Mapping hydration water molecules in the HIV-1 protease/DMP323 complex in solution by NMR spectroscopy.

A tetrahedrally hydrogen-bonded structural water molecule, water 301, is seen in the crystal structure of nearly every HIV-1 protease/inhibitor complex. Although the urea oxygen of the designed inhibitor, DMP323, mimics and replaces water 301, other water molecules are seen in the protease/DMP323 crystal structure. As a first step toward understanding how water molecules may contribute to inhibitor potency and specificity, we have recorded water-NOESY and water-ROESY spectra of the protease/ DMP323 complex. Cross relaxation rates derived from these spectra, together with interproton distances calculated from the crystal structure of the complex, were used to classify the exchange cross peaks as follows: (A) a direct NOE with a water proton, (B) an indirect NOE with water through a labile protein proton, and (C) direct exchange of an amide proton with water. Type A and B cross peaks were analyzed using three models of water dynamics: (1) two-site exchange, with water molecules randomly hopping between bound and free states, (2) bound water with internal motion, and (3) free diffusion. Using the two-site exchange model to analyze the relaxation data of the type A cross peaks, it was found that the water molecules had short residence times, ca. 500 ps. in contrast with the > 9 ns residence time estimated for water 301 in the protease/P9941 complex [Grzesiek et al. (1994) J. Am. Chem. Soc. 116, 1581-1582]. The NMR data are consistent with the X-ray observation that two symmetry-related water molecules, waters 422 and 456, are bound at the DMP323 binding site. Hence, these water molecules may help to stabilize the structure of the complex. Finally, it was found that three buried and hydrogen-bonded Thr hydroxyl protons were in slow exchange with solvent. In contrast, it was found that the DMP323 H4/H5 hydroxyl protons and the Asp25/125 carboxyl protons, which form a buried hydrogen-bonded network at the catalytic site of the protease, are in rapid exchange with solvent, suggesting that solvent can penetrate into the buried protein/inhibitor interface on the millisecond to microsecond time scale.