In dynamic combinatorial chemistry, reversible self-assembly of chemical building blocks is used to generate libraries of compounds. Allowing the library to equilibrate in the presence of a target protein creates a driving force that favours the formation of library members that bind tightly to the target — a self-screening process that could, in theory, greatly accelerate lead discovery. However, although this approach has been successful in simple systems in which the best-binding inhibitor is sufficiently favoured to allow easy identification, the difficulty of distinguishing inhibitors with similar binding constants hinders the application of the technique to large libraries with greater potential for lead discovery. Now, writing in the Journal of the American Chemical Society, Cheeseman et al. propose a promising strategy for addressing this problem — irreversibly destroying the weak-binding inhibitors to leave the best-binding inhibitor.

The authors tested the potential of their approach by using a small, non-dynamic library of dipeptide inhibitors of carbonic anhydrase. Using two compartments separated by a protein-impermeable membrane — each initially containing a mixture of dipeptides at the same concentration — they showed that the presence of carbonic anhydrase in one compartment could concentrate the dipeptides in that compartment in proportions that reflect their relative binding constants. However, as expected, the concentration differences between dipeptides of similar binding strength were small. But when a protease that could cleave the dipeptides was added to the compartment without the carbonic anhydrase, the ratio of the best binder to weaker binders was significantly enhanced over time, as dipeptides that bind more strongly to carbonic anhydrase were protected to a greater extent from proteolytic hydrolysis.

The results were in line with the theoretical model that was proposed by the authors, providing proof-of-principle for the strategy, and holding promise for its extension to larger dynamic libraries. For the approach to be applicable to large libraries, however, it will be important for the destruction process to be non-specific to avoid biasing the library owing to differences in destruction rate rather than differences in binding to the target. This could be achieved, for example, by physical methods, such as adsorption to a solid phase.