New drug production, from target identification to marketing approval, takes over 12 years and can cost around $2.6 billion. Furthermore, the COVID-19 pandemic has unveiled the urgent need for more powerful computational methods for drug discovery. Here, we review the computational approaches to predicting protein–ligand interactions in the context of drug discovery, focusing on methods using artificial intelligence (AI). We begin with a brief introduction to proteins (targets), ligands (e.g. drugs) and their interactions for nonexperts. Next, we review databases that are commonly used in the domain of protein–ligand interactions. Finally, we survey and analyze the machine learning (ML) approaches implemented to predict protein–ligand binding sites, ligand-binding affinity and binding pose (conformation) including both classical ML algorithms and recent deep learning methods. After exploring the correlation between these three aspects of protein–ligand interaction, it has been proposed that they should be studied in unison. We anticipate that our review will aid exploration and development of more accurate ML-based prediction strategies for studying protein–ligand interactions.

Aldehyde dehydrogenase 4A1 (ALDH4A1) catalyzes the final steps of both proline and hydroxyproline catabolism. It is a dual substrate enzyme that catalyzes the NAD + -dependent oxidations of L-glutamate-γ-semialdehyde to L-glutamate (proline metabolism), and 4-hydroxy-L-glutamate-γ-semialdehyde to 4-erythro-hydroxy-L-glutamate (hydroxyproline metabolism). Here we investigated the inhibition of mouse ALDH4A1 by the six stereoisomers of proline and 4-hydroxyproline using steady-state kinetics and X-ray crystallography. Trans-4-hydroxy-L-proline is the strongest of the inhibitors studied, characterized by a competitive inhibition constant of 0.7 mM, followed by L-proline (1.9 mM). The other compounds are very weak inhibitors (approximately 10 mM or greater). Insight into the selectivity for L-stereoisomers was obtained by solving crystal structures of ALDH4A1 complexed with trans-4-hydroxy-L-proline and trans-4-hydroxy-D-proline. The structures suggest that the 10-fold greater preference for the L-stereoisomer is due to a serine residue that hydrogen bonds to the amine group of trans-4-hydroxy-L-proline. In contrast, the amine group of the D-stereoisomer lacks a direct interaction with the enzyme due to a different orientation of the pyrrolidine ring. These results suggest that hydroxyproline catabolism is subject to substrate inhibition by trans-4-hydroxy-L-proline, analogous to the known inhibition of proline catabolism by L-proline. Also, drugs targeting the first enzyme of hydroxyproline catabolism, by elevating the level of trans-4-hydroxy-L-proline, may inadvertently impair proline catabolism by the inhibition of ALDH4A1.

No abstract