A ligand-observed mass spectrometry approach integrated into the fragment based lead discovery pipeline

Abstract

In fragment-based lead discovery (FBLD), a cascade combining multiple orthogonal technologies is required for reliable detection and characterization of fragment binding to the target. Given the limitations of the mainstream screening techniques, we presented a ligand-observed mass spectrometry approach to expand the toolkits and increase the flexibility of building a FBLD pipeline especially for tough targets. In this study, this approach was integrated into a FBLD program targeting the HCV RNA polymerase NS5B. Our ligand-observed mass spectrometry analysis resulted in the discovery of 10 hits from a 384-member fragment library through two independent screens of complex cocktails and a follow-up validation assay. Moreover, this MS-based approach enabled quantitative measurement of weak binding affinities of fragments which was in general consistent with SPR analysis. Five out of the ten hits were then successfully translated to X-ray structures of fragment-bound complexes to lay a foundation for structure-based inhibitor design. With distinctive strengths in terms of high capacity and speed, minimal method development, easy sample preparation, low material consumption and quantitative capability, this MS-based assay is anticipated to be a valuable addition to the repertoire of current fragment screening techniques.

Figure 1. Fragment library screening based on the ligand-observed LC/MS approach.

(A) Scheme of the MS-based fragment screening workflow. The protein is incubated with a mixture of the fragment library, ligand-bound complexes are purified by ultrafiltration and the dissociated ligands are identified and quantified by LC/MS. Positive fragment binding is indicated by its S/N ratio > 10. (B) Results of the primary screen of a 384-member fragment library in a single run. Solid and open circles designate data points acquired in the positive and negative modes of MS analysis respectively. Circles above the threshold line denote positive hits. S/N ratios are averages of experimental duplicates. (C) Results from the secondary screen of 7 subdivided fragment cocktails, each consisting of 50 or 86 members. A total of 12 hits were identified in this run. S/N ratios are averages of experimental duplicates. (D) Binding degrees of individual fragments measured from the 12-hit mixture at different protein/ligand ratios (P/L) while the protein concentration was held constant in incubation. Ten fragments were validated to bind to NS5B and their binding degrees indicate a relative order of binding affinity. Error bars denote s.d. from experimental duplicates.

Figure 2. Quantitative assessment of binding affinity of fragment hits.

(A) Chemical structures, estimated dissociation constants (K_d) and ligand efficiency (LE) of the 10 hits. K_d calculation is based on ligand concentration measurement from the third-round validation assay (see Methods and Results for details). (B) S/N factors measured in the validation assay and SPR responses of the ten fragment hits. S/N factors from two independent experiments are shown. In SPR analysis, the corrected response above 5 RU indicates positive binding and the ranking order is a measure of relative binding affinity.

Figure 3. X-ray crystal structures of five fragments found to bind the allosteric site at the palm pocket (A, B, C, D) or at the thumb pocket (E).

Crystal structures of bound fragments 114 (A), 204 (B), 117 (C), 328 (D) and 162 (E) are shown. The Fo–Fc omit electron density maps are shown as a yellow mesh contoured at 3σ around the fragments. Right panels in A, B, C, D and E depict the amino acids found to interact with the fragments, represented by dashed black lines. Carbon atoms are depicted by gray, oxygen by red, nitrogen by blue and sulfur by yellow colors.

Figure 4. Determination of binding specificity of fragment 204 to NS5B in solution.

(A) LC-MS chromatograms of fragment 204 detected from incubations containing the wild-type or specific mutants of NS5B, or from the protein-free control experiment. M414 and C366 are two residues observed to interact with the fragment in the crystal structure. M423 is speculated to have no interaction with the fragment. (B) Relative quantification for fragment 204 based on its MS intensity from (A) implies its interactions with M414T and C366A were significantly impaired whereas interaction with M423T was not, an evidence of binding specificity at these two sites. MS intensity percentages of fragment 204 from the mutant incubation relative to the wild-type incubation (defined as 100%) were averages from duplicate measurement and s.d. are shown accordingly.

Figure 5. A fragment screening cascade incorporating the ligand-observed MS approach for efficient fragment hit identification.

The MS approach shown in this study can play a major role at the stages of preliminary screens as well as hit validation and K_d estimation due to its supreme speed, selectivity, sensitivity and quantitative capability.