Scientific Report 2006

Molecular Biology

Computation and Visualization in Structural Biology

A.J. Olson, D.S. Goodsell, M.F. Sanner, S. Dallakyan, A. Gillet, R. Harris, Y. Hu, R. Huey, J. Huntoon, S. Karnati, W. Lindstrom, G.M. Morris, A. Omelchenko, M. Pique, B. Norledge, R. Rosenstein, M. Utsintong, G. Vareille, Q. Zhang, Y. Zhao

In the Molecular Graphics Laboratory, we develop novel computational methods to analyze, understand, and communicate the structure and interactions of complex biomolecular systems. This past year, we showed the effectiveness of 3-dimensional molecular models as a tangible human-computer interface in educational and research settings. Within our component-based visualization environment, we continue to develop methods for predicting biomolecular interactions, analyzing biomolecular structure and function, and presenting the biomolecular world in education and outreach.

We have applied these methods to several important systems in human health and welfare. In a novel distributed computing network, we continue the search for HIV protease inhibitors to fight the growing problem of drug resistance in HIV disease. We used AutoDock, a suite of programs for predicting bound conformations and binding energies for biomolecular complexes, in the virtual screening of large databases of compounds and ultimately identified new compounds for use in the treatment of cancer. We used methods for predicting protein interaction to probe the mechanism of blood coagulation.

Tangible Interfaces in Structural Biology

We have continued to develop autofabricated physical models (“solid printing”) of biological molecules and the components and assemblies of the molecules; our goal is to use the models in both research and education. We integrated computer graphics and computation with these physical models by using augmented reality to create custom interfaces to facilitate exploration and computation of molecular interactions. We have begun to use a self-assisted protein-folding model to teach elements of protein structure and assembly to our graduate students. We are continuing to develop the software that will enable the control of interactive computations through manipulation of the tangible models.

In collaboration with T. Herman, Milwaukee School of Engineering, Milwaukee, Wisconsin, we created a model of the active site of acetylcholinesterase that can be opened to show the buried active site and bound substrates (Fig. 1). This model was used, along with an interactive Internet guide to the structure, as part of a Waksman Challenge at Rutgers, the State University of New Jersey. Groups of teachers and students were asked to use the models and associated materials to explore problems with insecticide resistance in compounds that act on mosquito acetylcholinesterase.

Fig. 1. A tangible model of the active site of acetylcholinesterase. The model separates into 4 pieces, allowing students to fit different substrates into the buried active site.

Recently, we worked with Biomedical Graphics at Scripps Research to establish a solid-model printing service for researchers at Scripps and elsewhere. This service is now in operation and has made a number of molecular models for scientists working with molecular structures. The users’ responses have been positive, and the research community is beginning to see how a solid 3-dimensional model can provide tangible, multimodal feedback that mouse, keyboard, and image behind a glass screen cannot provide.

Advances in Computational Docking

We have just completed developing and testing a semiempirical free energy force field for use in AutoDock and similar grid-based docking methods. The force field is based on a comprehensive thermodynamic model that allows incorporation of intramolecular energies into the predicted free energy of binding. The model also incorporates a charge-based method for evaluating desolvation designed to use a typical set of atom types. The method was calibrated by using a set of 188 diverse protein-ligand complexes of known structure and binding energy and was tested by using a set of 100 complexes of ligands with retroviral proteases. Compared with the previous AutoDock force field, the new force field provides an improvement in redocking simulations.

AutoDock 4 has been modified to support more atom types and to use an improved atom-typing mechanism. Importantly, AutoDock 4 also now simulates the molecular system being docked in the unbound state, by generating and evaluating the extended conformation of the ligand and moving side chains in the receptor before the docking occurs. The unbound state is now considered in the calculation of the change in free energy upon binding. AutoDock’s companion graphical user interface, AutoDockTools, has been modified to support preparation of input files for AutoDock 4, in particular to allow the definition of flexible side chains in macromolecules. AutoDockTools has also been made easier to use by simplifying the menus.

On the basis of the AutoDock force field, we developed a method for locating and characterizing the optimal binding site for ligands on the surface of a protein of known structure. The method identifies the contiguous constant-volume region with the most favorable binding affinity. The optimal binding sites identify regions of primary binding affinity and regions of suboptimal binding strength, which can be used to predict the function of proteins if the function is unknown or to identify target locations for the design of new inhibitors. We showed the usefulness of the method in the design of inhibitors for HIV type 1 protease, and we are applying the method to a large database of protein structures.

Component-Based Visualization Environments

To facilitate the integration and interoperation of computational models and techniques from a wide variety of scientific disciplines, we continue to expand our component-based software environment. The environment is centered on Python, a high-level, object-oriented, interpretive programming language. This approach allows the compartmentalization and reuse of software components. Python provides a powerful computation “glue” for assembling computational components and, at the same time, a flexible language for the interactive scripting of new applications.

We released version 1.4.1 of our software components in March 2006. This release contains substantial enhancements, including a completely rewritten interface to the adaptive Poisson-Bolzman solver APBS, making it easy to produce high-quality pictures of electrostatic potentials on molecular surfaces. A new control panel provides a high-level interface for rapidly displaying molecular models in a variety of representations. This new release is also distributed with installer programs for computers running the Windows and Macintosh OS X operating systems. We also added a parser for macromolecular Crystallographic Information File that allow users to read and write files in the macromolecular Crystallographic Information File format. This addition helps overcome limitations in the Protein Data Bank format such as maximum number of atoms or chain IDs.

Modeling Protein Flexibility in Docking

We have developed a hierarchical and multiresolution representation of the flexibility of biological macromolecules that can be used in computational simulations. This treelike structure enables the computationally tractable encoding of a small subset of a protein’s conformational subspace. After implementing the core infrastructure of the Flexibility Tree and developing intuitive graphical interfaces for building such trees, we have started exploring the use of this data structure in the context of automated docking. We reproduced a cross-docking experiment carried out earlier with AutoDock in which 20 inhibitors of HIV protease I where docked systematically into the 20 conformations of the receptor. We showed that by adding receptor flexibility, we could increase the rate of successful cross docking from 72% to 98%.

Protein Docking

In collaboration with C. Bajaj, University of Texas, Austin, we are investigating a novel fast Fourier transform–based method for predicting the association of protein in complexes. In parallel with this docking method, we evaluated the effect on blurring molecular surfaces on the shape complementarity at the interface between proteins in a complex. We characterized the level of distortion introduced by blurring atomic spheres by using gaussian distributions and determined an optimal blurring level for docking purposes. In addition, we added software components for the calculating the curvature of meshes that are used in the docking procedure.

Fighting Drug Resistance in HIV Disease

As part of a program project, we continue our work on inhibitors to fight drug resistance in the treatment of AIDS. In collaboration with K.B. Sharpless and C.-H. Wong, Department of Chemistry, we have designed and optimized a series of inhibitors built around a triazole formed in a click chemistry reaction. We are also exploring larger issues of resistance via docking experiments with large chemical databases and large sets of mutant protease structures. These massive docking experiments are made possible by the resources available in the FightAIDS@Home distributed computing system. FightAIDS@Home enlists the worldwide community in a large computational effort to design effective therapeutic agents to fight AIDS. Personal computers are used by the program when the computers are not in use by their owners, providing an enormous, and largely untapped, computational resource. The current goal is to identify inhibitors that are effective against the wild-type virus and against common mutant forms of the virus.

In the past year, we moved FightAIDS@Home to the IBM World Community Grid. This transition involved working closely with the team at IBM to board our automated docking software, AutoDock, to be able to run on the Windows United Devices client and the Linux and Macintosh OS X Berkeley Open Infrastructure for Network Computing clients. This move has increased the number of available processors to about 300,000 and has enabled us to compute a complete scan of the National Cancer Institute diversity set (2000 compounds) against a panel of 200 mutant HIV proteases in a matter of 4 months. This computation required more than 2 quadrillion energy evaluations of ligand vs protein.

Interactions of Tissue Factor

We used our ligand-protein and protein-protein model to study the interactions of tissue factor (TF) in the initiation of blood coagulation and the related regulatory roles of the factor. TF plays a potential role in metastasis, growth, and angiogenesis of tumor cells via 2 distinct mechanisms: interaction of the complex consisting of TF and factor VIIa with protease-activated receptor 2 (PAR2) and interaction of the complex consisting of TF, factor VIIa, and factor Xa with PAR1 or PAR2. However, no PAR structures are available for studying these mechanisms. Because PAR2 is involved in both pathways, we performed protein homology modeling studies of this receptor.

We found 8 unique PAR2 sequences. For each unique sequence, we searched its homology sequences in the Protein Data Bank and chose as the homology template the sequence that has the highest-resolution x-ray crystal structure. Sequence alignment was then performed between the PAR2 sequence and the template sequence. The alignment was then input to MODELLER for building 10 homology structures, from which we chose the structure with the best quality as the homology model. The homology models have enabled us to use AutoDock to perform docking studies of PAR2-activating peptides and small molecules on PAR2. The discovered binding modes are being confirmed by our collaborator, W. Ruf, Department of Immunology.

Structure-Based Drug Design in Gaucher Disease

Gaucher disease is the most common lipid-storage disorder caused by activity-compromising mutations in glucosylceramidase and is the most common genetic disease affecting Ashkenazi Jews. In addition to causing great pain, anemia, and massive enlargement of the liver and spleen, Gaucher disease can lead to neurologic impairment or early demise. Among the most promising treatments, small-molecule chemical chaperones can rescue the enzyme activity of the misfolded glucosylceramidase. The experimentally identified deoxynojirimycin-type inhibitors have a narrow concentration range and cause a mild improvement in the activity of the mutant enzyme.

We did a molecule fragment–based virtual screening with the National Cancer Institute diversity data set.

A total of 72 compounds identified as the best compounds of interest by virtual screening were tested in the enzyme assays by our collaborator, J. Kelly, Department of Chemistry. A total of 13 compounds are insoluble in dimethyl sulfoxide at up to 10 mM; 25 precipitate in the assay buffer. Among the remaining 34 compounds examined by using in vitro enzyme assays, 16 show significant inhibition of the enzyme. The top compounds of interest have almost doubled the activities of the mutant enzymes N370S and G202R in vivo.

Protein Phosphatase 2C Inhibitors

In collaboration with P. Greengard, Rockefeller University, New York, New York, we used AutoDock to screen the National Cancer Institute diversity set against protein phosphatase 2C (PP2C), an enzyme that must remain active for tumor growth in breast cancer. Several compounds were identified as inhibitors of PP2C in the computational screen. The compounds were ordered from the National Cancer Institute and were assayed experimentally. Several were inhibitory at micromolar concentrations; the potency of the best was between 5 and 10 μM. The lead compounds discovered in this study are the first nonphosphate-based PP2C inhibitors reported.

Mechanistic Studies of Biocatalysts in Cocaine Antibodies

Currently, no effective treatment of cocaine addiction approved by the Food and Drug Administration is available. One possible treatment based on receptor design entails thorough investigation of cocaine hydrolysis by catalytic antibodies. Between 2 possible reaction pathways (an oxyanion hole for carbonyl or an hydrogen-bond trap for hydroxide ion formed by tyrosines at positions H50 and L94), quantum mechanical, molecular docking, and molecular dynamics free-energy calculations have shown no conclusive evidence for a dominant pathway between 2 possible ones. An explanation for the low turnover rate of the antibody is the failure of the antibody to promote any mechanism selectively because of a homogeneous microenvironment generated by eliciting against a hapten with 2 equivalent phosphorus-oxygen bonds. Computational modeling would help improve hapten design and thus improve the efficiency of the catalytic antibody.

Visual Methods from Atoms to Cells

Understanding structural molecular biology is essential to foster progress and critical decision making among students, policy makers, and the general public. In the past year, we continued our longstanding commitment to science education and outreach with a combination of presentations, popular and professional illustrations and animations, 3-dimensional tangible models, and a presence on the Worldwide Web. In these projects, we use the diverse visualization tools developed in the Molecular Graphics Laboratory to disseminate results that range from atomic structure to cellular function.

We also continued several regular features that informally present molecular structure and function. The “Molecule of the Month” at the Protein Data Bank entered its seventh year of providing an accessible introduction to the central database of biomolecular structure. Each month, a new molecule is presented with a description of the molecule’s structure, function, and relevance to health and welfare (Fig. 2). Visitors are then given suggestions about to how to begin their own exploration of the structures in the data bank. Other projects include “The Molecular Perspective,” articles in the journal The Oncologist that present structures of interest to clinical oncologists and provide a source of continuing education for physicians; “Recognition in Action,” a new series at the Journal of Molecular Recognition; and work with the Nanoscale Informal Science Network supported by the National Science Foundation to develop new materials for presenting the science of nanotechnology.

Fig. 2. ATP synthase was presented as the Molecule of the Month in 2005. The illustration of this complex molecular machine was constructed from 4 separate entries in the Protein Data Bank: 1c17, 1e79, 2a7u, and 1l2p.

Publications

Beuscher, A., Olson, A.J., Goodsell, D.S. Identifying protein binding sites and optimal ligands. Lett. Drug Des. Discov. 2:483, 2005.

Cheng, T.-J., Goodsell, D.S., Kan, C.-C. Identification of sanguinarine as a novel HIV protease inhibitor from high-throughput screening of 2,000 drugs and natural products with a cell-based assay. Lett. Drug Des. Discov. 2:364, 2005.

Dickerson, T.J., Beuscher, A.E. IV, Rogers, C.J., Hixon, M.S., Yamamoto, N., Xu, Y., Olson, A.J., Janda, K.D. Discovery of acetylcholinesterase peripheral anionic site ligands through computational refinement of a directed library. Biochemistry 44:14845, 2005.

Goodsell, D.S. Computational docking of biomolecular complexes with AutoDock. In: Protein-Protein Interactions: A Molecular Cloning Manual, 2nd ed. Golemis, E., Adams, P. (Eds.). Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY, 2005, p. 885.

Goodsell, D.S. The molecular perspective: c-Abl tyrosine kinase. Oncologist 10:758, 2005; Stem Cells 24:209, 2006.

Goodsell, D.S. The molecular perspective: cisplatin. Oncologist 11:316, 2006; Stem Cells 24:514, 2006.

Goodsell, D.S. The molecular perspective: double-stranded DNA breaks. Oncologist 10:361, 2005; Stem Cells 23:1021, 2005.

Goodsell, D.S. The molecular perspective: RAD51 and BRCA2. Oncologist 10:555, 2005; Stem Cells 23:1434, 2005.

Goodsell, D.S. The molecular perspective: tumor necrosis factor. Oncologist 11:83, 2006.

Goodsell, D.S. Recognition in action: DNA mimicry. J. Mol. Recognit. 18:427, 2005.

Goodsell, D.S. Representing structural information. In: Current Protocols in Bioinformatics. Baxeranis, A.D., Davison, D.B. (Eds.). Wiley & Sons, Hoboken, NJ, 2005, p. 5.4.1.

Huey, R., Morris, G.M., Olson, A.J., Goodsell, D.S. A semi-empirical free energy force field with charge-based desolvation. J. Comput. Chem., in press.

Rogers, J.P., Beuscher, A.E. IV, Flajolet, M., McAvoy, T., Nairn, A.C., Olson, A.J., Greengard, P. Discovery of protein phosphatase 2C inhibitors by virtual screening. J. Med. Chem. 49:1658, 2006.

Sanner, M., Stolz, M., Burkhard, P., Kong, X.-P., Min, G., Sun, T.-T., Driamov, S., Aebi, U., Stoffler, D. Nature at work from the nano to the macro scale. Nanobiotechnology 1:7, 2005.