A Machine Learning Approach to Heterologous Membrane Protein Overexpression

A Machine Learning Approach to Heterologous Membrane Protein Overexpression

Sunday, February 28, 2016 design of novel hKOR constructs and their complexes with agonists, antagonists, and biased ligands. 207-Plat Crystal Structu...

45KB Sizes 2 Downloads 34 Views

Sunday, February 28, 2016 design of novel hKOR constructs and their complexes with agonists, antagonists, and biased ligands. 207-Plat Crystal Structure of the Calcium ATPase SERCA in Complex to a Novel Anti-Cancer Agent that Targets Multidrug-Resistant Leukemia John K. Lee1, Joseph M. Autry1, Razvan Cornea1, Nicholas Bleeker2, Denise Casemore2, Chengguo Xing2, David D. Thomas1. 1 Department of Biochemistry, Molecular Biology and Biophysics, University of Minnesota, Minneapolis, MN, USA, 2Department of Medicinal Chemistry, University of Minnesota, Minneapolis, MN, USA. We have used x-ray crystallography to solve the crystal structure of the sarco/ endoplasmic reticulum Ca-ATPase (SERCA) in complex with a novel lead compound (CXL017) that targets multidrug-resistant leukemia (Hermanson et al., Mol Pharmacol 2009). The structure, solved at 3.01A resolution, reveals CXL017 bound to SERCA near the ATP binding site and suggests blockage of the ATP channel as the main mechanism of inhibition. This result confirms that CXL017 binds SERCA at a location that is distinct from the binding sites of other inhibitors of SERCA, and it explains the synergistic relationship that was uncovered between CXL017 and classic SERCA inhibitors (Bleeker et al., Mol. Pharmacol 2013). Further, it provides the platform for structureguided improvements on this new class of potential anti-cancer agent. This work was funded by a grant from NIH to DDT (R01 GM27906) and a grant from NIH to XC (R01 CA163864). 208-Plat Advances in in situ X-ray Crystallography of Membrane Proteins Jana Broecker1, Viviane Klingel2, Bryan T. Eger1, Oliver P. Ernst1,3. 1 Biochemistry, University of Toronto, Toronto, ON, Canada, 2Institute of Biomaterials and Biomolecular Systems, University of Stuttgart, Stuttgart, Germany, 3Molecular Genetics, University of Toronto, Toronto, ON, Canada. In the last decade, crystallisation of membrane proteins in lipidic cubic phases (LCP; in meso) boosted the numbers and resolution of membrane-protein structures. However, membrane-protein crystals are usually small and notoriously difficult to harvest from the highly viscous LCP. Recently, an in situ in meso approach has been introduced, in which crystals are not harvested and flashfrozen but placed in the X-ray beam within the mesophase at room temperature. In this study, we introduce novel in situ in meso plates that show significantly less background scattering, are easier to handle, and are variable with respect to the crystallisation volume. Moreover, we developed holders for either individual or up to four in situ wells at a time, which are easy and cheap to produce, easy to handle, reusable, and compatible with measurements at room temperature and under cryogenic conditions. Because the holders are attached to standard ALS-type goniometer bases, they allow for storage and shipping of entire wells (with typically several dozens of crystals) in Universal V1-Pucks under liquid nitrogen and for auto-mounting at synchrotrons. We validated the new setups using water-soluble hen egg lysozyme and the membrane protein bacteriorhodopsin. In conclusion, this study demonstrates the potential of combining in meso crystallisation with in situ diffraction and the possibility to store, ship, and measure crystals under cryogenic conditions for obtaining structural information on membrane proteins. In conjunction with the current developments at synchrotrons like smaller beams, faster detectors, and software for multi-crystal strategies, this approach promises high-resolution structural studies of membrane proteins to become faster and more routine. 209-Plat A Machine Learning Approach to Heterologous Membrane Protein Overexpression Shyam M. Saladi, Nauman Javed, Axel Mu¨ller, William M. Clemons. Division of Chemistry and Chemical Engineering, California Institute of Technology, Pasadena, CA, USA. Membrane protein production is difficult; their biogenesis does not stop with translation but also requires translocation and integration into a lipid bilayer. These additional steps hamper their heterologous expression which significantly impedes biophysical and structural studies. Detailed and anecdotal evidence in literature suggests that a variety nucleotide and amino-acid sequence level determinants may potentially support or hinder their biogenesis, e.g. mRNA pausing elements, codon adaptation, transmembrane segment hydrophobicity, ‘‘positive inside rule.’’ By training a preference-ranking Support Vector Machine, we have developed a statistical model that predicts a relative likelihood of a membrane protein’s successful expression using quantitative experimental data of overexpression. This model is rigorously validated against expression outcomes from smallscale laboratory experiments (e.g. expression tests that routinely precede struc-


tural studies) published in the literature as well as large-scale expression trials from a Protein Structure Initiative consortium facility. We show remarkable agreement between the predicted and experimental expression outcomes and propose our model, trained and cross-validated on the entire corpus of data, as a tool for the membrane protein biophysics community to streamline the process of overexpressing a target for study. Given the framework of our model, it can be trivially re-trained as additional experimental outcomes are gathered from past work or created from experiments. Furthermore, the relative weights gathered from parameters of the statistical model may help further characterize translocation mechanisms and suggest intriguing areas for further biophysical and computational experiments. 210-Plat Linking the Outer Membrane of E.Coli to the Cell Wall via OMPA & Braun’s Lipoprotein: Towards a Molecular Model of a Virtual Bacterial Cell Envelope Syma Khalid1, Maite Ortiz-Suarez2, Peter J. Bond3, Thomas Piggot4. 1 School of Chemistry, University of Southampton, UK, Southampton, United Kingdom, 2Centre for Molecular Informatics, University of Cambridge, Cambridge, United Kingdom, 3Bioinformatics institute, A *STAR research institute, Singapore, Singapore, 4Defence Science and Technology Laboratory, Salisbury, United Kingdom. In the last couple of years molecular dynamics simulations of bacterial membranes have reached unprecedented levels of detail. In particular the outer membrane of Gram-negative bacteria is now being routinely simulated as an asymmetric membrane, incorporating lipopolysaccharide (LPS) molecules in the outer leaflet and a realistic mixture of phospholipids in the inner leaflet, at the atomistic level of detail. However to move towards a more systems approach to simulating the whole cell envelope, the cell wall must be considered. Here we present our progress on simulating the full-length OmpA protein in its monomeric and dimeric forms anchored in the outer membrane but also bound to peptidoglycan of the cell wall. Our results reveal considerable flexibility in the OmpA linker, predict stable binding of the peptidoglycan to the Cterminal domain and reveal hitherto unexplored details of the dimerization interface in a realistic membrane environment Furthermore, we also discuss our models of Braun’s lipoprotein, which also provides a link between these two components of the cell envelope. In summary we present our progress towards detailed models of an entire bacterial organelle. 211-Plat Systematic Evaluation of the CS-Rosetta De Novo Structure Prediction Method for Membrane Proteins Katrin Reichel1,2, Olivier Fisette1, Tatjana Braun3, Gerhard Hummer2, Oliver Lange3, Lars Scha¨fer1. 1 Ruhr-University Bochum, Bochum, Germany, 2MPI of Biophysics, Frankfurt, Germany, 3TUM Munich, Munich, Germany. In silico structure prediction of proteins on the basis of their sequence is a major challenge in computational structural biology, even though the ca. 100.000 entries stored in the Protein Data Bank now give a broad sampling of folds. Membrane proteins are particularly challenging cases due to their size and complexity, and high-resolution structural data is often lacking due to experimental challenges. We have investigated how the recently developed RASREC CS-Rosetta methodology benefits from integrating sparse NMR data for de novo structure prediction of membrane proteins. In particular, we tested the 4-TM disulfide binding protein B and sensory rhodopsin (full-length and 4TM subdomain) for which both NMR and high-resolution X-ray crystal structures are available. We systematically investigated the effect of varying the type of data (chemical shifts, NOE distance restraints) and the amount of data (number of long-range NOEs) on the accuracy of the structure prediction. Our results show that RASREC CS-Rosetta can reliably predict membrane protein structures even with very sparse NMR data, and determine the minimal amount of NMR data needed. 212-Plat Insights into How Mutations Thermostabilize G-Protein-Coupled Receptors Nagarajan Vaidehi1, Sangbae Lee1, Supriyo Bhattacharya1, Manbir Sandhu1, Reinhard Grisshammer2, Christopher G. Tate3. 1 Immunology, Beckman Research Institute of City of Hope, Duarte, CA, USA, 2Membrane Protein Structure Function Unit, National Institute of Neurological Disorders and Stroke, Rockville, MD, USA, 3Cambridge Biomedical Campus, MRC Laboratory of Molecular Biology, Cambridge, United Kingdom.