Proteins at atomic resolution.

Experimental advances in data collection, including bright sources, cryogenic cooling and two-dimensional detectors, have made it tractable to record data to beyond 1.2 A for several proteins, yielding high-accuracy models and fine details of structure. For small metalloproteins, atomic-resolution data have enabled ab initio solution of the phase problem.

How nature deals with stereoisomers.

All natural proteins are composed of L-amino acids and are inherently chiral. The properties of both L- and chemically synthesized D-amino acids are identical except in optically asymmetric interactions. Structural studies of D-I racemic mixtures of crystallographic interest are discussed. The review also gives some recent examples of stereospecificity: how L-proteins deal with L- or D-substrates and how enzymes can function as racemases. Two particular examples of stereoselectivity are then discussed.

The benefits of atomic resolution.

After a long gestation, the elucidation of the crystal structures of proteins at atomic resolution is now maturing. The use of such data for both refinement and structure solution is advancing space. The necessary technology is generally available, in terms of data collection, computing hardware and software. The structures appearing in the literature mainly relate to demonstration projects on native proteins. The importance of these alone is already obvious. Biologically significant results, in terms of ligand complexes and prosthetic groups, are just starting to emerge.

[Three-dimensional structure of yellow fluorescent protein zYFP538 from Zoanthus sp. at the resolution 1.8 angstrom].

The three-dimensional structure of yellow fluorescent proteins zYFP538 (zFP538) from the button polyp Zoanthus sp. was determined at a resolution of 1.8 angstrom by X-ray analysis. The monomer of zYFP538 adopts a structure characteristic of the green fluorescent protein (GFP) family, a beta-barrel formed from 11 antiparallel beta segments and one internal alpha helix with a chromophore embedded into it. Like the TurboGFP, the beta-barrel of zYFP538 contains a water-filled pore leading to the chromophore Tyr67 residue, which presumably provides access of molecular oxygen necessary for the maturation process. The post-translational modification of the chromophore-forming triad Lys66-Tyr67-Gly68 results in a tricyclic structure consisting of a five-membered imidazolinone ring, a phenol ring of the Tyr67 residue, and an additional six-membered tetrahydropyridine ring. The chromophore formation is completed by cleavage of the protein backbone at the Calpha-N bond of Lys66. It was suggested that the energy conflict between the buried positive charge of the intact Lys66 side chain in the hydrophobic pocket formed by the Ile44, Leu46, Phe65, Leu204 and Leu219 side chains is the most probable trigger that induces the transformation of the bicyclic green form to the tricyclic yellow form. A stereochemical analysis of the contacting surfaces at the intratetramer interfaces helped reveal a group of conserved key residues responsible for the oligomerization. Along with others, these residues should be taken into account in designing monomeric forms suitable for practical application as markers of proteins and cell organelles.

Atomic resolution structure of the double mutant (K53,56M) of bovine pancreatic phospholipase A2.

The structure of the double mutant K53,56M has previously been refined at 1.9 A resolution using room-temperature data. The present paper reports the crystal structure of the same mutant K53,56M refined against 1.1 A data collected using synchrotron radiation. A total of 116 main-chain atoms from 29 residues and 44 side chains are modelled in alternate conformations. Most of the interfacial binding residues are found to be disordered and alternate conformations could be recognized. The second calcium ion-binding site residue Glu92 adopts two alternate conformations. The minor and major conformations of Glu92 correspond to the second calcium ion bound and unbound states.

Crystal structure of a bacterial chitinase at 2.3 A resolution.

BACKGROUND: Chitinases cleave the beta-1-4-glycosidic bond between the N-acetyl-D-glucosamine units of which chitin is comprised. Chitinases are present in plants, bacteria and fungi, but whereas structures are available for two prototypic plant enzymes, no structure is available for a bacterial or fungal chitinase. RESULTS: To redress this imbalance, the structure of native chitinase A from Serratia marcescens has been solved by multiple isomorphous replacement and refined at 2.3 A resolution, resulting in a crystallographic R-factor of 16.2%. The enzyme comprises three domains: an all beta-strand amino-terminal domain, a catalytic alpha/beta-barrel domain, and a small alpha+beta-fold domain. There are several residues with unusual geometries in the structure. Structure determination of chitinase A in complex with N,N',N",N"'-tetra-acetylo-chitotetraose, together with biochemical and sequence analysis data, enabled the positions of the active-site and catalytic residues to be proposed. CONCLUSIONS: The reaction mechanism seems to be similar to that of lysozyme and most other glycosylhydrolases, i.e. general acid-base catalysis. The role of the amino-terminal domain could not be identified, but it has similarities to the fibronectin III domain. This domain may possibly facilitate the interaction of chitinase A with chitin.

2.2 A resolution structure of the amino-terminal half of HIV-1 reverse transcriptase (fingers and palm subdomains).

BACKGROUND: HIV-1 reverse transcriptase (RT) catalyzes the transformation of single-stranded viral RNA into double-stranded DNA, which is integrated into host cell chromosomes. The molecule is a heterodimer of two subunits, p51 and p66. The amino acid sequence of p51 is identical to the sequence of the amino-terminal subdomains of p66. Earlier crystallographic studies indicate that the RT molecule is flexible, which may explain the difficulty in obtaining high-resolution data for the intact protein. We have therefore determined the structure of a fragment of RT (RT216), which contains only the amino-terminal half of the RT molecule ('finger' and 'palm' subdomains). RESULTS: The crystal structure of RT216 has been refined at 2.2 A resolution to a crystallographic R-value of 20.8%. The structure is very similar to that of the corresponding part of the p66 subunit in the p66/p51 heterodimer, although there is a small difference in the relative orientation of the two subdomains compared with the structure of an RT-DNA-antibody fragment complex. There are a large number of stabilizing contacts (mainly hydrogen bonds and hydrophobic interactions) between the subdomains. The locations of conserved amino acids and the position of some important drug-resistant mutations are described. CONCLUSIONS: The RT216 structure provides detailed three-dimensional information of one important part of HIV-1 RT (including the critical active site residues). We propose a model to explain the inhibitory effect of non-nucleoside inhibitors, which partially accounts for their effect in terms of conformational changes of active site residues.

Pancreatic spasmolytic polypeptide: first three-dimensional structure of a member of the mammalian trefoil family of peptides.

BACKGROUND: The trefoil peptides are a rapidly growing family of peptides, mainly found in the gastrointestinal tract. There is circumstantial evidence that they stabilize the mucus layer, and may affect the rate of healing of the mucosal epithelium. RESULTS: We have determined the structure of porcine pancreatic spasmolytic polypeptide (PSP) to 2.5 A resolution. The polypeptide contains two trefoil domains. The domain structure is compact, and is composed of a central short antiparallel beta-sheet with one short helix above and one below it. This is a novel motif. The two domains are related by two-fold symmetry, and each domain contains a cleft. CONCLUSIONS: The cleft within each domain could accommodate a polysaccharide chain, and may therefore be responsible for binding mucin glycoproteins. We suggest that PSP may cross-link glycoproteins, explaining its ability to stabilize the mucus layer.

Crystal structure of the human acyl protein thioesterase I from a single X-ray data set to 1.5 A.

BACKGROUND: Many proteins undergo posttranslational modifications involving covalent attachment of lipid groups. Among them is palmitoylation, a dynamic, reversible process that affects trimeric G proteins and Ras and constitutes a regulatory mechanism for signal transduction pathways. Recently, an acylhydrolase previously identified as lysophospholipase has been shown to function as an acyl protein thioesterase, which catalyzes depalmitoylation of Galpha proteins as well as Ras. Its amino acid sequence suggested that the protein is evolutionarily related to neutral lipases and other thioesterases, but direct structural information was not available. RESULTS: We have solved the crystal structure of the human putative Galpha-regulatory protein acyl thioesterase (hAPT1) with a single data set collected from a crystal containing the wild-type protein. The phases were calculated to 1.8 A resolution based on anomalous scattering from Br(-) ions introduced in the cryoprotectant solution in which the crystal was soaked for 20 s. The model was refined against data extending to a resolution of 1.5 A to an R factor of 18.6%. The enzyme is a member of the ubiquitous alpha/beta hydrolase family, which includes other acylhydrolases such as the palmitoyl protein thioesterase (PPT1). CONCLUSIONS: The human APT1 is closely related to a previously described carboxylesterase from Pseudomonas fluorescens. The active site contains a catalytic triad of Ser-114, His-203, and Asp-169. Like carboxylesterase, hAPT1 appears to be dimeric, although the mutual disposition of molecules in the two dimers differs. Unlike carboxylesterase, the substrate binding pocket and the active site of hAPT1 are occluded by the dimer interface, suggesting that the enzyme must dissociate upon interaction with substrate.

Ab initio determination of the crystal structure of cytochrome c6 and comparison with plastocyanin.

BACKGROUND: Electron transfer between cytochrome f and photosystem I (PSI) can be accomplished by the heme-containing protein cytochrome c6 or by the copper-containing protein plastocyanin. Higher plants use plastocyanin as the only electron donor to PSI, whereas most green algae and cyanobacteria can use either, with similar kinetics, depending on the copper concentration in the culture medium. RESULTS: We report here the determination of the structure of cytochrome c6 from the green alga Monoraphidium braunii. Synchrotron X-ray data with an effective resolution of 1.2 A and the presence of one iron and three sulfur atoms enabled, possibly for the first time, the determination of an unknown protein structure by ab initio methods. Anisotropic refinement was accompanied by a decrease in the 'free' R value of over 7% the anisotropic motion is concentrated at the termini and between residues 38 and 53. The heme geometry is in very good agreement with a new set of heme distances derived from the structures of small molecules. This is probably the most precise structure of a heme protein to date. CONCLUSIONS: On the basis of this cytochrome c6 structure, we have calculated potential electron transfer pathways and made comparisons with similar analyses for plastocyanin. Electron transfer between the copper redox center of plastocyanin to PSI and from cytochrome f is believed to involve two sites on the protein. In contrast, cytochrome c6 may well use just one electron transfer site, close to the heme unit, in its corresponding reactions with the same two redox partners.

Anomalous signal of solvent bromides used for phasing of lysozyme.

The anomalous signal of bromide ions, present in the crystal structure of tetragonal hen egg-white lysozyme through the substitution of NaCl by NaBr in the crystallization medium, was used for phasing of X-ray data collected to 1.7 A resolution with a wavelength near the absorption edge of bromine. Phasing of a single wavelength data set, based purely on anomalous deltaf " contribution, led to easily interpretable electron density, equivalent to the complete multiwavelength anonalous dispersion phasing based on four-wavelength data. The classic small-structure direct methods program SHELXS run against all anomalous differences gave a successful solution of six highest peaks corresponding to six bromide ions in the structure with data limited up to a resolution of 3.5 A. Interpretable maps were obtained at a resolution up to 3.0 A using programs MLPHARE and DM. Bromide ions occupy well ordered positions at the protein surface. Phasing based on the single wavelength signal of anomalous scatterers introduced into the ordered solvent shell can be proposed as a tool for solving structures of well diffracting crystals.

Crystallization of recombinant chitobiase from Serratia marcescens.

We are currently investigating the biochemical and structural properties of both chitin degrading enzymes chitinase and chitobiase from Serratia marcescens. Previously we have reported the first crystallization and characterization of chitinase crystals (Vorgias et al., 1992). In this communication we present the first crystallization of chitobiase. The protein was synthesized in Escherichia coli and purified to homogeneity using cation exchange chromatography and fast protein liquid chromatography. The crystals have the shape of small prisms and the space group is P2(1) with beta = 101.0 degrees and unit cell dimensions a = 63.2 A, b = 133.2 A, c = 55.1 A. They diffract X-rays to about 2.5 A resolution and are suitable for three-dimensional structural analysis.

Crystallization of recombinant chitinase from the cloned chiA gene of Serratia marcescens.

The chiA gene encoding for the chitinase enzyme from Serratia marcescens was efficiently overexpressed under the pL promoter and the enzyme was secreted into the growth medium. The chitinase was purified to homogeneity using affinity chromatography on a Phenyl-Sepharose column and the protein was successfully crystallized. The crystals are presently in the form of small needles in space group C222(1) and have unit cell dimensions a = 204(+/- 0.5) A, b = 134(+/- 0.5) A, c = 60(+/- 0.5) A. The crystals diffract X-rays to about 3 A resolution and are suitable for three-dimensional structural analysis.

Crystallization and preliminary diffraction studies of 5 S rRNA from the thermophilic bacterium Thermus flavus.

Crystals of purified 5 S rRNA from Thermus flavus have been obtained. The crystals diffract up to 8 A resolution, using synchrotron radiation, and have the monoclinic space-group C2. The unit cell has the dimensions a = 190 A, b = 110 A, c = 138 A and beta = 117 degrees. The cell volume suggests the presence of four 5 S rRNA molecules per asymmetric unit.

Crystal structure of the closed form of chicken cytosolic aspartate aminotransferase at 1.9 A resolution.

The crystal structure of chicken cytosolic aspartate aminotransferase (cAATase; EC 2.6.1.1) has been solved and refined at 1.9 A resolution. Orthorhombic crystals, space group P2(1)2(1)2(1), a = 56.4 A, b = 126.0 A and c = 142.3 A, were grown from polyethylene glycol solutions in the presence of maleate, a dicarboxylic inhibitor that forms a Michaelis-like complex. The pyridoxal form of the enzyme was used for crystallization. Diffraction data were collected using synchrotron radiation. The structure of the new orthorhombic crystal form was solved by molecular replacement using the partially refined 2.8 A resolution structure of the high-salt crystal form as a search model. The final value of the crystallographic R-factor after rigid body and restrained least-squares refinement is 0.175 with very good model geometry. The two 2-fold-related subunits of cAATase have distinct environments in the crystal lattice. Domain movement is strictly hindered by the lattice contacts in one subunit, while the second one possesses conformational freedom. Despite their different environments, both subunits were found in the closed conformation with one maleate molecule tightly bound in each active site. The present study allows a detailed comparison of the highly refined structures of the aspartate aminotransferase isozymes, and thus provide better insight into the role of conserved and variable residues in substrate recognition and catalysis.

High resolution structures of holo and apo formate dehydrogenase.

Three-dimensional crystal structures of holo (ternary complex enzyme-NAD-azide) and apo NAD-dependent dimeric formate dehydrogenase (FDH) from the methylotrophic bacterium Pseudomonas sp. 101 have been refined to R factors of 11.7% and 14.8% at 2.05 and 1.80 A resolution, respectively. The estimated root-mean-square error in atomic co-ordinates is 0.11 A for holo and 0.18 A for apo. X-ray data were collected from single crystals using an imaging plate scanner and synchrotron radiation. In both crystal forms there is a dimer in the asymmetric unit. Both structures show essentially 2-fold molecular symmetry. NAD binding causes movement of the catalytic domain and ordering of the C terminus, where a new helix appears. This completes formation of the enzyme active centre in holo FDH. NAD is bound in the cleft separating the domains and mainly interacts with residues from the co-enzyme binding domain. In apo FDH these residues are held in essentially the same conformation by water molecules occupying the NAD binding region. An azide molecule is located near the point of catalysis, the C4 atom of the nicotinamide moiety of NAD, and overlaps with the proposed formate binding site. There is an extensive channel running from the active site to the protein surface and this is supposed to be used by substrate to reach the active centre after NAD has already bound. The structure of the active site and a hypothetical catalytic mechanism are discussed. Sequence homology of FDH with other NAD-dependent formate dehydrogenases and some D-specific dehydrogenases is discussed on the basis of the FDH three-dimensional structure.

Crystal structure of dUTPase from equine infectious anaemia virus; active site metal binding in a substrate analogue complex.

The X-ray structures of dUTPase from equine infectious anaemia virus (EIAV) in unliganded and complexed forms have been determined to 1.9 and 2.0 A resolution, respectively. The structures were solved by molecular replacement using Escherichia coli dUTPase as search model. The exploitation of a relatively novel refinement approach for the initial model, combining maximum likelihood refinement with stereochemically unrestrained updating of the model, proved to be of crucial importance and should be of general relevance.EIAV dUTPase is a homotrimer where each subunit folds into a twisted antiparallel beta-barrel with the N and C-terminal portions interacting with adjacent subunits. The C-terminal 14 and 17 amino acid residues are disordered in the crystal structure of the unliganded and complexed enzyme, respectively. Interactions along the 3-fold axis include a water-containing volume (size 207 A3) which has no contact with bulk solvent. It has earlier been shown that a divalent metal ion is essential for catalysis. For the first time, a putative binding site for such a metal ion, in this case Sr2+, is established. The positions of the inhibitor (the non-hydrolysable substrate analogue dUDP) and the metal ion in the complex are consistent with the location of the active centre established for trimeric dUTPase structures, in which subunit interfaces form three surface clefts lined with evolutionary conserved residues. However, a detailed comparison of the active sites of the EIAV and E. coli enzymes reveals some structural differences. The viral enzyme undergoes a small conformational change in the uracil-binding beta-hairpin structure upon dUDP binding not observed in the other known dUTPase structures.

Structures of the lectin IV of Griffonia simplicifolia and its complex with the Lewis b human blood group determinant at 2.0 A resolution.

The structures of the fourth lectin isolated from Griffonia simplicifolia (GS4) and its complex with the methyl-glycoside of the Lewis b human blood group determinant (Le(b)-OMe) are reported at high resolution. The native GS4 crystal is isomorphous with the complexed GS4 crystal. The space group is P4(2)2(1)2 with unit cell dimensions a = 78.9 A, c = 89.1 A with one subunit of the lectin (bound to 1 Le(b)-OMe in the complex) in the crystallographic asymmetric unit. The native GS4 structure was solved by the molecular replacement technique and least-squares refined (PROLSQ and X-PLOR). The orientation of the Le(b)-OMe tetrasaccharide in the complex was established from a 2.8 A difference map with coefficients (Fcomplex--Fnative) and calculated phase angles from the native model. Both the final native and complex GS4 models consist of 1904 protein non-hydrogen atoms, one sulfate ion, one Ca ion, one Mn ion and three covalently-bound sugar residues N-linked to Asn18. In addition, the complex model has 47 Le(b)-OMe non-hydrogen atoms. The two structures have 135 water molecules in common in addition to eight and nine unique water molecules in the native and complex structures, respectively. The root-mean-square deviations from ideal bond distances and angles are 0.016 A, 3.2 degrees and 0.016 A, 3.0 degrees, for the native and complexed GS4, respectively. The R index for all unique data from 8 to 2.0 A is 0.187 for the native (19,204 reflections) and 0.181 for the complex (19,212 reflections). The tertiary structure of each subunit is similar to that of other leguminous lectins but the quaternary structure of the molecular dimer is different from that of any other lectin reported to date. The co-ordination about the Ca ion is pentagonal bipyramidal (with 1 long Ca(2+)-oxygen bond) and the co-ordination about the Mn ion is octahedral. Two conserved residues (Asp149 and Ser155) appear to be important because they are hydrogen-bonded to each other and to groups that co-ordinate the Mn ion. There are three cis-peptides in the polypeptide chain; two involve non-proline residues, one of which is homologous with other leguminous lectins and the other is unique to GS4. The two non-proline cis-peptides are located in the carbohydrate-binding site and are important for the specificity of the lectin. The molecular recognition of Le(b)-OMe by GS4 involves both polar and extensive non-polar interactions.(ABSTRACT TRUNCATED AT 400 WORDS)

The crystal structure of a parallel-stranded guanine tetraplex at 0.95 A resolution.

In both DNA and RNA, stretches of guanine bases can form stable four-stranded helices in the presence of sodium or potassium ions. Sequences with a propensity to form guanine tetraplexes have been found in chromosomal telomers, immunoglobulin switch regions, and recombination sites. We report the crystal structure at 0.95 A resolution of a parallel-stranded tetraplex formed by the hexanucleotide d(TG4T) in the presence of sodium ions. The four strands form a right-handed helix that is stabilized by hydrogen-bonding tetrads of co-planar guanine bases. Well-resolved sodium ions are found between and, at defined points, within tetrad planes and are coordinated with the guanine O6 groups. Nine calcium ions have been identified, each with a well-defined hepta-coordinate hydration shell. Hydrogen-bonding water patterns are observed within the tetraplex's helical grooves and clustered about the phosphate groups. Water molecules in the groove may form a hydrogen bond with the O4', and may affect the stacking behavior of guanine. Two distinct stacking arrangements are noted for the guanine tetrads. The thymine bases do not contribute to the four-stranded conformation, but instead stack to stabilize the crystal lattice. We present evidence that the sugar conformation is strained and propose that this originates from forces that optimize guanine base stacking. Discrete conformational disorder is observed at several places in the phosphodiester backbone, which results from a simple crankshaft rotation that requires no net change in the sugar conformation.

Crystallization and preliminary X-ray analysis of the complex between a mouse Fab fragment and a single IgG-binding domain from streptococcal protein G.

The Fab fragment of a mouse immunoglobulin G1, complexed with a single IgG-binding domain from streptococcal protein G, has been crystallized in a form suitable for analysis by X-ray diffraction. The needle-shaped crystals were grown from polyethylene glycol 4000 using vapour diffusion methods and diffract to 2.3 A resolution. The space group is P2(1)2(1)2(1) (a = 64.5 A, b = 70.5 A and c = 120.1 A), with one Fab-protein G domain complex in the asymmetric unit. Solution of the three-dimensional structure of the complex will permit a detailed analysis of the molecular interactions between protein G and the Fab portion of IgG.