In Fragment Based Drug Discovery1 , structural data are generally needed to develop smaller hits into high-affinity lead compounds. Here, we present a fast and efficient NMR-based method for structure determination of low-affinity protein-ligand complexes rapidly enough to support a program of medicinal chemistry.
Selective isotope labelling of I, L and V methyl groups is well established2 as a means of obtaining information about large protein-ligand complexes. We have employed an alternative procedure which provides methyl assignments for A and T in addition to those of ILV3. This requires two protein samples for completion of the necessary chemical shift assignments: (i) a triple-labelled sample is employed to obtain the backbone and methyl resonance assignments using non-uniform sampling4(NUS) and requiring only low protein concentrations (0.3 – 0.4 mM); (ii) a 10% 13C-labelled protein is used for stereo assignments using CT-[13C,1H]-HSQC. Once the resonance assignments are obtained, a uniformly [13C,15N]-labelled sample is needed for 3D 13Cali-filtered methyl NOESY experiments, using similarly low protein concentrations to generate intermolecular NOEs for protein-ligand complexes of interest. We note that methionine assignments from 3D 13Cali [1H,1H]-NOESY data are only possible if the protein structure is available. Finally, sparse intermolecular NOEs and ambiguous interaction constraints derived from chemical shift perturbations are used to generate the structures using HADDOCK5 .
Using our protocol, data acquisition, analysis and determination of the structure of the complex can be achieved within 2-3 weeks where the resonance assignments are unknown and in under 1 week once the protein has been assigned. To illustrate the approach, we have solved structures of Escherichia coli DsbA (EcDsbA)6 and Vibrio cholerae DsbA (VcDsbA)7in complex with small molecules, both of which were not amenable to characterisation by X-ray crystallography. Broader analysis of structural data for a range of protein-ligand complexes indicates the widespread applicability of this approach.