We have optimised the geometry of Cu(imidazole)2(SCH3)(S(CH3)2)+ in vacuum with the density functional B3LYP method. The results [21] are sensational [c.f. New Scientist, 5 July, 1997, p. 26-27]: the oxidised complex assumes a structure that is virtually identical to the crystal structure of plastocyanin. For the reduced system, the vacuum structure seems to be slightly more tetrahedral than the experimental one, but it can be changed to a structure very similar to the crystal structure at an expense of less than 5 kJ/mole, i.e. within the error limits of the method. Thus, the results give no support to the strain hypotheses; instead the protein seems to have minimised the reorganisation energy by an appropriate choice of ligands. This calls for an revaluation of the strain hypotheses, or as prof. B. Malmström, the founder of the induced-rack theory, puts it: "The results described in this paper have forced a complete re-orientation of the thinking about blue copper proteins" [Curr. Opin. Struct. Biol., 2, 1998, 286]. The implications of our results was the theme of a commentary in J. Biol. Inorg. Chem. 2000 [38]. We have also collected our results in three review articles [28,39,45].
Since protein strain can hardly explain the unique properties of the blue copper proteins, we have started to look for alternative explanations. First, we have investigated why copper in the blue proteins prefers a trigonal geometry, whereas small inorganic complexes normally assume a tetragonal structure [26]. We have optimised tetragonal and trigonal structures of Cu(NH3)3X, where X is a number of ligands related to cysteine and methionine, and compared their geometries, electronic structures, and relative energies. In tetragonal structures all four ligands form sigma bonds to the copper ion. Trigonal structures, on the other hand, arise when a ligand makes a pi bond with two lobes of the Cu 3d orbital, thereby formally occupying two positions in a square coordination. Large, soft, and easily polarisable ligands, such as SH- and SeH-, give rise to covalent copper-ligand bonds and structures close to a tetrahedron, which might be trigonal or tetragonal with approximately the same stability. On the other hand, small and hard ligands, such as NH3, H2O, and OH-, give ionic bonds and flattened tetragonal structures. The soft cysteine ligand is essential for the stabilisation of a structure that is close to a tetrahedron (either trigonal or tetragonal), which ensures a low reorganisation energy during electron transfer. The methionine ligand also have a similar (but smaller) effect. The protein is probably important for the geometry by protecting the copper ion from solvated molecules that may act as weak axial ligands.
Second, we have studied a small system, Cu(SH)+, as a minimal model for the Cu-Cys interaction that seems to be crucial for the special properties of the blue copper proteins [41]. We have examined how the method, basis sets, and active space affect the geometry and electronic structure of this system. The results indicate our density functional method performs quite well on the system We have also studied the influence of the theoretical method, basis sets, model size, and solvation on the structure of more realistic blue-copper models [41]. Our results show that for the Cu(II) systems, the results are very stable and accurate. For Cu(I), the MP2 method gives slightly different results, but this seems to be more a problem with MP2 than with the density functional method.
Third, we have in cooperation with Kristine Pierloot in Leuven studied the electronic spectrum of plastocyanin with the quantum chemical CASPT2 method (second-order perturbation theory on a multiconfigurational wave function) [22]. This is the first time a protein spectrum is interpreted with a high-level ab initio method. Again, the result is impressive: six spectral lines were assigned with an error less than 1900 cm-1. The ground state is a mixture between Cu 3d and SCys 3ppi and the dominant blue band in the spectrum is a cysteine to copper charge-transfer excitation. The surrounding enzyme and solvent (modelled by point charges) influence the spectrum appreciably (up to 2000 cm-1).
Fourth, we have studied the spectrum and structure of the so-called rhombic type 1 proteins, e.g. nitrite reductase, pseudoazurin, and cucumber blue protein. They have exactly the same copper ligands as plastocyanin, but they exhibit a different structure, spectrum, and ESR characteristics. Our results show that this is because they have a different electronic structure [26,30], viz. a strongly distorted tetragonal geometry with mainly sigma bonds between the copper ion and the four ligands. In cooperation with Leif Eriksson in Stockholm, we have calculated the ESR parameters for the two structures and showed that the tetragonal models give a rhombic spectrum in contrast to the axial plastocyanin spectrum. The two structures are distinct local minima with almost the same energy (within 7 kJ/mole), separated by a small barrier. In fact, all structural and spectroscopic properties of the various types of blue copper proteins (axial and rhombic type 1, type 1.5, and 2) can be explained by the copper coordination sphere going from trigonal to square-planar, with a concomitant change of the character of the bond between Cu and SCys from pi to sigma [26,30].
In order to explain why some proteins stabilise a trigonal structure whereas others favour the tetragonal structure of the same copper complex we have performed free energy perturbations of blue copper proteins, changing the structure from trigonal to tetragonal, and vice versa [32]. The result shows that the diverging preferences are caused by small changes in the protein (less than 16 kJ/mole), mainly in the angular and electrostatic interactions. Moreover, the results indicate that it is highly unlikely that the proteins may determine the structure and reduction potential of the copper complex by constraining the Cu-Met bond length as Solomon and Malmström have suggested lately.
Fifth, stellacyanin has been studied [27]. It is a blue copper protein where the methionine ligand is replaced by glutamine. It binds at a much shorter distance than methionine and this close interaction results in a trigonal structure that is similar to plastocyanin but a spectrum that is closer to the rhombic proteins. We also show that a change of the coordinating atom (from O to N) cannot explain the spectral shifts observed at elevated pH.
Sixth, we have studied the structure and spectroscopy of two metal-substituted proteins. Co(II)-azurin is characterised by a short Co-O bond and a long Co-SMet bond (compared to the native Cu protein). This is again a result of the electronic structure, being tetrahedral and ionic for Co, but covalent and trigonal for Cu. This difference is reflected in the spectrum of Co-azurin, for which nine excitations have been assigned with an error of less than 2400 cm-1 [34].
A typical blue copper site can be constructed in alcohol dehydrogenase by replacing the native catalytic zinc ion with a Cu(II) ion. This protein has been studied by crystallography and spectroscopy, but several aspects of the coordination are not understood. In cooperation with the experimental groups, we study the structure and electronic spectrum of this enzyme [?]. The results show that the crystal structure is valid (which has been questioned) and that the copper ion makes a pi bond to one thiolate ligand and a sigma bond to the other. The spectral variation observed with various ligands and coenzymes is due to differences in the electrostatic interactions near the copper site.
We have calculated the inner-sphere reorganisation energy of several copper complexes [29]. This gives an estimate of the sutability of the complex for electron transfer and a clue to how evolution has chosen the copper ligands. The reorganisation energy is low already for the complexes in vacuum (60 kJ/mole for the self-exchange inner-sphere reorganisation energy) and again both the cysteine and, to a smaller extent, the methionine ligands are important for the reduction of the reorganisation energy.
Interestingly, the reorganisation energy is even lower in the protein, about 30 kJ/mole. This has been shown by combined quantum chemical and molecular mechanical geometry optimisations [40]. The reason is that electrostatic effects and hydrogen bonds in the protein shorten the Cu-SCys and Cu-N bonds, but more in the reduced structure than in the oxidised one, so that the two structures become more similar to each other than in vacuum. All dihedral angles are also much more similar in the oxidised and reduced structures when optimised in the protein than in vacuum.
It is interesting to compare the blue copper proteins to other related proteins. We have studied the structure and reorganisation energies of the CuA site, found in cytochrome c oxidase and nitrous oxide reductase [44]. This site is essentially a dimer of blue copper sites, where each copper coordinates to a histidine ligand and two bridging cysteine ligands. In addition, they have a weaker axial ligand each, a methionine or a back-bone carbonyl ligand. We have shown that also this site arises naturally from the interactions between the copper ions and their ligands - no strain is necessary, contrary to what has been frequently suggested. In fact, differences in properties between the protein site and small inorganic models of it are caused by strain in the models. Variations in the Cu-Cu, Cu-O, and Cu-SMet interactions cannot change the structure or the reduction potential of the CuA site significantly.
There are two other groups of electron carriers commonly used in biological systems, the iron-sulphur clusters and the cytochromes. We have studied the structure and reorganisation energies also of these two types of sites [42,43]. The cytochromes have the lowest reorganisation energy, less than 10 kJ/mole if the axial ligands are uncharged (as they normally are in electron-carrier proteins). Steric strain in the rigid porphyrin ring reduce the reorganisation energy by about 10 kJ/mole. The iron-sulphur clusters, which consist of one to four iron ions, each surrounded by four sulphur ions, have higher reorganisation energies, 40-90 kJ/mole. However, NH-SCys hydrogen bonds in the protein reduce it significantly, in a similar way as in the blue copper proteins.
Finally, we have investigated why the reduction potentials of the blue copper proteins are higher than those of inorganic copper complexes. We have shown that the Cu-SMet interaction is much less important for the reduction potential than what has been suggested (less than 100 mV) [35]. Even if the other axial ligand (a back-bond carbonyl group), is included in the calculations and the methionine ligand is allowed to be mutated, the reduction potential does not change more than 140 mV. This is in accordance with mutation experiments.
Updated 2001-04-27