Hydroxy fatty acids and their derivatives represent a class of compounds of growing interest in many scientific areas. The resolution of hydroxyl fatty acids racemates by enzymatic acetylation was proposed as a hypothetical way to obtain enantiomerically enriched samples. To better understand the experimental behavior, and in order to identify the fast reacting enantiomer, a computational study was planned taking into account two steps:
1 tetrahedral intermediates study;
2 3D-QSAR prediction of enantioselectivity.
It should be noted that hydroxystearic acids are very different from the substrates used for training the original 3D-QSAR model, but at the same time this investigation provides information on the restrictions for the applications of 3D-QSAR quantitative predicting models. In agreement with the lipase biological role, it seems that hydroxyfatty acids are accommodated better in the acylic pocket, with the carboxylic group toward the catalytic machinery, rather than in the alcohol pocket. Indeed, the presence of the hydrophilic hydroxy group in the middle of the hydrophobic chain translates in scarce enantiodiscrimination. As a consequence, the best poses coming out from docking simulations were not compatible with observed reactivity.
Experimental data clearly show that CALB is able to accept such hindered substrates, despite a very low catalytic efficiency. MD simulations show some interesting behavior of the enzyme, which is able to enlarge the active site funnel for accommodating such hindered substrates.
The prediction of enantioselectivity is based on the previously developed PLS model, which was not trained with such hindered secondary alcohols like hydroxystearic acids. By applying the 3DQSAR model, the enantioselectivity is overestimated in respect to experimentally measured E. Since the predicted -log E is negative, the predicted fast reacting enantiomer is the R one.
The following conclusions can be drawn:
• The fast reacting enantiomer was correctly predicted
• Enantioselectivity was overestimated due to the diversity between hydroxystearic acids acetyl esters and the data set used to train the PLS model.
• Sterically hindered substrates require some analysis of possible “induced fitting” within the active site
• In order to obtain accurate quantitative predictions of E it would be necessary to include in the training set of the 3D-QSAR model more hindered secondary alcohols.
Nevertheless, this is the first attempt to get to enantioenriched hydroxystearic acids by the use of enzyme resolution and two hydroxyacids, namely (R)-7-HSA and (S)-9-HSA were obtained with about 55% e.e.
