Marjorie M. Harding

Metals are present in more than one-third of all proteins as they occur in nature, and are usually important in biological function or maintenance of the structure. Some are present as cations, directly associated with amino acid functional groups of the protein, others within small molecule cofactors associated with the protein. For the 10 metals commonly found as cations, Na, Mg, K, Ca, Mn, Fe, Co, Ni, Cu and Zn, a survey is given of occurrence, relative frequencies of both metal and donor atom or group type, and geometry of coordination. The survey is based on crystal structure information deposited in the Protein Data Bank (PDB) [Berman, H.; Henrick, K.; Nakamura, H.; Markley, J.L. The Worldwide Protein Data Bank (wwPDB): Ensuring a Single, Uniform Archive of PDB Data. Nucleic Acids Res. 2007, 35, D301–D303]. The precision and reliability of this information is assessed in detail. Illustrative examples are given for each metal, including, usually, details of the structure of the metal site in relation to the whole protein and to its function; there are comparisons between metals and descriptions of features such as binding to carboxylate and multiple metal sites close to each other. Metals found within cofactors which associate with the protein, most notably Mo, are included within these examples. Also included briefly are the prediction of metal sites in proteins resulting from genomic synthesis, information which can be derived from methods other than X-ray diffraction of crystals, and metal–protein systems which do not occur naturally.

A database with details of the geometry of metal sites in proteins has been set up. The data are derived from metalloprotein structures that are in the Protein Data Bank [PDB; Berman, Henrick, Nakamura & Markley (2006).Nucleic Acids Res.35, D301–D303] and have been determined at 2.5 Å resolution or better. The database contains all contacts within the crystal asymmetric unit considered to be chemical bonds to any of the metals Na, Mg, K, Ca, Mn, Fe, Co, Ni, Cu or Zn. The stored information includes PDB code, crystal data, resolution of structure determination, refinement program andRfactor, protein class (from PDB header), contact distances, atom names of metal and interacting atoms as they appear in the PDB, site occupancies,Bvalues, coordination numbers, information on coordination shapes, and metal–metal distances. This may be accessed by SQL queries, or by a user-friendly web interface which searches for contacts between specified types of atoms [for example Ca and carboxylate O of aspartate, Co and imidazole Nδ of histidine] or which delivers details of all the metal sites in a specified protein. The web interface allows graphical display of the metal site, on its own or within the whole protein molecule, and may be accessed at http://eduliss.bch.ed.ac.uk/MESPEUS/. Some applications are briefly described, including a study of the characteristics of Mg sites that bind adenosine triphosphate, the derivation of an average Mg—O_phosphatedistance and some problems that arise when average bond distances with high precision are required.

A new analysis has been made of distances around metal sites in protein structures in the Protein Data Bank determined with resolution < or =1.25 A and equivalent distances have been extracted from the Cambridge Structural Database. They are for the metals Na, Mg, K, Ca, Mn, Fe, Co, Cu, Zn and the donor atoms O of water, O of Asp and Glu, O of the main-chain carbonyl group, N of His and S of Cys. Some revisions are recommended to the tables of 'target distances' previously given. As well as small changes in many distances and a large improvement for Mg-O(carboxylate), the table includes an indication of how reliable each prediction may be. Special attention was given to carboxylate interactions. When the carboxylate group is monodentate, the M-O(carboxylate) distance is well defined, but for bidentate carboxylate groups a wide range of distances is allowable; when the metal is Co, Cu or Zn the M-O(1) and M-O(2) distances are clearly inversely correlated; for the more purely electrostatic interactions involving Na, K and Ca there is a wider scatter of distances and little correlation.

A set of tables is presented and a survey given of the architecture of metal coordination groups in a representative set of protein structures from the Protein Data Bank [Bernstein et al. (1977), J. Mol. Biol. 112, 535-542; Berman et al. (2000), Nucleic Acids Res. 28, 235-242]. The structures have been determined to a resolution of 2.5 A or better; the metals considered are Ca, Mg, Mn, Fe, Cu, Zn, Na and K, with particular emphasis on Ca and Zn and the exclusion of haem groups and Fe/S clusters; the proteins are a representative set in which none has more than 30% sequence identity with any other. In them the metal is coordinated by several donor groups from different amino-acid residues in the protein chain and often also by water or other small molecules. The tables, for approximately 600 metal coordination groups, include information on the conformations of the protein chain in the region around the metal and reliability indicators. They illustrate the wide variety of coordination numbers, chelate-loop sizes and other properties and the different characteristics of different metals. They show that glycine has a particular significance in the position adjacent to a donor residue, especially in Ca coordination groups. They also show that metal coordination does not appear to lead to significant distortions of the torsion angles phi, psi from their normally allowed values. Very few metal coordination groups occur more than once in the representative set and when they do they are usually related in fold and function; they have similar but not necessarily identical conformations. However, individual chelate loops, for example Zn(-C-X-X'-C-), in which both cysteines are coordinated to Zn through S, and X and X' are any amino acids, are repeated frequently in many different and unrelated proteins. Not all chelate loops with the same composition have the same conformation, but for smaller loops there are usually one or two strongly preferred and well defined conformations. Quite frequently more than one metal coordination group is associated with one protein chain; these proteins are identified.

In previous papers [Harding (2001), Acta Cryst. D57, 401-411, and references therein] the geometry of metal-ligand interactions was examined for six metals (Ca, Mg, Mn, Fe, Cu, Zn) using the Protein Data Bank and compared with information from accurately determined structures of relevant small-molecule crystals in the Cambridge Structural Database. Here, the environments of Na(+) and K(+) ions found in protein crystal structures are examined in an equivalent way. Target M(+).O distances are proposed and the agreement with observed distances is summarized. The commonest interactions are with water molecules and the next commonest with main-chain carbonyl O atoms.

The geometry of metal-ligand interactions in proteins is examined and compared with information for small-molecule complexes from the Cambridge Structural Database (CSD). The paper deals with the metals Ca, Mg, Mn, Fe, Cu, Zn and with metal-donor atom distances, coordination numbers and extent of distortion from ideal geometry (octahedral, tetrahedral etc.). It assesses the agreement between geometry found in all metalloprotein structures in the Protein Data Bank (PDB) determined at resolution < or = 1.6 A with that predicted from the CSD for ligands which are analogues of amino-acid side chains in proteins [Harding (1999), Acta Cryst. D55, 1432-1443; Harding (2000), Acta Cryst. D56, 857-867]. The agreement is reasonably good for these structures but poorer for many determined at lower resolution (examined to 2.8 A resolution). For metal-donor distances, the predictions from the CSD, with minor adjustments, provide good targets either for validation or for restraints in refinement of structures where only poorer resolution data is available. These target distances are tabulated and the use of restraints is recommended. Validation of angles or the use in refinement of restraints on angles at the metal atom is more difficult because of the inherent flexibility of these angles. A much simplified set of parameters for angle restraints with quite large standard deviations is provided. (Despite the flexibility of the angles, acceptable and preferred coordination numbers and shapes are well established and a summary table is provided.) An unusual and perhaps biochemically important feature of Zn coordination with carboxylate seen in the CSD examples is also clearly present in metalloprotein structures. With metals like Ca, carboxylate coordination is monodentate or bidentate (two M-O bonds of nearly equal length). In Zn carboxylates a continuous range between monodentate and bidentate coordination is found, with one Zn-O bond of normal length and another of any length between this and a van der Waals contact.

Geometrical data which could be of relevance in the structure determination, structure refinement, assessment or understanding of metalloproteins have been extracted from the Cambridge Structural Database (CSD). The CSD contains crystallographic data from 'small-molecule' structures determined by X-ray or neutron diffraction to an accuracy and precision much better than that of most current protein structure determinations. Structures of Mg, Mn, Fe, Cu and Zn complexes with ligands whose donor atoms may be only N, O, S or Cl have been selected and analysed in terms of the geometry of the metal coordination group - octahedral, tetrahedral, tetragonal pyramidal etc. The r.m.s. deviation of all the interbond angles around the metal atom provides a measure, delta, of the deviation from ideal geometry. Average values of delta are tabulated for the different metals in each type of complex. For simple non-chelated complexes of Mn, Fe and Zn, distortions of up to 5 degrees in octahedral complexes and 10 degrees in tetrahedral complexes are found to be normal and seem likely to be a consequence of packing effects, ligand bulk or intramolecular effects. Substantially larger distortions are found for some other metals and geometries and are common for chelated complexes. Brief comments on six-, seven- and eight-coordinate Ca complexes are included. Tables are also presented showing that for four- and five-coordinate complexes of Zn and Cu it is quite common to find additional weakly coordinated ligands, usually with N or O donor atoms and with M.N,O distances longer than a normal bond length but shorter than a van der Waals contact, e.g. in the range 2.4-3.0 A for Zn and 2.6-3.0 A for Cu. Although the contributions to bond valency or bonding energy of such interactions may not be great, their effect on geometry can be considerable; they can, for example, cause much larger distortions of tetrahedral Zn complexes than indicated above.

Geometrical data which could be of relevance in the structure determination, structure refinement, assessment or understanding of metalloproteins have been extracted from the Cambridge Structural Database (CSD). The CSD contains crystallographic data from `small-molecule' structures determined by X-ray or neutron diffraction to an accuracy much better than that of most current protein structure determinations. The structures selected have a crystallographic R factor ≤ 0.065 and contain Ca, Mg, Mn, Fe, Cu or Zn interacting with ligands which are analogues of the amino-acid side chains commonly found in proteins; they include carboxylate groups, alcohols, phenolates, thiolates, imidazole groups and also water molecules. For each pair, the mean metal–donor-atom distance, the sample standard deviation and the range of observed values are tabulated, using ∼4500 observations in all. Where practicable, subsets with different coordination numbers and/or oxidation states are given. Also included are inter-bond angles at the ligand donor atom, the orientation of carboxylate and imidazole groups with respect to the metal–donor-atom bond and some other aspects of ligand geometry. Thus, for example, target distances and their standard deviations could be easily looked up for the validation of a metalloprotein structure or for use in restrained refinement with low-resolution data.