Computational Modeling of pH-dependent Protein Structures

With the rising importance of proteomics research, current experimental methods of protein structure determination are too slow to map the millions of protein interactions observed in humans. In an effort to remedy this problem, computational methods of protein modeling are being developed. Changes in protein energetic and structural properties, caused by pH, are key to many interactions within the human body and other living organisms. However, current computational protein modeling methods, such as the Rosetta algorithm, cannot account for the effect of pH on the structural and energetic properties of protein-protein interactions. My goal was to add pH functionality to the Rosetta algorithm by building a routine to model the effects of pH on protein energy and structure. This routine uses a two-stage method: explicitly adding/removing hydrogen atoms on ionizable amino acid groups and modifying heavy atom partial charges to reflect the electrostatic effects of protonation/de-protonation. Alpha-helical hydrogel proteins, which are known to exhibit pH-dependency, were used as a preliminary test case to verify that this routine functions as intended. Computational results showed the electrostatic energy was 27% (28.69 kJ/mole) higher at low pH and 74% (78.65 kJ/mole) higher at high pH as compared to neutral pH. These results reflect experimental data; thus, the pH-routine has demonstrated that it is functional for these particular alpha-helical hydrogel proteins. Currently, a plan is being developed to assess the pH-routines accuracy on other pH-dependent structures. Additional ideas include accounting for pKa shifts in buried residues and pH-sensitive backbone movement. Ryan Harrison, 5-9-04 Baltimore Polytechnic Institute Baltimore Branch NAACP, 7009 Pg 2 of 8