Transcriptional regulation group
IntroductionThe theoretical study of transcriptional regulation has entered a new era with the availability of many fully sequenced genomes in conjunction with a number of new, high-throughput experimental techniques for the study of protein-DNA binding. The gene regulation group focuses on the delineation of regulatory motifs and interactions based on the integration and analysis of this variety of information sources.
Scientific overviewFrom CORG to TRAP: Transcription factor binding site prediction
(Thomas Manke, Helge Roider, Aditi Kanhere, Utz Pape)
A transcription factor tends to bind to particular DNA patterns which can be summarized by so-called Positional Weight Matrices (PWMs). After having explored the power and problems of matching PWMs to sequence in the CORG-database, we have developed an alternative biophysics-inspired approach. This "TRAP" method (for Transcription Factor Affinity Prediction) transforms the match between a sequence and a pattern into a binding probability and integrates over the region of interest, say a promoter region. The TRAP method has been validated by comparison to large scale DNA binding experiments (ChIP-chip and ChIP seq experiments) and shown to be successful in predicting novel target genes of transcription factors. Based on a statistical normalization of the affinity scores, the most likely binding factors to a particular promoter can be inferred. Together with the group of Stefan Haas, TRAP has further been utilized to recognize transcription factor binding sites which are over-represented in co-expressed or tissue-specific groups of genes. Ongoing work applies TRAP for predicting possible effects of regulatory SNPs. It has also been applied in the context of deriving gene regulatory networks. Estimating statistical significance of possible findings is crucial in the analysis of large data sets. To this end, we have derived probabilistic descriptions of the occurrence of hits to PWMs and of the distribution of TRAP scores. Significant efforts have gone into the development of measures of similarity among transcription factor binding sites, which in turn has proven instrumental in computing the probability of observing combinations of binding sites in a regulatory region, thus providing an instrument to study combinatorial regulation.
Epigenetic regulation(Ho-Ryun Chung, Rosa Karlic, Julia Lasserre, Irit Gat-Viks)
Experimental progress over recent years has driven home the point, that in eukaryotes, and in particular in mammals, transcription factors are not the only regulators of gene expression. Chromatin structure, histone modifications, and DNA methylation also play vital roles or are at least correlated to expression status. Our own interest in this epigenetic level of regulation focuses on the question, in how far the DNA sequence can provide us with information not only about transcription factor binding sites, but also about, e.g., chromatin structure. While the sequence dependence of nucleosome positioning is still under debate, one does see a strong division of promoters into those with high vs. low contents of CpG dinucleotides. We have recently published that this distinction governs the localization and type of transcription factor binding sites, and are currently working on establishing that histone modification patterns also depend on these sequence features.
Evolution of regulation(Tomasz Zemojtel, Szymon Kielbasa, Sarah Behrens, Andrew Hufton, Morgane Thomas-Chollier)
While protein evolution is nowadays generally described by a Markov process on the sequence positions with selection acting on the level of protein function, the evolution of regulatory DNA sequences is still badly understood. In collaboration with the Evolutionary Genomics Group of Peter Arndt, we have been studying the influence of the Cyotsine deamination on the appearance of binding sites. A mutation of a C in the context of a CpG dinucleotide due to deamination is much more likely than other mutations. A careful inspection of Alu repeats has shown that these transposable elements carry possible predecessors of binding sites, which are moved around the genome in the course of evolution, bearing the potential to become functional binding sites upon deamination. In this context we are now systematically investigating the role of transposable elements and numerous types of transcription factor binding sites as a possible source of novel binding sites in evolution. In contrast to this mechanism, studies in the time it would take for a binding site to evolve purely by point mutations indicate that this is not flexible enough a process to explain regulatory evolution. This theoretical work is further complemented by studies in ancient conserved elements in collaboration with the Poustka/Panoupoulou group (Dept. Lehrach) and studies in the evolution of Hox genes.
Gene networks(Ewa Szczurek, Thomas Manke, Anirban Banerjee, Lloyd Demetrius, Roman Brinzanik, Utz Pape)
Several projects, many of them in collaboration with experimentalist, try to delineate regulatory networks or study the general features of biological networks. In collaboration with C. Sers (Charité) we are studying the regulatory cascade downstream of the ras-triggered MAP kinase pathway. Gene regulatory networks in heart development were the topic of a collaboration S. Sperling (Dept. Lehrach). In a collaborative project with other Max Planck Institutes we are working on the delineation of regulatory networks integrating data from metabolomics and transcriptomics. The analysis of gene networks has also led us to propose an algorithmic framework to predict most informative experiments to elucidate regulatory dependencies. Following their earlier work on the evolution of complex networks, Lloyd Demetrius, Thomas Manke and Anirban Banerjee have continued to develop a graph-theoretical framework for the characterisation of biological networks. Applying an entropic formalism to large-scale protein interaction data, they investigated the relationships between the essentiality of a protein and its overall position in the molecular networks of yeast and nematode worm. In numerical studies on model networks they found that network entropy correlates positively with many heuristic measures of structural and dynamical robustness of networks, such as the percolation threshold and mixing rates. This approach has been complemented by work on the graph spectrum as an alternative characterisation of biological networks. The normalized graph Laplacian spectrum does not only provide insights into the modular organisation of networks, but also helps to measure the distance between networks with different sizes (cooperation with J. Jost, MPI-MIS Leipzig). Lloyd Demetrius has continued his work on ageing models, proposing that differences among organisms in the rate of ageing and life span are due to differences in metabolic stability, rather than differences in metabolic rate. J. Adjaye (Dept. Lehrach) provided experimental evidence in support of this theory.
CollaborationsIn addition to MPI-internal collaborations like the ones mentioned above, group members have collaborations within Berlin or Germany, as well as internationally. Frequently these collaborations will be in the context of a DFG-, BMBF- or EU-grant. While within Berlin we are closely cooperating with the FU bioinformatics group (K. Reinert) and with experimentalists at Charité (C. Sers), work together with A. Nordheim from Tübingen on SRF regulated genes has been very successful, too. On an international level, fruitful cooperation with B. Lenhard (Bergen, Norway), J. Tiuryn (Warsaw), E. Birney (Hinxton), and Fengzhu Sun (Los Angeles) need to be mentioned.
ContactProf. Dr. Martin Vingron
Gene Regulation Group
Department Computational Molecular Biology
Max Planck Institute for Molecular Genetics
14195 Berlin, Germany
Last Change: 05 Jan 2010