Validation of Machine Learning-assisted Screening of PKC Ligands: PKC Binding Affinity and Activation.

Abstract

Protein kinase C (PKC) is a family of serine/threonine kinases, and PKC ligands have the potential to be therapeutic seeds for cancer, Alzheimer's disease, and human immunodeficiency virus infection. However, in addition to desired therapeutic effects, most PKC ligands also exhibit undesirable pro-inflammatory effects. The discovery of new scaffolds for PKC ligands is important for developing less inflammatory PKC ligands, such as bryostatins. We previously reported that machine learning combined with our knowledge of the pharmacophore yielded 15 PKC ligand candidates, but we did not evaluate their PKC binding affinities fully. In this paper, PKC binding affinities of four candidates were examined to assess their potential as PKC ligands and to validate machine learning-assisted screening. Although compound 3' did not bind to PKC C1 domains, 1a, 2', and 4a exhibited moderate PKC binding affinities, suggesting that machine learning-assisted screening is advantageous in identifying new PKC ligand scaffolds.