Chapter 22.1. Protein surfaces and volumes: measurement and use

Protein volume can be defined in a straightforward sense through a particular geometric construction called the Voronoi polyhedron. In essence, this construction provides a useful way of partitioning space amongst a collection of atoms. Each atom is surrounded by a single convex polyhedron and allocated the space within it (Fig. 22.1.1.1). The faces of Voronoi polyhedra are formed by constructing dividing planes perpendicular to vectors connecting atoms, and the edges of the polyhedra result from the intersection of these planes.

Figure 22.1.1.1 | top | pdf | The Voronoi construction in two and three dimensions. Representative Voronoi polyhedra from 1CSE (subtilisin) are shown. (a) Six polyhedra around the atoms in a Phe ring. (b) A single polyhedron around the side-chain hydroxyl oxygen (OG) of a serine. (c) A schematic showing the construction of a Voronoi polyhedron in two dimensions. The broken lines indicate planes that were initially included in the polyhedron but then removed by the `chopping-down' procedure (see Fig. 22.1.1.4).

Voronoi polyhedra were originally developed by Voronoi (1908) nearly a century ago. Bernal & Finney (1967) used them to study the structure of liquids in the 1960s. However, despite the general utility of these polyhedra, their application to proteins was limited by a serious methodological difficulty. While the Voronoi construction is based on partitioning space amongst a collection of `equal' points, all protein atoms are not equal. Some are clearly larger than others. In 1974, a solution was found to this problem (Richards, 1974), and since then Voronoi polyhedra have been applied to proteins.

In the procedure outlined above, all atoms are considered equal, and the dividing planes are positioned midway between atoms (Fig. 22.1.1.3). This method of partition, called bisection, is not physically reasonable for proteins, which have atoms of obviously different size (such as oxygen and sulfur). It chemically misallocates volume, giving excess to the smaller atom.

Figure 22.1.1.3 | top | pdf | Positioning of the dividing plane. (a) The dividing plane is positioned at a distance d from the larger atom with respect to radii of the larger atom (R) and the smaller atom (r) and the total separation between the atoms (D). (b) Vertex error. One problem with using method B is that the calculation does not account for all space, and tiny tetrahedra of unallocated volume are created near the vertices of each polyhedron. Such an error tetrahedron is shown. The radical-plane method does not suffer from vertex error, but it is not as chemically reasonable as method B.

Two principal methods of repositioning the dividing plane have been proposed to make the partition more physically reasonable: method B (Richards, 1974) and the radical-plane method (Gellatly & Finney, 1982). Both methods depend on the radii of the atoms in contact (R for the larger atom and r for the smaller one) and the distance between the atoms (D). As shown in Fig. 22.1.1.3, they position the plane at a distance d from the larger atom. This distance is always set such that the plane is closer to the smaller atom.

22.1.1.2.3.3. `Chopping-down' method of finding vertices

| top | pdf |

Because of vertex error and the complexities in locating vertices, a different algorithm has to be used for volume calculation with method B. (It can also be used with bisection.) First, surround the central atom (for which a volume is being calculated) by a very large, arbitrarily positioned tetrahedron. This is initially the `current polyhedron'. Next, sort all neighbouring atoms by distance from the central atom and go through them from nearest to farthest. For each neighbour, position a plane perpendicular to the vector connecting it to the central atom according to the predefined proportion (i.e. from the method B formulae or bisection). Since a Voronoi polyhedron is always convex, if any vertices of the current polyhedron are on the other side of this plane to the central atom, they cannot be part of the final polyhedron and should be discarded. After this has been done, the current polyhedron is recomputed using the plane to `chop it down'. This process is shown schematically in Fig. 22.1.1.4. When it is finished, one has a list of vertices that can be traversed to calculate volumes, as in the basic Voronoi procedure.

Figure 22.1.1.4 | top | pdf | The `chopping-down' method of polyhedra construction. This is necessary when using method B for plane positioning, since one can no longer solve for the position of vertices. One starts with a large tetrahedron around the central atom and then `chops it down' by removing vertices that are outside the plane formed by each neighbour. For instance, say vertex 0214 of the current polyhedron is outside the plane formed by neighbour 6. One needs to delete 0214 from the list of vertices and recompute the polyhedron using the new vertices formed from the intersection of the plane formed by neighbour 6 and the current polyhedron. Using the labelling conventions in Fig. 22.1.1.2, one finds that these new vertices are formed by the intersection of three lines (021, 024 and 014) with plane 06. Therefore one adds the new vertices 0216, 0246 and 0146 to the polyhedron. However, there is a snag: it is necessary to check whether any of the three lines are not also outside of the plane. To do this, when a vertex is deleted, all the lines forming it (e.g. 021, 024, 014) are pushed onto a secondary list. Then when another vertex is deleted, one checks whether any of its lines have already been deleted. If so, this line is not used to intersect with the new plane. This process is shown schematically in two dimensions. For the purposes of the calculations, it is useful to define a plane created by a vector v from the central atom to the neighbouring atom using a constant K so that for any point u on the plane . If , u is on the wrong side of the plane, otherwise it is on the right side. A vertex point w satisfies the equations of three planes: , and . These three equations can be solved to give the components of w. For example, the x component is given by

Voronoi polyhedra are closely related (i.e. dual) to another useful geometric construction called the Delaunay triangulation. This consists of lines, perpendicular to Voronoi faces, connecting each pair of atoms that share a face (Fig. 22.1.1.5).

Figure 22.1.1.5 | top | pdf | Delaunay triangulation and its relation to the Voronoi construction. (a) A standard schematic of the Voronoi construction. The atoms used to define the Voronoi planes around the central atom are circled. Lines connecting these atoms to the central one are part of the Delaunay triangulation, which is shown in (b). Note that atoms included in the triangulation cannot be selected strictly on the basis of a simple distance criterion relative to the central atom. The two circles about the central atoms illustrate this. Some atoms within the outer circle but outside the inner circle are included in the triangulation, but others are not. In the context of protein structure, Delaunay triangulation is useful in identifying true `packing contacts', in contrast to those contacts found purely by distance threshold. The broken lines in (a) indicate planes that were initially included in the polyhedron but then removed by the `chopping-down' procedure (see Fig. 22.1.1.4).

Delaunay triangulation is described here as a derivative of the Voronoi construction. However, it can be constructed directly from the atom coordinates. In two dimensions, one connects with a triangle any triplet of atoms if a circle through them does not enclose any additional atoms. Likewise, in three dimensions one connects four atoms with a tetrahedron if the sphere through them does not contain any further atoms. Notice how this construction is equivalent to the specification for Voronoi polyhedra and, in a sense, is simpler. One can immediately see the relationship between the triangulation and the Voronoi volume by noting that the volume is the distance between neighbours (as determined by the triangulation) weighted by the area of each polyhedral face. In practice, it is often easier in drawing to construct the triangles first and then build the Voronoi polyhedra from them.

Delaunay triangulation is useful in many `nearest-neighbour' problems in computational geometry, e.g. trying to find the neighbour of a query point or finding the largest empty circle in a collection of points (O'Rourke, 1994). Since this triangulation has the `fattest' possible triangles, it is the choice for procedures such as finite-element analysis.

In terms of protein structure, Delaunay triangulation is the natural way to determine packing neighbours, either in protein structure or molecular simulation (Singh et al., 1996; Tsai et al., 1996, 1997). Its advantage is that the definition of a neighbour does not depend on distance. The alpha shape is a further generalization of Delaunay triangulation that has proven useful in identifying ligand-binding sites (Edelsbrunner et al., 1996, 1995; Edelsbrunner & Mucke, 1994; Peters et al., 1996).

When one is carrying out the Voronoi procedure, if a particular atom does not have enough neighbours the `polyhedron' formed around it will not be closed, but rather will have an open, concave shape. As it is not often possible to place enough water molecules in an X-ray crystal structure to cover all the surface atoms, these `open polyhedra' occur frequently on the protein surface (Fig. 22.1.1.6). Furthermore, even when it is possible to define a closed polyhedron on the surface, it will often be distended and too large. This is the problem of the protein surface in relation to the Voronoi construction.

Figure 22.1.1.6 | top | pdf | The problem of the protein surface. This figure shows the difficulty in constructing Voronoi polyhedra for atoms on the protein surface. If all the water molecules near the surface are not resolved in a crystal structure, one often does not have enough neighbours to define a closed polyhedron. This figure should be compared with Fig. 22.1.1.1, illustrating the basic Voronoi construction. The two figures are the same except that in this figure, some of the atoms on the left are missing, giving the central atom an open polyhedron. The broken lines indicate planes that were initially included in the polyhedron but then removed by the `chopping-down' procedure (see Fig. 22.1.1.4).

There are a number of practical techniques for dealing with this problem. First, one can use very high resolution protein crystal structures, which have many solvent atoms positioned (Gerstein & Chothia, 1996). Alternatively, one can make up the positions of missing solvent molecules. These can be placed either according to a regular grid-like arrangement or, more realistically, according to the results of molecular simulation (Finney et al., 1980; Gerstein et al., 1995; Richards, 1974).

More fundamentally, however, the `problem of the protein surface' indicates how closely linked the definitions of surface and volume are and how the definition of one, in a sense, defines the other. That is, the two-dimensional (2D) surface of an object can be defined as the boundary between two 3D volumes. More specifically, the polyhedral faces defining the Voronoi volume of a collection of atoms also define their surface. The surface of a protein consists of the union of (connected) polyhedra faces. Each face in this surface is shared by one solvent atom and one protein atom (Fig. 22.1.1.7).

Figure 22.1.1.7 | top | pdf | Definitions of the protein surface. (a) The classic definitions of protein surface in terms of the probe sphere, the accessible surface and the molecular surface. (This figure is adapted from Richards, 1977). (b) Voronoi polyhedra and Delaunay triangulation can also be used to define a protein surface. In this schematic, the large spheres represent closely packed protein atoms and the smaller spheres represent the small loosely packed water molecules. The Delaunay triangulation is shown by dotted lines. Some parts of the triangulation can be used to define surfaces. The outermost part of the triangulation of just the protein atoms forms the convex hull. This is indicated by the thick line around the protein atoms. For the convex-hull construction, one imagines that the water is not present. This is highlighted by the thick dotted line, which shows how Delaunay triangulation of the surface atoms in the presence of the water diverges from the convex hull near a deep cleft. Another part of the triangulation, also indicated by thick black lines, connects the first layer of water molecules (those that touch protein atoms). A time-averaged version of this line approximates the accessible surface. Finally, the light thick lines show the Voronoi faces separating the protein surface atoms from the first layer of water molecules. Note how this corresponds approximately to the molecular surface (considering the water positions to be time-averaged). These correspondences between the accessible and molecular surfaces and time-averaged parts of the Voronoi construction are understandable in terms of which part of the probe sphere (centre or point of tangency) is used for the surface definition. The accessible surface is based on the position of the centre of the probe sphere while the molecular surface is based on the points of tangency between the probe sphere and the protein atoms, and these tangent points are similarly positioned to Voronoi faces, which bisect interatomic vectors between solvent and protein atoms.

Another somewhat related definition is the convex hull, the smallest convex polyhedron that encloses all the atom centres (Fig. 22.1.1.7). This is important in computer-graphics applications and as an intermediary in many geometric constructions related to proteins (Connolly, 1991; O'Rourke, 1994). The convex hull is a subset of the Delaunay triangulation of the surface atoms. It is quickly located by the following procedure (Connolly, 1991): Find the atom farthest from the molecular centre. Then choose two of its neighbours (as determined by the Delaunay triangulation) such that a plane through these three atoms has all the remaining atoms of the molecule on one side of it (the `plane test'). This is the first triangle in the convex hull. Then one can choose a fourth atom connected to at least two of the three in the triangle and repeat the plane test, and by iteratively repeating this procedure, one can `sweep' across the surface of the molecule and define the whole convex hull.

Other parts of the Delaunay triangulation can define additional surfaces. The part of the triangulation connecting the first layer of water molecules defines a surface, as does the part joining the second layer. The second layer of water molecules, in fact, has been suggested on physical grounds to be the natural boundary for a protein in solution (Gerstein & Lynden-Bell, 1993c). Protein surfaces defined in terms of the convex hull or water layers tend to be `smoother' than those based on Voronoi faces, omitting deep grooves and clefts (see Fig. 22.1.1.7).

In the absence of solvent molecules to define Voronoi polyhedra, one can define the protein surface in terms of the position of a hypothetical solvent, often called the probe sphere, that `rolls' around the surface (Richards, 1977) (Fig. 22.1.1.7). The surface of the probe is imagined to be maintained at a tangent to the van der Waals surface of the model.

Various algorithms are used to cause the probe to visit all possible points of contact with the model. The locus of either the centre of the probe or the tangent point to the model is recorded. Either through exact analytical functions or numerical approximations of adjustable accuracy, the algorithms provide an estimate of the area of the resulting surface. (See Section 22.1.2 for a more extensive discussion of the definition, calculation and use of areas.)

Depending on the probe size and whether its centre or point of tangency is used to define the surface, one arrives at a number of commonly used definitions, summarized in Table 22.1.1.2 and Fig. 22.1.1.7.

For all the calculations outlined above, the hard-sphere approximation is used for the atoms. (One must remember that in reality atoms are neither hard nor spherical, but this approximation has a long history of demonstrated utility.) There are many lists of the radii of such spheres prepared by different laboratories, both for single atoms and for unified atoms, where the radii are adjusted to approximate the joint size of the heavy atom and its bonded hydrogen atoms (clearly not an actual spherical unit).

Some of these lists are reproduced in Table 22.1.1.1. They are derived from a variety of approaches, e.g. looking for the distances of closest approach between atoms (the Bondi set) and energy calculations (the CHARMM set). The differences between the sets often come down to how one decides to truncate the Lennard–Jones potential function. Further differences arise from the parameterization of water and other hydrogen-bonding molecules, as these substances really should be represented with two radii, one for their hydrogen-bonding interactions and one for their VDW interactions.

Table 22.1.1.1 | top | pdf | Standard atomic radii (Å) For `*' see following notes on specific sets of values. Bondi: Values assigned on the basis of observed packing in condensed phases (Bondi, 1968). Lee & Richards: Values adapted from Bondi (1964) and used in Lee & Richards (1971). Shrake & Rupley: Values taken from Pauling (1960) and used in Shrake & Rupley (1973). >C= value can be either 1.5 or 1.85. Richards: Minor modification of the original Bondi set in Richards (1974). (Rationale not given.) See original paper for discussion of aromatic carbon value. Chothia: From packing in amino-acid crystal structures. Used in Chothia (1975). Richmond & Richards: No rationale given for values used in Richmond & Richards (1978). Gelin & Karplus: Origin of values not specified. Used in Gelin & Karplus (1979). Dunfield et al.: Detailed description of deconvolution of molecular crystal energies. Values represent one-half of the heavy-atom separation at the minimum of the Lennard–Jones 6–12 potential functions for symmetrical interactions. Used in Nemethy et al. (1983) and Dunfield et al. (1979). ENCAD: A set of radii, derived in Gerstein et al. (1995), based solely on the ENCAD molecular dynamics potential function in Levitt et al. (1995). To determine these radii, the separation at which the 6–12 Lennard–Jones interaction energy between equivalent atoms was 0.25  was determined (0.15 kcal mol−1; 1 kcal = 4.184 kJ). CHARMM: Determined in the same way as the ENCAD set, but for the CHARMM potential (Brooks et al., 1983) (parameter set 19). Tsai et al.: Values derived from a new analysis (Tsai et al., 1999) of the most common distances of approach of atoms in the Cambridge Structural Database. Atom type and symbol Bondi (1968) Lee & Richards (1971) Shrake & Rupley (1973) Richards (1974) Chothia (1975) Richmond & Richards (1978) Gelin & Karplus (1979) Dunfield et al. (1979) ENCAD derived (1995) CHARMM derived (1995) Tsai et al. (1999) Aliphatic, methyl 2.00 1.80 2.00 2.00 1.87 1.90 1.95 2.13 1.82 1.88 1.88 Aliphatic, methyl 2.00 1.80 2.00 2.00 1.87 1.90 1.90 2.23 1.82 1.88 1.88 Aliphatic, CH — 1.70 2.00 2.00 1.87 1.90 1.85 2.38 1.82 1.88 1.88  Aromatic, CH — 1.80 1.85 * 1.76 1.70 1.90 2.10 1.74 1.80 1.76 Trigonal, aromatic 1.74 1.80 * 1.70 1.76 1.70 1.80 1.85 1.74 1.80 1.61 Amino, protonated — 1.80 1.50 2.00 1.50 0.70 1.75 — 1.68 1.40 1.64 Amino or amide 1.75 1.80 1.50 — 1.65 1.70 1.70 — 1.68 1.40 1.64 Peptide, NH or N 1.65 1.52 1.40 1.70 1.65 1.70 1.65 1.75 1.68 1.40 1.64 Carbonyl oxygen 1.50 1.80 1.40 1.40 1.40 1.40 1.60 1.56 1.34 1.38 1.42 Alcoholic hydroxyl — 1.80 1.40 1.60 1.40 1.40 1.70 — 1.54 1.53 1.46 Carboxyl oxygen — 1.80 1.89 1.50 1.40 1.40 1.60 1.62 1.34 1.41 1.42 Sulfhydryl — 1.80 1.85 — 1.85 1.80 1.90 — 1.82 1.56 1.77 Thioether or –S–S– 1.80 — — 1.80 1.85 1.80 1.90 2.08 1.82 1.56 1.77

Perhaps because of the complexities in defining VDW parameters, there are some great differences in Table 22.1.1.1. For instance, the radius for an aliphatic CH (>CH=) ranges from 1.7 to 2.38 Å, and the radius for carboxyl oxygen ranges from 1.34 to 1.89 Å. Both of these represent at least a 40% variation. Moreover, such differences are practically quite significant, since many geometrical and energetic calculations are very sensitive to the choice of VDW parameters, particularly the relative values within a single list. (Repulsive core interactions, in fact, vary almost exponentially.) Consequently, proper volume and surface comparisons can only be based on numbers derived through use of the same list of radii.

In the last column of the table we give a recent set of VDW radii that has been carefully optimized for use in volume and packing calculations. It is derived from analysis of the most common distances between atoms in small-molecule crystal structures in the Cambridge Structural Database (Rowland & Taylor, 1996; Tsai et al., 1999).

A series of surfaces can be described by using a probe sphere with a specified radius. Since this is to be a convenient mathematical construct in calculation, any numerical value may be chosen with no necessary relation to physical reality. Some commonly used examples are listed in Table 22.1.1.2.

The solvent-accessible surface is intended to be a close approximation to what a water molecule as a probe might `see' (Lee & Richards, 1971). However, there is no uniform agreement on what the proper water radius should be. Usually it is chosen to be about 1.4 Å.

Volume calculations are principally applied in measuring packing. This is because the packing efficiency of a given atom is simply the ratio of the space it could minimally occupy to the space that it actually does occupy. As shown in Fig. 22.1.1.8, this ratio can be expressed as the VDW volume of an atom divided by its Voronoi volume (Richards, 1974, 1985; Richards & Lim, 1994). (Packing efficiency also sometimes goes by the equivalent terms `packing density' or `packing coefficient'.) This simple definition masks considerable complexities – in particular, how does one determine the volume of the VDW envelope (Petitjean, 1994)? This requires knowledge of what the VDW radii of atoms are, a subject on which there is not universal agreement (see above), especially for water molecules and polar atoms (Gerstein et al., 1995; Madan & Lee, 1994).

Figure 22.1.1.8 | top | pdf | Packing efficiency. (a) The relationship between Voronoi polyhedra and packing efficiency. Packing efficiency is defined as the volume of an object as a fraction of the space that it occupies. (It is also known as the `packing coefficient' or `packing density'.) In the context of molecular structure, it is measured by the ratio of the VDW volume (, shown by a light grey line) and Voronoi volume (, shown by a dotted line). This calculation gives absolute packing efficiencies. In practice, one usually measures a relative efficiency, relative to the atom in a reference state: . Note that in this ratio the unchanging VDW volume of an atom cancels out, leaving one with just a ratio of two Voronoi volumes. Perhaps more usefully, when one is trying to evaluate the packing efficiency P at an interface, one computes , where p is packing efficiency of the reference data set (usually 0.74), is the actual measured volume of each atom i at the interface and is the reference volume corresponding to the type of atom i. (b) A graphical illustration of the difference between tight packing and loose packing. Frames from a simulation are shown for liquid water (left) and for liquid argon, a simple liquid (right). Owing to its hydrogen bonds, water is much less tightly packed than argon (packing efficiency of 0.35 versus ∼0.7). Each water molecule has only four to five nearest neighbours while each argon atom has about ten.

Knowing that the absolute packing efficiency of an atom is a certain value is most useful in a comparative sense, i.e. when comparing equivalent atoms in different parts of a protein structure. In taking a ratio of two packing efficiencies, the VDW envelope volume remains the same and cancels. One is left with just the ratio of space that an atom occupies in one environment to what it occupies in another. Thus, for the measurement of packing, standard reference volumes are particularly useful. Recently calculated values of these standard volumes are shown in Tables 22.1.1.3 and 22.1.1.4 for atoms and residues (Tsai et al., 1999).

Table 22.1.1.4 | top | pdf | Standard atomic volumes Tsai et al. (1999) and Tsai et al. (2001) clustered all the atoms in proteins into the 18 basic types shown below. Most of these have a simple chemical definition, e.g. `=O' are carbonyl carbons. However, some of the basic chemical types, such as the aromatic CH group (`'), need to be split into two subclusters (bigger and smaller), as is indicated by the column labelled `Cluster'. Volume statistics were accumulated for each of the 18 types based on averaging over 87 high-resolution crystal structures (in the same fashion as described for the residue volumes in Table 22.1.1.3). No. is the number of atoms averaged over. The final column (`Symbol') gives the standardized symbol used to describe the atom in Tsai et al. (1999). The atom volumes shown here are part of the ProtOr parameter set (also known as the BL+ set) in Tsai et al. (1999). Atom type Cluster Description Average volume (Å3) Standard deviation (Å3) No. Symbol >C= Bigger Trigonal (unbranched), aromatics 9.7 0.7 4184 C3H0b >C= Smaller Trigonal (branched) 8.7 0.6 11876 C3H0s Bigger Aromatic, CH (facing away from main chain) 21.3 1.9 2063 C3H1b Smaller Aromatic, CH (facing towards main chain) 20.4 1.7 1742 C3H1s >CH— Bigger Aliphatic, CH (unbranched) 14.4 1.3 3642 C4H1b >CH— Smaller Aliphatic, CH (branched) 13.2 1.0 7028 C4H1s —CH2— Bigger Aliphatic, methyl 24.3 2.1 1065 C4H2b —CH2— Smaller Aliphatic, methyl 23.2 2.3 4228 C4H2s —CH3   Aliphatic, methyl 36.7 3.2 3497 C4H3u >N—   Pro N 8.7 0.6 581 N3H0u >NH Bigger Side chain NH 15.7 1.5 446 N3H1b >NH Smaller Peptide 13.6 1.0 10016 N3H1s —NH2   Amino or amide 22.7 2.1 250 N3H2u   Amino, protonated 21.4 1.2 8 N4H3u =O   Carbonyl oxygen 15.9 1.3 7872 O1H0u —OH   Alcoholic hydroxyl 18.0 1.7 559 O2H1u —S—   Thioether or –S–S– 29.2 2.6 263 S2H0u —SH   Sulfhydryl 36.7 4.2 48 S2H1u

In analysing molecular systems, one usually finds that close packing is the default (Chandler et al., 1983), i.e. atoms pack like billiard balls. Unless there are highly directional interactions (such as hydrogen bonds) that have to be satisfied, one usually achieves close packing to optimize the attractive tail of the VDW interaction. Close-packed spheres of the same size have a packing efficiency of ∼0.74. Close-packed spheres of different size are expected to have a somewhat higher packing efficiency. In contrast, water is not close-packed because it has to satisfy the additional constraints of hydrogen bonding. It has an open, tetrahedral structure with a packing efficiency of ∼0.35. (This difference in packing efficiency is illustrated in Fig. 22.1.1.8b)

The protein core is usually considered to be the atoms inaccessible to solvent i.e. with an accessible surface area of zero or a very small number, such as 0.1 Ų. Packing calculations on the protein core are usually done by calculating the average volumes of the buried atoms and residues in a database of crystal structures. These calculations were first done more than two decades ago (Chothia & Janin, 1975; Finney, 1975; Richards, 1974). The initial calculations revealed some important facts about protein structure. Atoms and residues of a given type inside proteins have a roughly constant (or invariant) volume. This is because the atoms inside proteins are packed together fairly tightly, with the protein interior better resembling a close-packed solid than a liquid or gas. In fact, the packing efficiency of atoms inside proteins is roughly as expected for the close packing of hard spheres (0.74).

More recent calculations measuring the packing in proteins (Harpaz et al., 1994; Tsai et al., 1999) have shown that the packing inside of proteins is somewhat tighter (by ∼4%) than that observed initially and that the overall packing efficiency of atoms in the protein core is greater than that in crystals of organic molecules. When molecules are packed this tightly, small changes in packing efficiency are quite significant. In this regime, the limitation on close packing is hard-core repulsion, which is expected to have a twelfth power or exponential dependence, so even a small change is energetically quite substantial. Furthermore, the number of allowable configurations that a collection of atoms can assume without core overlap drops off very quickly as these atoms approach the close-packed limit (Richards & Lim, 1994).

The exceptionally tight packing in the protein core seems to require a precise jigsaw puzzle-like fit of the residues. This appears to be the case for the majority of atoms inside of proteins (Connolly, 1986). The tight packing in proteins has, in fact, been proposed as a quality measure in protein crystal structures (Pontius et al., 1996). It is also believed to be a strong constraint on protein flexibility and motions (Gerstein et al., 1993; Gerstein, Lesk & Chothia, 1994). However, there are exceptions, and some studies have focused on these, showing how the packing inside proteins is punctuated by defects, or cavities (Hubbard & Argos, 1994, 1995; Kleywegt & Jones, 1994; Kocher et al., 1996; Rashin et al., 1986; Richards, 1979; Williams et al., 1994). If these defects are large enough, they can contain buried water molecules (Baker & Hubbard, 1984; Matthews et al., 1995; Sreenivasan & Axelsen, 1992).

Surprisingly, despite the intricacies of the observed jigsaw puzzle-like packing in the protein core, it has been shown that one can simply achieve the `first-order' aspect of this, getting the overall volume of the core right rather easily (Gerstein, Sonnhammer & Chothia, 1994; Kapp et al., 1995; Lim & Ptitsyn, 1970). This has to do with simple statistics for summing random numbers and the fact that the distribution of sizes for amino acids usually found inside proteins is rather narrow (Table 22.1.1.3). In fact, the similarly sized residues Val, Ile, Leu and Ala (with volumes 138, 163, 163 and 89 ų ) make up about half of the residues buried in the protein core. Furthermore, aliphatic residues, in particular, have a relatively large number of adjustable degrees of freedom per  ų, allowing them to accommodate a wide range of packing geometries. All of this suggests that many of the features of protein sequences may only require random-like qualities for them to fold (Finkelstein, 1994).

Measuring the packing efficiency inside the protein core provides a good reference point for comparison, and a number of other studies have looked at this in comparison with other parts of the protein. The most obvious thing to compare with the protein inside is the protein outside, or surface. This is particularly interesting from a packing perspective, since the protein surface is covered by water, and water is packed much less tightly than protein and in a distinctly different fashion. (The tetrahedral packing geometry of water molecules gives a packing efficiency of less than half that of hexagonal close-packed solids.)

Calculations based on crystal structures and simulations have shown that the protein surface has intermediate packing, being packed less tightly than the core but not as loosely as liquid water (Gerstein & Chothia, 1996; Gerstein et al., 1995). One can understand the looser packing at the surface than in the core in terms of a simple trade-off between hydrogen bonding and close packing, and this can be explicitly visualized in simulations of the packing in simple toy systems (Gerstein & Lynden-Bell, 1993a,b).

Interactions between molecules are most likely to be mediated by the properties of residues at their surfaces. Surfaces have figured prominently in functional interpretations of macromolecular structure. Which residues are most likely to interact with other molecules? What are their properties: charged, polar, or hydrophobic? What would be the estimated energy of interaction? How do the shapes and properties complement one another? Which surfaces are most conserved among a homologous family? At the centre of these questions that are often asked at the start of a structural interpretation lies the calculation of the molecular and/or accessible surfaces.

Surface-area calculations are used in two ways. Graphical surface representations help to obtain a quick intuitive understanding of potential molecular functions and interactions through visualization of the shape, charge distribution, polarity, or sequence conservation on the molecular surface (for example). Quantitative calculations of surface area are used en route to approximations of the free energy of interactions in binding complexes.

Part of this subject area was the topic of an excellent review by Richards (1985), to which the reader is referred for greater coverage of many of the methods of calculation. This review will attempt to incorporate more recent developments, particularly in the use of graphics, both realistic and schematic.

The concept of molecular surface derives from the behaviour of non-bonded atoms as they approach each other. As indicated by the Lennard–Jones potential, strong unfavourable interactions of overlapping non-bonding electron orbitals increase sharply according to , and atoms behave almost as if they were hard spheres with van der Waals radii that are characteristic for each atom type and nearly independent of chemical context. Of course, when orbitals combine in a covalent bond, atoms approach much more closely. Lower-energy attractions between atoms, such as hydrogen bonds or aromatic ring stacking, lead to modest reductions in the distance of closest approach. The van der Waals surface is the area of a volume formed by placing van der Waals spheres at the centre of each atom in a molecule.

Non-bonded atoms of the same molecule contact each other over (at most) a very small proportion of their van der Waals surface. The surface is complicated with gaps and crevices. Much of this surface is inaccessible to other atoms or molecules, because there is insufficient space to place an atom without resulting in forbidden overlap of non-bonded van der Waals spheres (Fig. 22.1.2.1). These crevices are excluded in the molecular surface area. The molecular surface area, also known as the solvent-excluding surface, is the outer surface of the volume from which solvent molecules are excluded. Strictly, this would depend on the orientation of non-spherically symmetric solvents such as water. However, since hydrogen atoms are smaller than oxygen atoms, for current purposes it is sufficient to consider water as a sphere with a radius of 1.4 to 1.7 Å, approximating the `average' distance from the centre of the oxygen atom to the van der Waals surface of water. The practical definition of the molecular surface is, then, the area of the volume excluded to a spherical probe of 1.4 to 1.7 Å radius.

Figure 22.1.2.1 | top | pdf | Surfaces in a plane cut through a hypothetical molecule. The molecular surface consists of the sum of the atomic surfaces that can be contacted by solvent molecules and the surface of the space between atoms from which solvent molecules are excluded. The solvent-accessible surface is the surface formed by the set of the centres of spheres that are in closest contact with the molecular surface.

As an aside, it is important to note that surface-area calculations depend on inexact parameterization. For example, there is no radius of any hard-sphere model that can give a realistic representation of the solvent. Furthermore, the choice of van der Waals radii can depend on whether the distance of zero or minimum potential energy is estimated and the potential-energy function or experimental data used. (Tables of common values are given by Gerstein & Richards in Section 22.1.1.) Thus, calculations of molecular and accessible surfaces are approximate. However, when the errors are averaged over large areas of a macromolecule, the numbers can be precise enough to give important insights into function.

Fig. 22.1.2.1 shows that the molecular surface consists of two components. The contact surface is part of the van der Waals surface. The re-entrant surface encloses the interstitial volume and has components that are the exterior surfaces of atoms (contact surface) and parts of the surfaces of probes placed in positions where they are in contact with van der Waals surfaces of two or more atoms (re-entrant surface).

The occluded molecular surface is an approximate complement to the solvent-accessible surface. It is the part of the surface that would be inaccessible to solvent because of steric conflict with neighbouring macromolecular atoms. It is an approximation in that current calculations use van der Waals surfaces, ignoring the differences between atomic and re-entrant surfaces (see below), and the volume of the probe is not fully accounted for (Pattabiraman et al., 1995). Occluded area is defined as the atomic area whose normals cannot be extended 2.8 Å (the presumptive diameter of a water molecule) without intersecting the van der Waals volume of another atom. This crude approximation to the surface that is inaccessible to water not only increases the speed of calculation, but enables surface areas to be partitioned between the atoms. It is used primarily to evaluate model protein structures by comparing the fraction of each amino acid's surface area that is occluded with that calculated for the same residue types in a database of accurate structures.

Whether graphically displaying a molecule or examining potential docking interactions, it is usually the molecular surface or solvent-accessible surface that is used. However, macromolecules also interact through the small (solvent) molecules that are more or less tightly bound (Gerstein & Lynden-Bell, 1993c). There is a gradation of how tightly solvent molecules are bound and how many are bound around different side chains. With dynamics simulations, Gerstein & Lynden-Bell (1993c) showed that the second hydration shell was a reasonable, practical `average' limit to which water atoms should be considered significantly perturbed by the protein. They defined a hydration surface as the surface of this second shell and presented evidence that it approximates the boundary between bound and bulk solvent. They presented calculations that showed that molecules interact significantly when their hydration surfaces interact, and not just when they are close enough for their molecular surfaces to form contacts. It may be computationally impractical to perform the simulations required to calculate the hydration surfaces of many proteins, but this work reminds us that energetically significant interactions occur over a wider area than the commonly computed contact molecular-surface area.

The hydrophobic effect (Kauzmann, 1959; Tanford, 1997) has its origins in unfavourable entropic terms for water molecules immediately surrounding a hydrophobic group. In the bulk solvent, each water molecule can be oriented in a variety of ways with favourable hydrogen bonding. At the interface with a hydrophobic group, hydrogen bonds are possible only in some directions, with some configurations of the water molecules. When a hydrophobic group is embedded in water, the surrounding solvent molecules have a more restricted set of hydrogen-bonding configurations, resulting in an unfavourable entropic term. The magnitude of the entropic term should be proportional to the number of solvent molecules immediately surrounding the hydrophobic group. This integer number can be considered very approximately proportional to the area of the surface made by the centres of the set of possible solvent probes contacting the solute, i.e. the solvent-accessible surface area (Fig. 22.1.2.1). When large areas are considered, summed over many hydrophobic atoms, the errors of this non-integer approximation are insignificant. It is now common practice to estimate the hydrophobic effect free-energy contribution by multiplying the change in macromolecular surface area by an energy per unit area [~80 J mol⁻¹ Å⁻² (Richards, 1985), but see also below].

The first method to be discussed allows the calculation of an accessible surface. The first method for calculating molecular surface involved raining water down on a model of a macromolecule and constructing a surface by making a net under the spheres in their landing positions (Greer & Bush, 1978). This ignored overhangs and was replaced by the dot surface method. More recently, methods were developed to make polyhedral surfaces of triangles by contouring between lattice points or by delimiting with arcs the spherical and toroidal surfaces and then subdividing the piece-wise quartic molecular surface. The surface is then composed of patches whose areas can be precisely integrated. van der Waals surfaces consist of convex spherical triangles whose areas can be estimated by the Gauss–Bonnet theorem. Re-entrant surfaces are comprised of concave spherical triangles whose areas can be similarly estimated and toroidal saddle-shaped patches whose areas can be calculated by analytical geometry and calculus.

The first method for calculating the accessible surface area overlaid the molecule on a regular stack of finely spaced parallel planes (Lee & Richards, 1971). The advantage of this method was the ease with which the area could be calculated. The intersection of the atomic surfaces with the planes were circular arcs whose lengths were readily calculated and multiplied by the planar spacing to give an approximation to the surface area. Programs that are currently distributed use more sophisticated methods.

A molecular dot surface is a smooth envelope of points on the molecular surface. A probe sphere is placed at a set of approximately evenly spaced points so that the probe and van der Waals surfaces of a given atom are tangential. If the probe sphere does not overlap any other atom, the point is designated as surface. To define the re-entrant surface, sphere centres are also sampled that are tangential to both van der Waals spheres of a pair of neighbouring atoms and are equidistant from the interatomic axis. Arcs are then drawn between surface points and the arcs are subdivided into a set of finely spaced points to define the re-entrant surface. Similarly, spheres contacting triplets of neighbouring atoms are tested, and approximately evenly spaced points within the concave triangle defined by the three contact points are added to the re-entrant surface.

This is conceptually the simplest method and is used in the program GRASP (Nicholls et al., 1991). First, grid points of a cubic lattice overlaid on the molecule are segregated into `interior' and `exterior' as follows. All points farther from an atom than the sum of the van der Waals radius and a probe radius are flagged as external. External points with an internal neighbour are flagged as an approximate `accessible surface'. All grid points falling within probe spheres centred at each surface point now join the set of exterior points. Points that remain `interior' define the volume enclosed by the molecular surface.

All that remains is to contour the molecular surface that lies between interior and exterior grid points. It is a little complicated in three dimensions and is achieved by the marching-cube algorithm. Cubes containing adjacent grid points that are both interior and exterior are used to define potential polyhedral vertices. Triangles are defined by joining the midpoints of unit-cell edges that have one interior and one exterior point. The triangles are joined at their edges in a consistent manner to create a polyhedral surface.

Several algorithms start by dividing the surface into regions within which the surface is smooth and continuous. The surface can be efficiently described in terms of a set of arcs and their start and end points. In complete rolling, the probe is placed in all possible positions at which it contacts the van der Waals spheres of three neighbouring atoms. Those surrounding the same atom are paired as the start and end points of an arc. The complete rolling algorithm does not distinguish outer and inner (cavity) surfaces. In the connected rolling algorithm, the process starts at a triple contact point that is far from the centre of mass and therefore likely to be external. The probe is then rolled only along crevices between two atoms, pursuing all alternatives, stopping each pathway only when the probe returns to a place that has already been probed. This algorithm therefore produces only the outer surface.

An analytical method was also proposed for calculating approximate accessible areas (Wodak & Janin, 1980). It assumed random distributions of neighbouring atoms, but this can be a sufficient approximation when calculating the area of an entire molecule. The areas of spherical and toroidal pieces of surface can be calculated exactly by analytic and differential geometry (Richmond, 1984; Connolly, 1983). An advantage of analytical expressions over the prior numerical approximations is that analytical derivatives of the areas can be calculated, albeit with significant difficulty. This then provides the opportunity to optimize atomic positions with respect to surface area. Pseudo-energy functions that approximate the hydrophobic contribution to free energy with a term proportional to the accessible surface area (Richards, 1977) can therefore be incorporated in energy-minimization programs. Although rigorous, these methods are computationally cumbersome and are not used in all energy-minimization routines. Incorporation of solvent effects may become more universal with the Gaussian atom approximations discussed below.

The methods discussed above are computationally quite cumbersome, especially if they need to be repeated many times. Thus, they are not well suited to comparisons of many structures. They are also not well suited to the calculation of surface-area-dependent energy terms during dynamics simulation or energy minimization, which require the calculation of the derivatives of the surface area with respect to atomic position. It has been argued by several (including A. Nicholls and K. Sharp, personal communications) that simplifying approximations to the surface-area calculations are in order, because the common uses of surface area already embody crude ad hoc approximations, such as non-integer numbers of spherical solvent molecules.

In the treatments discussed earlier, the volume of the protein is (implicitly) described by a set of overlapping step functions that have a constant value if close enough to an atom, or zero if not. Several authors have replaced these step functions with continuous spherical Gaussian functions centred on each atom (Gerstein, 1992; Grant & Pickup, 1995) in treatments reminiscent of Ten Eyck's electron-density calculations (Ten Eyck, 1977). This speeds up the calculation and also facilitates the calculation of analytical derivatives of the surface area. A surface can be calculated for graphical display by contouring the continuous function at an appropriate threshold. The final envelope can be modified by using iterative procedures that fill cavities and crevices that are (nearly) surrounded by protein atoms (Gerstein, 1992).

Structures of macromolecules determined by X-ray crystallography rarely reveal the positions of the hydrogen atoms. It is, of course, possible to add explicit hydrogen atoms at the stereochemically most likely positions, but this is rarely done for surface-area calculations. Instead, their average effect is approximately and implicitly accounted for by increasing the heteroatom van der Waals radius by 0.1 to 0.3 Å. (It is not usual to smear atoms to account for thermal motion.)

As previously introduced, hydrophobic energies result primarily from the increased entropy of water molecules at the macromolecule–solvent interface and can be estimated from the accessible surface area. A number of different constants relating area to free energy of transfer from a hydrophobic to aqueous environment have been proposed in the range of 67 to 130 J mol⁻¹ Å⁻² (Reynolds et al., 1974; Chothia, 1976; Hermann, 1977; Eisenberg & McLachlan, 1986), but if a single value is to be used for all of the protein surface, the consensus among crystallographers has been about 80 J mol⁻¹ Å⁻² (Richards, 1985).

There are two widely used enhancements of the basic method. Atomic solvation parameters (ASPs, Δσ) remove the assumption that all protein atoms have equal potential influence on the hydrophobic free energy. Eisenberg & McLachlan (1986) determined separate ASPs for atom types C, N/O, O^.., N⁺ and S (treating hydrogen atoms implicitly) by fitting these constants to the experimentally determined octanol/water relative transfer free energies of the 20 amino-acid side chains of Fauchere & Pliska (1983), assuming standard conformations of the side chains. A much improved free energy change of solvation can then be estimated from , where the summation is over all atoms with accessible area A and is specific for the atom type. Their estimates of ASPs are given in Table 22.1.2.1. Use of ASPs rather than a single value for all atoms makes substantial differences to the estimated free energies of association of macromolecular assemblies (Xie & Chapman, 1996). Through calculation of the overall energy of solvation, calculations with ASPs also allow discrimination between proposed structures that are correctly folded (with hydrophobic side chains that are predominantly internal) and those that are not (Eisenberg & McLachlan, 1986).

The work of Sharp et al. (1991) indicates that hydrophobicity depends not only on surface area, but curvature. Sharp et al. were trying to reconcile long-apparent differences between microscopic and macroscopic measurements of hydrophobicity (Tanford, 1979). Microscopic measurements, the basis of all of our preceding discussions, are derived from the partitioning of dilute solutes between solvents. Macroscopic values can come from the measurements of the surface tension between a liquid bulk of the molecule of interest and water. Macroscopic values for aliphatic carbons are much higher, ~302 J mol⁻¹ Å⁻². Postulating that the entropic effects at the heart of hydrophobicity depended on the number of water molecules in contact with each other at the molecular surface (Nicholls et al., 1991), Sharp et al. pointed out that not all surfaces were equivalent. Relative to a plane, concave solute surfaces would accommodate fewer solvent molecules neighbouring the molecular surface, whereas convex surfaces would accommodate more. Their treatment could be considered to be a second-order approximation to the number of interfacial solvent molecules, compared to the prior first-order consideration of only area.

To calculate the curvature of point a on the accessible surface (relative to that of a plane), a sphere of twice the solvent radius is drawn (Nicholls et al., 1991). This represents the locus of the centres of solvent molecules that could be in contact with a solvent at a. A curvature correction, c, is the proportion of points on the spherical surface that are inside the inaccessible volume, relative to that for a planar accessible surface (). In calculating the free energy of transfer, each element of the accessible area is multiplied by its curvature correction. When this is done, the increasingly convex surfaces of small aliphatic molecules account for most of the discrepancy between microscopic and macroscopic hydrophobicities (Nicholls et al., 1991). Furthermore, it emphasizes that, just by their shape, concave surfaces can become relatively hydrophobic. This has been clearly illustrated with GRASP surface representations (see below) in which the accessible surface is coloured according to the local curvature (Nicholls et al., 1991). Consideration of curvature also indicates that the energy of macromolecular association is slightly less than it would otherwise be due to the generation of a concave collar at the interface between two binding macromolecules (Nicholls et al., 1991).

In a molecular association in which (as is often the case) hydrophobic interactions dominate, the binding energy can be estimated from the surfaces of the individual molecules that become buried upon association (Richards, 1985). The buried area is simply the sum of the surfaces of the two molecules (calculated independently) minus the surface of the complex, calculated as if one molecule. Usually, all heteroatoms are regarded as equivalent, and the buried area is multiplied by a uniform constant, say 80 J mol⁻¹ Å⁻² (Richards, 1985). It is only slightly more complicated to use the different ASPs (Eisenberg & McLachlan, 1986) for different atom types and/or to account for curvature (Nicholls et al., 1991). It should be noted that in many crystal structures, the distinction between atom types in some side chains remains indeterminate, e.g. N and C in histidines, O and O^.. in carboxylates, and N and N⁺ in arginines. In such cases, average values of the two ASPs can be used (Xie & Chapman, 1996). Such energy calculations have been put to several uses, including attempts to predict assembly and disassembly pathways for viral capsid assemblies (Arnold & Rossmann, 1990; Xie & Chapman, 1996, and citations therein).

Which are the amino acids most likely to interact with other molecules? It is reasonable to expect them to be surface-accessible. In determining which residues are most surface-exposed, it is necessary to partition molecular or accessible surfaces between atoms. Contact surfaces (Fig. 22.1.2.1) are atom specific. Re-entrant or accessible surfaces can be divided among surface atoms by proximity. Surface areas can then be summed over the atoms in a residue. Accessible surface areas are sometimes reported as accessibilities (Lee & Richards, 1971) – fractions of a maximum where the standard is evaluated from a tripeptide in which the residue of interest is surrounded by glycines. A different approach to assessing surface exposure is to ask what is the largest molecular fragment that could contact a given atom. This is commonly assayed by determining the largest sphere that can be placed tangentially to the van der Waals surface without intersecting any other atom. An alternative approach to locating functionally important surface regions was proposed in the mid-1980s, but is currently not used very often. The local irregularity of surface texture was characterized through measurement of the fractal dimension (Lewis & Rees, 1985).

Substrates, drugs and ligands often bind in clefts or pockets that are concave in shape. Conversely, it is the most exposed convex regions that are likely to be antigenic. The surface shape can be determined by placing a large (say 6 Å radius) sphere at each vertex of the polyhedral molecular surface. If more than half of the sphere's volume overlaps the molecular volume, then the surface is concave, while if less than half, the surface is convex.

Are there similarities in the shapes of surfaces at the interfaces of macromolecular complexes? For example, are there similarities between the shapes of evolutionary-related antigens or the hypervariable regions of antibodies that bind to them? Quantitative comparison of surface topologies is far from trivial, with questions of 3D alignment, the metrics to be used in quantifying topology etc. In addition to real differences between molecules, their surfaces may appear to differ due to the resolutions at which their structures were determined. Gerstein (1992) has proposed that comparisons be made in reciprocal space so that correlations can be judged as a function of resolution. Coordinates are aligned. Spherical Gaussian functions are placed at each atom, and an envelope is calculated at some threshold value and modified to remove cavities. Gerstein found that comparison of the envelope structure-factor vectors, obtained by Fourier transformation, led to a plausible classification of the hypervariable regions of known antibody structures.

For their work on viruses, Rossmann & Palmenberg (1988) introduced a highly schematic representation in which individual amino acids were labelled. The methods were extended by Chapman (1993) to other proteins and to the automatic display of features such as topology, sequence similarity and hydrophobicity. Roadmaps sacrifice a realistic impression of shape for the ability to show the locations and properties of constituent surface atoms or residues. This has been important in combining the power of structure and molecular biology in understanding function. Potential sites of mutation are readily identified without substantial molecular-graphics resources, and phenotypes of mutants are readily mapped to the surface and compared with the physiochemical properties to reveal structure–function correlations.

For a set of projection vectors, the intersection points with the first van der Waals (or solvent-accessible) surface of an atom are calculated by basic vector algebra. The atom is identified so that when the projection is mapped to a plane for display, the boundaries of each atom or amino acid can be determined. The atoms or amino acids can then be coloured, shaded, outlined, contoured, or labelled according to parameters that are either calculated from the coordinates (such as distance from the centre of mass), read from a file (such as sequence similarity), or follow properties that are dependent on the residue type (e.g. hydrophobicity) or atom type [e.g. atomic solvation parameters (Eisenberg & McLachlan, 1986)].

Several types of projections can be used. The simplest is similar to that used by most other surface-imaging programs. A set of parallel projection vectors is mapped to a 2D grid. An example is shown in Fig. 22.1.2.3. This view avoids distortions, but only one side of the molecule is visualized. Roadmaps are flat, two-dimensional projections that cannot be rotated in real time to reveal other views. Three-dimensionality is limited to an extension by Jean-Yves Sgro that maps the parallel projection of one view to a three-dimensional surface shell that can be rotated with interactive graphics and/or viewed with stereo imaging (Harber et al., 1995; Sgro, 1996). However, the schematic nature of roadmaps leads to the ability to view all parts of the molecule simultaneously.

Figure 22.1.2.3 | top | pdf | (a) Solvent-accessible surface topology of a rhinovirus 14–drug complex (Kim et al., 1993). The triangle shows one of the 60 symmetry-equivalent faces of an icosadeltahedron that constitute the entire virus surface. The surface is coloured and contoured according to distance from the centre of the virus, red being the most elevated. Residues are marked with dotted lines and labelled with residue type and number. A letter starting the residue label indicates a symmetry equivalent. The first numeral indicates the protein number (1 to 4), which is followed by the three-digit residue number. A depression, the `canyon', is where the cellular receptor is bound (Olson et al., 1993). The locations of the dominant neutralizing immunogenic (NIm) sites were determined through the sequencing of escape mutants (Sherry & Rueckert, 1985; Sherry et al., 1986) and are labelled `NIm'. (b) The same view is coloured according to sequence similarity (Palmenberg, 1989; Chapman, 1994), with blue being the most conserved rhinoviral amino acids and red being the most variable. Comparison of diagrams like these suggested the `canyon hypothesis' (Rossmann, 1989). The prediction has proved largely true in that the sites of receptor attachment in several picornaviruses would be depressed areas whose sequences could be more highly conserved because they were partially inaccessible to antibodies and therefore not under the same selective pressure to mutate. In this and other applications, the schematic nature of these diagrams has helped in the collation of structure with data arising from the known phenotypes of site-directed or natural mutants. Part (b) is reproduced from Chapman (1993). Copyright (1993) The Protein Society. Reprinted with the permission of Cambridge University Press.

To view all parts of the molecule, cylindrical projections are used that are similar to those used in atlases. This is possible because the representation is schematic (not realistic), and longitudinal distortion, similar to that near the poles in world maps, is acceptable. The surface is projected outwards radially onto a cylinder that wraps around the macromolecule (Fig. 22.1.2.4). Active-site clefts, drug or inhibitor binding sites and pores can be similarly illustrated by projecting their surfaces outward (from the axis) onto a cylinder that encloses the pore, pocket, or cleft. Such clefts are rarely straight, but with some distortion a satisfactory representation is possible by segmenting the cylinder, so that its axis follows the (curved) centre of the binding site or pore (Fig. 22.1.2.4).

Figure 22.1.2.4 | top | pdf | Different projections illustrated with lysozyme. (a) Lysozyme (Blake et al., 1965; Diamond, 1974) is sketched with MOLSCRIPT (Kraulis, 1991) and shown with a ribbon indicting the active-site cleft (Kelly et al., 1979). A cylindrical surface shown unrolled in (b) is shown in (a) wrapped around lysozyme. Vectors orthogonal to the now cylindrical illustrative surface are extended inwards until they intersect with the sphere. Vectors then run towards the centre of the molecule, and their intersections with the solvent-accessible surface are projected back upon the cylinder [unrolled in (b)]. (b) The surface is shaded according to distance from the centre, revealing the substrate-binding cleft as lighter shading. Details of active-site residues are revealed in (c) with a different type of projection. A segmented bent cylinder was traced along the substrate-binding cleft. The surface shows the projection outwards from points on the cylindrical axis. This reveals the amino acids likely to be in most intimate contact with the substrate. Similar plots, coloured according to charge, atomic solvation parameters, or hydrophobicity, can reveal the nature of predominant chemical interactions. This figure is reproduced from Chapman (1993). Copyright (1993) The Protein Society. Reprinted with the permission of Cambridge University Press.