The vast and growing volume of 3D protein structural data stored in the PDB contains abundant information about macromolecular complexes, and hence, data about protein interfaces. Non-covalent contacts between amino acids are the basis of protein interactions, and they are responsible for binding afinity and specificity in biological processes. In addition, water networks in protein interfaces can also complement direct interactions contributing significantly to molecular recognition, although their exact role is still not well understood.
It is estimated that protein complexes in the PDB are substantially underrepresented due to their crystallization dificulties. Methods for automatic classifification and description of the protein complexes are essential to study protein interfaces, and to propose putative binding regions. Due to this strong need, several protein-protein interaction databases have been developed. However, most of them do not take into account either protein-peptide complexes, solvent information or a proper classification of the binding regions, which are fundamental components to provide an accurate description of protein interfaces.
In the firest stage of my thesis, I developed the SCOWLP platform, a database and web application that structurally classifies protein binding regions at family level and defines accurately protein interfaces at atomic detail. The analysis of the results showed that protein-peptide complexes are substantially represented in the PDB, and are the only source of interacting information for several families. By clustering the family binding regions, I could identify 9,334 binding regions and 79,803 protein interfaces in the PDB. Interestingly, I observed that 65% of protein families interact to other molecules through more than one region and in 22% of the cases the same region recognizes different protein families. The database and web application are open to the research community (www.scowlp.org) and can tremendously facilitate high-throughput comparative analysis of protein binding regions, as well as, individual analysis of protein interfaces.
SCOWLP and the other databases collect and classify the protein binding regions at family level, where sequence and structure homology exist. Interestingly, it has been observed that many protein families also present structural resemblances within each other, mostly across folds. Likewise, structurally similar interacting motifs (binding regions) have been identified among proteins with different folds and functions. For these reasons, I decided to explore the possibility to infer protein binding regions independently of their fold classification. Thus, I performed the firest systematic analysis of binding region conservation within all protein families that are structurally similar, calculated using non-sequential structural alignment methods. My results indicate there is a substantial molecular recognition information that could be potentially inferred among proteins beyond family level. I obtained a 6 to 8 fold enrichment of binding regions, and identified putative binding regions for 728 protein families that lack binding information. Within the results, I found out protein complexes from different folds that present similar interfaces, confirming the predictive usage of the methodology. The data obtained with my approach may complement the SCOWLP family binding regions suggesting alternative binding regions, and can be used to assist protein-protein docking experiments and facilitate rational ligand design.
In the last part of my thesis, I used the interacting information contained in the SCOWLP database to help understand the role that water plays in protein interactions in terms of affinity and specificity. I carried out one of the firest high-throughput analysis of solvent in protein interfaces for a curated dataset of transient and obligate protein complexes. Surprisingly, the results highlight the abundance of water-bridged residues in protein interfaces (40.1% of the interfacial residues) that reinforces the importance of including solvent in protein interaction studies (14.5% extra residues interacting only water- mediated). Interestingly, I also observed that obligate and transient interfaces present a comparable amount of solvent, which contrasts the old thoughts saying that obligate protein complexes are expected to exhibit similarities to protein cores having a dry and hydrophobic interfaces. I characterized novel features of water-bridged residues in terms of secondary structure, temperature factors, residue composition, and pairing preferences that differed from direct residue-residue interactions. The results also showed relevant aspects in the mobility and energetics of water-bridged interfacial residues.
Collectively, my doctoral thesis work can be summarized in the following points:
1. I developed SCOWLP, an improved framework that identiffies protein interfaces and classifies protein binding regions at family level.
2. I developed a novel methodology to predict alternative binding regions among structurally similar protein families independently of the fold they belong to.
3. I performed a high-throughput analysis of water-bridged interactions contained in SCOWLP to study the role of solvent in protein interfaces. These three components of my thesis represent novel methods for exploiting existing structural information to gain insights into protein- protein interactions, key mechanisms to understand biological processes.
| Identifer | oai:union.ndltd.org:DRESDEN/oai:qucosa:de:qucosa:25412 |
| Date | 02 November 2010 |
| Creators | Teyra i Canaleta, Joan |
| Contributors | Pisabarro, Mayte, Schröder, Michael, Hoflack, Bernard, Technische Universität Dresden |
| Publisher | BIOTEC Dresden |
| Source Sets | Hochschulschriftenserver (HSSS) der SLUB Dresden |
| Language | English |
| Detected Language | English |
| Type | doc-type:doctoralThesis, info:eu-repo/semantics/doctoralThesis, doc-type:Text |
| Rights | info:eu-repo/semantics/openAccess |
Page generated in 0.0025 seconds