Dynamic analysis of the atomic vibrations of proteins, as exemplified by the binding of myristic acid to human serum albumin.

Prediction of the biological function of a protein from its three-dimensional structure is an important, still unsolved problem. A new approach to this objective, tried here, is use of crystallographic temperature factors, which contain the same information as IR and Raman spectra but lack their overlap problems. The hypothesis that atomic vibrations are evolutionally optimized for a particular function by adoption of collective modes governed by an attractor has been tested on 19 proteins with the result that strong correlation (r = 0.98) was found between the dimension of the attractor and the number of vectors needed to describe the function. The binding of five molecules of myristic acid (MA) to human serum albumin (HSA) at two sites accommodating two or three MA molecules, respectively, gave rise to four conformational changes in distinct regions. Two of these were located at the binding sites but the others occurred in segments far removed from the ligands both in the sequence and spatially. According to the statistical criteria employed, the conformational changes at the ligand-binding sites were not necessarily controlled by an attractor of low order, but the others were governed by one of dimension of 2-3. This was ascribed to entropic compensation. The results were tested using another ligand, an inhibitor of the BCL-2 family of proteins. The HSA underwent the same conformational changes with this ligand as with MA.