computer modeling in molecular biology.pdf

4 Molecular Dynamics and Free Energy Calculations 97Analyses of a number of “non-disruptive” mutations [89, 90, 107, 110, 113-1151where a large hydrophobic amino-acid was replaced by a smaller hydrophobicresidue, as in the Ile -, Ala case studies here, showed that the unfolding free energieswere highly variable, but that they were generally larger in magnitude than the correspondingdifference in transfer free energies. This can be attributed in part to theuse of transfer data obtained with different organic solvents [go, 1071. But it is morelikely to result from the fact that transfer experiments may not always be a goodmodel for protein folding. It is for example questionable that the protein interiorresembles non-polar or, even slightly polar, organic liquids. Proteins are moredensely packed than these liquids, with densities similar to those of crystals of smallmolecules [116]. The polymeric nature of proteins could furthermore severely constrainpacking re-arrangements needed to accommodate changes in residue size.Also, the environment of a given amino acid may differ from one protein to another,or in different sites within the same protein,Arguments in favor of these hypotheses have been recently provided. The analysisof six non-disruptive hydrophobic mutations in the core of phage T4 lysozyme [113],presents compelling evidence that the excess in stabilization free energy over thetransfer free energies observed in these mutants is linearly related to the size of thecavity created upon the mutation evaluated from the corresponding mutant crystalstructure. This suggests that the protein core may indeed differ from organic solventsby its reduced ability to adjust packing around the modified group. Rather convincingtheoretical arguments were also presented [117], that this excess can be attributedto contributions from cavity formation and protein reorganization processes.In view of the above considerations, what can be said about the Ile 96-+Alamutation in barnase? The measured change in unfolding free energy (3.214.0 kcal/mol) could concievable be in excess relative to the difference in the transfer freeenergies (1.5-3.11 kcal/mol) if the lower value (obtained for octanol) is considered.In this case the possibility that substitution of the bulky Ile sidechain by the smallerAla creates a cavity at the helix-sheet interface, should be examined. To this end,atomic volume calculations [54] were performed using coordinates from the endpoints of the simulation pathway. They show that a cavity of about 60-90 A3 isformed around residue 96 in the Ala containing mutant. Furthermore, rms atomicfluctuations were computed from several independent 30 ps vacuum moleculardynamics trajectories of the wild type and the mutant proteins. While these werefound to be similar on the average, fluctuations of mainchain and C, atoms ofresidue 96 were about twice as large in the Ala mutant, suggesting that the latter ismore mobile. These results suggests that the Ile 96 + ala mutant in barnase followsthe trends observed for other non-disruptive mutations. But given that this conclusionis based on simulated models, it must await conformation from a detailedanalysis of the mutant crystal structure.