Structure of azurin from Alcaligenes denitrificans refinement at 1.8 A resolution and comparison of the two crystallographically independent molecules

Construction of new ligand binding sites in proteins of known structure. II. Grafting of a buried transition metal binding site into Escherichia coli thioredoxin

In an accompanying paper a computational procedure is described, which introduces new ligand-binding sites into proteins of known structure. Here we describe the experimental implementation of one of the designs, which is intended to introduce a copper-binding site into Escherichia coli thioredoxin. The new binding site can be introduced with a minimum of four amino acid changes. The binding site is buried so that structural rules for making mutations in the hydrophobic core of a protein, as well as for the introduction of new functions, are being tested in this experiment. The mutant protein is folded even in the absence of metals, and variants that retain the original activity of thioredoxin can be isolated. The protein has gained a metal-binding site specific for transition metals. The metal co-ordination chemistry at the binding site varies depending on the metal that is introduced into it. Mercury(II) is co-ordinated in the expected manner. Copper(II) binds in a way that was not anticipated in the original design. It appears to use two of the four residues intended to form the co-ordination sphere, and two other residues that were not part of the original set of mutations. It is therefore necessary not only to introduce new functional groups to form a new site, but also to consider and remove alternative modes of binding.

Construction of new ligand binding sites in proteins of known structure. I. Computer-aided modeling of sites with pre-defined geometry

We have devised a molecular model building computer program (DEZYMER) which builds new ligand binding sites into a protein of known three-dimensional structure. It alters only the sequence and the side-chain structure of the protein, leaving the protein backbone fold intact by definition. The program searches for a constellation of backbone positions arranged such that if appropriate side-chains were placed there, they would bind the ligand according to a pre-defined geometry of interaction specified by the experimentalist. These binding sites are introduced by the program by taking into account simple rules such as steric hindrance, atomic close-packing and hydrogen bond patterns, which are known to maintain the integrity of a protein structure to a first approximation. A test case is presented in this paper where the copper binding site found in blue-copper proteins such as plastocyanin, azurin and cupredoxin is introduced into Escherichia coli thioredoxin. The model building of one of the solutions found by the program is presented in some detail. The experimental construction and properties of this new protein are described in an accompanying paper. It is hoped that this program provides a general method for the design of ligand binding sites and enzyme active sites, which can then be tested experimentally.

Resonance Raman spectra of plastocyanin and pseudoazurin: evidence for conserved cysteine ligand conformations in cupredoxins (blue copper proteins)

New resonance Raman (RR) spectra at 15 K are reported for poplar (Populus nigra) and oleander (Oleander nerium) plastocyanins and for Alcaligenes faecalis pseudoazurin. The spectra are compared with those of other blue copper proteins (cupredoxins). In all cases, nine or more vibrational modes between 330 and 460 cm-1 can be assigned to a coupling of the Cu-S(Cys) stretch with Cys ligand deformations. The fact that these vibrations occur at a relatively constant set of frequencies is testimony to the highly conserved ground-state structure of the Cu-Cys moiety. Shifts of the vibrational modes by 1-3 cm-1 upon deuterium exchange can be correlated with N-H...S hydrogen bonds from the protein backbone to the sulfur of the Cys ligand. There is marked variability in the intensities of these Cys-related vibrations, such that each class of cupredoxin has its own pattern of RR intensities. For example, plastocyanins from poplar, oleander, French bean, and spinach have their most intense feature at approximately 425 cm-1; azurins show greatest intensity at approximately 410 cm-1, stellacyanin and ascorbate oxidase at approximately 385 cm-1, and nitrite reductase at approximately 360 cm-1. These variable intensity patterns are related to differences in the electronic excited-state structures. We propose that they have a basis in the protein environment of the copper-cysteinate chromophore. A further insight into the vibrational spectra is provided by the structures of the six cupredoxins for which crystallographic refinements at high resolution are available (plastocyanins from P. nigra, O. nerium, and Enteromorpha prolifera, pseudoazurin from A. faecalis, azurin from Alcaligenes denitrificans, and cucumber basic blue protein). The average of the Cu-S(Cys) bond lengths is 2.12 +/- 0.05 A. Since the observed range of bond lengths falls within the precision of the determinations, this variation is considered insignificant. The Cys ligand dihedral angles are also highly conserved. Cu-S gamma-C beta-C alpha is always near -170 degrees and S gamma-C beta-C alpha-N near 170 degrees. As a result, the Cu-S gamma bond is coplanar with the Cys side-chain atoms and part of the polypeptide backbone. The coplanarity accounts for the extensive coupling of Cu-S stretching and Cys deformation modes as seen in the RR spectrum. The conservation of this copper-cysteinate conformation in cupredoxins may indicate a favored pathway for electron transfer.

Assignment of the 600-MHz 1H-NMR spectrum of amicyanin from Thiobacillus versutus by two-dimensional NMR methods provides information on secondary structure

The nearly complete assignment (pH 6.8; T 310 K) of the 1H-NMR spectrum of reduced amicyanin from Thiobacillus versutus is reported. Experimental evidence is presented, that the structure of the amicyanin contains two beta-sheets, a feature common to plastocyanins and azurins. The loops joining the beta-strands have also been identified. The loop F-G (Thr94-Phe98), together with the flanking residues Cys93 and Met99, comprises three of the four copper ligands and is short compared to similar loops in plastocyanin and azurin. His96 turns out to be the copper ligand that can be protonated. Amicyanin resembles plastocyanin in overall structure but differs from it on account of an N-terminal strand of 22 amino acids in front of strand A, shorter loops A-B, D-E and F-G and the absence of any alpha-helical segments.

Crystal structure analysis of oxidized Pseudomonas aeruginosa azurin at pH 5.5 and pH 9.0. A pH-induced conformational transition involves a peptide bond flip

The X-ray crystal structure of recombinant wild-type azurin from Pseudomonas aeruginosa was determined by difference Fourier techniques using phases derived from the structure of the mutant His35Leu. Two data sets were collected from a single crystal of oxidized azurin soaked in mother liquor buffered at pH 5.5 and pH 9.0, respectively. Both data sets extend to 1.93 A resolution. The two pH forms were refined independently to crystallographic R-factors of 17.6% (pH 5.5) and 17.5% (pH 9.0). The conformational transition previously attributed to the protonation/deprotonation of residue His35 (pKa(red) = 7.3, pKa(ox) = 6.2), which lies in a crevice of the protein close to the copper binding site, involves a concomitant Pro36-Gly37 main-chain peptide bond flip. At the lower pH, the protonated imidazole N delta 1 of His35 forms a strong hydrogen bond with the carbonyl oxygen from Pro36, while at alkaline pH the deprotonated N delta 1 acts as an acceptor of a weak hydrogen bond from HN Gly37. The structure of the remainder of the azurin molecule, including the copper binding site, is not significantly affected by this transition.

Detection of common three-dimensional substructures in proteins

We present a fully automatic algorithm for three-dimensional alignment of protein structures and for the detection of common substructures and structural repeats. Given two proteins, the algorithm first identifies all pairs of structurally similar fragments and subsequently clusters into larger units pairs of fragments that are compatible in three dimensions. The detection of similar substructures is independent of insertion/deletion penalties and can be chosen to be independent of the topology of loop connections and to allow for reversal of chain direction. Using distance geometry filters and other approximations, the algorithm, implemented in the WHAT IF program, is so fast that structural comparison of a single protein with the entire database of known protein structures can be performed routinely on a workstation. The method reproduces known non-trivial superpositions such as plastocyanin on azurin. In addition, we report surprising structural similarity between ubiquitin and a (2Fe-2S) ferredoxin.

Insulin hexamers: new conformations and applications

Recent studies on the structural and chemical properties of insulin have shown that the insulin hexamer is an allosteric protein capable of adopting three distinct conformations, designated T6, T3R3 and R6. Although the physiological consequences of this allostery are not established, new applications for the insulin hexamer as a model system for the study of allostery and for the study of zinc enzymes and copper proteins are emerging.

Beta-lactamase of Bacillus licheniformis 749/C. Refinement at 2 A resolution and analysis of hydration

The crystallographic and molecular structure of the class A beta-lactamase (penicillinase) of Bacillus licheniformis 749/C has been refined with X-ray diffraction data to 2 A resolution. For the 27,330 data with F greater than or equal to 3 sigma(F), the R factor is 0.15; for all 30,090 data, R is 0.16. The estimated co-ordinate error is 0.15 A. In the final model, the deviation of covalent bonds and angles from ideality is 0.012 A and 2.2 degrees, respectively. The model includes two molecules of 29,500 daltons each in the asymmetric unit of space group P2(1), 484 water molecules and two tetrahedral buffer anions. Overlay of the two protein molecules results in a root-mean-square difference of 0.17 A and 0.41 A for alpha-carbon atoms and for all atoms, respectively. Twenty-six water molecules fall within 0.25 A of matching water molecules associated with the second protein molecule. The reactive Ser70 is on a turn of 3(10) helix at the N terminus of a longer alpha-helix (72-83). The penicillin-binding site near this helix contains at least seven water molecules. Upon penicillin entry, a water molecule in the oxyanion hole, hydrogen-bonded between the N terminus of helix (80-83) and beta-strand (230-238), would be displaced by the oxygen atom of the beta-lactam carbonyl group. An unexpelled molecule of water is proposed to be the catalytic water required for penicillin hydrolysis. The water is hydrogen-bonded to Glu166, a conserved residue in all beta-lactamases, and it lies 3 A from the alpha-face of a previously modeled penicillin. The position of the water-Glu166 pair is stabilized in the active site by a cis peptide bond at Pro167.

A fast unbiased comparison of protein structures by means of the Needleman-Wunsch algorithm

A fast dynamic programming algorithm for the spatial superposition of protein structure without prior knowledge of an initial alignment has been developed. The program was applied to serine proteases, hemoglobins, cytochromes C, small copper-binding proteins, and lysozymes. In most cases the existing structural homology could be detected in a completely unbiased way. The results of the method presented are in general agreement with other studies. Applying our method, the different alignment results obtained by other authors for serine proteases and cytochromes C can be classified in terms of different alignment parameters such as gap penalties or cut-off length. Limitations of the method are discussed.

X-ray crystal structure of the two site-specific mutants His35Gln and His35Leu of azurin from Pseudomonas aeruginosa

The three-dimensional structures of two site-specific mutants of the blue copper protein azurin from Pseudomonas aeruginosa have been solved by a combination of isomorphous replacement and Patterson search techniques, and refined by energy-restrained least-squares methods. The mutations introduced by recombinant DNA techniques involve residue His35, which was exchanged for glutamine and leucine, to probe for its suggested role in electron transfer. The two mutants, His35Gln (H35Q) and His35Leu (H35L), crystallize non-isomorphously in the orthorhombic space group P2(1)2(1)2(1) with unit cell dimensions of a = 109.74 A, b = 99.15 A, c = 47.82 A for H35Q, a = 57.82 A, b = 81.06 A, c = 110.03 A for H35L. In each crystal form, there are four molecules in the asymmetric unit. They are arranged as a dimer of dimers in the H35Q case and are distorted from ideal C2 symmetry in H35L. The final crystallographic R-value is 16.3% for 20.747 reflections to a resolution of 2.1 A for H35Q and 17.0% for 32,548 reflections to 1.9 A for H35L. The crystal structures reported here represent the first crystallographically refined structures for azurin from P. aeruginosa. The structure is very similar to that of azurin from Alcaligenes denitrificans. The copper atom is located about 7 A below a hydrophobic surface region and is ligated by five donor groups in a distorted trigonal bipyramidal fashion. The implications for electron transfer properties of the protein are discussed in terms of the mutation site and the packing of the molecules within the tetramer.

Gene synthesis, expression, and mutagenesis of the blue copper proteins azurin and plastocyanin

Genes for the blue copper proteins Populus nigra var. italica plastocyanin and Pseudomonas aeruginosa azurin have been constructed by a stepwise procedure. The leader sequence for azurin has been placed before the genes directing plastocyanin and azurin transport to the periplasmic space when the genes are expressed in Escherichia coli. Site-saturation mutagenesis has been used to alter two copper-binding residues of azurin (Met-121 and His-46) and Met-92 of plastocyanin. While the plastocyanin mutants do not appear to bind copper, the azurin variants all bind copper and show characteristic type I blue copper centers. In particular, the electronic spectra reflect the dominance of the charge transfer interaction between copper and the thiolate of Cys-112, being relatively insensitive to changes in Met-121 or His-46. In contrast, removal of Met-121 appreciably alters the EPR spectra of the mutants, although, to a first order, the spectra of all mutants are themselves similar, suggesting a more distorted geometry around copper in the mutants than in the wild type.

EXAFS analysis of the pH dependence of the blue-copper site in amicyanin from Thiobacillus versutus

The room temperature Cu K-edge EXAFS (extended X-ray absorption fine structure) spectrum of reduced and oxidized amicyanin, the blue copper protein from Thiobacillus versutus, was measured at low and high pH. The data interpretation was partly based on independent NMR evidence for the occurrence of a ligand histidine protonation at low pH (pKa = 6.9) in the reduced protein. In the oxidized protein two nitrogen-donors (from two histidines; Cu-N distances 1.95-2.01 A and 1.86-1.89 A) and a sulfur-donor (from a cysteine; Cu-S distance 2.11-2.13 A) were identified and the coordination appears independent of pH. Upon reduction at high pH the Cu-S bond and one of the Cu-N bonds lengthen slightly (from 2.11 to 2.19 A and from 2.01 to 2.18 A, respectively). Upon lowering of the pH one of the N-donors of the Cu in reduced amicyanin disappears from the Cu EXAFS and a second S-donor (from a methionine) becomes visible at 2.41 A from the Cu. The Debye-Waller factors are compatible with a Cu-N vibrational stretch frequency in the range of 150-250 cm-1 and one greater than 285 cm-1, and a Cu-S vibrational stretch frequency of about 150 cm-1 (Cu-Smet; reduced amicyanin at low pH) and one in the range of 230-800 cm-1 (Cu-Scys).

Copper protein structures

The structural comparison of copper-containing proteins has provided a new dimension to the relationships suggested by sequence similarities. Ryden (1988) summarized the putative relationships, suggesting that a primordial single-domain cupredoxin evolved into the multidomain copper oxidases. The structures have revealed the fact that the differences reside primarily in insertions and deletions at junctions between secondary-structure elements. The mechanism of evolution (e.g., integration of new sequences into regions not essential to the Greek key fold) remains unknown. Which of the properties of a cupredoxin fold are necessary for function is the subject of site-directed mutagenesis studies. Can two of the ligands be interchanged (e.g., the upstream histidine and partially answered by the multidomain copper oxidase structure. The Tyr-Cys-Thr sequence in plastocyanin (in which threonine is a member of the hydrogen-bonding pair) is homologous with the His-Cys-His sequence in ascorbate oxidase. In the latter electron transfer is believed to flow from the type I copper (bound by the cysteine) to the trinuclear cluster, probably via these histidine residues. Hence, one might infer that the tyrosine and threonine have some role in electron transfer. Tyr-83 has been previously implicated in NMR studies as a primary site of electron transfer. The multi-copper protein structures have revealed interesting new features. The extra coppers are bound at domain interfaces, and can be single metals or the novel trinuclear cluster, depending on the availability of liganding histidines. A structural model of ceruloplasmin suggests that it will have at least two type I sites and, possibly, a third type I site such as stellacyanin (no methionine ligand), as well as a binding site for a trinuclear cluster. The similarity of the sequences of N2O reductases and a domain of cytochrome oxidase to the sequences of proteins with known structures suggests that these, too, will have Greek key domains. Galactose oxidase and hemocyanin do not have Greek key folds in their functional domains, although each does have a Greek key domain. The need for a Greek key fold remains obscure. The apoproteins are clearly stable without metals; there are examples other than immunoglobulins of Greek key folds. So far copper seems to be found in a very limited subset of structures; other chapters in this volume show that zinc, for example, has a much wider variety of environments in proteins, as does iron. It may be that the copper-containing Greek key proteins represent a very small evolutionary niche.(ABSTRACT TRUNCATED AT 400 WORDS)

Hydrogen bond stereochemistry in protein structure and function

Fifty high resolution protein structures from the Brookhaven Protein Data Bank have been analyzed for recurring motifs in hydrogen bond stereochemistry. Although an exhaustive analysis of hydrogen bond statistics has been presented by Baker & Hubbard, a detailed stereochemical analysis of classical donor (N-H, O-H, or S-H) and acceptor (N:, O:, or S:) structure within proteins is lacking. Here, we describe the preferential hydrogen bond stereochemistry for the side-chains of glutamate and aspartate (carboxylate), glutamine and asparagine (carboxamide), arginine (guanidinium), histidine (imidazole/imidazolium), tryptophan (indole), tyrosine (phenolic hydroxyl), lysine (ammonium), serine and threonine (alkyl hydroxyl), cysteine (thiol), methionine (thioether) and cystine (disulfide). Preferential hydrogen bond stereochemistry is governed by (1) the electronic configuration of acceptor atoms, (2) the steric accessibility of donor atoms and (3) the conformation of amino acid side-chains. Applications of hydrogen bond stereochemistry are useful in the interpretation of protein structure, function and stability. Additionally, this stereochemistry is a prerequisite to the interpretation of protein-other molecule recognition and biological catalysis.

A new tool for the qualitative and quantitative analysis of protein surfaces using B-spline and density of surface neighborhood

To improve the qualitative and quantitative analysis of surfaces of protein, two new methods are proposed: one that smoothes the MS surface of Connolly with B-spline smoothing functions to highlight the significant features of the surface, and one that computes the density of surface neighborhood to allow quantitative comparison.

Improvements in protein secondary structure prediction by an enhanced neural network

Computational neural networks have recently been used to predict the mapping between protein sequence and secondary structure. They have proven adequate for determining the first-order dependence between these two sets, but have, until now, been unable to garner higher-order information that helps determine secondary structure. By adding neural network units that detect periodicities in the input sequence, we have modestly increased the secondary structure prediction accuracy. The use of tertiary structural class causes a marked increase in accuracy. The best case prediction was 79% for the class of all-alpha proteins. A scheme for employing neural networks to validate and refine structural hypotheses is proposed. The operational difficulties of applying a learning algorithm to a dataset where sequence heterogeneity is under-represented and where local and global effects are inadequately partitioned are discussed.

Isolation and sequencing of the Alcaligenes denitrificans azurin-encoding gene: comparison with the genes encoding blue copper proteins from Pseudomonas aeruginosa and Alcaligenes faecalis

The gene (azu) encoding azurin from Alcaligenes denitrificans has been cloned and sequenced. The gene codes for a pre-protein with a 19-aa signal peptide. Comparison with the sequences coding for the blue copper proteins from Pseudomonas aeruginosa and Alcaligenes faecalis reveals the presence of ntrA and fnr boxes in front of all three genes, instead of a regular [-10, -35]-promoter. In P. aeruginosa, the azu gene is terminated by a bidirectional terminator and flanked by open reading frames on the opposite strand.

Unusual stability properties of a reptilian ceruloplasmin

Ceruloplasmin from the turtle Caretta caretta was isolated to purity by using the single-step procedure recently developed by us to purify sheep and chicken ceruloplasmin. It has a Mr of ca. 145,000 and a total copper content of 5.1 +/- 0.2 atoms of copper per molecule, 50% of which are detectable by EPR. The spectroscopic features include an absorption maximum at 603 nm in the electronic spectrum and the total absence of any resonance attributable to Type 2 copper in the EPR spectrum. Turtle ceruloplasmin was found to be unusually resistant to aging and proteolysis, when compared to ceruloplasmins isolated from other species. p-Phenyl-endiamine oxidase activity measurements revealed an unusually low catalytic efficiency, while the kinetic parameters of Fe(II) oxidation were consistent with those reported for other species of ceruloplasmin.

Novel method for the rapid evaluation of packing in protein structures

There has been considerable effort to predict the structure of proteins from their amino acid sequences. A major problem in all prediction efforts has been that, short of a direct comparison with crystallographic co-ordinates, it is often difficult to evaluate the merit of a model, or "proposed" protein structure. Here, we present a method for evaluating proposed protein structures that does not require a structural model of complete atomic detail. Our method evaluates residue-residue packing density using a simplified model of the polypeptide chain where amino acids are represented as one, two (histidine, tyrosine and phenylalanine), or three (tryptophan) spheres. This method also gives a measure of the appropriateness of residue-residue contacts, thus giving a measure of the amino acid distribution throughout the protein. Amino acid packing and amino acid distribution, as evaluated by this technique, are consistent with the accuracy of model-built structures. We have been able to select the best structures from a set of combinatorially generated models using this method, and we anticipate that it will be useful as a general tool for model-building.

Beta-lactamase of Bacillus licheniformis 749/C at 2 A resolution

Two crystal forms (A and B) of the 29,500 Da Class A beta-lactamase (penicillinase) from Bacillus licheniformis 749/C have been examined crystallographically. The structure of B-form crystals has been solved to 2 A resolution, the starting model for which was a 3.5 A structure obtained from A-form crystals. The beta-lactamase has an alpha + beta structure with 11 helices and 5 beta-strands seen also in a penicillin target DD-peptidase of Streptomyces R61. Atomic parameters of the two molecules in the asymmetric unit were refined by simulated annealing at 2.0 A resolution. The R factor is 0.208 for the 27,330 data greater than 3 sigma (F), with water molecules excluded from the model. The catalytic Ser-70 is at the N-terminus of a helix and is within hydrogen bonding distance of conserved Lys-73. Also interacting with the Lys-73 are Asn-132 and the conserved Glu-166, which is on a potentially flexible helix-containing loop. The structure suggests the binding of beta-lactam substrates is facilitated by interactions with Lys-234, Thr-235, and Ala-237 in a conserved beta-strand peptide, which is antiparallel to the beta-lactam's acylamido linkage; an exposed cavity near Asn-170 exists for acylamido substituents. The reactive double bond of clavulanate-type inhibitors may interact with Arg-244 on the fourth beta-strand. A very similar binding site architecture is seen in the DD-peptidase.

A retroviral Cys-Xaa2-Cys-Xaa4-His-Xaa4-Cys peptide binds metal ions: spectroscopic studies and a proposed three-dimensional structure

Retroviral gag gene-encoded core nucleic acid binding proteins contain either one or two sequences of the form Cys-Xaa2-Cys-Xaa4-His-Xaa4-Cys. Previously, one of us has proposed that these sequences form metal-binding domains in analogy with the "zinc finger" domains first observed in transcription factor IIIA. We report that an 18-amino acid peptide derived from the core nucleic acid binding protein from Rauscher murine leukemia virus binds metal ions such as Co2+ and Zn2+. The absorption spectrum of the peptide-Co2+ complex is highly suggestive of tetrahedral coordination involving three cysteinates and one histidine. Titration experiments indicate that the dissociation constant for the peptide-Co2+ complex is 1.0 microM and that Zn2+ binds more tightly than Co2+. A detailed three-dimensional structure for this domain based on conserved substructures in other crystallographically characterized metalloproteins and on a detailed analysis of the Cys-Xaa2-Cys-Xaa4-His-Xaa4-Cys sequences from retroviruses and other related sources is proposed.