Cn. Hodge et al., CALCULATED AND EXPERIMENTAL LOW-ENERGY CONFORMATIONS OF CYCLIC UREA HIV PROTEASE INHIBITORS, Journal of the American Chemical Society, 120(19), 1998, pp. 4570-4581
One important factor influencing the affinity of a flexible ligand for
a receptor is the internal strain energy required to attain the bound
conformation. Calculation of fully equilibrated ensembles of bound an
d free Ligand and receptor conformations are computationally not possi
ble for most systems of biological interest; therefore, the qualitativ
e evaluation of a novel structure as a potential high-affinity ligand
for a given receptor can benefit from taking into account both the bou
nd and unbound (usually aqueous) low-energy geometries of the Ligand a
nd the difference in their internal energies. Although many techniques
for computationally generating and evaluating the conformational pref
erences of small molecules are available, there are a limited number o
f studies of complex organics that compare calculated and experimental
ly observed conformations. To assess our ability to predict a priori f
avored conformations of cyclic HIV protease (HIV-1 PR) inhibitors, con
formational minima for nine 4,7-bis(phenylmethyl)-2H-1,3-diazepin-2-on
es I (cyclic ureas) were calculated using a high temperature quenched
dynamics (QD) protocol. Single crystal X-ray and aqueous NMR structure
s of free cyclic ureas were obtained, and the calculated low-energy co
nformations compared with the experimentally observed structures. in e
ach case the ring conformation observed experimentally is also found i
n the lowest energy structure of the QD analysis, although significant
ly different ring conformations are observed at only slightly higher e
nergy. The 4- and 7-benzyl groups retain similar orientations in calcu
lated and experimental structures, but torsion angles of substituents
on the urea nitrogens differ in several cases. The data on experimenta
l and calculated cyclic urea conformations and their binding affinitie
s to HIV-1 PR are proposed as a useful dataset for assessing affinity
prediction methods.