Abstract
The fragment-based drug design approach consists of screening libraries of fragment-like ligands, to identify hits that typically bind the protein target with weak affinity ( 100μM –5 mM). The determination of the protein–fragment complex 3D structure constitutes a crucial step for uncovering the key interactions responsible for the protein–ligand recognition, and for growing the initial fragment into potent active compounds. The vast majority of fragments are aromatic compounds that induce chemical shift perturbations (CSP) on protein NMR spectra. These experimental CSPs can be quantitatively used to guide the ligand docking, through the comparison between experimental CSPs and CSP back-calculation based on the ring current effect. Here we implemented the CSP back-calculation into the scoring function of the program PLANTS. We compare the results obtained with CSPs measured either on amide or aliphatic protons of the human peroxiredoxin 5. We show that the different kinds of protons lead to different results for resolving the 3D structures of protein–fragment complexes, with the best results obtained with the Hα protons.