A hierarchical screening approach to enantiomeric separation

Enantioselective UHPLC Screening Combined with In Silico Modeling for Streamlined Development of Ultrafast Enantiopurity Assays

Enantioselective chromatography has been the preferred technique for the determination of enantiomeric excess across academia and industry. Although sequential multicolumn enantioselective supercritical fluid chromatography screenings are widespread, access to automated ultra-high-performance liquid chromatography (UHPLC) platforms using state-of-the-art small particle size chiral stationary phases (CSPs) is an underdeveloped area. Herein, we introduce a multicolumn UHPLC screening workflow capable of combining 14 columns (packed with sub-2 μm fully porous and sub-3 μm superficially porous particles) with nine mobile phase eluent choices. This automated setup operates under a vast selection of reversed-phase liquid chromatography, hydrophilic interaction liquid chromatography, polar-organic mode, and polar-ionic mode conditions with minimal manual intervention and high success rate. Examples of highly efficient enantioseparations are illustrated from the integration of chiral screening conditions and computer-assisted modeling. Furthermore, we describe the nuances of in silico method development for chiral separations via second-degree polynomial regression fit using LC simulator (ACD/Labs) software. The retention models were found to be very accurate for chiral resolution of single and multicomponent mixtures of enantiomeric species across different types of CSPs, with differences between experimental and simulated retention times of less than 0.5%. Finally, we illustrate how this approach lays the foundation for a streamlined development of ultrafast enantioseparations applied to high-throughput enantiopurity analysis and its use in the second dimension of two-dimensional liquid chromatography experiments.

Modeling Enantiomeric Separations as an Interfacial Process Using Amylose Tris(3,5-dimethylphenyl carbamate) (ADMPC) Polymers Coated on Amorphous Silica

Chiral high-performance liquid chromatography (HPLC) is commonly performed to isolate the biologically active enantiomer of a drug from the ineffective or even harmful ones. Understanding the molecular-level recognition that underlies this process is necessary for trimming down the very large number of possible combinations of chiral stationary phases, solvent systems, and other experimental HPLC conditions, a particularly important consideration when only small quantities of the racemate are available. Fully atomistic molecular dynamics (MD) simulation is a useful tool to provide this molecular-level understanding and predict experimental separation factors under a given set of conditions. To predict the chiral separation results for drug enantiomers by amylose tris(3,5-dimethylphenyl carbamate) (ADMPC) chiral stationary phase, we design a model of multiple ADMPC polymer strands coated on an amorphous silica slab. Using various MD techniques, we successfully coat ADMPCs onto the surface without losing the structural character of the backbone in the presence of the solvent system. Not only is this model more representative of the polymer surface on a solid support that is encountered by the enantiomers, but it also provides more opportunities for chiral molecules interacting with ADMPC, provides the possibility for large drug molecules to interact with two polymer strands at the same instant, and provides better agreement with experiment when we use the overall average quantities as the predictive metric. For a better understanding of why some metrics are better predictors than others, we use charts of the distribution of hydrogen-bonding lifetimes for various donor-acceptor pairs that contribute to the interaction events determining the relative retention times for the enantiomers. We also examine the contribution of ring-ring interactions to the molecular recognition process and ultimately to the differential retention of enantiomers. The results are more consistent than previous models and resolve the problematic case of two drugs, thalidomide and valsartan.