A spectroscopic and computational study of stereochemistry in 2-hydroxymutilin

Abstract

A computational and spectroscopic study of two epimers of 2-hydroxymutilin was conducted to obtain assignments for the configuration of the carbon at the 2-position. Gas-phase structural models for ( R)- and ( S)-2-hydroxymutilin were optimized using density functional theory (DFT). These models were validated using internuclear distances measured from 1H nuclear Overhauser effect (NOE) build-up rates in CDCl 3 solution. The models were subsequently used to predict NMR parameters using the hybrid B3LYP density functional. After linear scaling, the agreement between predicted gas-phase and observed solution-state chemical shifts was sufficient for relative stereochemical assignments, with an average absolute error of 0.16 and 1.47 ppm for 1H and 13C shifts, respectively. 1H– 1H and 1H– 13C coupling constants were also predicted using the B3LYP functional with the aug-cc-p-VTZ-J basis set and compared to experimental values (measured by J-resolved and gradient J-HMBC experiments). The predicted coupling constants were within 10% of their experimental values in most cases, and like the chemical shielding, allowed for relative stereochemical assignments of the isomers. DFT prediction of coupling constants gave more useful information than empirical parameterizations. Although just one chiral center changes between the two isomers, with no consequent conformational change and only two affected NOE restraints, predictable effects on 20 different chemical shielding and J-coupling values were observed. Including these values in the stereochemical assignment allows for additional confirmation of the stereochemical results, using data already available in 1D 1H spectra and 2D 1H– 13C correlation spectra. Electronic and vibrational circular dichroism (ECD and VCD) spectra of the two isomers are also measured and compared with theoretical predictions. VCD was especially useful for determining the absolute stereochemistry of 2-hydroxymutilin and was found to be superior to ECD in this role. The combination of computational and spectroscopic work is shown to be a useful tool in the analysis of unknown mutilin derivatives or other compounds, such as those present as minor impurities or metabolites in pharmaceutical samples.