Fragment-based drug discovery (FBDD) has become prevalent in the pharmaceutical industry as a way to identify molecular scaffolds that can be developed efficiently into lead compounds. By screening small molecular weight fragments (~100-300 Da), the researcher can sample more chemical space with low affinity molecular frameworks that bind protein targets of interest. Subsequently, these “hits” can be combined chemically resulting in a higher affinity lead drug molecule. Label-free biophysical platforms such as biosensors, NMR, and X-ray crystallography are commonly employed to screen fragment libraries for binders. The BiOptix 404pi biosensor based on Surface Plasmon Enhanced Common Path Interferometry (SPE-CPI) is an important and useful tool in the screening of fragment libraries due to its excellent detection capabilities, low sample consumption, and ability to collect kinetic or steady-state data in real time. We demonstrate below the screening of 330 fragments from the commercially available Maybridge Ro3 Diversity Fragment Library against two proteins, bovine carbonic anhydrase II (CAII) and NeutrAvidin (NA).
The running buffer in all experiments was phosphate-buffered saline with 0.05% Tween 20 (PBS-T), pH 7.4 with 2% DMSO. All fragments were injected for 12 seconds (s) and dissociation was followed for 12 s. No regeneration of the surfaces was required. Sensorgram data were processed and fit in Scrubber 2.0 software. The steady-state titrations for the confirmatory studies on the NA fragment hits were fit in Prism 7 to a steady-state model with a baseline offset. CAII was from Sigma and NA was from Thermo Fisher. Fragments were purchased from Maybridge and consisted of 330 fragments chosen randomly from the Ro3 Diversity Fragment Library.
Results and Discussion
Pre-screen to identify fragments binding to the sensor surface. All 330 fragments were injected over bare and activated and blocked CMD200m surfaces to eliminate fragments that bound to the carboxymethyldextran hydrogel before the primary binding screen to target protein was performed. The BiOptix 404pi was run in 4×1 mode for this large initial screen since no reference surface is required to identify sticky compounds. Fragments that bound to the sensor surfaces were found by taking a report point 58 seconds after the end of the 12 s injection. Any fragment that had >10 RU of binding signal remaining at 58 s was deemed a sticky binder and was excluded from the subsequent binding screen to target protein. Figure 1A and B show the fragments that had response units >10 RU (shown in blue). Table 1 shows that 23 fragments bound to the bare surfaces while only 19 fragments bound to the activated and blocked surfaces. In addition, 95% of the sticky fragments (shown in bold) observed binding to the activated and blocked surfaces were the same as those seen binding to the bare surfaces. Five additional fragments were also excluded from subsequent binding screens because they were insoluble in PBS-T, 2% DMSO and are given in Table 2. For all subsequent screens, activated and blocked reference surfaces were used since fewer fragments bound to those surfaces and were a more proper reference surface since the protein-immobilized surfaces had been activated and blocked.