A study of prolyl hydroxylase domain 3 (PHD3): protein biochemistry and the development of inhibitors

Abstract

Prolyl hydroxylase domain 1-3 (PHD1-3) enzymes form a distinct subfamily of 2-oxoglutarate-dependent dioxygenases best known for their role in oxygen-sensing. Since the seminal discovery around the 2000s that PHDs transduce environmental oxygen into a cellular signal via the hypoxia-inducible factor (HIF) family, much attention has been focused on PHD2, the apparent key mediator of this response. However, subsequent reports of non-HIF substrates suggest that PHDs have broader roles beyond oxygen homeostasis. As some of these new substrates are implicated in disease, pharmacologically targeting the PHDs may offer a novel way to treat unmet medical needs. In this work, we undertook computational and biochemical approaches to analyze the biochemistry and identify inhibitors of the less-characterized isoform, PHD3. First, we established a high-yield, high-purity affinity chromatography protocol for recombinant maltose-binding protein (MBP)-tagged PHD3. This convenient method eliminates the need for downstream processing (eg. affinity tag removal, dialysis or gel filtration) to yield active PHD3. Second, we developed two novel isoform- and substrate-independent assays to measure PHD activity, taking advantage of the PHD hydroxylation reaction that consumes oxygen and alpha-ketoglutarate and produces carbon dioxide and succinate. The first is a sensitive metabolomics mass spectrometry-based succinate detection assay. The second is a high-throughput, colorimetric assay that utilizes 2,4-dinitrophenylhydrazine (2,4-DNPH) to derivatize a carbonyl group present in alpha-ketoglutarate but not succinate. This 2,4-DNPH assay was employed to: 1) detect PHD2 and PHD3 inhibition by several inhibitors that are in clinical trials, and 2) assess the ability of PHD3 to hydroxylate a newly-reported substrate, acetyl-CoA carboxylase 2 (ACC2), compared to the canonical HIF-1alpha substrate. We found that ACC2 is a bona fide substrate for PHD3. Finally, we developed a hybrid PHD3 inhibitor screening pipeline, where we performed a small molecule in silico screen using a PHD3 homology model followed by a biochemical screen of the selected library (Enamine 2) with the 2,4-DNPH alpha-ketoglutarate detection assay. We found that the in silico screen significantly enriched for biochemical hits, yielding several novel PHD3 inhibitors. In summary, this work offers a variety of new biochemical tools and platforms supporting the validation of PHD3 targets and discovery of potential therapeutics.