Sequence-based prediction of pathological mutations.

The development of methods to assess the impact of amino acid mutations on human health has become an important goal in biomedical research, due to the growing number of nonsynonymous SNPs identified. Within this context, computational methods constitute a valuable tool, because they can easily process large amounts of mutations and give useful, almost cost-free, information on their pathological character. In this paper we present a computational approach to the prediction of disease-associated amino acid mutations, using only sequence-based information (amino acid properties, evolutionary information, secondary structure and accessibility predictions, and database annotations) and neural networks, as a model building tool. Mutations are predicted to be either pathological or neutral. Our results show that the method has a good overall success rate, 83%, that can reach 95% when trained for specific proteins. The methodology is fast and flexible enough to provide good estimates of the pathological character of large sets of nonsynonymous SNPs, but can also be easily adapted to give more precise predictions for proteins of special biomedical interest.

Classical molecular interaction potentials: improved setup procedure in molecular dynamics simulations of proteins.

The latest version of the classical molecular interaction potential (CMIP) has the ability to predict the position of crystallographic waters in several proteins with great accuracy. This article analyzes the ability of the CMIP functional to improve the setup procedure of the molecular system in molecular dynamics (MD) simulations of proteins. To this end, the CMIP strategy is used to include both water molecules and counterions in different protein systems. The structural details of the configurations sampled from trajectories obtained using the CMIP setup procedure are compared with those obtained from trajectories derived from a standard equilibration process. The results show that standard MD simulations can lead to artifactual results, which are avoided using the CMIP setup procedure. Because the CMIP is easy to implement at a low computational cost, it can be very useful in obtaining reliable MD trajectories.

Use of surface area computations to describe atom-atom interactions.

Accessible surface (ASA) and atomic contact (ACA) areas are powerful tools for protein structure analysis. However, their use for analysis purposes could be extended if a relationship between them and protein stability could be found. At present, this is the case only for ASAs, which have been used to assess the contribution of the hydrophobic effect to protein stability. In the present work we study whether there is a relationship between atomic contact areas and the free energy associated to atom-atom interactions. We utilise a model in which the contribution of atomic interactions to protein stability is expressed as a linear function of the accessible surface area buried between atom pairs. We assess the validity of this hypothesis, using a set of 124 lysozyme mutants (Matthews, 1995, Adv Protein Chem, 249-278) for which both the X-ray structure and the experimental stability are known. We tested this assumption for residue representations with increasing numbers of atom types. Our results indicate that for simple residue representations, with only 4 to 5 atom types, there is not a clear linear relationship between stability and buried accessible area. However, this relationship is observed for representations with 6 to 9 atom types, where gross heterogeneities in the atom type definition are eliminated. Finally, we also study a version of the linear model in which the atom- atom interactions are represented utilising a simple function for the buried accessible area, which may be useful for protein structure prediction studies.

Factors limiting the performance of prediction-based fold recognition methods.

In the past few years, a new generation of fold recognition methods has been developed, in which the classical sequence information is combined with information obtained from secondary structure and, sometimes, accessibility predictions. The results are promising, indicating that this approach may compete with potential-based methods (Rost B et al., 1997, J Mol Biol 270:471-480). Here we present a systematic study of the different factors contributing to the performance of these methods, in particular when applied to the problem of fold recognition of remote homologues. Our results indicate that secondary structure and accessibility prediction methods have reached an accuracy level where they are not the major factor limiting the accuracy of fold recognition. The pattern degeneracy problem is confirmed as the major source of error of these methods. On the basis of these results, we study three different options to overcome these limitations: normalization schemes, mapping of the coil state into the different zones of the Ramachandran plot, and post-threading graphical analysis.

A new procedure for constructing peptides into a given Calpha chain.

BACKGROUND: In ab initio protein folding studies, it is often advantageous to build the Calpha chain first and then to construct the full structure by filling in the peptide groups and the sidechains. Many algorithms have been reported for constructing peptide groups on the Calpha chain, but most are unsuitable for use in such studies; some are too slow for screening a large number of trial Calpha chains and others use only the local geometry and ignore the effects of specific non-neighbor interactions, which can be crucial for proper folding. We needed a fast procedure for constructing the peptide groups that does not ignore the effects of long-range, specific interactions. RESULTS: We first found rich correlations between the peptide orientation angle and both the local Calpha-chain geometry and the type of the flanking amino acid residues. These correlations can be used to greatly limit the range of possible peptide orientation angles. We devised a simple peptide construction procedure in which all orientations within this reduced range are systematically examined and the orientation is selected that minimizes a suitable energy function that includes long-range, specific interactions. When tested on known structures, the method is found to be among the fastest of known methods and attains an accuracy comparable with or better than most methods. CONCLUSIONS: The new method is fast and takes into account both the local and non-local specific interactions. It therefore appears to be suitable for use in ab initio protein folding studies, wherein a large number of Calpha chains are screened.

Discrete representations of the protein C alpha chain.

BACKGROUND: When a large number of protein conformations are generated and screened, as in protein structure prediction studies, it is often advantageous to change the conformation in units of four consecutive residues at a time. The internal geometry of a chain of four consecutive C alpha atoms is completely described by means of the three angles theta 1, tau, and theta 2, where tau is the virtual torsion angle defined by the four atoms and theta 1 and theta 2 are the virtual bond angles flanking the torsion angle on either side. In this paper, we examine the quality of the protein structures that can be obtained when they are represented by means of a set of discrete values for these angles (discrete states). RESULTS: Different models were produced by selecting various different discrete states. The performance of these models was tested by rebuilding the C alpha chains of 139 high-resolution nonhomologous protein structures using the build-up procedure of Park and Levitt. We find that the discrete state models introduce distortions at three levels, which can be measured by means of the 'context-free', 'in-context', and the overall root-mean-square deviation of the C alpha coordinates (crms), respectively, and we find that these different levels of distortions are interrelated. As found by Park and Levitt, the overall crms decreases smoothly for most models with the complexity of the model. However, the decrease is significantly faster with our models than observed by Park and Levitt with their models. We also find that it is possible to choose models that perform considerably worse than expected from this smooth dependence on complexity. CONCLUSIONS: Of our models, the most suitable for use in initial protein folding studies appears to be model S8, in which the effective number of states available for a given residue quartet is 6.5. This model builds helices, beta-strands, and coil/loop structures with approximately equal quality and gives the overall crms value of 1.9 A on average with relatively little variation among the different proteins tried.

A 15 amino acid stretch close to the Grb2-binding domain defines two differentially expressed hSos1 isoforms with markedly different Grb2 binding affinity and biological activity.

We compared structure, expression and functional properties of two hSos1 cDNA isoforms (IsfI and Isf II) isolated, respectively, from human fetal brain and adult skeletal muscle libraries. IsfI and IsfII nucleotide sequences differ only by the presence in IsfII of an inframe 45 hp insertion located near the first proline-rich motif required for Grb2 binding. Some human tissues express only one isoform whereas others express different proportions of both in fetal and adult stages. In vitro binding assays and in vivo functional studies showed that MI exhibits significantly higher Grb2 binding affinity and biological activity than IsfI. These results suggest that functionally different hSos1 isoforms, with differential tissue expression and distribution, play important regulatory roles in the mechanisms controlling Ras activation in different tissues and/or developmental stages.

The structural homology between uteroglobin and the pore-forming domain of colicin A suggests a possible mechanism of action for uteroglobin.

Although the exact physiological function of uteroglobin is not known, it has been suggested that it may function by inhibiting phospholipase A2. We have found that the uteroglobin fold is embedded in that of the poreforming domain of colicin A. Colicin A is an antibiotic protein that kills sensitive Escherichia coli cells by forming a pore in their phospholipid membrane. The RMS deviation between the C alpha atoms after the structural alignment is 2.39 A for the 52 superimposed residues. In the alignment, uteroglobin helices 1, 2, 3, and 4 align with colicin A helices 6, 7, 3, and 4, respectively. The motif is strongly amphipathic in both proteins. On the basis of this common structural motif and of known experimental data on both proteins, we propose that UG binds to the membrane surface by lying on it monotopically. The phospholipase A2 inhibition would follow this initial binding step.

Investigation of shape variations in the antibody binding site by molecular dynamics computer simulation.

Molecular dynamics simulations have been used to investigate the flexibility and variations in the shape of the binding site of an antibody against human Rhinovirus serotype 2 (HRV2) and its complex with a 15 amino acid oligopeptide, the structure of which has been recently determined by X-ray crystallography. During the simulation of the unbound antibody the binding site, defined in terms of the hypervariable regions or complementarity determining regions (CDRs), shows significant fluctuations in shape. For the complex such variations in the shape of the binding site were reduced. The largest fluctuations in the unbound antibody occurred within the CDR-H3. The largest differences between the bound and unbound crystal structures are also associated with CDR-H3. The relative displacements of the loops have been analysed in terms of internal distortions, rigid body motions of the loops and changes with respect to the framework regions. The degree to which the motions of the loops are correlated and the variation in the volume of the binding pocket during the simulation have also been examined.

Representation of noncovalent interactions in protein structures.

The energetics of solvent-atom and atom-atom nonbonded interactions can be described, for protein structures, in terms of the accessible and the contact atomic surface areas, respectively. This type of description emphasizes the importance of the local environment around groups in the three-dimensional structure of protein molecules. The graphical representation of nonbonded interactions according to this description allows one to visualize the spatial extent and distribution of these interactions and the relative stability of atoms or atomic groups in known or modified protein conformations. Applications of this short range description and of its graphical representation will be discussed.