Protein structures and information extraction from biological texts: the PASTA system.

MOTIVATION: The rapid increase in volume of protein structure literature means useful information may be hidden or lost in the published literature and the process of finding relevant material, sometimes the rate-determining factor in new research, may be arduous and slow. RESULTS: We describe the Protein Active Site Template Acquisition (PASTA) system, which addresses these problems by performing automatic extraction of information relating to the roles of specific amino acid residues in protein molecules from online scientific articles and abstracts. Both the terminology recognition and extraction capabilities of the system have been extensively evaluated against manually annotated data and the results compare favourably with state-of-the-art results obtained in less challenging domains. PASTA is the first information extraction (IE) system developed for the protein structure domain and one of the most thoroughly evaluated IE system operating on biological scientific text to date. AVAILABILITY: PASTA makes its extraction results available via a browser-based front end: http://www.dcs.shef.ac.uk/nlp/pasta/. The evaluation resources (manually annotated corpora) are also available through the website: http://www.dcs.shef.ac.uk/nlp/pasta/results.html.

Crystal structure of manganese catalase from Lactobacillus plantarum.

BACKGROUND: Catalases are important antioxidant metalloenzymes that catalyze disproportionation of hydrogen peroxide, forming dioxygen and water. Two families of catalases are known, one having a heme cofactor, and the other, a structurally distinct family containing nonheme manganese. We have solved the structure of the mesophilic manganese catalase from Lactobacillus plantarum and its azide-inhibited complex. RESULTS: The crystal structure of the native enzyme has been solved at 1.8 A resolution by molecular replacement, and the azide complex of the native protein has been solved at 1.4 A resolution. The hexameric structure of the holoenzyme is stabilized by extensive intersubunit contacts, including a beta zipper and a structural calcium ion crosslinking neighboring subunits. Each subunit contains a dimanganese active site, accessed by a single substrate channel lined by charged residues. The manganese ions are linked by a mu1,3-bridging glutamate carboxylate and two mu-bridging solvent oxygens that electronically couple the metal centers. The active site region includes two residues (Arg147 and Glu178) that appear to be unique to the Lactobacillus plantarum catalase. CONCLUSIONS: A comparison of L. plantarum and T. thermophilus catalase structures reveals the existence of two distinct structural classes, differing in monomer design and the organization of their active sites, within the manganese catalase family. These differences have important implications for catalysis and may reflect distinct biological functions for the two enzymes, with the L. plantarum enzyme serving as a catalase, while the T. thermophilus enzyme may function as a catalase/peroxidase.

Protein docking using a genetic algorithm.

A genetic algorithm (GA) for protein-protein docking is described, in which the proteins are represented by dot surfaces calculated using the Connolly program. The GA is used to move the surface of one protein relative to the other to locate the area of greatest surface complementarity between the two. Surface dots are deemed complementary if their normals are opposed, their Connolly shape type is complementary, and their hydrogen bonding or hydrophobic potential is fulfilled. Overlap of the protein interiors is penalized. The GA is tested on 34 large protein-protein complexes where one or both proteins has been crystallized separately. Parameters are established for which 30 of the complexes have at least one near-native solution ranked in the top 100. We have also successfully reassembled a 1,400-residue heptamer based on the top-ranking GA solution obtained when docking two bound subunits.

The high-resolution X-ray crystallographic structure of the ferritin (EcFtnA) of Escherichia coli; comparison with human H ferritin (HuHF) and the structures of the Fe(3+) and Zn(2+) derivatives.

The high-resolution structure of the non-haem ferritin from Escherichia coli (EcFtnA) is presented together with those of its Fe(3+) and Zn(2+) derivatives, this being the first high-resolution X-ray analysis of the iron centres in any ferritin. The binding of both metals is accompanied by small changes in the amino acid ligand positions. Mean Fe(A)(3+)-Fe(B)(3+) and Zn(A)(2+)-Zn(B)(2+) distances are 3.24 A and 3.43 A, respectively. In both derivatives, metal ions at sites A and B are bridged by a glutamate side-chain (Glu50) in a syn-syn conformation. The Fe(3+) derivative alone shows a third metal site (Fe( C)( 3+)) joined to Fe(B)(3+) by a long anti-anti bidentate bridge through Glu130 (mean Fe(B)(3+)-Fe(C)(3+) distance 5.79 A). The third metal site is unique to the non-haem bacterial ferritins. The dinuclear site lies at the inner end of a hydrophobic channel connecting it to the outside surface of the protein shell, which may provide access for dioxygen and possibly for metal ions shielded by water. Models representing the possible binding mode of dioxygen to the dinuclear Fe(3+) pair suggest that a gauche micro-1,2 mode may be preferred stereochemically. Like those of other ferritins, the 24 subunits of EcFtnA are folded as four-helix bundles that assemble into hollow shells and both metals bind at dinuclear centres in the middle of the bundles. The structural similarity of EcFtnA to the human H chain ferritin (HuHF) is remarkable (r.m.s. deviation of main-chain atoms 0.66 A) given the low amino acid sequence identity (22 %). Many of the conserved residues are clustered at the dinuclear centre but there is very little conservation of residues making inter-subunit interactions.

Structure-function relationships of a novel bacterial toxin, hemolysin E. The role of alpha G.

The novel pore-forming toxin hemolysin E (HlyE, ClyA, or SheA) consists of a long four-helix bundle with a subdomain (beta tongue) that interacts with target membranes at one pole and an additional helix (alpha(G)) that, with the four long helices, forms a five-helix bundle (tail domain) at the other pole. Random amino acid substitutions that impair hemolytic activity were clustered mostly, but not exclusively, within the tail domain, specifically amino acids within, adjacent to, or interacting with alpha(G). Deletion of amino acids downstream of alpha(G) did not affect activity, but deletions encompassing alpha(G) yielded insoluble and inactive proteins. In the periplasm Cys-285 (alpha(G)) is linked to Cys-87 (alpha(B)) of the four-helix bundle via an intramolecular disulfide. Oxidized HlyE did not form spontaneously in vitro but could be generated by addition of Cu(II) or mimicked by treatment with Hg(II) salts to yield inactive proteins. Such treatments did not affect binding to target membranes nor assembly into non-covalently linked octameric complexes once associated with a membrane. However, gel filtration analyses suggested that immobilizing alpha(G) inhibits oligomerization in solution. Thus once associated with a membrane, immobilizing alpha(G) inhibits HlyE activity at a late stage of pore formation, whereas in solution it prevents aggregation and consequent inactivation.

E. coli hemolysin E (HlyE, ClyA, SheA): X-ray crystal structure of the toxin and observation of membrane pores by electron microscopy.

Hemolysin E (HlyE) is a novel pore-forming toxin of Escherichia coli, Salmonella typhi, and Shigella flexneri. Here we report the X-ray crystal structure of the water-soluble form of E. coli HlyE at 2.0 A resolution and the visualization of the lipid-associated form of the toxin in projection at low resolution by electron microscopy. The crystal structure reveals HlyE to be the first member of a new family of toxin structures, consisting of an elaborated helical bundle some 100 A long. The electron micrographs show how HlyE oligomerizes in the presence of lipid to form transmembrane pores. Taken together, the data from these two structural techniques allow us to propose a simple model for the structure of the pore and for membrane interaction.

Crystal structure of E.coli RuvA with bound DNA Holliday junction at 6 A resolution.

Here we present the crystal structure of the Escherichia coli protein RuvA bound to a key DNA intermediate in recombination, the Holliday junction. The structure, solved by isomorphous replacement and density modification at 6 A resolution, reveals the molecular architecture at the heart of the branch migration and resolution reactions required to process Holliday intermediates into recombinant DNA molecules. It also reveals directly for the first time the structure of the Holliday junction. A single RuvA tetramer is bound to one face of a junction whose four DNA duplex arms are arranged in an open and essentially four-fold symmetric conformation. Protein-DNA contacts are mediated by two copies of a helix-hairpin-helix motif per RuvA subunit that contact the phosphate backbone in a very similar manner. The open structure of the junction stabilized by RuvA binding exposes a DNA surface that could be bound by the RuvC endonuclease to promote resolution.

Structural similarities between Escherichia coli RuvA protein and other DNA-binding proteins and a mutational analysis of its binding to the holliday junction.

Comparison of the structure of Escherichia coli RuvA with other proteins in the Protein Data Bank gives insights into the probable modes of association of RuvA with the Holliday junction during homologous recombination. All three domains of the RuvA protein possess striking structural similarities to other DNA-binding proteins. Additionally, the second domain of RuvA contains two copies of the helix-hairpin-helix (HhH) structural motif, which has been implicated in non-sequence-specific DNA binding. The two copies of the motif are related by approximate 2-fold symmetry and may form a bidentate DNA-binding module. The results described provide support for the organization of the arms of the DNA in our RuvA/Holliday junction complex model and support the involvement of the HhH motifs in DNA binding.

Representation and searching of carbohydrate structures using graph-theoretic techniques.

This paper describes how the carbohydrate structures in the Complex Carbohydrate Structure Database (CCSD) can be represented by labelled graphs, in which the nodes and edges of a graph are used to denote the residues and the inter-residue linkages, respectively, of a carbohydrate. These graph representations are then searched with a subgraph-isomorphism algorithm. We describe the use of one such algorithm, that due to Ullmann, and demonstrate that it provides a very precise way of searching the structures in CCSD. We also describe the use of screening techniques that can eliminate many of the CCSD structures from the subgraph-isomorphism search, with a consequent increase in the speed of the search.

Prokaryotic 5'-3' exonucleases share a common core structure with gamma-delta resolvase.

The three dimensional crystal structure of T5 5'-3' exonuclease was compared with that of two other members of the 5'-3' exonuclease family: T4 ribonuclease H and the N-terminal domain of Thermus aquaticus DNA polymerase I. Though these structures were largely similar, some regions of these enzymes show evidence of significant molecular flexibility. Previous sequence analysis had suggested the existence of a helix-hairpin-helix motif in T5 exonuclease, but a distinct, though related structure is actually found to occur. The entire T5 exonuclease structure was then compared with all the structures in the complete Protein Data Bank and an unexpected similarity with gamma-delta (gamma delta) resolvase was observed. 5'-3' exonucleases and gamma delta resolvase are enzymes involved in carrying out quite different manipulations on nucleic acids. They appear to be unrelated at the primary sequence level, yet the fold of the entire catalytic domain of gamma delta resolvase is contained within that of the 5'-3'exonuclease. Different large-scale helical structures are used by both families to form DNA binding sites.

Clique-detection algorithms for matching three-dimensional molecular structures.

The representation of chemical and biological molecules by means of graphs permits the use of a maximum common subgraph (MCS) isomorphism algorithm to identify the structural relationships existing between pairs of such molecular graphs. Clique detection provides an efficient way of implementing MCS detection, and this article reports a comparison of several different clique-detection algorithms when used for this purpose. Experiments with both small molecules and proteins demonstrate that the most efficient of these particular applications, which typically involve correspondence graphs with low edge densities, is the algorithm described by Carraghan and Pardalos. This is shown to be two to three times faster than the Bron-Kerbosch algorithm that has been used previously for MCS applications in chemistry and biology. However, the latter algorithm enables all substructures common to a pair of molecules to be identified, and not just the largest ones, as with the other algorithms considered here. The two algorithms can usefully be combined to increase the efficiency of database-searching systems that use the MCS as a measure of structural similarity.

Comparison of the three-dimensional structures of recombinant human H and horse L ferritins at high resolution.

Mammalian ferritins are 24-mers assembled from two types of polypeptide chain which provide the molecule with different functions. H(eavy) chains catalyse the first step in iron storage, the oxidation of iron(II). L(ight) chains promote the nucleation of the mineral ferrihydrite enabling storage of iron(III) inside the protein shell. We report here the comparison of the three-dimensional structures of recombinant human H chain (HuHF) and horse L chain (HoLF) ferritin homopolymers, which have been refined at 1.9 A resolution. There is 53% sequence identity between these molecules, and the two structures are very similar, the H and L subunit alpha-carbons superposing to within 0.5 A rms deviation with 41 water molecules in common. Nevertheless, there are significant important differences which can be related to differences in function. In particular, the centres of the four-helix bundles contain distinctive groups of hydrophilic residues which have been associated with ferroxidase activity in H chains and enhanced stability in L chains. L chains contain a group of glutamates associated with mineralisation within the iron storage cavity of the protein.

The aconitase family: three structural variations on a common theme.

The aconitase family contains a diverse group of iron-sulphur (Fe-S) isomerases and two types of iron regulatory protein (IRP). Structural comparisons have revealed three architecturally distinct variants in which one of the four structural domains is covalently linked at either the amino- or carboxy-terminal end of a single polypeptide or else this domain exists as an independent subunit.

Comparison of protein surfaces using a genetic algorithm.

A genetic algorithm (GA) is described which is used to compare the solvent-accessible surfaces of two proteins or fragments of proteins, represented by a dot surface calculated using the Connolly algorithm. The GA is used to move one surface relative to the other to locate the most similar surface region between the two. The matching process is enhanced by the use of the surface normals and shape terms provided by the Connolly program and also by a simple hydrogen-bonding descriptor and an additional shape descriptor. The algorithm has been tested in applications ranging from the comparison of small surface patches to the comparison of whole protein surfaces, and it has performed correctly in all cases. Examples of the matches are given and a quantitative analysis of the quality of the matches is performed. A number of possible future enhancements to the program are described which would allow the GA to be used for more complex surface comparisons.

Insights into the mechanisms of homologous recombination from the structure of RuvA.

The recent structure determination of RuvA has provided the first insights into the structural basis for its interaction with Holliday junction DNA. Multiple copies of a helix-hairpin-helix motif which line the four grooves between the monomers in the tetrameric structure are thought to be involved in the interaction of the protein with its DNA target. This suggests that the four arms of the junction are held by RuvA in a fourfold symmetric arrangement and has fuelled ideas on the way in which components of the Ruv complex combine to catalyse the process of homologous recombination.

Crystal structure of DNA recombination protein RuvA and a model for its binding to the Holliday junction.

The Escherichia coli DNA binding protein RuvA acts in concert with the helicase RuvB to drive branch migration of Holliday intermediates during recombination and DNA repair. The atomic structure of RuvA was determined at a resolution of 1.9 angstroms. Four monomers of RuvA are related by fourfold symmetry in a manner reminiscent of a four-petaled flower. The four DNA duplex arms of a Holliday junction can be modeled in a square planar configuration and docked into grooves on the concave surface of the protein around a central pin that may facilitate strand separation during the migration reaction. The model presented reveals how a RuvAB-junction complex may also accommodate the resolvase RuvC.

Biotin carboxylase comes into the fold.

Extensive three-dimensional structural resemblances between biotin carboxylase and the ADP-forming peptide synthetases, represented by glutathione synthetase and D-Ala:D-Ala ligase, reveal a previously unsuspected evolutionary relationship between two major families of ADP-forming ligases.