Warfarin inhibits VKORC1 by competitive slow tight binding

K. J. Czogalla1, A. Biswas1, K. Höning1, V. Hornung1,2, K. Liphardt1, M. Watzka1, J. Oldenburg1 (1Bonn, Germany, 2Munich, Germany)

Date: 16.02.2017,
Time: 11:00 - 12:00

Objective: Vitamin K 2,3-epoxide reductase complex subunit 1 (VKORC1) catalyzes the reduction of vitamin K quinone (K) and vitamin K 2,3-epoxide (K>O). This process is essential for further modification of vitamin K dependent coagulation factors leading to their biological activity. VKORC1 is also a drug target of 4-hydroxycoumarins, as warfarin is used in therapy and prevention of thrombosis. Over-anticoagulation results in serious bleedings but can be reversed by administration of K. Since current literature reports that warfarin binds to VKORC1 in a non-competitive manner, antagonism of warfarin by K presents a poorly understood mechanism.

Methods: Models of human VKORC1 in different oxidized/reduced states were generated based on the X-ray crystallographic structure of a prokaryotic VKOR enzyme using YASARA 13.1.1. Docking of warfarin, K and K>O were performed on these models using Autodock software. K reductase (VKR) and K>O reductase (VKOR) activity of predicted residues was measured in CRISPR/Cas9 engineered VKORC1 deficient HEK293T cells. Additionally, warfarin dose responses of mutated VKORC1 residues were measured.

Results: Best docking results for warfarin were generated in the oxidized form of VKORC1, where all loop cysteines required for active site regeneration and active site cysteines are oxidized (model-I). In model-I warfarin is accommodated in a big hydrophobic pocket with aromatic stacking interactions between the warfarin conjugated rings and the aromatic ring of Phe55 of VKORC1. For K and K>O best docking poses were generated when one loop cysteine is disulfide-bonded to an active site cysteine (model-II). Compared to the warfarin bound model, the naphthoquinone head groups of K and K>O interact with the opposite faces of the Phe55 residue. The in silico data was verified by in vitro measured dose-responses for F55A that resulted in complete resistance. Interestingly, F55G exhibited neither VKOR nor VKR activity.

Conclusion: Our results challenge the current concept of non-competitive warfarin inhibition since we demonstrate that K vitamers and warfarin share binding sites with F55 as central residue binding either the substrate or inhibitor. We suggest a competitive slow tight binding mechanism where warfarin binds VKORC1 in the oxidized form and is outcompeted by vitamin K when loop cysteines get reduced by oxidoreductase Partners.