Highly Parallel Translation of DNA Sequences into Small Molecules

A large body of in vitro evolution work establishes the utility of biopolymer libraries comprising 10(10) to 10(15) distinct molecules for the discovery of nanomolar-affinity ligands to proteins.[1,2,3,4,5] Small-molecule libraries of comparable complexity will likely provide nanomolar-affinity small-molecule ligands.[6,7] Unlike biopolymers, small molecules can offer the advantages of cell permeability, low immunogenicity, metabolic stability, rapid diffusion and inexpensive mass production. It is thought that such desirable in vivo behavior is correlated with the physical properties of small molecules, specifically a limited number of hydrogen bond donors and acceptors, a defined range of hydrophobicity, and most importantly, molecular weights less than 500 Daltons.[8] Creating a collection of 10(10) to 10(15) small molecules that meet these criteria requires the use of hundreds to thousands of diversity elements per step in a combinatorial synthesis of three to five steps. With this goal in mind, we have reported a set of mesofluidic devices that enable DNA-programmed combinatorial chemistry in a highly parallel 384-well plate format. Here, we demonstrate that these devices can translate DNA genes encoding 384 diversity elements per coding position into corresponding small-molecule gene products. This robust and efficient procedure yields small molecule-DNA conjugates suitable for in vitro evolution experiments.