Christodoulos Mitrogiannis, Sebastian Moes, Ottilia Andersson, S. Pauliina Turunen, Christian Sköld, Anders Karlén, Johan Wannberg, Oscar Verho

ACS Medicinal Chemistry Letters

DOI: 10.1021/acsmedchemlett.6c00385

Abstract

DNA-encoded chemical library technology offers a cost-effective method for screening billions of compounds in a single experiment. However, achieving broad coverage of lead-like chemical space remains challenging. Here, we describe the design and synthesis of a three-cycle, peptide-like DEL comprising 403,200 compounds, including an in silico-guided strategy for building block additions. A chemical space model based on 4,774 commercially available amino acid building blocks was used to identify amino acids underrepresented in the original design, in order to expand the library’s physicochemical diversity. Following split-and-pool synthesis, affinity selections against three targets were performed to assess library performance. Selections against carbonic anhydrase II successfully identified a known sulfonamide-containing pharmacophore. These results demonstrate that a computationally optimized, medium-sized DEL can effectively explore lead-like chemical space and identify specific binding motifs for purified protein targets.