Kinases are enzymes that transfer phosphoryl groups to a target molecule. In the case ofMAP kinases, they often form elab orate cascades in which each kinase trans fers a phosphoryl group to another enzyme that is itself a kinase. The transfer activates the second kinase, enabling it to bind to its target enzyme, transfer a phosphoryl group to activate that enzyme, and thus form a cascade. Signals from receptors on the surface of cells trigger these cascades, which, after a series of steps, phosphorylate proteins in the nucleus to produce changes in protein synthesis. The cascade allows signals from outside the cell to be amplified and coordinated, but the activ ity of each individual kinase needs to be carefully regulated so that the right signal reaches the nucleus. MAP kinases phosphorylate target pro teins at serine and threonine residues that are followed by proline, Goldsmith ex plained. However, short peptides con taining this motif don't bind to the kinase. Instead, the target protein must contain a docking-site sequence for the kinase to phosphorylate it. Goldsmith and her col leagues determined the structure of MAP kinase p38 in the presence of docking-site sequences previously identified by several laboratories. THE DOCKING SEQUENCES bind outside of the enzyme's active site in a groove on the enzyme's surface, Goldsmith found. This binding causes "unexpected confor mational changes, both locally in the peptide-binding groove and in the active site," Goldsmith said. By wrapping an "arm" around the outside of the enzyme, the target protein in some way seems to cause the active site to shift to become more accessible. The arm doesn't actually have to be at tached to the target enzyme, Goldsmith noted. The presence of a short peptide containing the docking sequence improves the phosphorylation of substrates lacking the docking site. It's not just the substrate that docks in to the kinase in a groove on its surface, Goldsmith said. The enzyme that activates the MAP kinase docks to the same site as well. Thus the enzymes that precede and follow the kinase in a particular cascade both interact with it in the same location. "It's a very interesting way to preserve path way specificity," Goldsmith said. AsTainer remarked, "Proteins like these are chemomechanical devices. They have the functional group chemistry that does reactions, but the rest of the protein is im portant, too." ■ HTTP://PUBS.ACS.ORG/CEN
HOW NOT TO FALL INTO A DEEP WELL Deep minima in potential energy surfaces are not inevitable traps, simulations show
A
N ASSUMPTION OF STATISTICAL
theories ofreaction rates is that a reactive system is drawn to wells—or minima—in the po tential energy surface. The sys tem, it is further assumed, stays long enough in a well to sample all its energy states. And through statistical mechanics, its mecha nism and kinetics can be described. Now, a clear example of a reaction that does not meet this expectation has been discovered by William L. Hase, a chem istry professor at Wayne State University, Detroit. With Lipeng Sun, a graduate stu dent, and Kihyung Song, a visiting chem istry professor from the Korea National University of Education, he has shown that the bimolecular nucleophilic substitution (S N 2) reaction of hydroxide ion with methyl fluoride to give methanol and fluor ide ion avoids a deep potential energy well in the reaction path to products [Science, 2 9 6 , 875 (2002)]. T h e study points to the need "to probe carefully to under stand how a reaction is occurring and to verify whether the statistical models are valid," Hase says. Hase and coworkers used direct dy namics simulations to simulate the motion of atoms on the potential energy surface. The technique is not new, but it has not been easy to perform before. "Ifou need two pieces of information to do these sim ulations," Hase explains. "%u need the po tential energy at every possible configura tion of the atoms for any possible set of Cartesian coordinates and the derivative of the potential energy with respect to each of the Cartesian coordinates at every pos sible point." Hase says these calculations have be come much easier with faster computers and computer programs such as Gaussian, developed by quantum chemists. "Quan tum chemistry algorithms are so efficient that all of the methodology of classical tra jectory simulations can be coupled direct ly with programs such as Gaussian, and we can use quantum chemistry theory directly to simulate the motion of atoms." Hase and coworkers found that a high level of quantum chemistry theory pre dicts a deep potential well corresponding
to an intermediate in which the departing fluoride is hydrogen bonded to methanol. The well is there, Hase says, but "90% of the time, the atoms don't see it." Instead, the transition state, {HO • • • CH 3 • • F}, usu ally dissociates directly to products. Avoidance of the deep well is due to the uneven distribution of energies among the reaction products as the transition state dissociates, Hase says. If the motions of the fluoride ion were coupled to the vi brational motions of the methanol prod uct, the coupling would have pulled the system into the well. As it happens, the translational energy of the fluoride ion is not drained to the vibrational motions of the methanol. The fluoride ion doesn't slow
NOT 'NSYNC The transition state of the reaction OH" + CH3F -> CH30H + F" dissociates directly (left), because products moving at different timescales preclude formation of a hydrogen bond (right). down. It flies away before the methanol has time to catch it. Hase suspects that many more exam ples of deep-well avoidance will be dis covered as more researchers apply the technique he and his coworkers have used. The place to look, he says, is systems where some motions are happening on a fast timescale and others are occurring on a longer timescale, for example, enzyme-cat alyzed reactions.-MAUREEN ROUHI C & E N / MAY 13, 2 0 0 2
45
