Target identification in the cellular context is a critical step for both target-based and phenotypic drug discovery. Limited proteolysis coupled with mass spectrometry (LiP-MS) has emerged as a powerful technique for target deconvolution of small molecules or peptides in cell lysate without compound modification or genetic manipulation of cell lines [1,2,3,4]. Utilizing a non-specific protease under well-controlled conditions, LiP-MS exploits drug-induced structural alteration or steric hindrance on protein targets and utilizes quantitative mass spectrometry to probe over 250"000 peptides covering more than 9000 proteins in the proteome. To aid target ID, this workflow incorporates a 7-concentration dose response experiment and a machine-learning framework [1] to compute a LiP-score to rank identified target proteins, as well as predict potential binding site.

In this study, we present target deconvolution results on four small molecule compounds and one peptide compound with very distinct pharmacological profile: 1) staurosporine, a broad-specific kinase inhibitor; 2) roniciclib, a failed clinical-stage pan-CDK inhibitor; 3) selumetinib, an allosteric, non-ATP-competitive MEK 1 and 2 inhibitor; 4) NST-628, a newly developed non-degradating molecular glue inhibiting the RAS/MAPK pathway; and 5) Acetyl-calpastatin, a 24 amino-acid polypeptide inhibitor of calpain 1 and 2.

Target ID experiment with staurosporine and roniciclib yielded > 150 and 68 kinase targets were identified respectively. For the pan-kinase inhibitor roniciclib, both cell-cycle CDKs (CDK1/cyclin B, CDK2/cyclin E, CDK4 cyclin D) and transcriptional CDK9 were among the targets. For Selumetinib, LiP-score ranking identified MEK1 and MEK2 as the top two targets, and only kinases among a total of 23 targets. Furthermore, mapping the LiP-peptides from this experiment onto MEK1 showed that they are in close proximity to the selumetinib allosteric binding site on co-crystal structure. Last but not least, we assessed the target landscape of NST-628, a novel non-degrading molecular glue and identified MEK1 and MEK2 clearly as the primary targets. This finding is in line with orthogonal data obtained from X-ray crystallography [5].

Collectively, this data demonstrates that LiP-MS can be deployed to effectively identify protein drug targets and predict binding sites in complex cellular millieu independent of the compound"s mechanism of action and without compound modification or labeling. These capabilities make LiP-MS a powerful addition to the target deconvolution toolbox. In addition, LiP-MS protocol can be fully automated to enable small molecule target ID following phenotypic drug screening or at lead optimization stage with higher throughput.

[1] Beaton et al. Nature Comm 2020

[2] Hendricks et al. ACS Chem Bio 2021

[3] AACR 2022

[4] AACR 2023

[5] AACR-NCI-EORTC 2023