Structure-based design of new antifungal drug leads
Amy Anderson - University of Connecticut, School of Pharmacy
Candida glabrata is emerging as a lethal fungal pathogen and is inherently resistant to several of the available antifungal therapeutics, making it critical to develop new agents to treat C. glabrata infections. Building on previous success to target dihydrofolate reductase (DHFR) in pathogenic organisms, we have synthesized a class of versatile DHFR inhibitors. Several of an initial series of our compounds are potent and selective inhibitors of the C. glabrata DHFR enzyme and exhibit antifungal activity in cultures of the organisms. In order to design more effective antifungal compounds, we determined a crystal structure of C. glabrata DHFR bound to NADPH and a lead inhibitor at 1.6 Å resolution. Not only does the structure reveal the details of interactions between the ligand and protein but it also reveals a hydrophobic region flanked by residues in a loop near the active site that can be exploited for increased potency and selectivity. Docking the lead inhibitor series to the crystal structure elucidated additional details of structure-activity relationships. Second generation inhibitors were synthesized and shown to bind the C. glabrata DHFR enzyme with subnanomolar potency, display greater than 2,000-fold levels of selectivity over the human enzyme and inhibit the growth of C. glabrata at levels observed with clinically employed antifungal therapeutics. Crystal structures of the enzyme bound to the new inhibitors complete the cycle of structure-based design and pave the way toward the third generation of antifungal drug leads.
Computational Research Needs for Renewable and Alternative Energy:
Studies of Natural and Artificial Photosynthesis
Victor Batista - Yale University
Most of the atomspheric oxygen that sustains life on earth has been generated by plants during the light period of photosynthesis. Each oxygen molecule has been generated by the reaction of two water molecules: 2 H2O -> O2 + 4 H+ + 4 e- catalyzed by an oxomanganese complex embedded in the subunit D1 of photosystem II (a transmembrane complex of about 20 proteins found in the thylakoid membranes of green plant chloroplasts). Both the reaction mechanism and the structure of the catalytic center responsible for this important reaction remain poorly understood. This talk will present recent advances in experimental and computational studies towards the development of rigorous models of the oxomangenese complex and the catalytic cycle responsible for oxygen evolution, as well as recent progress on studies of biomimetic systems for artificial photosynthesis.
Protein Flexibility and Drug Discovery: Trying to Hit a Moving Target
Heather A. Carlson - University of Michigan, Ann Arbor
In the multiple protein structure (MPS) method, a collection of protein conformations is used to represent the ensemble of states available to a flexible receptor. Each conformation is mapped with probe molecules and then combined to identify "consensus" regions where the requirements for complementing the receptor are consistent over many conformations. The probes in the consensus regions are translated into a pharmacophore model which describes the essential interactions but does not introduce any limits in the flexible areas. Experimental testing has shown that the technique is useful for pushing discovery into new chemical space with inhibitors of different sizes, shapes, chemical content and scaffolds. Applications to HIV-1 protease and the cancer target p53-MDM2 will be shown. In particular, inhibitors have been identified for a new pocket in HIV-1 protease; the inhibitors are half the molecular weight of existing therapies and may eventually yield new therapeutics with better pharmacokinetic properties.
New Methods for Connecting Protein NMR Data to Structure and Dynamics
David Case - Rutgers University
The "classic" route to protein structure determination from NMR is well established, and continues to produce many useful entries in the Protein Data Bank every year. There is continuing interest in expanding these NOE-based methods in two directions. On the low-resolution end, efforts are underway to extract useful structural data from limited data sets (involving chemicalshifts or one-bond coupling constants) that are relatively easy to obtain,even for large or ill-behaved systems. At the high-resolution end, one canhope to extract a lot of detail about both structure and dynamics (includinginformation about conformational heterogeneity) for small, well-characterizedproteins. These high- and low-resolution extensions have a lot in common,as they rely on computational methods that are based on molecular dynamicssimulations to assist in the interpretation of chemical shifts and residualdipolar couplings. I will discuss some recent developments in these areas,concentrating on new methods to relate MD simulations to chemical shifts,dipolar couplings, and to more conventional NMR relaxation measurements.
Modularity and analogies in protein-ligand binding sites
Manuela Helmer-Citterich - Università di Roma Tor Vergata, Italy
I will describe our work on the analysis of protein-ligand binding sites of known structure by computational methods.
The structural analysis of protein ligand binding sites can provide information relevant for assigning functions to unknown proteins, to guide the drug discovery process and to infer relations among distant protein folds. Previous approaches to the comparative analysis of binding pockets have usually been focused either on the ligand or the protein component. In the former case the analysis is restricted to binding pockets interacting with similar ligands, while in the latter it is difficult to systematically check whether the observed structural similarities have a functional significance.
I will discuss a new approach, based on the search for local structure similarity, that simultaneously takes into account both the binding pocket and the ligand and permits to exploit the whole dataset of protein ligand binding sites. Thanks to this method, we identified structural motifs whose functional significance is explained by the presence of shared features in the interacting ligands, even in the context of molecules with a different overall structure. In addition some of these motifs are present in a high number of evolutionarily unrelated proteins. I will propose some evolutionary hypotheses and the possibility to define a protein-ligand interaction code.
Algorithms for Protein Design and Drug Design
Bruce Donald - Duke University
I'll discuss some recent results from my lab, including new algorithms for protein design and drug design. Protein redesign aims at improving target protein properties, such as increasing the stability, switching an enzyme's specificity towards a non-cognate substrate, or redesigning the protein so that it will perform a completely novel function. Exhaustively testing protein mutations in vitro is infeasible, due to the enormous size of the space of possiblemutations. Computational (in silico) approaches can efficiently and accurately explore the combinatorial space of candidate solutions, and have proven valuable for protein redesign and protein engineering. Finally, I will describe experimental results from applying our algorithms to to switch the substrate specificity of a non-ribosomal peptide synthetase (NRPS) enzyme, and to design allosteric small-molecule inhibitors of protein:protein interactions that play a critical role in the development of acute myeloid leukemia and acute lymphocytic leukemia.
Molecular Simulations of Macromolecular Behavior in Physiological Conditions
Adrian H. Elcock - University of Iowa
Attempting to model the behavior of biological macromolecules in physiological conditions presents a number of difficulties not encountered in modeling their behavior in vitro. This talk will describe our laboratory's efforts to tackle some of these issues andto develop a realistic simulation methodology, based on Brownian dynamics ideas, that is capable of modeling the dynamics of very large macromolecular systems. Example applications of our simulation methodology that will be discussed include synthesis of proteinsin a dynamic model of a polyribosome, and diffusion of macromolecules in a realistic model of the cytoplasm of Escherichia coli.
Short Structured Functional Peptide motifs
Martin Schiller - University of Connecticut Health Center
Short peptide sequence motifs are important for binding to other molecules post-translational modifications, and protein trafficking. Our lab has built a database of minimotifs called Minimotif Miner that can be used to predict these functions in protein queries. To show the utility of minimotif analysis we are analyzing motifs in HIV proteins. We mapped the evolutionary conservation level of HIV sequences onto thestructures of HIV proteins to identify spatially conserved regions. Minimotifs that are localized to these regions may serve as new targets for therapeutic innervations as viruses exploit minimotifs as a means to acquire host machinery.
Simulations of Protein Aggregation
Joan-Emma Shea - University of California, Santa Barbara
A number of diseases, known as amyloid diseases, are associated with pathological protein folding. In the case of Alzheimer’s disease, incorrectly folded Amyloid-beta (Ab) proteins self-assemble into a variety of neurotoxic aggregate species, ranging from small soluble oligomers to amyloid fibrils. An attractive therapeutic approach to combat amyloid diseases lies in the development of strategies to inhibit or reverse aggregation. In the first part of my talk, I will present fully atomistic molecular dynamics simulations of the interaction of aggregation inhibiting peptides with Ab amyloid fibrils. In the second part, I will introduce a novel off-lattice coarse-grained model for the Ab protein and discuss the kinetics and thermodynamics of aggregation and fibrillogenesis inhibition.