Host–Guest Binding Constants Can Be Estimated Directly from the Product Distributions of Dynamic Combinatorial Libraries

Abstract

In the last decade, dynamic combinatorial chemistry has developed into a powerful method for the discovery of new sensor systems, catalysts, ligands for biomolecules, and synthetic receptors. Until now, determining the binding strength for individual library members has required their separation from the library and subsequent analysis. We now show that it is possible to obtain most host–guest binding affinities, even in 30+ component libraries, directly from product distributions by using a numerical data-fitting procedure. We have verified our method by using a threecomponent experimental system and showing that the binding constants computed from the behavior of the dynamic mixture agree well with those obtained independently by binding studies on the isolated library members. Dynamic combinatorial libraries (DCLs) of interconverting oligomers are made by mixing a set of building blocks that can reversibly combine with each other to give a dynamic equilibrium. The composition of the DCL is determined by the relative stability of all its members, so it is responsive to external stimuli that can affect these stabilities. Thus, addition of a guest molecule to a library of potential receptors should result in the stabilization and amplification of those library members that can form favorable interactions with the guest molecule. Ideally, the good receptors are amplified at the expense of the other oligomers. Quantifying their binding strength thus far entailed painstaking isolation of the selected compounds from the library. This process often requires switching off the exchange reaction through a substantial change in the pH value, or even covalent modification so that the protonation state or structure of the isolated species may no longer correspond to that of the species that was selected in the library. Also, under certain experimental conditions, the intuitive correlation between amplification and binding affinity can break down. Thus, there is a strong incentive to develop a method that allows the determination of binding affinities directly from the behavior of DCLs without requiring studies on isolated library members. We reasoned that the response of a DCL of potential receptors to changes in the concentrations of building blocks and guest molecules should contain enough information about the underlying host–guest binding constants that it should be possible to extract these by using a multivariable fitting procedure. We first tested this hypothesis on a simple set of seven libraries by using building block 1 (2 mm ; Figure 1) in the