Mapping protein binding sites by photoreactive fragment pharmacophores

Abstract Fragment screening is a popular strategy of generating viable chemical starting points especially for challenging targets. Although fragments provide a better coverage of chemical space and they have typically higher chance of binding, their weak affinity necessitates highly sensitive and often resource-intensive biophysical assays. Here, we introduce a new screening concept that combines evolutionary optimized fragment pharmacophores with the use of a photoaffinity handle that enables high hit rates by LC-MS-based detection. The sensitivity of our screening protocol was further improved by a target-conjugated photocatalyst. We have assembled, synthesized, and screened 100 diazirine-tagged fragments against three protein targets of different tractability. Our targets included a conventional enzyme, the first bromodomain of BRD4, a protein-protein interaction represented by the oncogenic KRasG12D protein, and the yet unliganded N-terminal domain of the STAT5B transcription factor. We have discovered several new fragment hits against all three targets and identified their binding sites via enzymatic digestion, structural studies and modelling. Our results revealed that this protocol outperforms screening traditional photoaffinity fragments in better exploration of the available binding sites and higher hit rates observed for even difficult targets.