Estimation of affinities of ligands in mixtures via magnetic recovery of target-ligand complexes and chromatographic analyses: chemometrics and an experimental model

Background: The combinatorial library strategy of using multiple candidate ligands in mixtures as library members is ideal in terms of cost and efficiency, but needs special screening methods to estimate the affinities of candidate ligands in such mixtures. Herein, a new method to screen candidate ligands present in unknown molar quantities in mixtures was investigated. Results: The proposed method involves preparing a processed-mixture-for-screening (PMFS) with each mixture sample and an exogenous reference ligand, initiating competitive binding among ligands from the PMFS to a target immobilized on magnetic particles, recovering target-ligand complexes in equilibrium by magnetic force, extracting and concentrating bound ligands, and analyzing ligands in the PMFS and the concentrated extract by chromatography. The relative affinity of each candidate ligand to its reference ligand is estimated via an approximation equation assuming (a) the candidate ligand and its reference ligand bind to the same site(s) on the target, (b) their chromatographic peak areas are over five times their intercepts of linear response but within their linear ranges, (c) their binding ratios are below 10%. These prerequisites are met by optimizing primarily the quantity of the target used and the PMFS composition ratio. The new method was tested using the competitive binding of biotin derivatives from mixtures to streptavidin immobilized on magnetic particles as a model. Each mixture sample containing a limited number of candidate biotin derivatives with moderate differences in their molar quantities were prepared via parallel-combinatorial-synthesis (PCS) without purification, or via the pooling of individual compounds. Some purified biotin derivatives were used as reference ligands. This method showed resistance to variations in chromatographic quantification sensitivity and concentration ratios; optimized conditions to validate the approximation equation could be applied to different mixture samples. Relative affinities of candidate biotin derivatives with unknown molar quantities in each mixture sample were consistent with those estimated by a homogenous method using their purified counterparts as samples. Conclusions: This new method is robust and effective for each mixture possessing a limited number of candidate ligands whose molar quantities have moderate differences, and its integration with PCS has promise to routinely practice the mixture-based library strategy.