Session Z41: Focus Session: Non-Covalent Protein Interactions

Novel Aspects of Hydrogen Bonding in Protein Function: Active Site Ionic Hydrogen Bonds

We use photoactive yellow protein (PYP), a bacterial photoreceptor, to explore novel aspects of the role of hydrogen bonding in protein function. PYP exhibits photochemical activity based on its ionized p-coumaric acid (pCA) chromophore, which is hydrogen bonded to Tyr42 and Glu46. We report that these active site ionic hydrogen bonding interactions cause unexpected molecular and functional properties of PYP. First, we describe a novel spectroscopic isotope effect (SIE) in which dissolving PYP in D2O causes a red-shift in its electronic absorbance spectrum. We assign this SIE to the ionic hydrogen bond between pCA and Glu46, which--in contrast to standard hydrogen bonds--is weakened upon H/D exchange. These findings extend the effects of H/D exchange from kinetic isotope effects to include shifts in absorbance spectrum, and illustrate the biological relevance of ionic hydrogen bonding to protein active sites. Secondly, we examine how the protein environment achieves the unusual strong preference of the pCA to remain ionized in the protein interior. We use the rescue of the Y42F mutant of PYP by incorporation of a trans-locked analog of pCA to dissect the contributions of active site hydrogen bonding to the large down-shift in the pKa of the pCA. Together, the Tyr42 and Glu46 hydrogen bonds to the pCA account for $\sim$80\% of this shift, which can be quantitatively explained by the loss of ionic hydrogen bonding upon pCA protonation from the solvent. Since ionic hydrogen bonds occur in many proteins, this mechanism of pKa tuning is likely to be of general relevance.   

Computational design of protein interactions: designing proteins that neutralize influenza by inhibiting its hemagglutinin surface protein

Molecular recognition underlies all life processes. Design of interactions not seen in nature is a test of our understanding of molecular recognition and could unlock the vast potential of subtle control over molecular interaction networks, allowing the design of novel diagnostics and therapeutics for basic and applied research. We developed the first general method for designing protein interactions. The method starts by computing a region of high affinity interactions between dismembered amino acid residues and the target surface and then identifying proteins that can harbor these residues. Designs are tested experimentally for binding the target surface and successful ones are affinity matured using yeast cell surface display. Applied to the conserved stem region of influenza hemagglutinin we designed two unrelated proteins that, following affinity maturation, bound hemagglutinin at subnanomolar dissociation constants. Co-crystal structures of hemagglutinin bound to the two designed binders were within 1Angstrom RMSd of their models, validating the accuracy of the design strategy. One of the designed proteins inhibits the conformational changes that underlie hemagglutinin's cell-invasion functions and blocks virus infectivity in cell culture, suggesting that such proteins may in future serve as diagnostics and antivirals against a wide range of pathogenic influenza strains. We have used this method to obtain experimentally validated binders of several other target proteins, demonstrating the generality of the approach. We discuss the combination of modeling and high-throughput characterization of design variants which has been key to the success of this approach, as well as how we have used the data obtained in this project to enhance our understanding of molecular recognition. References: Science 332:816 JMB, in press Protein Sci 20:753   

The power of simple hard-sphere models in protein structure prediction

There are several force-fields that are currently used to describe the potential energy of biological macromolecules such as proteins. These typically include many parameters, derived from a combination of statistical, experimental sources. These work on average to describe a protein, but the large number of parameters moves this description further away from a true physical understanding than is desirable. Our approach is to investigate to what extent simple hard sphere models can be used to model and predict the behavior of different aspects of protein structure. We present the results of specific calculations. The distributions of the side-chain dihedral angle chi1 of Val and Thr in proteins of known structure show distinctive features: Val side chains predominantly adopt dihedral angle, chi1, of 180, whereas Thr side chains typically adopt a dihedral angle, chi1, of 60 or 300. Several hypotheses have been proposed to explain these differences, including inter-residue steric clashes and hydrogen-bonding interactions. In contrast, we show that the observed side-chain dihedral angle distributions for both Val and Thr can be explained using only local steric interactions in a dipeptide mimetic. Our results emphasize the power of a simple physics-based approaches and their importance for future advances in protein engineering and design.   

Acetylation of LYS-16 of H4 Histone Tail May Sequester the Tail and Inhibit its Interactions with Neighboring Nucleosomes

Histone tails are highly flexible N terminal protrusions of histone proteins, which help to fold DNA into dense superstructures known as chromatin. On a molecular scale histone tails are poly-electrolites with high degree of conformational disorder, allowing them to function as bio-molecular ``switches,'' regulating various genetic regulatory processes via diverse types of covalent modifications. Because of being intrinsically disordered, the structural and dynamical aspects of histone tails are still poorly understood. Using multiple explicit solvent and coarse-grained MD simulations we have investigated the impact of the acetylation of LYS-16 residue on the conformational and DNA-binding propensities of H4 histone tail. The potential of mean force computed as a function of distance between a model DNA and histone tail center of mass showed a dramatic enhancement of binding affinity upon mono-acetylation of the H4 tail. The estimated binding free energy gain for the wild type is 2kT, while for the acetylated it reaches 4-5 kT. Additionally our structural analysis shows that acetylation is driving the chain into collapsed states, which get enriched in secondary structural elements upon binding to the DNA. We suggest a non-electrostatic mechanism that explains the enhanced binding affinity of the acetylated H4 tail. At last our findings lead us to propose a hypothesis that can potentially account for the celebrated chromatin ``fiber loosening effects'' observed in many experiments.   

Exploring copper chelation in Alzheimer's disease protein

Alzheimer's disease (AD) is a neurodegenerative disorder affecting millions of aging people in the U.S. alone. Clinical studies have indicated that metal chelation is a promising new approach in alleviating the symptoms of AD. Our study explores the as yet undetermined mechanism of copper chelation in amyloid-$\beta$, a protein implicated in AD. The structure of amyloid-$\beta$ is derived from experimental results and incorporates a planar copper-ion-binding structure in a semi-solvated state. We investigate the chelation process using the nudged elastic band method implemented in our {\it ab initio} real-space multigrid code. We find that an optimal sequence of unbonding and rebonding events as well as proton transfers are required for a viable chelation process. These findings provide fundamental insight into the process of chelation that may lead to more effective AD therapies.   

Transition-metal prion protein attachment: Competition with copper

Prion protein, PrP, is a protein capable of binding copper ions in multiple modes depending on their concentration. Misfolded PrP is implicated in a group of neurodegenerative diseases, which include ``mad cow disease'' and its human form, variant Creutzfeld-Jacob disease. An increasing amount of evidence suggests that attachment of non-copper metal ions to PrP triggers transformations to abnormal forms similar to those observed in prion diseases. In this work, we use hybrid Kohn-Sham/orbital-free density functional theory simulations to investigate copper replacement by other transition metals that bind to PrP, including zinc, iron and manganese. We consider all known copper binding modes in the N-terminal domain of PrP. Our calculations identify modes most susceptible to copper replacement and reveal metals that can successfully compete with copper for attachment to PrP.   

Steric clashes determine differences in side chain dihedral angle distributions: A study of Thr versus Val

With the long-term goal to improve the design of protein-protein interactions, we develop a simple hard sphere model for dipeptides that can predict the side-chain dihedral angle distributions of Val and Thr in both the $\alpha$-helix and $\beta$-sheet backbone conformations. We find that it is essential to include the non-polar hydrogens in the model; indeed interatomic clashes involving the non-polar hydrogens largely determine the form of side-chain dihedral angle distributions. Further, we are able to explain key differences in the side-chain dihedral angle distributions for Val and Thr from intra-residue steric clashes rather than inter-residue steric clashes or hydrogen bonding. These results are the crucial first step in developing computational models that can predict the side chain conformations of residues at protein-peptide interfaces.