Protein Engineering vol. 16 no. 9 pp. 665-671, 2003
© 2003 Oxford University Press
How can free energy component analysis explain the difference in protein stability caused by amino acid substitutions? Effect of three hydrophobic mutations at the 56th residue on the stability of human lysozyme
1Institute for Protein Research, Osaka University, Yamadaoka, Suita, Osaka 565-0871 and 2Institute of Molecular and Cellular Biosciences, University of Tokyo, Bunkyo-ku, Tokyo 113-0032, Japan 4Present address: Riken Harima Institute, HTPF Kout, Mikazuki-chou, Sayon-gun, Hyogo 679-5148, Japan
3 To whom correspondence should be addressed. e-mail: kitao{at}iam.u-tokyo.ac.jp
To elucidate the molecular mechanism of thermal stability, it is essential to determine what are the major free energy components that contribute significantly to the total free energy difference caused by amino acid mutations. In this work, we carried out free energy calculations based on all-atom molecular dynamics simulations to investigate the effect of three hydrophobic mutations at the same position, I56A, I56V and I56F of human lysozyme. The calculated free energy differences are in good agreement with the experimental values in all cases. From free energy component analysis, we found that small changes in stability in the I56A and I56V mutants originate from the short-range Lennard–Jones interactions, whereas the I56F mutant is largely destabilized owing to the changes in the long-range electrostatic interactions. The calculated results are also compared with the free energy components determined by an empirical relationship based on the native-state structure and thermodynamic data. Although this relationship has been shown to be very successful in reproducing the stability changes caused by various amino acid substitutions in several proteins, the changes of stability in I56V and I56F mutants are not reproduced very well. By comparing the free energy components calculated by these two approaches, we showed that the effect of the long-range interaction on the stability changes may be underestimated in the empirical relationships when the structural change caused by mutation is relatively small, as in I56F. It is also suggested that estimation of the change in accessible surface area, 
ASA, may be overestimated if the structure around the mutation site in the denatured state is native-like, which would cause overestimation of the free energy change as in the case of I56V. Our results clearly show that the combined approach of the free energy calculation based on the all-atom molecular dynamics simulation and the empirical relationships is very useful for understanding the detailed mechanism of protein stability.
Received April 24, 2003; revised July 7, 2003; accepted July 18, 2003.