In the SFB 618-A5 project, supported by the DFG, we developed innovative high throughput technologies and novel bioinformatics approaches for construction and analysis of protein-protein interaction networks. The aim was to provide a basis, both experimentally as well as computationally, for the systematic mapping and examination of the human protein interactome. To be able to decipher the intriguing complexity of the produced networks, we designed and implemented new tools for their analysis. A focus was set on the study of the dynamics and the design principles such as robustness and modularity of interaction networks. Finally, integration and analysis of protein interaction networks provided us with a new framework for the study of the molecular machinery in health and disease. This helped us to predict human disease genes, to define sets of biomarkers and to model biological processes on a systems level.
The project was led concertedly by Prof. Erich Wanker (MDC, Berlin), who is responsible for the performed experiments, and Dr. Matthias Futschik, who is responsible for the conducted computational investigations. The results of the computational investigations will described below.
After completion of a first map of the human interactome based on systematic Y2H-screens, (Stelzl et al, 2005, Cell), we set out to develop a first comprehensive human protein-protein interaction database. By 2006, ten large datasets on human protein-protein interactions had been created, including our network and another Y2H network completed shortly after. The other eight datasets have been derived by manual curation, predictions based on orthology and computational text mining approaches.To use them, scientists were required to carry out time-consuming manual searches.
To facilitate direct, easy access, we have developed and implemented a central web-based database which we named Unified Human Interactome (UNIHI, NAR 2007, 2009). UniHI integrates human interaction data from various sources with additional information. The database offers both experimentalists and bioinformaticians a single entry gate to the human interactome. The design and architecture of UniHI is based on state-of-the-art software tools to successfully accommodate for the rapid growth, fragmentation and complexity of stored data (JIB, 2007).
At the moment, UniHI is one of the most comprehensive databases for human protein-protein interactions and related information world-wide.As the interactions integrated in UniHI were derived by different experimental and computational techniques, we first evaluated the benefits and the quality of such integration. Statistical analysis revealed that the constructed interaction map shows high internal functional coherency and therefore provides a useful basis for biomedical investigation (Genome Informatics, 2006). Furthermore, our evaluation demonstrated that currently available interaction maps are highly complementary an observation that strongly supports our efforts to consolidate the distributed interaction data. Notably, we also showed that interactions derived from small-scale studies are strongly biased towards popular proteins (Bioinformatics, 2007). Furthermore, the structure of the current protein interaction maps are highly divergent (Genome Informatics, 2009). This underscores the necessity of our approach to construct a systematic and thus unbiased map of the human interactome.
To obtain a more detailed view of the human interactome, we proceeded with the examination of local structures (JIB 2007, BMC Systems Biology, 2010) . A focus was put on the identification of modular structures, as many molecular functions require the tight interactions of multiple proteins. It has been realized that modularity is major design principle of biological networks. Notably, our study represented the first effort to obtain a global overview of modular structures in the human interactome. It demonstrated that the human interaction network is highly modular. Inspection showed that these modules do not only include numerous known protein complexes, but also many novel and interesting associations between proteins. Of special interest here are proteins that link distinct modules. Based on these linker proteins, we were able to construct a meta-network of modules giving us a first intriguing image of the complex interplay between different components of the cellular machinery (JIB 2007).
Finally, protein networks can differ from tissue to tissue even if the same proteins are involved. Therefore, we established a compendium of tissue-specific expression pattern (BMC Genomics 2010), which we used to identify new tissue-specific molecular modulators in Huntington's disease, a fatal neurodegenerative disorder without existing cure. Applying a simple but efficient filtering approach for protein interaction network, we could identify neuron-specific collapsin response mediator protein (CRMP1) as a potential molecular modifier of Huntington's disease (Genome Research, 2015). Subsequent experiments confirmed the computational prediction and demonstrated that CRMP1 is a powerful suppressor of protein misfolding and neurotoxicity.
For our interdisciplinary study of the human interactome, we were awarded in 2008 with the prestigious Erwin-Schrödinger-Prize of the Helmholtz Association, one of the biggest scientific associations in Europe.
M. Stroedicke, Y. Bounab, N. Strempel, K. Klockmeier, S. Yigit, RP Friedrich, G. Chaurasia, S. Li, F. Hesse, SP Riechers, J. Russ, C. Nicoletti, A. Boeddrich, T. Wiglenda, C. Haenig, S. Schnoegl, D. Fournier, RK Graham, MR Hayden, S. Sigrist, GP Bates, J. Priller, Ma Andrade-Navarro, Matthias E. Futschik and EE Wanker (2015) Systematic interaction network filtering identifies CRMP1 as a novel suppressor of huntingtin misfolding and neurotoxicity. Genome Research 25: 701-713 (pdf + html)
Jenny Russ and Matthias E. Futschik, Comparison and consolidation of microarray data sets of human tissue expression, BMC Genomics, 11:305, 2010 (html+pdf)
Elisabetta Marras, Antonella Travaglione, Gautam Chaurasia, Matthias Futschik and Enrico Capobianco (2010) Inferring modules from human protein interactome classes, BMC Systems Biology, 4:102 (html+pdf)
Gautam Chaurasia, Soniya Malhotra, Jenny Russ, Sigrid Schnoegl, Christian. Hänig, Erich. E. Wanker and Matthias. E. Futschik. UniHI 4: New tools for query, analysis and visualization of the human protein-protein interactome, Nucleic Acids Research, 37, Database issue: D657-D660, 2009 (html+pdf)
M. Futschik, Anna Tschaut, Gautam Chaurasia, and Hanspeter Herzel; Graph-Theoretical Comparison Reveals Structural Divergence of Human Protein Interaction Networks, Genome Informatics, 18, 141-151, 2007 (pdf)
M. E. Futschik, Gautam Chaurasia, Anna Tschaut, Jenny Russ, M. Madan Babu and Hanspeter Herzel, Functional and Transcriptional Coherency of Modules in the Human Protein Interaction Network, Journal of Integrative Bioinformatics, 4(3):76, 2007 (pdf)
P. Umbach, M. Futschik, U.Stelzl and E.Wanker, Funktion durch Netzwerke von Proteinen und ihren Wechselwirkungen, Laborwelt, 8(2):9-11, 2007 (ps)
G. Chaurasia, Y. Iqbal, C. Hänig, H. Herzel, E.E. Wanker and M.E. Futschik, Flexible web-based integration of distributed large-scale human protein interaction maps, Journal of Integrative Bioinformatics, 4 (1):51, 2007 (Abstract, pdf)
G. Chaurasia, Y. Iqbal, C. Hänig, H. Herzel, E.E. Wanker and M.E. Futschik, UniHI: an entry gate to the human protein interactome, Nucleic Acids Research, Database issue, D590-4, 2007 (html+pdf)
M.E. Futschik, G. Chaurasia, E. Wanker and H. Herzel, Comparison of Human Protein-Protein Interaction Maps, Lecture Notes in Informatics, P-83, 21-32, 2006 (preprint-pdf)
G. Chaurasia, H. Herzel, E.E. Wanker and M.E. Futschik, Systematic functional assessment of human proteins-protein interaction maps, Genome Informatics, 17(1), 36-45, 2006 (pdf)