Ligand-observed FBS methods
Fragment or ligand-observed strategies are the first choice given a known protein target. These techniques rely on observation of changes to ligand spectra upon protein – ligand interactions. Any number of ligands can be screened, most commonly sets or libraries of 10-1000 compounds are used either in the form of a fragment based library or as separate samples of e.g. natural products, drug candidates or peptides. The Swedish NMR Centre offers access to two main fragment libraries; the Maybridge Ro3 diversity fragment library core set containing 800 fragments and the Bionet 19F fragment library obtained from Key Organics containing 428 fragments, with compounds in both of these libraries adhering to ‘rule of 3’ and without reactive groups or PAINS. There is also the possibility to conduct screens using any user supplied library or set of compounds.
An FBS screen starts with quality control of the fragments. Customarily, each substance is solubilized in DMSO-d6, diluted with buffer and a subsequent proton NMR spectrum is then sufficient to establish the integrity, solubility and purity of each substance. These spectra serve as reference spectra for each individual fragment. With the library data in hand, fragments are pooled together and added to the target protein in 10-fold or higher excess. Ten or more fragments may then be pooled in each sample. This allows for low amounts of protein to be consumed compared to the protein target observed approaches (see below), and spectra to be recorded without the need of protein isotope labeling. In addition, the signal line widths are not limited by the size of the protein target, and ligand-observed methods are thus applicable to large molecular weight systems. The protein – ligand mixes are analyzed by one or a combination of the following techniques: water-ligand observed via gradient spectroscopy (Water-LOGSY), saturation transfer difference (STD) or T2 or T1,rho relaxation experiments. Screening using 19F-detected experiments have emerged as an interesting complement to 1H screening because the simplicity of the spectra and wide 19F chemical shift range reduces the risk of overlap in fragment cocktails. The large difference in chemical shift between the free and bound states and large CSA of 19F also result in a larger binding response in T2-filtered experiments compared to 1H, hence lowering the required amount of protein needed even further.
Individual hit fragments should be validated either by use of a combination of the techniques above, with runs of individual fragments and, when applicable, in conjunction with a known inhibitor. The Swedish NMR Centre is currently using the 'Bruker FBS tool' to analyze these large-scale data sets in a systematic fashion. The FBS screen results may be used to extend, link or modify the fragments to obtain lead molecules with higher binding affinities, which in turn can be subjected to a range of protein target-observed methodologies.