.

At first glance, the surface of a protein determined by X-ray crystallography appears randomly populated by a layer of water molecules. A careful analysis, however, reveals that the arrangement of water molecules on protein surfaces is not random. In folded proteins, individual water molecules participate in a variety of structural and functional roles, ranging from filling small cavities that are not fully occupied by protein atoms to allowing flexibility, such as in the case of charged surface side chains that can move freely while continuously maintaining hydrogen-bonding partners. Water molecules can fill deep crevices on the protein surface, or they can play a crucial role in the thermodynamics of ligand binding. The mobility as well as the number and strength of hydrogen-bonding partners that are observed for water molecules bound to protein surfaces vary considerably, and it is becoming increasingly apparent that these factors are correlated with functional roles. The atomic coordinates for any protein should not be considered complete without those bound solvent molecules that can be observed, for they are part of the structure.

Bound water molecules have been implicated and studied in the context of substrate specificity and affinity (Quiocho et al., 1989; Herron et al., 1994; Ladbury, 1996), catalysis (Privé et al., 1992; Singer et al., 1993; Komives et al., 1995), mediation of protein–DNA interactions (Clore et al., 1994; Shakked et al., 1994; Morton & Ladbury, 1996), cooperativity (Royer et al., 1996), conformational stability (Bhat et al., 1994), and drug design (Poormina & Dean, 1995a,b,c). One of the challenges now is to translate the structural information observed into a thermodynamic understanding of the water contribution to the various processes. In some cases, an attempt has been made to relate changes in water structure between two forms of a protein (e.g. ligated and unligated or native and mutant) to changes in the measured heat capacity (Holdgate et al., 1997) or to measurements of enthalpy and entropy changes by titration calorimetry (Bhat et al., 1994). Thermodynamic solvent isotope effects have also been reported, where the thermodynamics of association of several binding processes were evaluated calorimetrically in light and heavy water (Chervenak & Toone, 1994). In other cases, the three-dimensional structures were directly interpreted in terms of thermodynamic contributions (Quiocho et al., 1989; Morton & Ladbury, 1996). Ultimately, a thorough understanding of the thermodynamics and kinetics underlying solvent structure will lead to powerful predictive methods. Theoreticians, on the one hand, have developed models based on physical principles and use experimental knowledge to assess whether their predictions are correct. Experimentalists, on the other hand, attempt to explain the observed phenomena in terms of the well established physical theories that govern the natural world. Progress is being made on both fronts, but a large gap still remains between the two. A bridge is being built from both sides of the gap and when the two sides meet at a common point, the many pieces of this complicated puzzle will have been deciphered and put in their proper places, so that a global view of molecular processes in water can be obtained from whatever perspective one wishes to take: chemical, physical, or biological.

The present chapter summarizes the empirical information gathered over the last decade or two on the structure of water molecules bound to proteins. The focus will be on structures solved by X-ray crystallography, although complementary techniques of obtaining solvent structure will be discussed briefly and, when appropriate, particular examples will be given. Section 23.4.2 is concerned with the methods by which solvent structure can be observed, Section 23.4.3 summarizes knowledge derived from database analysis of large numbers of proteins, Section 23.4.4 focuses on particular examples of groups of well studied protein structures, Section 23.4.5 discusses the contribution of protein models obtained at very high resolution to the understanding of solvent structure, and Section 23.4.6 contains an analysis of water molecules as mediators of complex formation. Finally, Section 23.4.7 presents a conclusion and a perspective regarding the direction in which this information can lead in building a cohesive understanding of the roles played by solvent in the structural integrity and biological function of macromolecules.