The Prediction and Interpretation of Protein pKa’s using QM/MM

2012-02-26T13:32:28Z (GMT) by Jan Halborg Jensen
<p>NSF proposal 7/15/02 - 6/30/06</p> <p>The objective of the proposed research is to establish a QM/MM methodology for the prediction and interpretation of protein pKa's that have unusual values and/or are difficult to model with current methodologies. The methodology will be applied to three proteins (turkey ovomucoid third domain, ubiquitin, and xylanase) for which detailed experimental studies of the factors that influence pKa's have been performed.</p> <p><br>A three-layered computational (QM/MM/LPBE) methodology for protein pKa prediction will be developed. The ionizable residue and its immediate environment will be treated by ab initio electronic structure (QM) methods (including energy minimization and harmonic vibrational analysis). The rest of the protein will be treated with a polarizable, multipole-based electrostatic model (MM) derived specifically for each protein by separate ab initio calculations. The bulk solvent will treated by a very accurate solution of the linearized Poisson-Boltzmann equation (LPBE).</p> <p><br>The successful application of a molecular modeling methodology, which combines an ab initio description of the ionizable residue and its immediate environment (including crystallographic water molecules) and accurate representation of the rest of the protein and bulk solvent, to pKa prediction will present a significant contribution to computational pKa prediction. Heretofore, computational protein pKa prediction has been primarily focussed on a priori prediction of all pKa values within a protein. This requires an efficient, and therefore relatively simple, description of intermolecular interactions. This approach is clearly adequate for the majority of ionizable groups in a protein. However, a more accurate, and therefore more computationally demanding, method is also needed to rationalize select pKa's where traditional approaches fail. Such a method will provide very detailed information about the forces that govern protein pKa's, which will aid in the improvement of current pKa prediction methods. The combined use of this new method with current methods will contribute significantly to the experimental design of proteins with greater stability or new functions.</p> <p><br>The proposed research is at the intersection of molecular physics, quantum chemistry, and structural biology, and will therefore offer important cross-disciplinary training to the postdoctoral associate, graduate student, and undergraduate students involved. Their training is further enhanced by tight collaboration with experimental research groups. The resulting scientists are thus well versed in both the mathematical foundations and practical aspects of quantum chemistry (including algorithm development and parallelization) and macromolecular modeling within the context of an interdisciplinary research environment.</p> <p><br>Some of the methodological advances will be incorporated as computational experiments in an introduction to molecular modeling class for senior undergraduate and graduate students taught by the PI, as has been done in the past. Furthermore, all algorithmic developments will be implemented in the quantum chemistry program GAMESS, which is distributed free of charge to the scientific community.</p>