OvervieW

Computational, biophysical, structural and proteomic studies of protein-protein interactions.

We are studying the specificity of protein-protein interactions in a research program that combines bioinformatic analysis, structural modeling, computational design and experimental characterization. Our aim is to understand, at a high level of detail, how the interaction properties of proteins are encoded in their sequences and structures. Most of our work is focused on two protein families that are important for human health: the a-helical coiled coil and the Bcl-2 family of apoptosis-regulating proteins.

Research Summary

Protein-protein interactions establish the architecture of the cell, regulate biological signaling, underlie the assembly of macromolecular machines and mediate chemical transformations of both small and large molecules. Although we now have fairly complete lists of the proteins found in various organisms, our knowledge of which proteins interact with one another, as well as how and why they interact as they do, is quite limited. In the Keating lab, we are particularly interested in the question of interaction specificity; i.e., how a protein selects a particular interaction partner out of a large number of closely related alternatives. Both computational and experimental methods are needed to accelerate discovery and understanding in this area. Our lab integrates both approaches, tackling the complex problems of characterizing, analyzing, designing and predicting protein-protein interaction specificity by studying individual protein domains that have relatively simple structures.

Coiled coils
The α-helical coiled coil is the simplest of all protein-protein interaction motifs. Coiled coils consist of two or more α-helices that wrap around each other with a superhelical twist. They are characterized by a repeating sequence of seven amino acids, (abcdefg)_n, in which the a- and d-position residues are predominantly hydrophobic and the e- and g-position residues are usually polar or charged. The regular sequence makes it possible to predict the occurrence of coiled coils in genomic sequence data. We estimate that >5% of all proteins in S. cerevisiae, C. elegans, A. thaliana and D. melanogaster contain a coiled-coil region. It is likely that many of these coiled coils mediate protein-protein interactions or oligomerization. An important, unanswered question about coiled coils is how their interaction specificity is encoded in their sequence. We call this the “partnering problem” for coiled coils and are studying it using both computational and experimental approaches.

Our experimental approach to the partnering problem started with an analysis of human bZIP transcription factor interactions. In these proteins, the coiled-coil region determines the homo- or heterodimerization specificity of the transcription factor, which in turn influences its DNA-binding properties and biological function. To determine how sequence encodes interaction preferences in the bZIPs, we used protein microarray technology to measure all of the pair-wise interactions between 49 human and 10 yeast bZIP peptides. We found that the interactions are very specific, and that interaction profiles are largely, but not universally, conserved within bZIP subfamilies. We also measured the relative stabilities of several bZIP homo- and heterodimers in solution, and found excellent agreement between the array studies and solution results. This work establishes the protein microarray as a powerful method for generating large amounts of high quality interaction data, and we are now testing it for other types of domains.

In addition to providing a wealth of data about important transcription factor interactions, the bZIP microarray data provide an opportunity for testing and improving computational models. We have used this information to develop and/or test several different methods for predicting coiled-coil interactions. A machine-learning algorithm trained on the literature shows excellent performance in detecting correct bZIP pairings. We have also used structure-based methods for prediction. Because the coiled coil has a very simple structure, it is particularly amenable to molecular modeling; the core structure of many coiled coils can be predicted with good accuracy from sequence alone. We have shown that structural modeling can be used in conjunction with learning models to provide improved predictions of bZIP coiled-coil interactions. We are now applying structure-based methods more broadly to the problem of predicting coiled-coil interaction specificity.

Another way to understand factors that mediate protein association is through the process of design. The field of protein design has seen exciting advances in the past ten years with the application of fast search algorithms to the problem of side-chain selection and positioning. This has allowed, for example, the design of proteins with new folds and functions. We are applying methods developed for the computational design of stable protein folds to the study of protein interaction specificity. In one study we designed and characterized a mini-protein heterotetratmer in collaboration with Barbara Imperiali’s group at MIT. More recently, we have designed coiled-coil peptides that bind specifically to native bZIP transcription factor targets.

Bcl-2 family proteins
The Bcl-2 family comprises ~25 proteins important for controlling apoptosis. Critical junctures that govern cellular life-vs-death decisions are regulated by specific interactions among pro- and anti-apoptotic members of this family. The delicate balance between these is often disrupted in cancers. Five mammalian anti-apoptotic family members have a conserved helical, globular structure, and all known family members share a weakly conserved short BH3 (Bcl-2 homology 3) sequence. Peptides corresponding to the BH3 region have been shown in several instances to adopt an α-helical structure and to bind into a hydrophobic groove on the surface of anti-apoptotic proteins. We are interested in how the interaction specificity of Bcl-2 family proteins is determined by sequence and structure and are exploring this using x-ray crystallography, mutational analysis and computational protein design. Using new computational methods for varying the backbone structure of α-helices, we have designed several novel ligands for the anti-apoptotic protein Bcl-xL. Solution binding studies confirm that many of these designed peptides bind with low- to mid- nanomolar affinity and have specificity profiles that differ from those of known native BH3s.

Selected Publications
Fu, X., Apgar, J., Keating. A.E. (2007) “Modeling backbone flexibility to achieve sequence diversity: The design of novel alpha-helical ligands for Bcl-xL” J. Mol. Bio. in press. doi:10.1016/j.jmb.2007.04.069

Grigoryan, G., Zhou, F., Lustig, S.R., Ceder, G., Morgan, D., Keating, A.E. (2006) “Ultra-fast evaluation of protein energies directly from sequence.” PLoS Comp. Biol., 2, e63.

Grigoryan, G., Keating, A.E. (2006) “Structure-based prediction of bZIP interaction specificity”, J. Mol. Bio. 355, 1125-1142.

Taylor, C., Keating, A.E. (2005) “Orientation and Oligomerization Specificity of the Bcr Coiled-coil Oligomerization Domain.” Biochemistry 44, 16246-16256.

Ali, M.H., Taylor, C.M., Grigoryan, G., Allen, K.N., Imperiali, B., Keating, A.E. (2005) “Design of a heterospecific, tetrameric, 21-residue miniprotein with mixed α/β structure.” Structure 13, 225-234.

Fong, JH, Keating, A.E., Singh M. (2004) “Predicting specificity in bZIP coiled-coil protein interactions,” Genome Biology 5(2): R11.

Newman, J.R.S., Keating, A.E. (2003) “Comprehensive identification of human bZIP interactions with protein microarrays,” Science 300, 2097-2101.

Search PubMed for Keating Lab publications.