Ultrafast Dielectric Response of Proteins from Dynamics Stokes Shifting of Coumarin in Calmodulin Pascale Changenet-Barret, ²,‡ Christin T. Choma, § Edward F. Gooding, ² William F. DeGrado, § and Robin M. Hochstrasser* ,² Department of Chemistry and Department of Biochemistry and Biophysics, UniVersity of PennsylVania, Philadelphia, PennsylVania 19104-6323 ReceiVed: April 27, 2000; In Final Form: July 10, 2000 The contributions of protein motions to the dielectric response in solvent-inaccessible regions of calmodulin are examined. Two-pulse and three-pulse stimulated-emission experiments are used to examine the Stokes shift dynamics of coumarins in their ground and excited states. The results allow an evaluation of linear response theory. For coumarin 153, the ground- and excited-state solvation dynamics are similar to those for acetonitrile and methanol solvents. The solvation of coumarin 343 peptide in calmodulin is noticeably different in ground and excited states, indicating that the linear response picture is not appropriate in this case. The major difference is a more prominent fast component of solvation in the ground-state solvation dynamics. The ground-state energy gap correlation function is conjectured to be the best representation of protein relaxation, while that for the excited state seems to be influenced by the coupling of the protein modes to the large dipole created by excitation. Introduction A quantitative description of solvent effects in electron- transfer reactions has been the goal of numerous studies during the past twenty years. Marcus first pointed out that solvent reorganization is decisive in ET reactions. 1 The role of the solvent reorganization dynamics following the formation of ions or dipoles in polar media was established a few years later. 2-4 From the earliest experimental observations of solvation dynam- ics 4 and the theoretical approaches describing the solvent as a dielectric continuum, 2,3 it was clear that solvation dynamics are strongly correlated with the dielectric response of the solvent. Since these pioneering investigations, solvation dynamics have been intensely studied both through experimental works and through the development of several theoretical models (for reviews, see refs 5-9). Molecular dynamics (MD) simulations 10,11 which allow a molecular description of the solvent have contributed to the understanding of solvation dynamics. 10-15 After the perturbation of the solute charge distribution by electronic excitation, the solvent molecules relax toward a new equilibrium distribution with a lower free energy of solvation. MD gave evidence for the first time that polar solvation in room-temperature liquids is a bimodal process which involves inertial motions of solvent molecules (50-150 fs time scale) and slower motions in the picosecond regime, attributed to the diffusion of the solvent molecules. For example, it has been found theoretically and experimentally that for a small polar nonprotic solvent like acetonitrile, most of the energy of solvation is dissipated via the inertial motion of the solvent molecules. 14,16,17 In the theoretical description of the solvation process, the concept of linear response theory has been often used. 10,11,14 This approach is based on Onsager’s linear regression hypothesis and assumes that the relaxation of a macroscopic nonequilibrium system can be described by the spontaneous microscopic fluctuations of an equilibrium system. 18 This prediction is expected to be accurate when the perturbation induced by the charge shift in the electronically excited state is small. The solvation dynamics can then be considered independent of the solute. The solvation free energy surfaces are then expected to be the same for different electronic states of the solute. The linear response assumption has been tested with various MD simulations. 11-15,19 Its validity must depend on the charge and size of the solute and on the importance of the inertial component to the solvation process. One way to test experimentally the validity of the linear response theory with MD simulations is measuring the solvation dynamics in both the ground state and electronically excited state of a fluorescent charge-sensitive probe. In this article, we use pump-probe and pump-dump-probe experiments to characterize the excited-state and ground-state solvation dynam- ics of coumarin dyes in acetonitrile, methanol, and a protein matrix. Other pump-dump-probe experiments have been reported in the past in the gas phase or in coherent Raman processes, but only recently this technique has been extended to optical transitions in the solution phase in the femtosecond regime. 20-22 Pump-dump-probe experiments have the advan- tage of providing a direct observation of the nonequilibrated ground-state kinetics of the dye. In such experiments, a pump pulse produces a nonequilibrium population of excited states which, after solvent relaxation, are suddenly stimulated back onto the ground-state surface (“dumped”), where they are probed via their ground-state absorption. For this study, we chose coumarins 153 and 343 as the solvation probes. These coumarins were often employed as probes for solvation dynamics, 17,23-26 since they have rigid structures that prevent large contributions of intramolecular reorganization of the solute to the dynamical * Corresponding author. ² Department of Chemistry, University of Pennsylvania. ‡ Present address: Laboratoire de Photophysique Mole ´culaire du CNRS (UPR 3361), Ba ˆt 210 Universite ´ Paris-Sud, 91405 Orsay Cedex, France. § Department of Biochemistry and Biophysics, University of Pennsyl- vania. 9322 J. Phys. Chem. B 2000, 104, 9322-9329 10.1021/jp001634v CCC: $19.00 © 2000 American Chemical Society Published on Web 09/13/2000