SH3 Domain Family Research Articles

Most yeast SH3 domains bind peptide targets with high intrinsic specificity.

A need exists to develop bioinformatics for predicting differences in protein function, especially for members of a domain family who share a common fold, yet are found in a diverse array of proteins. Many domain families have been conserved over large evolutionary spans and representative genomic data during these periods are now available. This allows a simple method for grouping domain sequences to reveal common and unique/specific binding residues. As such, we hypothesize that sequence alignment analysis of the yeast SH3 domain family across ancestral species in the fungal kingdom can determine whether each member encodes specific information to bind unique peptide targets. With this approach, we identify important specific residues for a given domain as those that show little conservation within an alignment of yeast domain family members (paralogs) but are conserved in an alignment of its direct relatives (orthologs). We find most of the yeast SH3 domain family members have maintained unique amino acid conservation patterns that suggest they bind peptide targets with high intrinsic specificity through varying degrees of non-canonical recognition. For a minority of domains, we predict a less diverse binding surface, likely requiring additional factors to bind targets specifically. We observe that our predictions are consistent with high throughput binding data, which suggests our approach can probe intrinsic binding specificity in any other interaction domain family that is maintained during evolution.

Comprehensive Analysis of the Human SH3 Domain Family Reveals a Wide Variety of Non-canonical Specificities

SH3 domains are protein modules that mediate protein-protein interactions in many eukaryotic signal transduction pathways. The majority of SH3 domains studied thus far act by binding to proline-rich sequences in partner proteins, but a growing number of studies have revealed alternative recognition mechanisms. We have comprehensively surveyed the specificity landscape of human SH3 domains in an unbiased manner using peptide-phage display and deep sequencing. Based on ∼70,000 unique binding peptides, we obtained 154 specificity profiles for 115 SH3 domains, which reveal that roughly half of the SH3 domains exhibit non-canonical specificities and collectively recognize a wide variety of peptide motifs, most of which were previously unknown. Crystal structures of SH3 domains with two distinct non-canonical specificities revealed novel peptide-binding modes through an extended surface outside of the canonical proline-binding site. Our results constitute a significant contribution toward a complete understanding of the mechanisms underlying SH3-mediated cellular responses.

Monte Carlo simulation of a statistical mechanical model of multiple protein sequence alignment.

A grand canonical Monte Carlo (MC) algorithm is presented for studying the lattice gas model (LGM) of multiple protein sequence alignment, which coherently combines long-range interactions and variable-length insertions. MC simulations are used for both parameter optimization of the model and production runs to explore the sequence subspace around a given protein family. In this Note, I describe the details of the MC algorithm as well as some preliminary results of MC simulations with various temperatures and chemical potentials, and compare them with the mean-field approximation. The existence of a two-state transition in the sequence space is suggested for the SH3 domain family, and inappropriateness of the mean-field approximation for the LGM is demonstrated.

Bayesian Modeling of the Yeast SH3 Domain Interactome Predicts Spatiotemporal Dynamics of Endocytosis Proteins

Author SummarySignificant diversity exists in protein structure and function, yet certain structural domains are used repeatedly across species to execute similar functions. The SH3 domain is one such common structural domain. It is found in signaling proteins and mediates protein–protein interactions by binding to short peptide sequences generally composed of proline. To investigate both the generality and selectivity of peptide binding by SH3 domains, we examined peptide specificity for almost all SH3 domains encoded within the proteome of the budding yeast, Saccharomyces cerevisiae, using a range of experimental methods. We found that although most of the intrinsic binding specificity for SH3 domains can be summarized by the two previously described canonical binding modes, each individual SH3 domain that we studied utilizes unique features of its cognate ligand to achieve binding selectivity. Moreover, some domains exhibit binding specificities that are distinct from the two canonical classes. We integrated peptide-SH3 domain binding data from three complementary screening techniques using a Bayesian statistical model to generate a protein–protein interaction network for the budding yeast SH3 domain family. This network was highly enriched in endocytosis proteins and their interactions. By examining these interactions in detail, we show that our SH3 domain network can be used to predict the temporal localization of several previously uncharacterized proteins to dynamic complexes that orchestrate the process of endocytosis.

Structure of the SH3 domain of human osteoclast-stimulating factor at atomic resolution

Osteoclast-stimulating factor (OSF) is an intracellular signaling protein, produced by osteoclasts themselves, that enhances osteoclast formation and bone resorption. It is thought to act via an Src-related signaling pathway and contains SH3 and ankyrin-repeat domains which are involved in protein-protein interactions. As part of a structure-based anti-bone-loss drug-design program, the atomic resolution X-ray structure of the recombinant human OSF SH3 domain (hOSF-SH3) has been determined. The domain, residues 12-72, yielded crystals that diffracted to the ultrahigh resolution of 1.07 A. The overall structure shows a characteristic SH3 fold consisting of two perpendicular beta-sheets that form a beta-barrel. Structure-based sequence alignment reveals that the putative proline-rich peptide-binding site of hOSF-SH3 consists of (i) residues that are highly conserved in the SH3-domain family, including residues Tyr21, Phe23, Trp49, Pro62, Asn64 and Tyr65, and (ii) residues that are less conserved and/or even specific to hOSF, including Thr22, Arg26, Thr27, Glu30, Asp46, Thr47, Asn48 and Leu60, which might be key to designing specific inhibitors for hOSF to fight osteoporosis and related bone-loss diseases. There are a total of 13 well defined water molecules forming hydrogen bonds with the above residues in and around the peptide-binding pocket. Some of those water molecules might be important for drug-design approaches. The hOSF-SH3 structure at atomic resolution provides an accurate framework for structure-based design of its inhibitors.

A novel structure-based encoding for machine-learning applied to the inference of SH3 domain specificity

Unravelling the rules underlying protein-protein and protein-ligand interactions is a crucial step in understanding cell machinery. Peptide recognition modules (PRMs) are globular protein domains which focus their binding targets on short protein sequences and play a key role in the frame of protein-protein interactions. High-throughput techniques permit the whole proteome scanning of each domain, but they are characterized by a high incidence of false positives. In this context, there is a pressing need for the development of in silico experiments to validate experimental results and of computational tools for the inference of domain-peptide interactions. We focused on the SH3 domain family and developed a machine-learning approach for inferring interaction specificity. SH3 domains are well-studied PRMs which typically bind proline-rich short sequences characterized by the PxxP consensus. The binding information is known to be held in the conformation of the domain surface and in the short sequence of the peptide. Our method relies on interaction data from high-throughput techniques and benefits from the integration of sequence and structure data of the interacting partners. Here, we propose a novel encoding technique aimed at representing binding information on the basis of the domain-peptide contact residues in complexes of known structure. Remarkably, the new encoding requires few variables to represent an interaction, thus avoiding the 'curse of dimension'. Our results display an accuracy >90% in detecting new binders of known SH3 domains, thus outperforming neural models on standard binary encodings, profile methods and recent statistical predictors. The method, moreover, shows a generalization capability, inferring specificity of unknown SH3 domains displaying some degree of similarity with the known data.

A neural strategy for the inference of SH3 domain-peptide interaction specificity

BackgroundThe SH3 domain family is one of the most representative and widely studied cases of so-called Peptide Recognition Modules (PRM). The polyproline II motif PxxP that generally characterizes its ligands does not reflect the complex interaction spectrum of the over 1500 different SH3 domains, and the requirement of a more refined knowledge of their specificity implies the setting up of appropriate experimental and theoretical strategies. Due to the limitations of the current technology for peptide synthesis, several experimental high-throughput approaches have been devised to elucidate protein-protein interaction mechanisms. Such approaches can rely on and take advantage of computational techniques, such as regular expressions or position specific scoring matrices (PSSMs) to pre-process entire proteomes in the search for putative SH3 targets.In this regard, a reliable inference methodology to be used for reducing the sequence space of putative binding peptides represents a valuable support for molecular and cellular biologists.ResultsUsing as benchmark the peptide sequences obtained from in vitro binding experiments, we set up a neural network model that performs better than PSSM in the detection of SH3 domain interactors. In particular our model is more precise in its predictions, even if its performance can vary among different SH3 domains and is strongly dependent on the number of binding peptides in the benchmark.ConclusionWe show that a neural network can be more effective than standard methods in SH3 domain specificity detection. Neural classifiers identify general SH3 domain binders and domain-specific interactors from a PxxP peptide population, provided that there are a sufficient proportion of true positives in the training sets. This capability can also improve peptide selection for library definition in array experiments. Further advances can be achieved, including properly encoded domain sequences and structural information as input for a global neural network.

Directed discovery of bivalent peptide ligands to an SH3 domain.

The Caenorhabditis elegans SEM-5 SH3 domains recognize proline-rich peptide segments with modest affinity. We developed a bivalent peptide ligand that contains a naturally occurring proline-rich binding sequence, tethered by a glycine linker to a disulfide-closed loop segment containing six variable residues. The glycine linker allows the loop segment to explore regions of greatest diversity in sequence and structure of the SH3 domain: the RT and n-Src loops. The bivalent ligand was optimized using phage display, leading to a peptide (PP-G(4)-L) with 1000-fold increased affinity for the SEM-5 C-terminal SH3 domain over that of a natural ligand. NMR analysis of the complex confirms that the peptide loop segment is targeted to the RT and n-Src loops and parts of the beta-sheet scaffold of this SH3 domain. This binding region is comparable to that targeted by a natural non-PXXP peptide to the p67(phox) SH3 domain, a region not known to be targeted in the Grb2 SH3 domain family. PP-G(4)-L may aid in the discovery of additional binding partners of Grb2 family SH3 domains.

Unusual Binding Properties of the SH3 Domain of the Yeast Actin-binding Protein Abp1: STRUCTURAL AND FUNCTIONAL ANALYSIS

Abp1p is an actin-binding protein that plays a central role in the organization of Saccharomyces cerevisiae actin cytoskeleton. By a combination of two-hybrid and phage-display approaches, we have identified six new ligands of the Abp1-SH3 domain. None of these SH3-mediated novel interactions was detected in recent all genome high throughput protein interaction projects. Here we show that the SH3-mediated association of Abp1p with the Ser/Thr kinases Prk1p and Ark1p is essential for their localization to actin cortical patches. The Abp1-SH3 domain has a rather unusual binding specificity, because its target peptides contain the tetrapentapeptide +XXXPXXPX+PXXL with positive charges flanking the polyproline core on both sides. Here we present the structure of the Abp1-SH3 domain solved at 1.3-A resolution. The peptide-binding pockets in the SH3 domain are flanked by two acidic residues that are uncommon at those positions in the SH3 domain family. We have shown by site-directed mutagenesis that one of these negatively charged side chains may be the key determinant for the preference for non-classical ligands.

SH3 Search Committee

Src-homology 3 domains (SH3 domains) are so named for their similarity to a region in the proto-oncogene protein kinase Src. SH3 domain family members bind to protein domains characterized by paired proline residues, and so mediate protein-protein interactions in cell signaling and cytoskeleton reorganization. Tong et al. (see the Perspective by Gerstein et al. ) used phage display to search peptide libraries of random sequences to identify preferred binding motifs for the various SH3 domains. They then searched the predicted proteome of Saccharomyces cerevisiae for potential binding partners, and also used a two-hybrid assay to identify proteins that interacted with the various SH3 domains when expressed in yeast. The common interactions from the networks predicted by each of these procedures were then assessed for likely biological significance. A. H. Y. Tong, B. Drees, G. Nardelli, G. D. Bader, B. Brannetti, L. Castagnoli, M. Evangelista, S. Ferracuti, B. Nelson, S. Paoluzi, M. Quondam, A. Zucconi, C. W. V. Hogue, S. Fields, C. Boone, G. Cesareni, A combined experimental and computational strategy to define protein interaction networks for peptide recognition modules. Science 295 , 321-324 (2002). [Abstract] [Full Text] M. Gerstein, N. Lan, R. Jansen, Integrating interactomes. Science 295 , 284-287 (2002). [Abstract] [Full Text]

Can we infer peptide recognition specificity mediated by SH3 domains?

Protein interaction domain families that modulate the formation of macromolecular complexes recognize specific sequence or structural motifs. For instance SH3 and WW domains bind to polyproline peptides while SH2 and FHA domains bind to peptides phosphorylated in Tyr and Thr respectively. Within each family, variations in the chemical characteristics of the domain binding pocket modulate a finer peptide recognition specificity and, as a consequence, determine the selection of functional protein partners in vivo. In the proteomic era there is the need for reliable inference methods to help restricting the sequence space of the putative targets to be confirmed experimentally by more laborious experimental approaches. Here we will review the published data about the peptide recognition specificity of the SH3 domain family and we will propose a classification of SH3 domains into eight classes. Finally, we will discuss whether the available information is sufficient to infer the recognition specificity of any uncharacterized SH3 domain.

Native protein sequences are close to optimal for their structures.

How large is the volume of sequence space that is compatible with a given protein structure? Starting from random sequences, low free energy sequences were generated for 108 protein backbone structures by using a Monte Carlo optimization procedure and a free energy function based primarily on Lennard-Jones packing interactions and the Lazaridis-Karplus implicit solvation model. Remarkably, in the designed sequences 51% of the core residues and 27% of all residues were identical to the amino acids in the corresponding positions in the native sequences. The lowest free energy sequences obtained for ensembles of native-like backbone structures were also similar to the native sequence. Furthermore, both the individual residue frequencies and the covariances between pairs of positions observed in the very large SH3 domain family were recapitulated in core sequences designed for SH3 domain structures. Taken together, these results suggest that the volume of sequence space optimal for a protein structure is surprisingly restricted to a region around the native sequence.

SH3-SPOT: an algorithm to predict preferred ligands to different members of the SH3 gene family

We have developed a procedure to predict the peptide binding specificity of an SH3 domain from its sequence. The procedure utilizes information extracted from position-specific contacts derived from six SH3/peptide or SH3/protein complexes of known structure. The framework of SH3/peptide contacts defined on the structure of the complexes is used to build a residue-residue interaction database derived from ligands obtained by panning peptide libraries displayed on filamentous phage. The SH3-specific interaction database is a multidimensional array containing frequencies of position-specific contacts. As input, SH3-SPOT requires the sequence of an SH3 domain and of a query decapeptide ligand. The array, that we call the SH3-specific matrix, is then used to evaluate the probability that the peptide would bind the given SH3 domain. This procedure is fast enough to be applied to the entire protein sequence database. Panning experiments were performed to search putative specific ligands of different SH3 domains in a database of decapeptides, or in a database of protein sequences. The procedure ranked some of the natural partners of interaction of a number of SH3 domains among the best ligands of the ∼5.6 × 10 9 different decapeptides in the SWISSPROT database. We expect the predictive power of the method to increase with the enrichment of the SH3-specific matrix by interaction data derived from new complex structures or from the characterization of new ligands. The procedure was developed using the SH3 domain family as test case but its application can easily be extended to other families of protein domains (such as, SH2, MHC, EH, PDZ, etc.).

Mutagenesis of a buried polar interaction in an SH3 domain: sequence conservation provides the best prediction of stability effects.

The SH3 domain from the Fyn tyrosine kinase possesses a buried hydrogen bond between the side chains of a glutamate (Glu24) and a serine (Ser41) residue. Multiple amino acid substitutions were made at these positions to determine the role of this interaction in the stability and conformational specificity of the domain and to assess the relationship between the thermodynamic stability of mutants and sequence conservation seen in the SH3 domain family. Analysis of single and double alanine mutations indicated that the Glu24-Ser41 interaction contributes 0.50 kcal/mol to the stability of the domain. However, disruption of the Glu24-Ser41 interaction did not impair peptide binding function, suggesting that the interaction is not critical for conformational specificity. The stability of the domain was not increased by the replacement of these residues with different combinations of hydrophobic residues or with potential salt bridge forming residues. Despite their similar structural roles in the Fyn SH3 domain, the Ser41 position was considerably more tolerant to substitution than was the Glu24 position. An alignment of >350 different SH3 domains has been completed in our laboratory. A statistically significant correlation was found between the conservation data for the Glu24 and Ser41 positions and the thermodynamic stabilities of the mutants constructed at these positions. Surprisingly, our analysis of sequence alignment data provided a more accurate prediction of the stability of mutants than did examination of the three-dimensional structure of the domain.