KNI-272 is a peptidomimetic transition state analog inhibitor, having very high specificity and binding affinity for the HIV-1 protease. In order to understand the interactions that enhance drug binding to the protease, we recorded 2D water/NOESY and water/ROESY spectra to identify water molecules that bind tightly to the protease/KNI-272 complex. Well-ordered water molecules are observed at the protease/inhibitor interface in the crystal structure of the complex that have short interproton distances to the Ile50/150, Ala28/128, and Asp29/129 amide protons. The cross peaks between these protein protons and water protons, observed in water/NOESY and water/ROESY spectra, provide strong evidence that these water molecules are present in the solution structure of the complex. Analysis of measured NOE and ROE cross relaxation rates indicates that, in solution, these water molecules have long residence times, at least 1 ns and possibly greater than 7 ns. The presence of long-lived hydration water molecules at the protein/inhibitor interface suggests that interactions involving these water molecules contribute to the potency of the inhibitor. Hence, consideration of the potential role of hydration water molecules in stabilizing protein/inhibitor structures could contribute to improved drug design and to a better understanding of the mechanisms of drug resistance.

The HIV-1 protease is a 22 kDa homodimeric protein essential for function of the AIDS virus, and protease inhibitors have been developed into effective HIV drugs. In order to better understand HIV-1 protease−inhibitor interactions, we have investigated amide backbone dynamics by correlated 1H−15N NMR spectroscopy. To date, HIV-1 protease/inhibitor complexes studied by NMR spectroscopy have been limited to C 2 symmetric structures, consisting of the protease bound to a symmetric inhibitor. Herein we report studies of the dynamics of HIV-1 protease complexed to KNI-272, a potent (K i ca. 5 pM), asymmetric inhibitor which lifts the chemical shift degeneracy of the protease monomers and allows us to ascertain if the individual protease monomers have significantly different backbone motions. Using isotope filtered/edited spectra of 15N/13C protease complexed with unlabeled KNI-272, together with distances derived from the protease/KNI-272 X-ray structure, we obtained monomer specific NMR signal assignments. We derived information about monomer dynamics from a Lipari-Szabo analysis of amide 15N T 1, T 2, and NOE values. Modeling the complex as an axially symmetric rotor yielded an average overall correlation time of 9.65 ns and an anisotropy, D ||/D ⊥, of 1.27. Over 90% of the backbone amide sites are highly ordered with the squared order parameter, averaged over all measured residues, being 0.85. High amplitude internal motions are observed in several loops in the protease, especially those in the elbows of the flaps, while millisecond to microsecond time scale motion is observed at the flap-tips. While these results are similar to those reported for complexes with symmetric inhibitors, we find differences in internal motions between several residues in the flap of one monomer and the corresponding residues on the other monomer. Residue Gly 149 has a significantly larger order parameter than Gly 49; in addition, the motions on the chemical exchange time scale contribute to the relaxation of Gly 152 and Phe 153 but not to the relaxation of Gly 52 and Phe 53. These differences in flexibility correlate with differences in interactions made by these residues with KNI-272, as seen in the crystal structure. We also find that the average of the order parameters measured for residues in monomer 1 is less than for monomer 2, a result that correlates with the observation that average B factor for these residues is less in monomer 2 than in monomer 1.