AGMIAL: implementing an annotation strategy for prokaryote genomes as a distributed system.

We have implemented a genome annotation system for prokaryotes called AGMIAL. Our approach embodies a number of key principles. First, expert manual annotators are seen as a critical component of the overall system; user interfaces were cyclically refined to satisfy their needs. Second, the overall process should be orchestrated in terms of a global annotation strategy; this facilitates coordination between a team of annotators and automatic data analysis. Third, the annotation strategy should allow progressive and incremental annotation from a time when only a few draft contigs are available, to when a final finished assembly is produced. The overall architecture employed is modular and extensible, being based on the W3 standard Web services framework. Specialized modules interact with two independent core modules that are used to annotate, respectively, genomic and protein sequences. AGMIAL is currently being used by several INRA laboratories to analyze genomes of bacteria relevant to the food-processing industry, and is distributed under an open source license.

The complete genome sequence of Lactobacillus bulgaricus reveals extensive and ongoing reductive evolution.

Lactobacillus delbrueckii ssp. bulgaricus (L. bulgaricus) is a representative of the group of lactic acid-producing bacteria, mainly known for its worldwide application in yogurt production. The genome sequence of this bacterium has been determined and shows the signs of ongoing specialization, with a substantial number of pseudogenes and incomplete metabolic pathways and relatively few regulatory functions. Several unique features of the L. bulgaricus genome support the hypothesis that the genome is in a phase of rapid evolution. (i) Exceptionally high numbers of rRNA and tRNA genes with regard to genome size may indicate that the L. bulgaricus genome has known a recent phase of important size reduction, in agreement with the observed high frequency of gene inactivation and elimination; (ii) a much higher GC content at codon position 3 than expected on the basis of the overall GC content suggests that the composition of the genome is evolving toward a higher GC content; and (iii) the presence of a 47.5-kbp inverted repeat in the replication termination region, an extremely rare feature in bacterial genomes, may be interpreted as a transient stage in genome evolution. The results indicate the adaptation of L. bulgaricus from a plant-associated habitat to the stable protein and lactose-rich milk environment through the loss of superfluous functions and protocooperation with Streptococcus thermophilus.

Identification of amino-acids in the alpha-subunit first and third loops that are crucial for the heterospecific follicle-stimulating hormone activity of equid luteinizing hormone/choriogonadotropin.

OBJECTIVE: To identify amino-acids in the alpha-subunit important for expression of heterospecific FSH activity of horse (e) LH/choriogonadotropin (CG) (eLH) and donkey (dk) LH/CG (dkLH) (FSH/LH ratio ten times higher for eLH than for dkLH); this FSH activity absolutely requires an equid (donkey or horse) alpha-subunit combined with an equid beta-LH subunit. DESIGN: Chimeric alpha-subunits possessing the first 63 amino-acids of the porcine (p) and the last 33 amino-acids of the donkey alpha-subunit (alphap-dk) and the inverse (alphadk-p) were constructed. Porcine-specific amino-acids were introduced by mutagenesis in donkey alpha-subunit at positions 70, 85, 89, 93 and 96 (alphadk5xmut), 18 (alphadkK18E) or 78 (alphadkI78A). METHODS: These different alpha-subunits were co-transfected in COS-7 cells with beta-eLH, beta-dkLH and beta-eFSH. The LH and FSH bioactivities of the dimers were then assessed in two heterologous in vitro bioassays. RESULTS: alphap-dk or alphadk-p exhibited FSH activity when co-expressed with beta-eLH but not with beta-dkLH. alphadkK18E or alphadkI78A gave hybrids with no FSH activity and important LH activity when expressed with beta-dkLH. alphadkI78A/betaeLH displayed an FSH/LH ratio as low as that of dkLH. However, mutation at 78 in alpha-dk had no effect on FSH bioactivity when co-expressed with beta-eFSH. CONCLUSIONS: Amino-acids present in both the first two-thirds and the last third of the alpha-subunit of equid LHs are involved in their heterologous biospecificity. Ile alpha78 exerts as strong an influence on it as the beta102-103 residues. By contrast, this residue plays no role in the FSH specificity of eFSH.

A molecular approach to dominance in hypophosphatasia.

Hypophosphatasia is an inherited disorder characterized by defective bone mineralization and a deficiency of tissue-nonspecific alkaline phosphatase (TNSALP) activity. The disease is highly variable in its clinical expression, because of various mutations in the TNSALP gene. In approximately 14% of the patients tested in our laboratory, only one TNSALP gene mutation was found, despite exhaustive sequencing of the gene, suggesting that missing mutations are harbored in intron or regulatory sequences or that the disease is dominantly transmitted. The distinction between these two situations is of importance, especially in terms of genetic counseling, but dominance is sometimes difficult to conclusively determine by using familial analysis since expression of the disease may be highly variable, with parents of even severely affected children showing no or extremely mild symptoms of the disease. We report here the study of eight point mutations (G46 V, A99T, S164L, R167 W, R206 W, G232 V, N461I, I473F) found in patients with no other detectable mutation. Three of these mutations, G46 V, S164L, and I473F, have not previously been described. Pedigree and/or serum alkaline phosphatase data suggested possible dominant transmission in families with A99T, R167 W, and G232 V. By means of site-directed mutagenesis, transfections in COS-1 cells, and three-dimensional (3D) modeling, we evaluated the possible dominant effect of these eight mutations. The results showed that four of these mutations (G46 V, A99T, R167 W, and N461I) exhibited a negative dominant effect by inhibiting the enzymatic activity of the heterodimer, whereas the four others did not show such inhibition. Strong inhibition resulted in severe hypophosphatasia, whereas partial inhibition resulted in milder forms of the disease. Analysis of the 3D model of the enzyme showed that mutations exhibiting a dominant effect were clustered in two regions, viz., the active site and an area probably interacting with a region having a particular biological function such as dimerization, tetramerization, or membrane anchoring.

Sheep cytosolic branched-chain amino acid aminotransferase: cDNA cloning, primary structure and molecular modelling and its unique expression in muscles.

This paper presents the cloning and the molecular modelling of the cytosolic branched-chain amino acid aminotransferase (BCATc) from sheep brain. The sheep BCATc cDNA (3 kb) encodes a mature polypeptide of 385 amino acids with a calculated molecular mass of 43072.93 Da. The sequence of the sheep BCATc cDNA is more similar to other mammalian BCATc cDNAs (53-87% identical) than to the sheep mitochondrial branched-chain amino acid aminotransferase (52%). Sheep BCATc belongs to the IV Folding class of pyridoxal-5'-phosphate-depending enzymes. Based on the known structure of the branched-chain amino acid aminotransferase (BCAT) from Escherichia coli, a molecular model of sheep BCATc (amino acid residues 62-385) was built. This is the first three-dimensional model of any mammalian BCAT. It suggests that the enzymatic mechanism of sheep BCATc and likely of all mammalian BCAT is very similar to the mechanism of the E. coli BCAT and confirms the hypotheses regarding to the substrate binding sites of E. coli BCAT. Sheep skeletal muscle, which is the main in vivo site for transamination of branched-chain amino acids, exhibits an unique expression of BCATc.

A 3D model of human P450c21: study of the putative effects of steroid 21-hydroxylase gene mutations.

In order to better understand the disease-causing role of missense mutations found in the CYP21 gene from patients affected with congenital adrenal hyperplasia (CAH) due to steroid 21-hydroxylase deficiency, we built two three-dimensional (3D) models of human P450c21 using all known 3D structures of P450s. For each residue affected by a missense mutation, its location in the 3D structure and the putative changes in terms of biochemical properties brought about by the mutation were analyzed. Most of the severe alleles were found to affect residues located in functionally important regions of the molecule such as substrate recognition sites (SRS) or the heme region, whereas moderate mutations were mostly found in less crucial regions of the molecule. Thus, there is a good correlation between the 3D structure study and clinical data and mutagenesis experiments previously reported. In one case, however, the observed clinical severity of the mutation (E380D) did not match its expected severity deduced from the model, pointing to a potential functionally important region of the molecule. Our 3D human models provide a basic model for further studies of mutations responsible for 21-hydroxylase, and for identification of important residues involved in the specific activity of the enzyme.

Correlations of genotype and phenotype in hypophosphatasia.

Hypophosphatasia, a rare inherited disorder characterized by defective bone mineralization, is highly variable in its clinical expression. The disease is due to various mutations in the tissue-non-specific alkaline phosphatase ( TNSALP ) gene. We report here the use of clinical data, site-directed mutagenesis and computer-assisted modelling to propose a classification of 32 TNSALP gene mutations found in 23 European patients, 17 affected with lethal hypophosphatasia and six with non-lethal hypophosphatasia. Transfection studies of the missense mutations found in non-lethal hypophosphatasia showed that six of them allowed significant residual in vitro enzymatic activity, suggesting that these mutations corresponded to moderate alleles. Each of the six patients with non-lethal hypophosphatasia carried at least one of these alleles. The three-dimensional model study showed that moderate mutations were not found in the active site, and that most of the severe missense mutations were localized in crucial domains such as the active site, the vicinity of the active site and homodimer interface. Some mutations appeared to be organized in clusters on the surface of the molecule that may represent possible candidates for regions interacting with the C-terminal end involved in glycosylphosphatidylinositol (GPI) attachment or with other dimers to form tetramers. Finally, our results show a good correlation between clinical forms of the disease, mutagenesis experiments and the three-dimensional structure study, and allowed us to clearly distinguish moderate alleles from severe alleles. They also confirm that the extremely high phenotypic heterogeneity observed in patients with hypophosphatasia was due mainly to variable residual enzymatic activities allowed by missense mutations found in the human TNSALP gene.

In unison: regularization of protein secondary structure predictions that makes use of multiple sequence alignments.

We present a method whose purpose is to post-process the fuzzy results of secondary structure prediction methods that use multiple sequence alignments, in order to obtain 'realistic' secondary structures, i.e., secondary structure elements whose length is greater than or equal to some predefined minimum length. This regularization helps with interpretation of the secondary structure prediction.

Surprising similarities in structure comparison.

Examination of a protein's structural 'neighbors' can reveal distant evolutionary relationships that are otherwise undetectable, and perhaps suggest unsuspected functional properties. In the past, such analyses have often required specialized software and computer skills, but new structural comparison methods, developed in the past two years, increasingly offer this opportunity to structural and molecular biologists in general. These methods are based on similarity-search algorithms that are fast enough to have effectively removed the computer-time limitation for structure-structure search and alignment, and have made it possible for several groups to conduct systematic comparisons of all publicly available structures, and offer this information via the World Wide Web. Furthermore, and perhaps surprisingly given the difficulty of the structure-comparison problem, these groups seem to have converged on quite similar approaches with respect to both fast search algorithms and the identification of statistically significant similarities.

Threading a database of protein cores.

We present an analysis of 10 blind predictions prepared for a recent conference, "Critical Assessment of Techniques for Protein Structure Prediction." The sequences of these proteins are not detectably similar to those of any protein in the structure database then available, but we attempted, by a threading method, to recognize similarity to known domain folds. Four of the 10 proteins, as we subsequently learned, do indeed show significant similarity to then-known structures. For 2 of these proteins the predictions were accurate, in the sense that a similar structure was at or near the top of the list of threading scores, and the threading alignment agreed well with the corresponding structural alignment. For the best predicted model mean alignment error relative to the optimal structural alignment was 2.7 residues, arising entirely from small "register shifts" of strands or helices. In the analysis we attempt to identify factors responsible for these successes and failures. Since our threading method does not use gap penalties, we may readily distinguish between errors arising from our prior definition of the "cores" of known structures and errors arising from inherent limitations in the threading potential. It would appear from the results that successful substructure recognition depends most critically on accurate definition of the "fold" of a database protein. This definition must correctly delineate substructures that are, and are not, likely to be conserved during protein evolution.

Influence of the local amino acid sequence upon the zones of the torsional angles phi and psi adopted by residues in proteins.

A set of parameters is derived to express the influence of the local amino acid sequence on the torsional angles phi and psi adopted by each residue in a protein. The formalism used, which is based on information theory, evaluates the probability for a given residue to be in a particular zone of the Ramachandran map. Comparisons with crystallographic structures suggest that the method can extract almost all of the available information from the local sequence and show that the local sequence carries only, on average, about 65% of the information necessary for specifying the conformation of a given residue in a protein. The rest is specified by long-range interactions that are specific for each protein fold. The parameters derived here provide a more detailed description of the prediction than other methods in allowing the allocation of the torsional angles for residues having an aperiodic structure and are intended to be used for directing the conformational search in a subsequent simulation of the three-dimensional structure. This method should also predict segments of the polypeptide chain that are the most stable and thus less sensitive to long-range interactions.

Normal mode analysis of human lysozyme: study of the relative motion of the two domains and characterization of the harmonic motion.

A normal mode analysis of human lysozyme has been carried out at room temperature. Human lysozyme is an enzyme constituted of two domains separated by an active site cleft, the motion of which is thought to be relevant for biological function. This motion has been described as a hinge bending motion. McCammon et al. have determined the characteristics of the hinge bending motion but they assumed a prior knowledge of the hinge axis. In this work we propose a method which is free from this assumption and determines the hinge axis and root mean square (rms) rotation angle which give the best agreement with the pattern of changes in all the distances between nonhydrogen atoms in the two domains, obtained by the normal mode analysis. The hinge axis we found is notably different from the one previously determined and goes, roughly, through the C alpha 55 and C alpha 76, i.e., it is located at the base of the beta-sheet of the second domain. The rms value for the rotation angle is also twice as large as the previous one: 3.37 degrees. It is shown that this hinge bending motion provides a fairly good approximation of the dynamics of human lysozyme and that the normal mode with the lowest frequency has a dominating contribution to this hinge bending motion. A study of the accessible surface area of the residues within the cleft reveals that the motion does not result in a better exposure to the solvent of these residues. A characterization of the thermally excited state (under the hypothesis of the harmonicity of the potential energy surface) has been done using the concept of topology of atom packing. Under this hypothesis the thermal fluctuations result only in a small change of the topology of atom packing, leading therefore to nearly elastic deformations of the protein.

Secondary structure prediction and protein design.

For non-homologous proteins, and after cross-validation, the methods reviewed in this article exhibit a probability index (percentage of correctly predicted residues per predicted residues) of 59-65.5% according to the methods employed with a standard deviation of 7% for three conformational states: alpha-helix, beta-strand and coil. These present limitations in the accuracy of prediction are related both to the limited number of known structures and to the effect of long-range interactions. The methods based on sequence similarity can improve the accuracy of prediction by expressing the homology of the protein to be predicted explicitly with proteins in the database. Under these circumstances, the probability index can reach 87% with a standard deviation of 6.6%. This property was used for modelling homologous proteins by assisting in amino acid sequence alignments. Examples will be given for the alignment of serine proteinases.

Model-building of Fnr and FixK DNA-binding domains suggests a basis for specific DNA recognition.

The DNA-binding C-terminal domains of the regulatory proteins Fnr from Escherichia coli and FixK from Rhizobium meliloti have been modelled on the basis of their homologies to the CAP protein from E. coli. Residues Glu181, Thr182 and Arg185 of CAP, which are exposed residues of the DNA-recognition helix alpha F, are conserved in Fnr and FixK. However, Arg180 and Gly184 are substituted by Val and Ser respectively in Fnr. We propose that this valine makes a Van der Waals' contact with the first thymine in the Fnr consensus TTGA-N6-TCAA, and that the serine contributes to the binding by displacing a thymine-bound water molecule. The corresponding residues in FixK, Ile and Ser allow the same interactions with a thymine. Therefore we predict that FixK may recognize the same sites as Fnr. This is supported experimentally by showing that Fnr can substitute for FixK in activating the fixN gene in E. coli.

Secondary structure prediction: combination of three different methods.

A combination of three complementary secondary structure prediction methods is presented. The methods used are the GOR III method, the Homologue method and a new method, the bit pattern method, which is based on hydrophilic/hydrophobic residue patterns. For this purpose a hydropathy scale was developed and is presented here. The combination algorithm (Combine method) was designed to take the best results of each method and use their differences in order to improve the prediction. The combination yields 65.5% correctly predicted residues in three states: alpha-helix (H), beta-strand (E) and aperiodic structure (C) which is an improvement ranging from 2.5 to 6.5% compared with the individual methods when tested with a 67-polypeptide chain database. Seventy-five per cent of the regular secondary structure (H and E) runs are correctly located and beta-sheet runs are much better located by the Combine method in comparison to the other methods.

Further developments of protein secondary structure prediction using information theory. New parameters and consideration of residue pairs.

We have re-evaluated the information used in the Garnier-Osguthorpe-Robson (GOR) method of secondary structure prediction with the currently available database. The framework of information theory provides a means to formulate the influence of local sequence upon the conformation of a given residue, in a rigorous manner. However, the existing database does not allow the evaluation of parameters required for an exact treatment of the problem. The validity of the approximations drawn from the theory is examined. It is shown that the first-level approximation, involving single-residue parameters, is only marginally improved by an increase in the database. The second-level approximation, involving pairs of residues, provides a better model. However, in this case the database is not big enough and this method might lead to parameters with deficiencies. Attention is therefore given to overcoming this lack of data. We have determined the significant pairs and the number of dummy observations necessary to obtain the best result for the prediction. This new version of the GOR method increases the accuracy of prediction by 7%, bringing the amount of residues correctly predicted to 63% for three states and 68 proteins, each protein to be predicted being removed from the database and the parameters derived from the other proteins. If the protein to be predicted is kept in the database the accuracy goes up to 69.7%.

Prediction of the conformation and antigenic determinants of the V-sis viral oncogene product homologous with human platelet-derived growth factor.

In an extensive study involving unification of secondary structure prediction methods to predict peptide backbone angles, folding simulations, and interactive graphics we have calculated conformations for a 160 residue p28sis protein fragment. We demonstrate that the state of the art is sufficient to aid artificial vaccine design.