Integrative bioinformatics: making sense of the networks

The focus of biology has shifted from the investigation of individual genes and proteins to the study of large complex networks featuring interactions between tens of thousands of molecular and cellular components. Information on these networks is obtained from genome-scale experimental and theoretical analyses, which yield valuable but noisy data, on biological processes that are still poorly understood. The new exciting developments in bioinformatics show great promise in meeting the challenge of extracting biological insight from these data.:

Transcriptional regulation of protein complexes in yeast

BACKGROUND

Multiprotein complexes play an essential role in many cellular processes. But our knowledge of the mechanism of their formation, regulation and lifetimes is very limited. We investigated transcriptional regulation of protein complexes in yeast using two approaches. First, known regulons, manually curated or identified by genome-wide screens, were mapped onto the components of multiprotein complexes. The complexes comprised manually curated ones and those characterized by high-throughput analyses. Second, putative regulatory sequence motifs were identified in the upstream regions of the genes involved in individual complexes and regulons were predicted on the basis of these motifs.

RESULTS

Only a very small fraction of the analyzed complexes (5-6%) have subsets of their components mapping onto known regulons. Likewise, regulatory motifs are detected in only about 8-15% of the complexes, and in those, about half of the components are on average part of predicted regulons. In the manually curated complexes, the so-called 'permanent' assemblies have a larger fraction of their components belonging to putative regulons than 'transient' complexes. For the noisier set of complexes identified by high-throughput screens, valuable insights are obtained into the function and regulation of individual genes.

CONCLUSIONS

A small fraction of the known multiprotein complexes in yeast seems to have at least a subset of their components co-regulated on the transcriptional level. Preliminary analysis of the regulatory motifs for these components suggests that the corresponding genes are likely to be co-regulated either together or in smaller subgroups, indicating that transcriptionally regulated modules might exist within complexes.

Combining pattern discovery and discriminant analysis to predict gene co-regulation

MOTIVATION

Several pattern discovery methods have been proposed to detect over-represented motifs in upstream sequences of co-regulated genes, and are for example used to predict cis-acting elements from clusters of co-expressed genes. The clusters to be analyzed are often noisy, containing a mixture of co-regulated and non-co-regulated genes. We propose a method to discriminate co-regulated from non-co-regulated genes on the basis of counts of pattern occurrences in their non-coding sequences.

METHODS

String-based pattern discovery is combined with discriminant analysis to classify genes on the basis of putative regulatory motifs.

RESULTS

The approach is evaluated by comparing the significance of patterns detected in annotated regulons (positive control), random gene selections (negative control) and high-throughput regulons (noisy data) from the yeast Saccharomyces cerevisiae. The classification is evaluated on the annotated regulons, and the robustness and rejection power is assessed with mixtures of co-regulated and random genes.

Prediction of protein–protein interactions: the CAPRI experiment, its evaluation and implications

Given the increasing interest in protein-protein interactions, the prediction of these interactions from sequence and structural information has become a booming activity. CAPRI, the community-wide experiment for assessing blind predictions of protein-protein interactions, is playing an important role in fostering progress in docking procedures. At the same time, novel methods are being derived for predicting regions of a protein that are likely to interact and for characterizing putative intermolecular contacts from sequence and structural data. Together with docking procedures, these methods provide an integrated computational approach that should be a valuable complement to genome-scale experimental studies of protein-protein interactions.

The aMAZE LightBench: a web interface to a relational database of cellular processes

The aMAZE LightBench (http://www.amaze.ulb. ac.be/) is a web interface to the aMAZE relational database, which contains information on gene expression, catalysed chemical reactions, regulatory interactions, protein assembly, as well as metabolic and signal transduction pathways. It allows the user to browse the information in an intuitive way, which also reflects the underlying data model. Moreover links are provided to literature references, and whenever appropriate, to external databases.

ACLAME: a CLAssification of Mobile genetic Elements

The ACLAME database (http://aclame.ulb.ac.be) is a collection and classification of prokaryotic mobile genetic elements (MGEs) from various sources, comprising all known phage genomes, plasmids and transposons. In addition to providing information on the full genomes and genetic entities, it aims to build a comprehensive classification of the functional modules of MGEs at the protein, gene and higher levels. This first version contains a comprehensive classification of 5069 proteins from 119 DNA bacteriophages into over 400 functional families. This classification was produced automatically using TRIBE-MCL, a graph-theory-based Markov clustering algorithm that uses sequence measures as input, and then manually curated. Manual curation was aided by consulting annotations available in public databases retrieved through additional sequence similarity searches using Psi-Blast and Hidden Markov Models. The database is publicly accessible and open to expert volunteers willing to participate in its curation. Its web interface allows browsing as well as querying the classification. The main objectives are to collect and organize in a rational way the complexity inherent to MGEs, to extend and improve the inadequate annotation currently associated with MGEs and to screen known genomes for the validation and discovery of new MGEs.

Assessment of blind predictions of protein-protein interactions: current status of docking methods

The current status of docking procedures for predicting protein-protein interactions starting from their three-dimensional structure is assessed from a first major evaluation of blind predictions. This evaluation was performed as part of a communitywide experiment on Critical Assessment of PRedicted Interactions (CAPRI). Seven newly determined structures of protein-protein complexes were available as targets for this experiment. These were the complexes between a kinase and its protein substrate, between a T-cell receptor beta-chain and a superantigen, and five antigen-antibody complexes. For each target, the predictors were given the experimental structures of the free components, or of one free and one bound component in a random orientation. The structure of the complex was revealed only at the time of the evaluation. A total of 465 predictions submitted by 19 groups were evaluated. These groups used a wide range of algorithms and scoring functions, some of which were completely novel. The quality of the predicted interactions was evaluated by comparing residue-residue contacts and interface residues to those in the X-ray structures and by analyzing the fit of the ligand molecules (the smaller of the two proteins in the complex) or of interface residues only, in the predicted versus target complexes. A total of 14 groups produced predictions, ranking from acceptable to highly accurate for five of the seven targets. The use of available biochemical and biological information, and in one instance structural information, played a key role in achieving this result. It was essential for identifying the native binding modes for the five correctly predicted targets, including the kinase-substrate complex where the enzyme changes conformation on association. But it was also the cause for missing the correct solution for the two remaining unpredicted targets, which involve unexpected antigen-antibody binding modes. Overall, this analysis reveals genuine progress in docking procedures but also illustrates the remaining serious limitations and points out the need for better scoring functions and more effective ways for handling conformational flexibility.

CAPRI: a Critical Assessment of PRedicted Interactions

CAPRI is a communitywide experiment to assess the capacity of protein-docking methods to predict protein-protein interactions. Nineteen groups participated in rounds 1 and 2 of CAPRI and submitted blind structure predictions for seven protein-protein complexes based on the known structure of the component proteins. The predictions were compared to the unpublished X-ray structures of the complexes. We describe here the motivations for launching CAPRI, the rules that we applied to select targets and run the experiment, and some conclusions that can already be drawn. The results stress the need for new scoring functions and for methods handling the conformation changes that were observed in some of the target systems. CAPRI has already been a powerful drive for the community of computational biologists who development docking algorithms. We hope that this issue of Proteins will also be of interest to the community of structural biologists, which we call upon to provide new targets for future rounds of CAPRI, and to all molecular biologists who view protein-protein recognition as an essential process.

Activation of CCR5 by chemokines involves an aromatic cluster between transmembrane helices 2 and 3

CCR5 is a G protein-coupled receptor responding to four natural agonists, the chemokines RANTES (regulated on activation normal T cell expressed and secreted), macrophage inflammatory protein (MIP)-1 alpha, MIP-1 beta, and monocyte chemotactic protein (MCP)-2, and is the main co-receptor for the macrophage-tropic human immunodeficiency virus strains. We have previously identified a structural motif in the second transmembrane helix of CCR5, which plays a crucial role in the mechanism of receptor activation. We now report the specific role of aromatic residues in helices 2 and 3 of CCR5 in this mechanism. Using site-directed mutagenesis and molecular modeling in a combined approach, we demonstrate that a cluster of aromatic residues at the extracellular border of these two helices are involved in chemokine-induced activation. These aromatic residues are involved in interhelical interactions that are key for the conformation of the helices and govern the functional response to chemokines in a ligand-specific manner. We therefore suggest that transmembrane helices 2 and 3 contain important structural elements for the activation mechanism of chemokine receptors, and possibly other related receptors as well.

Automatic sequence design of major histocompatibility complex class I binding peptides impairing CD8+ T cell recognition

An automatic protein design procedure was used to compute amino acid sequences of peptides likely to bind the HLA-A2 major histocompatibility complex (MHC) class I allele. The only information used by the procedure are a structural template, a rotamer library, and a well established classical empirical force field. The calculations are performed on six different templates from x-ray structures of HLA-A0201-peptide complexes. Each template consists of the bound peptide backbone and the full atomic coordinates of the MHC protein. Sequences within 2 kcal/mol of the minimum energy sequence are computed for each template, and the sequences from all the templates are combined and ranked by their energies. The five lowest energy peptide sequences and five other low energy sequences re-ranked on the basis of their similarity to peptides known to bind the same MHC allele are chemically synthesized and tested for their ability to bind and form stable complexes with the HLA-A2 molecule. The most efficient binders are also tested for inhibition of the T cell receptor recognition of two known CD8(+) T effectors. Results show that all 10 peptides bind the expected MHC protein. The six strongest binders also form stable HLA-A2-peptide complexes, albeit to varying degrees, and three peptides display significant inhibition of CD8(+) T cell recognition. These results are rationalized in light of our knowledge of the three-dimensional structures of the HLA-A2-peptide and HLA-A2-peptide-T cell receptor complexes.

An overview of data models for the analysis of biochemical pathways

Biochemical pathways such as metabolic, regulatory or signal transduction pathways can be viewed as interconnected processes forming an intricate network of functional and physical interactions between molecular species in the cell. The amount of information available on such pathways for different organisms is increasing very rapidly. This is offering the possibility of performing various analyses on the structure of the full network of pathways for one organism as well as across different organisms, and has therefore generated interest in developing databases for storing and managing this information. Analysing these networks remains far from straightforward owing to the nature of the databases, which are often heterogeneous, incomplete or inconsistent. Pathway analysis is hence a challenging problem in systems biology and in bioinformatics. Various forms of data models have been devised for the analysis of biochemical pathways. This paper presents an overview of the types of models used for this purpose, concentrating on those concerned with the structural aspects of biochemical networks. In particular, the different types of data models found in the literature are classified using a unified framework. In addition, how these models have been used in the analysis of biochemical networks is described. This enables us to underline the strengths and weaknesses of the different approaches, as well as to highlight relevant future research directions.

Folding free energy function selects native-like protein sequences in the core but not on the surface

An automatic protein design procedure is used to select amino acid sequences that optimize the folding free energy function for a given protein. The only information used in designing the sequences is a set of known backbone structures for each protein, a rotamer library, and a well established classical empirical force field, which relies on basic physical chemical principles that underlie molecular interactions and protein stability, and has not been adjusted to yield native-like sequences. Applying the procedure to 7 different known protein folds, representing a total of 45 different native protein structures, yields ensembles of designed sequences displaying remarkable similarity to their natural counterparts in the protein core, but which are distinctly non-native on the protein surface. We show that natural and designed sequences for a given fold score significantly higher than random sequences against profiles derived from both, designed and natural sequence ensembles. Furthermore, we find that designed sequence profiles can be used to retrieve the native sequences for many of the analyzed proteins using standard PSI-BLAST searches in sequence databases. These findings may have important implications for our understanding the selection pressures operating on natural protein sequences and hold promise for improving fold recognition.

Conserved water molecules in MHC class-I molecules and their putative structural and functional roles

A set of conserved water positions making direct contacts with the alpha1 and alpha2 domains of the MHC class-I protein was identified by a cluster analysis in 12 high-resolution crystal structures of proteins from different allele types and different species, comprising human, mouse and rat. The analysis revealed a total of 63 clusters, corresponding to water molecules, whose positions are conserved in half or more of the analyzed structures. Analysis of these clusters shows that the most conserved water positions-those appearing in the largest fraction of the structures-were also the most accurately defined, as measured by their normalized crystallographic B-factor. Not too surprisingly, these positions displayed better overlap and formed more H-bonds with the protein. In a second part of this work, a detailed analysis is presented of three of the most conserved water positions and their putative structural and functional roles are discussed. The most highly conserved of the three appears to play an important role in stabilizing the conformation of a twisted beta-turn between residues 118 and 122 (numbering of HLA-B3501, PDB code 1A1N). An equivalent water molecule was found to be associated with a similar beta-turn in 43 unrelated structures surveyed in the PDB, leading to the suggestion that this water molecule plays an important structural role in this type of turn. The second water molecule makes hydrogen bonds with residues lining pocket B in the peptide-binding groove and is suggested to play a role in modulating peptide recognition. The third highly conserved water molecule is located at the first kink of the alpha2 helix, possibly playing a role in determining the position of the N-terminal segment of that helix, which also carries side chains in contact with the bound peptide. This information on conserved water positions in MHC class-I molecules should be helpful in modeling interactions with bound peptide antigens and in designing new peptides with tailor-made affinities.

Pescador: the PEptides in Solution ConformAtion Database: Online Resource

In recent years a large body of data has been obtained from Nuclear Magnetic Resonance and Circular Dichroism experiments on the influence of the amino acid sequence and various other parameters on the conformational state of peptides in solution. Interpreting the experimental data in terms of the conformational populations of the peptides remains a key problem, for which current solutions leave appreciable room for improvement. Considering that making this body of data available for surveys and analysis should be instrumental in tackling the problem, we undertook the development of Pescador: The 'PEptides in Solution ConformAtion Database: Online Resource'. Pescador contains data from NMR and CD spectroscopy on peptides in solution as well as information on the structural parameters derived from these data. It also features specialized Web-based tools for data deposition, and means for readily accessing the stored information for analysis purposes. To illustrate the use of the database in deriving information for the conformational analysis of peptides, we show how the alpha proton delta-values stored in Pescador and measured by NMR for different peptides in different laboratories can be used to derive a new set of 'random coil' chemical shift values. Firstly, we show these values to be very similar to those obtained experimentally for model peptides in water, and their variation with increasing Tri-Fluoro-Ethanol (TFE) concentration is similar to that reported for model peptides. We show, furthermore, that the chemical shift data in Pescador can be used to derive correction factors that take into account effects of neighboring residues. These correction factors compare favorably with those recently derived from a series of model GGXGG peptides (Schwarzinger et al., 2001). These encouraging results suggest that, as the quantity of NMR data on peptide deposited in Pescador increases, surveys of these data should be a valuable means of deriving key parameters for the analysis of peptide conformation.

Advanced pairwise structure alignments of proteins and analysis of conformational changes

MOTIVATION

Comparing the 3D structures of two proteins or analyzing the structural changes undergone by a protein upon ligand binding or when it crystallizes under different conditions, can be both tricky and tedious, especially when the two proteins are distantly related, or when the structural changes are complex. Readily accessible tools for performing these tasks automatically and reliably should therefore be welcome.

RESULTS

We describe a web interface to several automatic procedures for performing pairwise structure superposition in a flexible manner, for detailed analyses of conformational changes and for displaying the results in a pictorial fashion.

AVAILABILITY

This interface can be accessed at the Brussels and Cuba Web sites, respectively: http://www.ucmb.ulb.ac.be/SCMBB/Tools.htmland http://bio.cigb.edu.cu.

Automatic procedures for protein design

This review describes computational procedures for deriving the amino acid sequences that are compatible with a given protein backbone structure. Such procedures can be used to gain insight into the constraints imposed by the 3D structure of the protein sequence, or to design proteins that are likely to adopt a given backbone conformation. We start by presenting a short overview of the various types of approaches to protein design developed over more than a decade. This is followed by a more detailed presentation of a recently developed sequence selection procedure DESIGNER. This latter presentation illustrates the basic principles underlying this type of procedures, described what they may teach us when applied to small proteins, and highlights issues that need to be addressed in order to go forward.

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.

A conserved Asn in transmembrane helix 7 is an on/off switch in the activation of the thyrotropin receptor

The thyrotropin (TSH) receptor is an interesting model to study G protein-coupled receptor activation as many point mutations can significantly increase its basal activity. Here, we identified a molecular interaction between Asp(633) in transmembrane helix 6 (TM6) and Asn(674) in TM7 of the TSHr that is crucial to maintain the inactive state through conformational constraint of the Asn. We show that these residues are perfectly conserved in the glycohormone receptor family, except in one case, where they are exchanged, suggesting a direct interaction. Molecular modeling of the TSHr, based on the high resolution structure of rhodopsin, strongly favors this hypothesis. Our approach combining site-directed mutagenesis with molecular modeling shows that mutations disrupting this interaction, like the D633A mutation in TM6, lead to high constitutive activation. The strongly activating N674D (TM7) mutation, which in our modeling breaks the TM6-TM7 link, is reverted to wild type-like behavior by an additional D633N mutation (TM6), which would restore this link. Moreover, we show that the Asn of TM7 (conserved in most G protein-coupled receptors) is mandatory for ligand-induced cAMP accumulation, suggesting an active role of this residue in activation. In the TSHr, the conformation of this Asn residue of TM7 would be constrained, in the inactive state, by its Asp partner in TM6.

Checking nucleic acid crystal structures

The program SFCHECK [Vaguine et al. (1999), Acta Cryst. D55, 191-205] is used to survey the quality of the structure-factor data and the agreement of those data with the atomic coordinates in 105 nucleic acid crystal structures for which structure-factor amplitudes have been deposited in the Nucleic Acid Database [NDB; Berman et al. (1992), Biophys. J. 63, 751-759]. Nucleic acid structures present a particular challenge for structure-quality evaluations. The majority of these structures, and DNA molecules in particular, have been solved by molecular replacement of the double-helical motif, whose high degree of symmetry can lead to problems in positioning the molecule in the unit cell. In this paper, the overall quality of each structure was evaluated using parameters such as the R factor, the correlation coefficient and various atomic error estimates. In addition, each structure is characterized by the average values of several local quality indicators, which include the atomic displacement, the density correlation, the B factor and the density index. The latter parameter measures the relative electron-density level at the atomic position. In order to assess the quality of the model in specific regions, the same local quality indicators are also surveyed for individual groups of atoms in each structure. Several of the global quality indicators are found to vary linearly with resolution and less than a dozen structures are found to exhibit values significantly different from the mean for these indicators, showing that the quality of the nucleic acid structures tends to be rather uniform. Analysis of the mutual dependence of the values of different local quality indicators, computed for individual residues and atom groups, reveals that these indicators essentially complement each other and are not redundant with the B factor. Using several of these indicators, it was found that the atomic coordinates of the nucleic acid bases tend to be better defined than those of the backbone. One of the local indicators, the density index, is particularly useful in spotting regions of the model that fit poorly in the electron density. Using this parameter, the quality of crystallographic water positions in the analyzed structures was surveyed and it was found that a sizable fraction of these positions have poorly defined electron density and may therefore not be reliable. The possibility that cases of poorly positioned water molecules are symptomatic of more widespread problems with the structure as a whole is also raised.