Design, Synthesis, and Validation of a β-Turn Mimetic Library Targeting Protein–Protein and Peptide–Receptor Interactions

The design and synthesis of a β-turn mimetic library as a key component of a small molecule library targeting the major recognition motifs involved in protein–protein interactions is described. Analysis of a geometric characterization of 10,245 β-turns in the protein data bank (PDB) suggested that trans-pyrrolidine-3,4-dicarboxamide could serve as an effective and synthetically accessible library template. This was confirmed by initially screening select compounds against a series of peptide-activated GPCRs that recognize a β-turn structure in their endogenous ligands. This validation study was highlighted by identification of both nonbasic and basic small molecules with high affinities (Ki = 390 nM and 23 nM, respectively) for the κ-opioid receptor (KOR). Consistent with the screening capabilities of collaborators and following the design validation, the complete library was assembled as 210 mixtures of 20 compounds, providing a total of 4,200 compounds designed to mimic all possible permutations of 3 of the 4 residues in a naturally occurring β-turn. Unique to the design and because of the C2 symmetry of the template, a typical 20 × 20 × 20-mix (8,000 compounds prepared as 400 mixtures of 20 compounds) needed to represent 20 variations in the side chains of three amino acid residues reduces to a 210 × 20-mix, thereby simplifying the library synthesis and subsequent screening. The library was prepared using a solution-phase synthetic protocol with liquid–liquid or liquid–solid extractions for purification and conducted on a scale that insures its long-term availability for screening campaigns. Screening the library against the human opioid receptors (KOR, MOR, and DOR) identified not only the activity of library members expected to mimic the opioid receptor peptide ligands, but also additional side chain combinations that provided enhanced receptor binding selectivities (>100-fold) and affinities (as low as Ki = 80 nM for KOR). A key insight to emerge from the studies is that the phenol of Tyr in endogenous ligands bearing the H-Tyr-Pro-Trp/Phe-Phe-NH2 β-turn is important for MOR binding, but may not be important for KOR (accommodated, but not preferred) and that the resulting selectivity for KOR observed with its removal can be increased by replacing the phenol OH with a chlorine substituent further enhancing KOR affinity