Abstract: Investments in genomics, functional genomics, machine learning and artificial intelligence have enhanced the identification and validation of novel targets. This has resulted in an increased proportion of intractable targets, which often lack tool compounds, lack known binding sites, contain intrinsic disordered domains and have no catalytic activity, thus posing a challenge for traditional lead generation approaches. The concomitant expansion in the number of therapeutic modalities provides opportunities to target biological pathways in completely novel ways, including new small molecule modalities such as PROTACs, molecular glues and direct degraders. Prosecuting these challenging targets requires screening in innovative ways to identify potential new mechanisms for disease intervention. CETSA (cellular thermal shift assay) provides an opportunity to identify small molecules that engage with the target protein in the more relevant context of the cellular environment. Within the high-throughput screening (HTS) centre at AstraZeneca, we have begun to implement HiBiT CETSA for large-scale screening. We have developed a fully automated HiBiT isothermal CETSA protocol, enabling screening of 25,000 compounds per day. So far, we have applied HiBiT CETSA to two intractable targets in succession, the first a scaffold protein and the second a transcription factor. Here, we describe the learnings from screening more than 1.5 million wells across these HTS campaigns. We describe how we used modelling in the absence of a tool compound to predict a range of hypothetical CETSA stabilisation events, which allowed us to select the optimal screening temperature for primary screening. We also optimised the screening concentration in order to reduce the false positive rate whilst maximising the chance of identifying true actives. These approaches were both implemented in the transcription factor screening campaign, and led to a 5-fold reduction in the primary screen active rate, whilst remaining sensitive to weak binders. We describe the application of an automated high-throughput HiBiT CETSA melt curve protocol, and compare and contrast this approach to concentration-response curves for hit triage. We highlight the learnings derived from the execution and analyses of these large screens, providing our perspective on the key considerations for a HiBiT CETSA screening cascade. Finally, we will summarise our findings by providing a vision for the future application of HiBiT CETSA in HTS, and the implications for early drug discovery.
.png)
