A new protein-protein docking scoring function based on interface residue properties.

MOTIVATION: Protein-protein complexes are known to play key roles in many cellular processes. However, they are often not accessible to experimental study because of their low stability and difficulty to produce the proteins and assemble them in native conformation. Thus, docking algorithms have been developed to provide an in silico approach of the problem. A protein-protein docking procedure traditionally consists of two successive tasks: a search algorithm generates a large number of candidate solutions, and then a scoring function is used to rank them. RESULTS: To address the second step, we developed a scoring function based on a Voronoï tessellation of the protein three-dimensional structure. We showed that the Voronoï representation may be used to describe in a simplified but useful manner, the geometric and physico-chemical complementarities of two molecular surfaces. We measured a set of parameters on native protein-protein complexes and on decoys, and used them as attributes in several statistical learning procedures: a logistic function, Support Vector Machines (SVM), and a genetic algorithm. For the later, we used ROGER, a genetic algorithm designed to optimize the area under the receiver operating characteristics curve. To further test the scores derived with ROGER, we ranked models generated by two different docking algorithms on targets of a blind prediction experiment, improving in almost all cases the rank of native-like solutions. AVAILABILITY: http://genomics.eu.org/spip/-Bioinformatics-tools-

Encapsulation of mono- and oligo-nucleotides into aqueous-core nanocapsules in presence of various water-soluble polymers.

In a previous study, we have shown that cidofovir (CDV) and azidothymidine-triphosphate (AZT-TP) were poorly encapsulated in poly(iso-butylcyanoacrylate) (PIBCA) aqueous-core nanocapsules. This was attributed to the rapid leakage of these small and hydrophilic molecules through the thin polymer wall of the nanocapsules. In the present study, we have selected various water-soluble polymers as increasing Mw adjuvants and investigated their influence on the entrapment of mononucleotides (CDV, AZT-TP) as well as of oligonucleotides (ODN) into these PIBCA aqueous-core nanocapsules. We show here that the presence of cationic polymers (i.e. poly(ethyleneimine) (PEI) or chitosan) in the nanocapsule aqueous compartment allowed successful encapsulation of AZT-TP and ODN.

Encapsulation of antiviral nucleotide analogues azidothymidine-triphosphate and cidofovir in poly(iso-butylcyanoacrylate) nanocapsules.

Nucleoside analogues are widely used in the treatment of various viral infections. However, the poor in vivo conversion of the nucleoside analogues like azidothymidine (AZT) into their active triphosphate nucleotide counterpart limits their pharmacological efficacy. This could be overcome by the direct administration of azidothymidine triphosphate (AZT-TP), but it requires an appropriate drug delivery approach. Besides nucleoside analogues, nucleotide analogues like cidofovir (CDV) are also used in the treatment of viral infections. CDV has raised recent interest because of its promising activity against smallpox, but its use is limited by its poor bioavailability and nephrotoxicity. Here again, a proper drug delivery system should address these issues. In this study, we investigated the encapsulation of the nucleotide analogues AZT-TP and CDV into poly(iso-butylcyanoacrylate) aqueous core nanocapsules, known to efficiently entrap oligonucleotides. We show here that the encapsulation of these mono-nucleotides is less efficient than with oligonucleotides and that a rapid release of AZT-TP from the nanocapsules occurred in vitro. This highlights the importance of the molecular weight of the entrapped molecules which, if they are too small, are diffusing through the thin polymer membrane of the nanocapsules. On the other hand, a good protection of the encapsulated AZT-TP was observed.

Standard atomic volumes in double-stranded DNA and packing in protein--DNA interfaces.

Standard volumes for atoms in double-stranded B-DNA are derived using high resolution crystal structures from the Nucleic Acid Database (NDB) and compared with corresponding values derived from crystal structures of small organic compounds in the Cambridge Structural Database (CSD). Two different methods are used to compute these volumes: the classical Voronoi method, which does not depend on the size of atoms, and the related Radical Planes method which does. Results show that atomic groups buried in the interior of double-stranded DNA are, on average, more tightly packed than in related small molecules in the CSD. The packing efficiency of DNA atoms at the interfaces of 25 high resolution protein-DNA complexes is determined by computing the ratios between the volumes of interfacial DNA atoms and the corresponding standard volumes. These ratios are found to be close to unity, indicating that the DNA atoms at protein-DNA interfaces are as closely packed as in crystals of B-DNA. Analogous volume ratios, computed for buried protein atoms, are also near unity, confirming our earlier conclusions that the packing efficiency of these atoms is similar to that in the protein interior. In addition, we examine the number, volume and solvent occupation of cavities located at the protein-DNA interfaces and compared them with those in the protein interior. Cavities are found to be ubiquitous in the interfaces as well as inside the protein moieties. The frequency of solvent occupation of cavities is however higher in the interfaces, indicating that those are more hydrated than protein interiors. Lastly, we compare our results with those obtained using two different measures of shape complementarity of the analysed interfaces, and find that the correlation between our volume ratios and these measures, as well as between the measures themselves, is weak. Our results indicate that a tightly packed environment made up of DNA, protein and solvent atoms plays a significant role in protein-DNA recognition.

Fine organization of Bombyx mori fibroin heavy chain gene.

The complete sequence of the Bombyx mori fibroin gene has been determined by means of combining a shotgun sequencing strategy with physical map-based sequencing procedures. It consists of two exons (67 and 15 750 bp, respectively) and one intron (971 bp). The fibroin coding sequence presents a spectacular organization, with a highly repetitive and G-rich (approximately 45%) core flanked by non-repetitive 5' and 3' ends. This repetitive core is composed of alternate arrays of 12 repetitive and 11 amorphous domains. The sequences of the amorphous domains are evolutionarily conserved and the repetitive domains differ from each other in length by a variety of tandem repeats of subdomains of approximately 208 bp which are reminiscent of the repetitive nucleosome organization. A typical composition of a subdomain is a cluster of repetitive units, Ua, followed by a cluster of units, Ub, (with a Ua:Ub ratio of 2:1) flanked by conserved boundary elements at the 3' end. Moreover some repeats are also perfectly conserved at the peptide level indicating that the evolutionary pressure is not identical along the sequence. A tentative model for the constitution and evolution of this unusual gene is discussed.

Proteins with a ring.

Proteins come in all sizes and shapes. Those which fold into a ring with a large hole in the middle may act as a clamp on DNA, a polysaccharide or another protein.

Wet and dry interfaces: the role of solvent in protein-protein and protein-DNA recognition.

Water molecules are found in abundance in protein-protein and protein-DNA interfaces. Although interface solvent molecules exchange quickly with the bulk solvent, structural and biochemical data suggest that water-mediated interactions are as important as direct hydrogen bonds in the stability and specificity of recognition.

Structure of the globular region of the prion protein Ure2 from the yeast Saccharomyces cerevisiae.

BACKGROUND: The [URE3] non-Mendelian element of the yeast S. cerevisiae is due to the propagation of a transmissible form of the protein Ure2. The infectivity of Ure2p is thought to originate from a conformational change of the normal form of the prion protein. This conformational change generates a form of Ure2p that assembles into amyloid fibrils. Hence, knowledge of the three-dimensional structure of prion proteins such as Ure2p should help in understanding the mechanism of amyloid formation associated with a number of neurodegenerative diseases. RESULTS: Here we report the three-dimensional crystal structure of the globular region of Ure2p (residues 95--354), also called the functional region, solved at 2.5 A resolution by the MAD method. The structure of Ure2p 95--354 shows a two-domain protein forming a globular dimer. The N-terminal domain is composed of a central 4 strand beta sheet flanked by four alpha helices, two on each side. In contrast, the C-terminal domain is entirely alpha-helical. The fold of Ure2p 95--354 resembles that of the beta class glutathione S-transferases (GST), in line with a weak similarity in the amino acid sequence that exists between these proteins. Ure2p dimerizes as GST does and possesses a potential ligand binding site, although it lacks GST activity. CONCLUSIONS: The structure of the functional region of Ure2p is the first crystal structure of a prion protein. Structure comparisons between Ure2p 95--354 and GST identified a 32 amino acid residues cap region in Ure2p exposed to the solvent. The cap region is highly flexible and may interact with the N-terminal region of the partner subunit in the dimer. The implication of this interaction in the assembly of Ure2p into amyloid fibrils is discussed.

X-ray structure of human nucleoside diphosphate kinase B complexed with GDP at 2 A resolution.

BACKGROUND: Nucleoside diphosphate (NDP) kinases provide precursors for DNA and RNA synthesis. In mammals, these enzymes are also involved in cell regulations. Human NDP kinase B, product of the human nm23-H2 gene, is both an enzyme and a transcription factor. It activates transcription of the c-myc oncogene independently of its catalytic function, by binding to its promoter DNA. How do the two functions coexist? RESULTS: Recombinant human NDP kinase B was co-crystallized with GDP. The X-ray structure was solved at 2.0 A resolution by molecular replacement from the homologous Drosophila Awd protein. Both enzymes are homo-hexamers with a characteristic beta alpha beta beta alpha beta fold. GDP binds near the active site His118. The guanine base is in a surface cleft and interacts with the C terminus of another subunit. CONCLUSIONS: The beta alpha beta beta alpha beta fold, also present in the 'palm' domain of Escherichia coli DNA polymerase I and HIV reverse transcriptase, is both a mononucleotide- and a polynucleotide-binding fold. If NDP kinase B binds DNA in the same way as the polymerases, the enzyme must undergo a conformation change in order to carry out gene activation.

Crystal structure of the Awd nucleotide diphosphate kinase from Drosophila.

BACKGROUND: Nucleotide diphosphate kinase (NDP kinase) is a phosphate transfer enzyme involved in cell regulation and in animal development. Drosophila NDP kinase is the product of the abnormal wing disc (awd) developmental gene, a point mutation in which can produce the killer of prune (K-pn) conditional lethal phenotype. The highly homologous mammalian genes control metastasis and a human NDP kinase acts as a transcription factor. RESULTS: The X-ray structure of the Awd protein prepared from Drosophila was solved at 2.4 A resolution by molecular replacement from the homologous Dictyostelium protein. Both are hexamers, and both have the same fold and the same active site. Subunit contacts differ as a result of sequence changes in the carboxy-terminal segment and in the loop that is the site of the K-pn mutation. CONCLUSIONS: Regulatory properties of animal NDP kinases depend on interactions with other macromolecules, such as DNA and the product of the Drosophila prune gene. The Awd structure suggests an allosteric mechanism of action of NDP kinase where DNA is the effector and the protein undergoes a major conformational change, possibly dissociating to dimers.

Structures of escherichia coli CMP kinase alone and in complex with CDP: a new fold of the nucleoside monophosphate binding domain and insights into cytosine nucleotide specificity.

BACKGROUND: . Nucleoside monophosphate kinases (NMP kinases) catalyze the reversible transfer of a phosphoryl group from a nucleoside triphosphate to a nucleoside monophosphate. Among them, cytidine monophosphate kinase from Escherichia coli has a striking particularity: it is specific for CMP, whereas in eukaryotes a unique UMP/CMP kinase phosphorylates both CMP and UMP with similar efficiency. RESULTS: . The crystal structure of the CMP kinase apoenzyme from E. coli was solved by single isomorphous replacement and refined at 1.75 A resolution. The structure of the enzyme in complex with CDP was determined at 2.0 A resolution. Like other NMP kinases, the protein contains a central parallel beta sheet, the strands of which are connected by alpha helices. The enzyme differs from other NMP kinases in the presence of a 40-residue insert situated in the NMP-binding (NMPbind) domain. This insert contains two domains: one comprising a three-stranded antiparallel beta sheet, the other comprising two alpha helices. CONCLUSIONS: . Two features of the CMP kinase from E. coli have no equivalent in other NMP kinases of known structure. Firstly, the large NMPbind insert undergoes a CDP-induced rearrangement: its beta-sheet domain moves away from the substrate, whereas its helical domain comes closer to it in a motion likely to improve the protection of the active site. Secondly, residues involved in CDP recognition are conserved in CMP kinases and have no counterpart in other NMP kinases. The structures presented here are the first of a new family of NMP kinases specific for CMP.

Crystal structure of the S. cerevisiae D-ribose-5-phosphate isomerase: comparison with the archaeal and bacterial enzymes.

Ribose-5-phosphate isomerase A has an important role in sugar metabolism by interconverting ribose-5-phosphate and ribulose-5-phosphate. This enzyme is ubiquitous and highly conserved among the three kingdoms of life. We have solved the 2.1 A resolution crystal structure of the Saccharomyces cerevisiae enzyme by molecular replacement. This protein adopts the same fold as its archaeal and bacterial orthologs with two alpha/beta domains tightly packed together. Mapping of conserved residues at the surface of the protein reveals strong invariability of the active site pocket, suggesting a common ligand binding mode and a similar catalytic mechanism. The yeast enzyme associates as a homotetramer similarly to the archaeal protein. The effect of an inactivating mutation (Arg189 to Lys) is discussed in view of the information brought by this structure.

Crystal structure of a barnase-d(GpC) complex at 1.9 A resolution.

The ribonuclease excreted by Bacillus amyloliquefaciens, Barnase, was co-crystallized with the deoxy-dinucleotide d(GpC). The crystal structure was determined by molecular replacement from a model of free Barnase previously derived by Mauguen et al. Refinement was carried out using data to 1.9 A resolution. The final model, which has a crystallographic R factor of 22%, includes 869 protein atoms, 38 atoms from d(GpC), a sulfate ion and 73 water molecules. Only minor differences from free Barnase are seen in the protein moiety, the root-mean-square C alpha movement being 0.45 A. The dinucleotide has a folded conformation. It is located near the active site of the enzyme, but outside the protein molecule and making crystal packing contacts with neighboring molecules. The guanine base is stacked on the imidazole ring of active site His102, rather than binding to the so-called recognition loop as it does in other complexes of guanine nucleotides with microbial nucleases. The deoxyguanosine is syn, with the sugar ring in C-2'-endo conformation; the deoxycytidine is anti and C-4'-exo. In addition to the stacking interaction, His102 hydrogen bonds to the free 5' hydroxyl, which is located near the position where the 3' phosphate group is found in other inhibitors of microbial ribonucleases. While the mode of binding observed with d(GpC) and Barnase would be non-productive for a dinucleotide substrate, it may define a site for the nucleotide product on the 3' side of the hydrolyzed bond.

Surface, subunit interfaces and interior of oligomeric proteins.

The solvent-accessible surface area (As) of 23 oligomeric proteins is calculated using atomic co-ordinates from high-resolution and well-refined crystal structures. As is correlated with the protein molecular weight, and a power law predicts its value to within 5% on average. The accessible surface of the average oligomer is similar to that of monomeric proteins in its hydropathy and amino acid composition. The distribution of the 20 amino acid types between the protein surface and its interior is also the same as in monomers. Interfaces, i.e. surfaces involved in subunit contacts, differ from the rest of the subunit surface. They are enriched in hydrophobic side-chains, yet they contain a number of charged groups, especially from Arg residues, which are the most abundant residues at interfaces except for Leu. Buried Arg residues are involved in H-bonds between subunits. We counted H-bonds at interfaces and found that several have none, others have one H-bond per 200 A2 of interface area on average (1 A = 0.1 nm). A majority of interface H-bonds involve charged donor or acceptor groups, which should make their contribution to the free energy of dissociation significant, even when they are few. The smaller interfaces cover about 700 A2 of the subunit surface. The larger ones cover 3000 to 10,000 A2, up to 40% of the subunit surface area in catalase. The lower value corresponds to an estimate of the accessible surface area loss required for stabilizing subunit association through the hydrophobic effect alone. Oligomers with small interfaces have globular subunits with accessible surface areas similar to those of monomeric proteins. We suggest that these oligomers assemble from preformed monomers with little change in conformation. In oligomers with large interfaces, isolated subunits should be unstable given their excessively large accessible surface, and assembly is expected to require major structural changes.

A soft, mean-field potential derived from crystal contacts for predicting protein-protein interactions.

We derive a series of novel mean-field potentials from statistical analyses of protein-protein contact regions in crystal structures. These potentials are parameterized in terms of the number of contacts made by an atom in an interface region. Such an explicit number dependence avoids the pairwise assumption and is intrinsically softer than distance-based approaches. It appears well suited to protein-protein docking applications, for which detailed interface geometry is generally lacking. In tests including protein complex reconstitution and docking of independently determined protein structures, we show that a hydrophobic potential of this type performs remarkably well, identifying native-like complexes by their favourable potential energies and in several cases demonstrating a recognition energy gap of 4-8 kcal/mol according to the system.

Crystal packing in six crystal forms of pancreatic ribonuclease.

We compare the molecular packing of bovine pancreatic ribonuclease A (RNase A) in six crystal forms, two grown with alcohol, three with high salt and one with polyethylene glycol as a precipitant. The six packings differ in the number of molecules in contact and in the extent of the contacts, which bury 1570 A2 to 2790 A2 of the RNase surface. Regions of the protein surface involved in the six packings cover almost the whole RNase molecule. The abundance of polar interactions, about one per 200 A2, is the same in all types of precipitants. All molecule-to-molecule contacts are different in the six crystal forms, except for the one that forms a RNase dimer. The dimer has a large interface covering 1800 A2 and eight to ten polar interactions. Its presence in the three salt-grown crystal forms suggests that it is an intermediate in salt induced crystallization. In contrast, the two alcohol-grown forms contain only small interfaces, implying a different mechanism of nucleation.

The atomic structure of protein-protein recognition sites.

The non-covalent assembly of proteins that fold separately is central to many biological processes, and differs from the permanent macromolecular assembly of protein subunits in oligomeric proteins. We performed an analysis of the atomic structure of the recognition sites seen in 75 protein-protein complexes of known three-dimensional structure: 24 protease-inhibitor, 19 antibody-antigen and 32 other complexes, including nine enzyme-inhibitor and 11 that are involved in signal transduction.The size of the recognition site is related to the conformational changes that occur upon association. Of the 75 complexes, 52 have "standard-size" interfaces in which the total area buried by the components in the recognition site is 1600 (+/-400) A2. In these complexes, association involves only small changes of conformation. Twenty complexes have "large" interfaces burying 2000 to 4660 A2, and large conformational changes are seen to occur in those cases where we can compare the structure of complexed and free components. The average interface has approximately the same non-polar character as the protein surface as a whole, and carries somewhat fewer charged groups. However, some interfaces are significantly more polar and others more non-polar than the average. Of the atoms that lose accessibility upon association, half make contacts across the interface and one-third become fully inaccessible to the solvent. In the latter case, the Voronoi volume was calculated and compared with that of atoms buried inside proteins. The ratio of the two volumes was 1.01 (+/-0.03) in all but 11 complexes, which shows that atoms buried at protein-protein interfaces are close-packed like the protein interior. This conclusion could be extended to the majority of interface atoms by including solvent positions determined in high-resolution X-ray structures in the calculation of Voronoi volumes. Thus, water molecules contribute to the close-packing of atoms that insure complementarity between the two protein surfaces, as well as providing polar interactions between the two proteins.

Interior and surface of monomeric proteins.

The solvent-accessible surface area (As) of 46 monomeric proteins is calculated using atomic co-ordinates from high-resolution and well-refined crystal structures. The As of these proteins can be determined to within 1 to 2% and that of their individual residues to within 10 to 20%. The As values of proteins are correlated with their molecular weight (Mr) in the range 4000 to 35,000: the power law As = 6.3 M0.73 predicts protein As values to within 4% on average. The average water-accessible surface is found to be 57% non-polar, 24% polar and 19% charged, with 5% root-mean-square variations. The molecular surface buried inside the protein is 58% non-polar, 39% polar and 4% charged. The buried surface contains more uncharged polar groups (mostly peptides) than the surface that remains accessible, but many fewer charged groups. On average, 15% of residues in small proteins and 32% in larger ones may be classed as "buried residues", having less than 5% of their surface accessible to the solvent. The accessibilities of most other residues are evenly distributed in the range 5 to 50%. Although the fraction of buried residues increases with molecular weight, the amino acid compositions of the protein interior and surface show no systematic variation with molecular weight, except for small proteins that are often very rich in buried cysteines. From amino acid compositions of protein surfaces and interiors we calculate an effective coefficient of partition for each type of residue, and derive an implied set of transfer free energy values. This is compared with other sets of partition coefficients derived directly from experimental data. The extent to which groups of residues (charged, polar and non-polar) are buried within proteins correlates well with their hydrophobicity derived from amino acid transfer experiments. Within these three groups, the correlation is low.

Haemoglobin: the surface buried between the alpha 1 beta 1 and alpha 2 beta 2 dimers in the deoxy and oxy structures.

Using the newly available refined co-ordinates of deoxy and oxyhaemoglobin, we have re-examined and compared the interfaces between the dimers alpha 1 beta 1 and alpha 2 beta 2. The most extensive monomer-monomer contacts are between alpha 1 and beta 2, and, symmetrically, alpha 2 and beta 1. In oxyhaemoglobin these interfaces bury 700 A2 less protein surface than in deoxyhaemoglobin. The alpha 1 alpha 2 interface involves similar salt bridges in both forms, but in oxyhaemoglobin buries 240 A2 more surface than in deoxyhaemoglobin. There is a loosely packed beta 1 beta 2 interface burying 320 A2 of surface in oxyhaemoglobin; there is no beta 1 beta 2 interface in deoxyhaemoglobin. The greater stability of the deoxy form, in the absence of ligands, can be attributed to a combination of hydrophobic, van der Waals' and electrostatic interactions.

Refined X-ray structure of Dictyostelium discoideum nucleoside diphosphate kinase at 1.8 A resolution.

The X-ray structure of the nucleoside diphosphate kinase (NDP kinase) from Dictyostelium discoideum has been refined at 1.8 A resolution from a hexagonal crystal form with a 17 kDa monomer in its asymmetric unit. The atomic model was derived from the previously determined structure of a point mutant of the protein. It contains 150 amino acid residues out of 155, and 95 solvent molecules. The R-factor is 0.196 and the estimated accuracy of the average atomic position, 0.25 A. The Dictyostelium structure is described in detail and compared to those of Drosophila and Myxococcus xanthus NDP kinases. The protein is a hexamer with D3 symmetry. Residues 8 to 138 of each subunit form a globular alpha/beta domain. The four-stranded beta-sheet is antiparallel; its topology is different from other phosphate transfer enzymes, and also from the HPr protein which, like NDP kinase, carries a phosphorylated histidine. The same topology is nevertheless found in several other proteins that bind mononucleotides, RNA or DNA. Strand connections in NDP kinase involve alpha-helices and a 20-residue segment called the Kpn loop. The beta-sheet is regular except for a beta-bulge in edge strand beta 2 and a gamma-turn at residue Ile120 just preceding strand beta 4. The latter may induce strain in the main chain near the active site His122. The alpha 1 beta 2 motif participates in forming dimers within the hexamer, helices alpha 1 and alpha 3, the Kpn loop and C terminus, in forming trimers. The subunit fold and dimer interactions found in Dictyostelium are conserved in other NDP kinases. Trimer interactions probably occur in all eukaryotic enzymes. They are absent in the bacterial Myxococcus xanthus enzyme which is a tetramer, even though the subunit structure is very similar. In Dictyostelium, contacts between Kpn loops near the 3-fold axis block access to a central cavity lined with polar residues and filled with well-defined solvent molecules. Biochemical data on point mutants highlight the contribution of the Kpn loop to protein stability. In Myxococcus, the Kpn loops are on the tetramer surface and their sequence is poorly conserved. Yet, their conformation is maintained and they make a similar contribution to the substrate binding site.