The highly complementary investigators' expertise, the large variety of methodologies and tools developed by the Center to dissect and map cellular pathways, and their integration into a modular and interoperable software platform provides a powerful framework to tackle a variety of relevant biomedical problems. We have identified a number of Driving Biological Projects (DBPs) that represent a wide range of biomedical problems that can benefit from the MAGNet activities at different levels of granularity. Our selection process was based on a number of relevant factors:
Biomedical relevance and scientific merit
: DBPs should address biomedically relevant problems that will uniquely benefit from the ability to dissect specific intra- or inter-cellular interactions.Ability to drive computational biology and bioinformatics innovation
: DBPs should drive the creation of new computational algorithms and data analysis methodologies, with specific emphasis on the integration of distinct data modalities.Addressing all levels of granularity of the Center's Network Centric vision
: different DBPs should address distinct ranges of granularity of the underlying biological processes, from the study of specific structure based interactions, to the elucidation of cellular pathways on model organisms, to the identification of modular control of cellular functions and processes, all the way to the dissection of complex disease based on genomic pathway information.Feasibility and pertinence to the MAGNet Center scientific mission
: DBPs should uniquely benefit from the MAGNet tools and investigator expertise.
Based on these criteria, the Center selected seven DBPs (briefly discussed below) as the focus of our research activity. The first 4 of these DBPs were part of the original Center application and were completed on July 2008. Work on the last 3 commenced on August 2008.
STRUCTURAL AND ENERGETIC BASIS OF CADHERIN BINDING SPECIFICITY
The goal of this project is to understand the structural and energetic basis of cell-cell adhesion mediated by the cadherin family of cell adhesion proteins. Adhesion between cadherin family members, primarily through selective homophilic interaction, provides a key driving force in the development of tissue architecture. It is the subtle differences among family members that are responsible for this crucial selectivity, which guides tissue development. Neither sequence analysis nor observations on known crystal structures have been able to reveal the structural and energetic origins of cadherin specificity. It has become clear that computational studies are required both to generate testable hypotheses and to interpret experimental results once they are obtained. We propose a joint program involving the energetic analysis of protein-protein interfaces, structure prediction, x-ray crystallography, biophysical measurements of binding and cell-sorting experiments. Possible specificity determinants will be identified, mutant proteins will be designed and evaluated for their binding propensities, and new crystal structures will be determined. The experiments will provide data to test quantitative theoretical predictions as well as qualitative concepts. In parallel, the theory will play a crucial role in driving and interpreting experiments. These questions and the nature of the experimental and theoretical challenges are quite general in nature. Indeed understanding how specificity is coded within protein families is a central molecular question that relates to the understanding of biological networks, the common theme of the MAGNet Center we propose to establish.
REGULATORY MODULES IN NORMAL AND TRANSFORMED B-CELLS
This project addresses a broad biological systems problem by attempting to identify conserved functional modules within Human B-Cell gene regulatory networks especially in relation to the Germinal Centers, a key structure for antibody mediated immune response and a target of transformation in B-Cell lymphomas. This will be accomplished by integrating several MAGNet developed bioinformatics methodologies. We plan to integrate (a) clues from information theoretic reverse-engineering methods applied to large microarray expression profile set (~400 microarrays) of normal, tumor-related, and experimentally manipulated B cells, (b) clues from advanced natural language-based data-mining algorithms applied to a large literature collection, and (c) clues from protein-protein and protein-DNA interaction databases. Modules that are disregulated in a variety of Germinal Center-related tumors will then be identified using expression analysis algorithms supported in geWorkbench. Finally, candidate module control genes will be identified using a combination of in-silico and invitro techniques.
GENOMIC AND BIOINFORMATICS SOLUTIONS TO THE SEARCH FOR GENETIC DETERMINANTS OF COMMON, HERITABLE DISORDERS
We will develop new disease gene pathway based bioinformatics approaches developed in the Center. The specific goal will be the identification of genes, and gene pathways, that harbor heritable determinants of two common, debilitating disorders of the nervous system, Alzheimer's disease (AD) and Autism plus related spectrum disorders (ASD). The approach will include the use of convergent bioinformatics, computational, genetic and genomic data to predict genes and gene combinations likely to harbor disease-related genetic variation, together with statistical and computational methods to detect and evaluate such datasets. For both studies, we will deploy a battery of bioinformatics and computational approaches that will be incorporated into the geWorkbench platform, to detect and evaluate disease-related genetic variation.
UNDERSTANDING AND PREDICTING TRANSCRIPTION FACTOR SPECIFICITIES
The long-term goal of this project is to understand how in vivo specificities are achieved by eukaryotic transcription factors. The project has two main components. The first uses evolutionary conservation of genomic DNA sequences to identify putative cis-regulatory modules (CRMs) and the transcription factors that bind these modules. The overall progression for this arm of the project is to develop novel computation prediction methods followed by their in vivo validation. This 'DNA-centric' approach is complemented by a 'protein-centric' approach which constitutes the second component of this project. Here, we use a combination of biochemical, modeling, structural, and in vivo methods to understand the DNA-binding specificities of the Hox family of homeodomain proteins. As with all homeodomain proteins, the Hox protein family exhibits a high degree of functional specificity in vivo but low specificity in in vitro DNA binding assays. Our recent results provide new insights into this paradox, which we plan to further expand upon in the future. For both arms of this project, our model system is the fruit fly, Drosophila melanogaster, with which we have ample experience using both in vitro and in vivo methods to study transcription factor specificity and function.
MICRORNA ANALYSIS IN NORMAL AND NEOPLASTIC HUMAN B CELL PHENOTYPES
microRNAs (miRNA) have been shown to play a key role in oncogenesis by post-transcriptional regulation of genes involved in regulatory and signaling pathways.The discovery of miRNA is still very much an on-going process, especially within specific cellular contexts. For example, only about a third of the miRNAs expressed in various stages of mature B cell differentiation was previously known. A rational strategy for the investigation of pathways that are dysregulated in B cell malignancies could be greatly impacted by the availability of a cell-context specific, genome-wide map of (a) their upstream regulators, (b) their downstream targets, and (c) their interplay with key oncogenes and tumor suppressors. We propose to understand the role of miRNA in normal B cell physiology and lymphomagenesis by enriching the existing B Cell Interactome with miRNA related interactions and by using it to identify key regulators of normal and tumor related processes.
COMPUTATIONAL AND FUNCTIONAL DISSECTION OF DRUG TARGETS IN MELANOMA
We aim to combine computational modeling and functional genomics to advance cancer therapy, using melanoma as our experimental model. We will use our Bayesian Network framework to integrate heterogeneous data including genotype, gene expression under "perturbations" and drug resistance to: (1) Detect which genetic alterations in tumors drive proliferation and drug resistance; (2) Model how these alterations perturb normal cell growth/survival; (3) Understand the adaptive/feedback mechanisms that attenuate drug efficacy, even with target inhibition; and (4) Identify additional target pathways for combinatorial drug treatment. The majority of melanomas are characterized by point mutations that activate the BRAF and NRAS oncogenes, suggesting that MAP kinase pathway inhibition may offer an appealing targeted therapeutic avenue in this malignancy. Despite robust preclinical evidence that the BRAF (V600E) mutation enacts an exquisite sensitivity to MAP kinase pathway inhibitors the available clinical data suggests that additional mutations feedback and other adaptations of the signaling network to pathway inhibition can attenuate these therapeutic effects. Thus, effective treatment may require a combination of novel therapeutics together with RAF or MED inhibition.
AN INTEGRATED ANALYSIS OF STRUCTURAL ORGANIZATION, DESIGN PRINCIPLES, AND EVOLUTION ACROSS MULTIPLE GENOMES OF A MODEL DEVELOPMENTAL NETWORK
The bacterial sporulation network is a unique model for understanding systems biology of developmental organization, transcriptional regulation, and evolution. The system is one of the largest known developmental networks in any organism; it includes more than 500 genes and about 20 known regulators. The ability to rapidly generate high-throughput experimental data, a complete collection of single-gene B. subtilis mutants, and the recent genome sequencing of more than 20 sporulating bacteria make this network a perfect model system. Understanding the functional organization of the sporulation network is also of the major public health and bio-defense importance due to its central role in the virulence of B. anthracis (close relative of B. subtilis). We propose to use a combination of computational and experimental methods in order to reconstruct missing regulatory, signaling, and metabolic interactions essential for the function of the sporulation network. We will build a mechanistic computational model of the regulatory network centered at the transcription factor Spo0A and we will use it to understand the interplay between the onset of sporulation, bacterial cannibalism, and biofilm formation.
