Scholar Profile

David A. Agard

Department of Biochemistry and Biophysics
University of California, San Francisco
San Francisco, CA 94143
Voice: 415-476-2521
Fax: 415-476-1902
Personal Homepage
1982 Searle Scholar

Research Interests

Structural Basis for Enzyme Specificity and Chromosome Structure

The research of our laboratory is devoted to structural studies of biological problems in an effort to understand the fundamental relationships between structure and function at the molecular and cellular levels. At the molecular level, we are probing how proteins achieve the folded state as well as the complex relationship between protein structure and function. At the cellular level, our interest focuses on understanding underlying principles of organization of supramolecular structures, and ultimately how this organization influences function. Currently, there are several areas of investigation being actively pursued: 1) understanding the mechanism of pro region catalyzed folding of a-lytic protease; 2) probing the structural basis of substrate specificity using a-lytic protease as a model system; 3) structure of Hsp-90 and its interaction with steroid hormone receptors; 4) molecular recognition in cholesterol metabolism; 5) three-dimensional structural analysis of chromosome condensation and 6) centrosome structure.

Understanding the Mechanism of Pro Region Catalyzed Protein Folding

An unexpected benefit of choosing a-lytic protease for the specificity studies described below was revealed when the gene was cloned and found to be synthesized as a preproenzyme. The 166 amino acid pro region is absolutely required for the proper folding of the 198 amino acid protease domain both in vivo and in vitro. However it forms no part of the active protease. Thus it must be required for the proper folding of the protease domain. Significantly, the covalent linkage between the pro region and the protease domain is not required for function. This enabled us to develop a powerful in vitro folding system where the pro region is added as a separate protein. In principle, the pro region could function by reducing the rate of off-pathway folding reactions such as aggregation (as suggested for the molecular chaperonins), or by increasing the rate of a limiting on-pathway reaction. It turns out that the pro region is a potent inhibitor of the mature protease. This suggests first, that the pro region directly facilitates an on-pathway reaction, and second, that the rate-limiting folding transition state has a native-like conformation.

Remarkably, refolding denatured protease in the absence of the pro region results in the accumulation of a stable folding intermediate. The intermediate rapidly chases to the native state upon addition of the pro region, but cannot fold by itself. Detailed kinetic analysis proves that the pro region functions by directly stabilizing the rate-limiting folding transition state, and thereby speeds folding by at least a factor of 106. The ability to isolate a stable folding intermediate paves the way for detailed structural characterization of the pro region-dependent folding intermediate and of the pro region-protease complex. A combination of solution NMR, H-D exchange NMR, and X-ray crystallographic experiments are now underway. Soon we will have the crystal structure of the complex between the pro region and the native enzyme. Other crystallographic efforts are also underway.

The interactions between the pro region and the protease are also being probed using genetics and molecular genetics approaches. By mutation, we have identified mutations within either the pro region or within the protease domain that primarily effect the rate of folding; thus telling us about interactions in the transition state. We are now screening for suppressors to these mutations in the other partner. Eventually, we would like to develop a protease that could fold without the pro region.

Taken together, these studies should provide insight into the conformational changes and energetics involved in a single-folding partial reaction. This system provides a rare opportunity to structurally and functionally examine the folding processes in detail.

Understanding the Structural Basis of Enzyme Specificity

One of the fundamental functions of an enzyme is to be specific, that is, to limit the number of substrates on which it can act. Unfortunately, how an enzyme accomplishes this task is quite unclear. A number of years ago, we chose a-lytic protease as an ideal model system to investigate structural and energetic aspects of specificity because its binding pocket is made of the side chains of 3 amino acids; providing a large volume which could be experimentally manipulated. Our approach is to combine solution kinetic analysis, X-ray crystallographic structural analysis, site-directed mutagenesis, and multidimensional heteronuclear nuclear NMR methods. Of key importance for the structural studies has been the availability of very tight binding peptide boronic acids, which provide an excellent model for the reaction transition state or nearby intermediates. Thus, these inhibitors allow us to examine key interactions taking place between enzyme and substrate at the defining step in catalysis.

By mutation, we have been able to dramatically alter the pattern of substrate specificity while maintaining or increasing enzyme activity. A large family of mutant proteins have been explored kinetically and structurally and have indicated the crucial role that protein flexibility plays in substrate selectivity. Our results indicate that, to understand enzyme specificity, the dynamical behavior of the enzyme must be considered. Current efforts are aimed at mapping the conformational energy surface for the enzyme and its peptide inhibitor complexes.

To better understand the dynamical behavior of the protease we are using powerful multi-dimensional NMR methods. In collaboration with Vladimir Basus (UCSF, Pharm Chem.) we have fully assigned the backbone of this 20KDa protein and are now examining hydrogen exchange and dynamics of the native state.

A key test of one's understanding is to be able to predict the effect of mutations on substrate specificity. This is a first step in being able to rationally design proteins. We have brought this goal dramatically closer through a newly developed algorithm that combines the side-chain rotamer concept developed by Ponder and Richards with a complete force field and solvent model. This approach is also very effective for modeling the structure of an unknown protein based on a homologous structure.

Structure of Hsp-90 and its Interactions with Steroid Hormone Receptors

The heat-shock protein Hsp-90 is highly conserved from E. coli to man. It accounts for ~ 1% of soluble protein in all cells and is believed to be important in late stages of protein folding. While its major housekeeping role in the cell is unclear, it plays a key role in the regulation of the steroid hormone receptors such as the glucocorticoid receptor (GR, see Yamamoto) as well as numerous kinase signaling proteins. For GR and related receptors, Hsp-90 forms a tight complex with and stabilizes the receptor in the absence of hormone. In this state the receptor is inactive. Upon addition of ligand, Hsp-90 is displaced, and the receptor is active to enter the nucleus and regulate transcription.

We are in the process of determining the structure of Hsp-90 by x-ray crystallography. In collaboration with the Yamamoto lab we are investigating the functional and structural interactions of Hsp-90 with steroid hormone receptors.

Molecular Recognition in Cholesterol Metabolism

Because it is so insoluble, cholesterol is transported throughout the blood in the form of lipoprotein particles such as VLDL (Very Low Density Lipoprotein), LDL (Low Density Lipoprotein), HDL (High Density Lipoprotein), etc. Cellular uptake of these particles is governed by the interaction of ligands on the lipoproteins (either apoE or apoB) and specific cell surface receptors such as the LDL receptor and its family members. Previously we solved the structure by x-ray crystallography of the receptor binding domain of the most common form of human apo-E, apo-E3. In addition we determined the structures of the two most common human apoE mutations, apoE2 which has dramatically reduced receptor binding and apoE4 which has altered lipoprotein particle distributions and is strongly implicated in Alzheimer's disease. These structures revealed a new mechanism of "action at a distance" through concerted salt bridge re-arrangements.

Current efforts focus on obtaining fragments of the ligand binding domain of the LDL receptor and the binding domain of apoB for structural analysis.

Three-Dimensional Analysis of Chromosome Condensation

Our studies on chromosome structure are done in close collaboration with John Sedat and his group. The primary aim of our research in this area is to provide a physical basis for understanding chromosome behavior and function by directly determining the three-dimensional structure of eukaryotic chromosomes as a function of both transcriptional state and the cell cycle stage. Owing to the complexity and variability of such super-molecular structures, the goal is to understand the structural patterns and themes that underlie chromosomal organization. To accomplish this task, we are attempting to answer the following broad questions: 1) how are fibers of nucleosomes folded into compact chromosomes; 2) what is the nature of the structural rearrangements that accompany chromatin condensation and decondensation; and 3) how is chromatin organized into bands and interbands in polytene chromosomes-this should help provide information on the relationship of chromatin structure to transcription.

To accomplish these aims, we have had to develop the necessary technologies (hardware and software) to allow us to examine complex non-crystalline specimens in three dimensions using electron and light microscopy. State-of-the-art imaging methods for both kinds of microscopes have been coupled with advanced three-dimensional image reconstruction and image processing methods. Three-dimensional electron microscopy (EM) analysis is performed by the method of electron microscopic tomography, which, by analogy with computed tomography (CT) methods, allows us to look inside chromosomes (or other cellular structures) and examine their internal arrangements in three dimensions at about 40-75 Angstrom resolution. The Howard Hughes Medical Institute's intermediate voltage electron microscope equipped with an ultrahigh angle tilt stage and a 1024 x 1024 pixel-cooled charge-coupled device provides ideal imaging capabilities for EM tomography of specimens up to .7 µ thick. Recent efforts have been aimed at automating the complex and laborious task of collecting three-dimensional tomographic data and calculating the three-dimensional reconstructions. This approach leads to a great reduction in the electron dose and improved resolution. Another area of basic research is to understand the mechanism of image formation for thick specimens in order to properly relate the digital images to the physical properties of the specimen. Surprisingly, we have found that phase contrast is the dominant mode of image formation even for 0.5 µm thick specimens. We are now developing an optimal strategy for combining the information in several images taken at different focus levels to produce a single corrected image. Our goal is to make this powerful new technology accessible to the cell biologist.

Three-Dimensional EM Provides Insights into Higher-Order Chromosome Structure

Over the past several years, we have made significant progress in analyzing the details of higher-order chromosome structure. In collaboration with Professor Chris Woodcock (U. Mass., Amherst) we have solved the structure of the "30nm" fiber which is generally thought to be the next highest level of structure above the nucleosome. Surprisingly, the structure is not organized as a solenoid as shown in most text books, but as an irregular 3D zig-zag. The structure is remarkably open, with no nucleosome-nucleosome contact. Instead, the structure sees to depend on the exit angle and length of the linker DNA.

Much of the chromatin within chromosomes is actually organized as a structure ~130nm in diameter, which in metaphase becomes folded into a still higher order structure. High resolution three-dimensional reconstructions (100-150 views, ± 70 degrees) of sections <=.5 µ thick from interphase, metaphase, and telophase chromosomes have provided significant insights into the exceedingly complex problem of chromosome structure. Recently we have begun using a Xenopus extract in vitro chromosome condensation system developed in the Mitchison lab. This should allow us to sample different stages of condensation. Moreover, as key components become defined biochemically in the Mitchison lab, we will be able to examine where they are located and to see what abortive structures are formed when they are depleted.

Centrosome Structure

Centrosomes are complex cellular organelles that control the array of microtubules during interphase and the spindle during mitosis. In collaboration with the Albert's lab, we have begun to investigate the structure of isolated Drosophila embryonic centrosomes by EM Tomography. The goal is to understand the structure and organization of those components that nucleate microtubule (MT) assembly. From the tomography it has been possible to trace the MTs within the complex pericentriolar material (PCM) and locate their minus ends (where polymerization starts). The nucleating structures seem to be randomly oriented and located within the PCM, and are not part of a well define structure. Instead, the organizer must be comparable in size to the MT diameter. Examination of reconstructions without MTs, show the presence of a large number of small ring structures that are the same diameter as MTs, and 10-13nm in length.

Immunolocalization studies indicate that g-tubulin which is know to be important for MT assembly is localize to the rings, and can be found at the minus ends of MTs. We are now pursuing higher resolution reconstructions and immunolocalizations of other known proteins.