BIOZENTRUM

Structural Bioinformatics
Schwede Group

People

Projects

Teaching

Publications

Open Positions

Contact


SIB

 

Projects & Collaborations

SWISS-MODEL expert system for protein structure modeling

Comparative protein structure modeling (or homology modeling) predicts the three-dimensional structure of a given protein sequence (target) based on an alignment of this protein to one or more homologues proteins of known structure (templates). It is anticipated that within the next decade, most structural folds of soluble non-membrane proteins will be known, and homology modeling can be applied to most protein sequences. Facing the huge amount of data originating from genome sequencing projects and structural genomics studies, it is crucial to develop stable and reliable expert systems and fully automated methods.

SWISS-MODEL is one of the most widely used web-based modeling servers with more than 1'500 daily modeling requests. The SWISS-MODEL Repository is a database of annotated comparative protein structure models. It contains the results of automated large-scale modeling for sequences from the UniProt database using the SWISS-MODEL server pipeline. The integration of structural models with biochemical knowledge databases, such as explicit cross-linking with SwissProt or InterPro is a crucial step in any computational analysis of sequence-structure-function relationships. The SWISS-MODEL Workspace is a web based graphical user interface. Software tools required for comparative modeling and structure quality evaluation can be invoked from within the web interface. [Schwede et al., NAR 31, 3381-338; Kiefer et al. NAR 37, D387-D392; Arnold et al., Bioinformatics, 22,195-201].

PMP - the PSI SGKB Protein Model Portal

One of the challenges in using model information effectively has been to access all models available for a specific protein in heterogeneous formats at different sites using various incompatible accession code systems. The goal of the Protein Model Portal (PMP) is to provide a single portal which gives access to all available models and experimental protein structures for a given protein. A single interface allows querying simultaneously pre-computed models across various sites, and provides links to interactive services for template selection, target-template alignment, model building, and quality assessment.

The Protein Model Portal is part of the PSI-Nature Structural Genomics Knowledgebase, a comprehensive resource produced in collaboration between the Protein Structure Initiative (PSI) and Nature Publishing Group (NPG), to offer an easy way of keeping abreast of developments both by the PSI and more generally in the fields of structural genomics and structural biology. [Arnold et al., J Struct Funct Genomics, 10, 1–8; Berman et al. NAR 37, D365–368.]

Discovery of Novel Inhibitors of Dengue Virus Methyltransferase by Virtual Screening on a Grid

Diseases which affect mainly the developing countries are of limited commercial potential. As a consequence, these diseases are not the primary focus of the pharmaceutical industry, and alternative efforts in the public domain – with support by the industry which holds the Drug Discovery know how – are needed to drive the discovery of new medicines for these diseases. In this context, we have initiated a project which is aimed at the computational screening of small compounds available in public compound libraries to identify very early drug candidates. The first disease we will address in this approach will be Dengue Fever.
During the past months, we have developed a GRID based architecture that allows us routinely perform in silico docking simulations distributed on the GRID partner resources. The most promising compounds identified by computational screening will be transferred to interested parties with the capability to further analyze their value in a Drug Discovery program. This study is a collaboration with the Novartis Institute for Tropical Diseases (NITD) in Singapore and Schrödinger LLC.
Dengue fever is a viral disease that affects 50−100 million people annually and is one of the most important emerging infectious diseases in many areas of the world. Currently, neither specific drugs nor vaccines are available. In this study, we report on the discovery of new inhibitors of the viral NS5 RNA methyltransferase, a promising flavivirus drug target. We have used a multistage molecular docking approach to screen a library of more than 5 million commercially available compounds against the two binding sites of this enzyme. In 263 compounds chosen for experimental verification, we found 10 inhibitors with IC50 values of <100 μM, of which four exhibited IC50 values of <10 μM in in vitro assays. The initial hit list also contained 25 nonspecific aggregators. [J Med Chem, (in press).]

Mechanisms of cyclic di-GMP signalling

Living cells employ small diffusible molecules, so-called second messengers, to signal environmental cues from sensory proteins to cellular receptors. Only recently, it has become apparent that bacteria utilize the cyclic dinucleotide c-di-GMP as a ubiquitous second messenger to switch between rapidly growing single cells and a quiescent life style, called biofilm. In pathogenic bacteria, this switch is often accompanied by the transition from an acute to a chronic phase of infection. This makes c-di-GMP signal transduction an attractive target for novel antibiotics that interfere with bacterial persistence. The cellular concentration of c-di-GMP is the result of the opposing activities of diguanylate cyclases that synthesize c-di-GMP from two GTP molecules, and phosphodiesterases that degrade the compound. These two key enzymatic activities regulate c-di-GMP and thus the state of the various c-di-GMP receptors and their associated activities within the cell.

To uncover the molecular mechanisms of the c-di-GMP signaling network, this Sinergia project aims at combining in vivo studies with pathogenic and non-pathogenic bacterial model systems with the analysis of the isolated and purified proteins of the network. This involves their enzymatic and biophysical characterization, 3D-structure determination by X-ray crystallography and the study of their dynamic properties by fluorescence energy transfer measurements. Furthermore, the vast amount of bioinformatic data available will be exploited for structure prediction and the identification of yet unrecognized members of the network. The results will further our general knowledge about c-di-GMP mediated cell signaling and behavior and contribute important information towards the successful control of chronic infections by animal and human pathogens. This collaboration with Urs Jenal, Tilman Schirmer and Dagmar Klostermeier is funded by the SNSF Swiss National Science Foundation. [Genes Dev, 23(1), 93–104.].

Structural and functional basis of odorant receptor mediated signaling

Olfaction, the detection of odorous compounds, is among the oldest of the sensory systems. It is ideally suited to study the complex mechanism of transducing chemical into neuronal signals generating behavioral responses, which are essential for the survival of most mammals. The interaction of volatile molecules with distinct G protein coupled olfactory receptors (OR), about 350 in human, is the central molecular event of detecting and discriminating thousands of odorants. We have entered collaboration with the Vogel group (EPFL) in an integrated experimental and bioinformatics approach to elucidate the molecular basis of olfaction and its decoding into cellular responses: Functional screening of chemical libraries is used to develop computational models of the molecular specificity of distinct ORs which are not yet characterized for their ligand specificity. Iterative cycles of site-directed mutagenesis of receptor ligand binding regions combined with remodeling will refine the accuracy of the predicted receptor structures and specificities. First experimental and modeling results pointing to dual functions of some odorant compounds activating both ORs and endocrine receptors. [J Biol Chem, 284(44), 30547–30555.]

OpenStructure

We have developed OpenStructure as an open-source, modular, flexible, molecular modelling and visualization environment. It is targeted at interested users and method developers in the field of structural bioinformatics. OpenStructure forms the framework for several new developments in our group, e.g. QMEAN model quality assessment, or BEscore for scoring protein-ligand interactions. Future versions of SWISS-MODEL will be based on this framework. The first beta release of OpenStructure under the lGPL license is available at http://www.openstructure.org.

DeepView - Swiss-PdbViewer

DeepView - aka Swiss-PdbViewer - is a program for protein structure visualization, manipulation and modeling developed by Nicolas Guex. It's easy to use and available for Mac and MS Windows. We are collaborating with with Nicolas Guex (SIB Vital-IT) in the development of the DeepView program.

Previous projects

Variations, mutations, SNPs: Structural basis of human disease

Effects of genetic variations in the human genome range from neutral mutations over increased susceptibly for complex diseases, individual variation in drug response, to rare single-allele mendelian inherited diseases. Three dimensional protein structure models are valuable to provide insights in the molecular basis of these mutations. While visual inspection of the structural context of mutations provides valuable insight in single cases, the development of general objective scoring functions is still an unsolved problem. Our preliminary studies on mutations of human phenylalanine hydroxylase causing phenylketonuria showed that computational approaches based on protein structure models could significantly contribute to an objective quantification of mutational effects. This project is funded by the Swiss National Science Foundation (SNF).

CASP7: 7th Community Wide Experiment on the Critical Assessment of Techniques for Protein Structure Prediction

We have assessed the accuracy of models submitted in the template based modeling and disorder prediction categories of the seventh Critical Assessment of Techniques for Protein Structure Prediction (CASP7). In the TBM category, the accuracy of predicted protein models for 108 target domains was evaluated based on detailed numerical comparison between the experimental and predicted structures. The assessment was performed using numerical measures for backbone and structural alignment accuracy, and by scoring correctly modeled hydrogen bond interactions in the predictions. Based on these criteria, our statistical analysis identified a number of groups whose predictions were on average significantly more accurate. All predictions and numerical evaluation results are available on the CASP7 homepage, the evaluation results are published in a special issue of Proteins.

Sulfonylureas and Glinides Exhibit Peroxisome Proliferator-Activated Receptor gamma Activity: A Combined Virtual Screening and Biological Assay Approach

Most drugs currently employed in the treatment of type 2 diabetes either target the sulfonylurea receptor stimulating insulin release (sulfonylureas, glinides), or target the peroxisome proliferator-activated receptor (PPAR{gamma}) improving insulin resistance (thiazolidinediones). Our work shows that sulfonylureas and glinides additionally bind to PPAR{gamma} and exhibit PPAR{gamma} agonistic activity. This activity was predicted in silico by virtual screening and confirmed in vitro in a binding assay, a transactivation assay, and by measuring the expression of PPAR{gamma} target genes. Among the measured compounds, gliquidone and glipizide (two sulfonylureas), as well as nateglinide (a glinide), exhibit PPAR{gamma} agonistic activity at concentrations comparable with those reached under pharmacological treatment. The most active of these compounds, gliquidone, is shown to be as potent as pioglitazone at inducing PPAR{gamma} target gene expression. This dual mode of action of sulfonylureas and glinides may open new perspectives for the molecular pharmacology of antidiabetic drugs, because it provides evidence that drugs can be designed that target both the sulfonylurea receptor and PPAR{gamma}. Targeting both receptors could increase pancreatic insulin secretion and improve insulin resistance. Glinides, sulfonylureas, and other acidified sulfonamides may be promising leads in the development of new PPAR{gamma} agonists. In addition, we provide a unified concept of the PPAR{gamma} binding ability of seemingly disparate compound classes. [Scarsi et al. Molecular Pharmacology 71, 398-406.]

The Plug Domain of Yeast Sec61p Is Important for Efficient Protein Translocation, but Is Not Essential for Cell Viability

The Sec61/SecY translocon mediates translocation of proteins across the membrane and integration of membrane proteins into the lipid bilayer. The structure of the translocon revealed a plug domain blocking the pore on the lumenal side. It was proposed to be important for gating the protein conducting channel and for maintaining the permeability barrier in its unoccupied state. In collaboration with group of Martin Spiess (Biozentrum), we built a model for the yeast Sec61 complex which served as basis for designing destabilizing point mutations in the plug domain or of its partial or complete deletion. Unexpectedly, even when the entire plug domain was deleted, cells were viable without growth phenotype. They showed an effect on signal sequence orientation of diagnostic signal-anchor proteins, a minor defect in cotranslational and a significant deficiency in posttranslational translocation. [Junne et al. Molecular Biology of the Cell , 17, 4063-4068.]

The SwissBioGrid

the SwissBioGrid is an initiative of national scope that will support large-scale computational applications in bioinformatics, biosimulation, chemoinformatics and bio-medical sciences by utilizing distributed high-performance computing, high speed networks, massive data collections and archives, as well as the necessary software tools and data integration capabilities. The SwissBioGrid initiative is coordinated by the Swiss National Supercomputing Center (CSCS), with the member sites: Biozentrum Basel, Friedrich Miescher Institut Basel (FMI), Novartis Basel, the Functional Genomics Center Zürich (FGCZ), and SIB / VitalIT (Lausanne). [Podvinec et al IEEE e-Science 2006, 148.]

InterPro3D

InterPro is an integrated documentation resource for protein families, domains and sites, developed initially as a means of rationalizing the complementary efforts of the PROSITE, PRINTS, Pfam and ProDom database projects, and has now been extended to include SMART and TIGRFAMs. Each combined InterPro entry includes functional descriptions and literature references, and links are made back to the relevant member database(s). Our goal is to include 3-dimensional structures models to allow a spatial interpretation of the observed sequence motifs. Some aspects of this project were funded by the European Commission as the TEMBLOR, contract-no. QLRI-CT-2001-00015 under the RTD program "Quality of Life and Management of Living Resources".

Homology Modeling and Structure-based Drug Development

Structure-based and structure-guided drug design methods have made significant impact on the development of drugs in recent years. Although several compounds discovered with the help of structural information have successfully passed clinical trials and have become approved drugs, one of the bottlenecks of structure-based methods is the availability of experimental structures of the target proteins. In these cases, protein structure homology models can provide a valuable alternative. One of the open questions in this context is how errors and inaccuracies of the homology models influence the molecular modeling of the protein-ligand interaction. In this project we collaborate with with the group of Markus Meuwly (Theoretical Chemistry, University Basel) and Vincent Zoete (SIB Lausanne) to explore to which extend homology models can be used to study protein-ligand interactions. By systematically introducing typical errors in the protein structure before simulation, we can explore the influence of inaccuracies of homology models on molecular modeling. [Thorsteinsdottir et al. Proteins 65 , 407-423.]

 

Last modified : 06-Sep-2011 by TS .