A Proteomic Approach to Identification of Prodrug-Activating Serine Hydrolases: Activation of Valacyclovir

Abstract

Prodrug discovery and development in the pharmaceutical industry have been hampered by a lack of knowledge on activation pathways. Such knowledge would de-risk prodrug campaigns by enabling proper selection of preclinical animal models, prediction of pharmacogenomic variability, and identification of drug-drug interactions. Technologies for annotation of activating enzymes have not kept pace with the growing need for it. Activity-based protein profiling (ABPP) has matured considerably in recent decades, leading to widespread use in drug discovery. Here, we report the extension of competitive ABPP to prodrug-activating enzyme (PAE) identification in SILAC cell lysates using a modified fluorophosphonate (FP) probe and MS-based identification. This was followed by characterization of the new PAE in vitro, as well as in mouse tissue lysates and by single-pass intestinal perfusions (SPIP). Focusing on the ester prodrug valacyclovir (VACV), we modified the FP probe using an aminomethylbenzyl phosphonate ester structurally based on the serine protease inhibitor 4-(2-aminoethyl)benzenesulfonyl fluoride, which extinguishes VACV activation in Caco-2 cells, our proteome of interest, and inhibits BPHL, the only confirmed VACV PAE to date. The novel probe, AMB-FP, identified serine hydrolase RBBP9 as another activating enzyme in Caco-2 cells, which was validated via the selective inhibitor emetine. Selective concentrations of emetine abolished VACV activation in Caco-2 lysates, suggesting that RBBP9, not BPHL, is the major presystemic VACV-activating enzyme. Kinetic characterization of RBBP9 revealed a single-binding stoichiometry and a catalytic efficiency comparable to that of BPHL (kcat∙KM-1 = 104 mM-1∙s-1). Emetine inhibition data from VACV hydrolysis assays in jejunum and liver soluble lysates of wildtype (WT) and Bphl-knockout (KO) mice revealed that RBBP9 and BPHL are co-equal mediators of VACV activation in the liver, but that RBBP9 is the predominant PAE in the jejunum. Furthermore, the data indicated that these two enzymes are likely the only major enzymes involved in VACV activation in these tissues. Finally, SPIP data from WT mice with and without emetine pre-perfusion confirmed the RBBP9-dependence of VACV presystemic activation as portal vein samples showed >50% increase in unactivated VACV in emetine-treated mice. Importantly, this was the first evidence of RBBP9 VACV activation in a more physiologically relevant system than lysates. Based on these results, we envision that others might use the cABPP approach in the future for global, rapid, and efficient discovery of prodrug-activating enzymes in situations where traditional solutions are inadequate or prohibitively inefficient.