Computational covalent warhead design: QM/ML modeling of acrylamide warhead reactivity toward cysteine targets

In low-molecular weight pharmaceutical development, covalent inhibition can offer many advantages over non-covalent inhibition such as longer residence times, reduced competition with endogenous substrates, and lower dosing. However, the innately reactive functional group that forms the covalent bond to the target, known as the covalent warhead, can lead to several unfavorable side effects like glutathione scavenging, CYP metabolism, and off-target binding. Despite these difficulties, recent interest in covalent inhibitors continues to grow as they can provide new pathways to drug problematic targets. Accordingly, a workflow to understand and control the reactivities of covalent warheads could mitigate unwanted side effects and greatly assist in covalent inhibitor development. While warhead reactivities are commonly measured with assays, a computational model to predict warhead reactivities could be useful for several aspects of the covalent inhibitor design process.

Studies have shown correlations between covalent warhead reactivities and quantum mechanic (QM) properties that describe important aspects of the covalent reaction mechanism. However, the models from these studies are often linear regression equations and can have limitations associated with their usage. Applications of machine learning (ML) models to predict covalent warhead reactivities with QM descriptors are not extensively seen in the literature. This study uses QM descriptors, calculated at different levels of theory, to train ML models to predict reactivities of covalent acrylamide warheads. The QM/ML models are compared with linear regression models built upon the same QM descriptors and with ML models trained on structure-based features like Morgan fingerprints and RDKit descriptors. Experiments show that the QM/ML models outperform both the linear regression models and the structure-based ML models. Literature test sets are used to demonstrate the power of the QM/ML models to predict reactivities of unseen acrylamide warhead scaffolds.

This QM/ML workflow is a fast and accurate means to predict the relative reactivities of covalent inhibitors with acrylamide warheads. The process can likely be tailored to model reactivities of other types of covalent warheads as well. QM/ML models generated in this manner may aid medicinal chemists in tuning covalent warhead reactivity and can ultimately expedite the design of new covalent inhibitors.