Cyclic peptide library screening technologies show immense promise for identifying drug leads and chemical probes for challenging targets. However, the structural and functional diversity encoded within such libraries is largely undefined. We have systematically profiled the affinity, selectivity, and structural features of library-derived cyclic peptides selected to recognize three closely related targets: the acetyllysine-binding bromodomain proteins BRD2, -3, and -4. We report affinities as low as 100 pM and specificities of up to 10⁶-fold. Crystal structures of 13 peptide–bromodomain complexes reveal remarkable diversity in both structure and binding mode, including both α-helical and β-sheet structures as well as bivalent binding modes. The peptides can also exhibit a high degree of structural preorganization. Our data demonstrate the enormous potential within these libraries to provide diverse binding modes against a single target, which underpins their capacity to yield highly potent and selective ligands.

环肽库筛选技术在识别具有挑战性的目标的药物先导和化学探针方面显示出巨大的希望。然而，在这些库中编码的结构和功能多样性在很大程度上是不确定的。我们系统地分析了库衍生环肽的亲和力、选择性和结构特征，这些环肽被选择用于识别三个密切相关的目标：乙酰赖氨酸结合溴域蛋白 BRD2、-3 和 -4。我们报告的亲和力低至 100 pM，特异性高达 10 ⁶-折叠。13 种肽-溴结构域复合物的晶体结构显示出结构和结合模式的显着多样性，包括 α-螺旋和 β-折叠结构以及二价结合模式。肽还可以表现出高度的结构预组织。我们的数据证明了这些文库中提供针对单个靶标的多种结合模式的巨大潜力，这巩固了它们产生高效和选择性配体的能力。