Desi gn of a structurally diverse fragment library suitable for fragment-based chemogenomics

In this study we present the first comprehensive fragment-based chemogenomics analysis of eight protein targets belonging to four diverse target classes, GPCRs, LGICs, phosphodiesterases and kinases. Biochemical fragment screening with a chemically diverse library of 1010 fragment-like molecules, of which the construction and analyses is described, yielded diverse sets of fragment hits for all targets with hit rates varying between 1 and 10%. The hits include many novel (and selective) fragment scaffolds. Interestingly, fragment cross-reactivity was often larger between unrelated proteins (e.g., H4R and 5HT3R) than between proteins with higher sequence similarity (ADRB2 and H4R). Chemoinformatics analysis of fragment screening shows that fragments that bind to more than one target are more complex than selective fragments, but are not more hydrophobic as has been postulated for drug-like bioactive molecules. Finally, fragment bioaffinity profiles across related and unrelated protein targets are explored to identify ligand-affinity cliffs and molecular selectivity switches, demonstrating that fragment screening can be used to gain new insights into details at the atomic level of (selective) protein-ligand interactions.