The catalytic mechanism of vitamin K epoxide reduction in a cellular environment

Abstract

Vitamin K epoxide reductases (VKORs) constitute a major family of integral membrane thiol oxidoreductases. In humans, VKOR sustains blood coagulation and bone mineralization through the vitamin K cycle. Previous chemical models assumed that the catalysis of human VKOR (hVKOR) starts from a fully reduced active site. This state, however, constitutes only a minor cellular fraction (5.6%). Thus, the mechanism whereby hVKOR catalysis is carried out in the cellular environment remains largely unknown. Here we use quantitative mass spectrometry (MS) and electrophoretic mobility analyses to show that KO likely forms a covalent complex with a cysteine mutant mimicking hVKOR in a partially oxidized state. Trapping of this potential reaction intermediate suggests that the partially oxidized state is catalytically active in cells. To investigate this activity, we analyze the correlation between the cellular activity and the cellular cysteine status of hVKOR. We find that the partially oxidized hVKOR has considerably lower activity than hVKOR with a fully reduced active site. Although there are more partially oxidized hVKOR than fully reduced hVKOR in cells, these two reactive states contribute about equally to the overall hVKOR activity, and hVKOR catalysis can initiate from either of these states. Overall, the combination of MS quantification and biochemical analyses reveals the catalytic mechanism of this integral membrane enzyme in a cellular environment. Furthermore, these results implicate how hVKOR is inhibited by warfarin, one of the most commonly prescribed drugs.

Keywords: covalent intermediate; integral membrane enzyme; mixed inhibition; quantitative mass spectrometry; thiol oxidoreductase; vitamin K epoxide; vitamin K epoxide reductase; warfarin.

Figure 1

Discrepancy between previous catalytic model of hVKOR and its cellular redox states.A, conventional model of hVKOR catalysis (11). Previous studies assume that KO reduction initiates with reduced Cys132/Cys135. The reduction results in the oxidation of Cys132/Cys135 to form a disulfide. The catalysis involves an intermediate state in which Cys135 is covalently linked to 3-hydroxyl vitamin K. Electron transfer from Cys132-SH resolves this covalent complex. Subsequently, the 3-hydroxyl group is protonated. The leaving of water completes the KO to K reduction. B, the cellular redox states of hVKOR. SH: reduced cysteine, S–S: disulfide bond. Quantitative MS shows that the state with reduced Cys132/Cys135 (R state) constitutes only 5.6% of the total cellular hVKOR. The remainder of hVKOR contains either a partially oxidized (PO state; Cys132 forms a disulfide with Cys51 and Cys135 reduced) or fully oxidized (O state; Cys132-Cys135 disulfide) active site. The TM region of hVKOR is shown as green, blue, and purple cyclinders for the R, PO, and O states, respectively. The luminal region is shown in pink hemisphere. Cysteines circled in dashed line indicate minor fraction of a certain state. The catalytic activity of partially oxidized hVKOR is unclear from previous studies.

Figure 2

Capture of a covalent reaction intermediate using a Cys43Ala mutant.A, Top, the anti-flag immunoblot under the nonreducing condition (-DTT) shows that KO (10 μM) treatment increases the apparent MW of the Cys43Ala mutant. Bottom, under the reducing condition (+DTT, 100 mM), KO treatment cannot induce this MW change of Cys43Ala. B, explanation of trapping of Cys43Ala by the covalent complex. Left, a proposed catalytic pathway with hVKOR in PO state. Cys135 attacks KO to form the covalent intermediate that Cys135 is linked to 3-hydroxyl vitamin K (KOH). Electron transfer from Cys43 generates a reduced Cys132 that subsequently resolves the covalent complex to generate the reaction product, K. Right, the Cys43Ala mutation blocks the electron-transfer process required for resolving the covalent complex. C, trapping the covalent complex requires the free thiol group of Cys135. The additional Cys135Ser or Cys135Ala mutation abolishes the KO-induced mobility change of Cys43Ala. The immunoblots are conducted in the same way as in A.

Figure 3

Quantitative MS analysis of the cellular cysteine status suggests covalent complex formation and hVKOR conformational change. The NEM protection level of cysteines is compared for wild-type and mutant hVKOR with and without KO (10 μM) treatment. A, for the Cys43Ala mutant, Cys135 can be protected from NEM modification when Cys135 forms a covalent complex after the KO treatment (inset). Alternatively, Cys135 may be protected owing to the formation of the Cys132–Cys135 disulfide. Cys16 and Cys85 become protected owing to the reduced structural flexibility of hVKOR with the covalently bound complex; similar protection was observed with warfarin binding (13). B, wild-type hVKOR (WT). Cys43 and Cys135 become protected after KO treatment owing to the formation of the Cys43–Cys51 and Cys132–Cys135 disulfide bonds (inset). With excess KO driving the reaction and electron transfer from Cys43, most of the covalently complex is resolved in the wild-type hVKOR. C–E, control experiments with other cysteine mutants. See text for explanations. Eror bars are standard deviations. Two-way Student's t-test (∗∗p < 0.01, ∗p < 0.05, ns, not significant) was performed for each cysteine from samples with and without KO treatment. The errors are calculated from a combination of three biological repeats and MS analyses of different peptides containing the same cysteine (six peptides for Cys16; three peptides for Cys43 and Cys51; and one peptide for Cys85, Cys96, Cys132, and Cys135). Detailed analysis of the MS data are presented in Tables S1 and S2, Fig. S2, and the Supplementary Excel File.

Figure 4

Formation of the covalent complex reduces conformational flexibility of the hVKOR protein.Left, the TMs of hVKOR are flexible in absence of a bound ligand. Owing to this structural flexibility, Cys16 on TM1 and Cys85 on TM2 can be readily labeled by NEM. Right, upon warfarin binding or formation of the covalent complex with KO (orange hexagon), the TMs become less flexible, and Cys16 and Cys85 become protected from NEM modification.

Figure 5

Correlation between the cellular fraction of reduced Cys132/Cys135 and the cellular activity of hVKOR.A, cellular fraction of reduced Cys132/Cys135 in wild-type hVKOR and cysteine mutants. The cellular fraction of reduced Cys132/Cys135 is measured by quantitative MS analyses of peptides containing both cysteines. This fraction cannot be measured for Cys132Ala and Cys135Ala owing to these mutations. B, cellular activity of wild-type hVKOR and cysteine mutants. The activity of Cys43Ala and Cys51Ala roughly correlates with their content of reduced Cys132/Cys135 (shown in A). Compared with the Cys51Ala mutant, wild-type hVKOR has higher activity but lower content of reduced Cys132/Cys135. C, model explaining the cellular activity and cysteine status. Because Cys43Ala and Cys51Ala block the electron transfer via Cys43/Cys51, the fraction of reduced Cys132/Cys135 determines the activity of these mutants. In contrast, both the reduced Cys132/Cys135 and the partially oxidized state contribute to the cellular activity of wild-type hVKOR.

Figure 6

Mechanisms of hVKOR catalysis and warfarin inhibition in the cellular environment. The TM barrel is shown in green, blue, and purple for the hVKOR active site in R, PO, and O states, respectively. These states can be interconverted by electron-transfer steps via thiol-disulfide relays (curved arrows), by oxidation after reacting with KO, or by reduction with the involvement of other molecules. Identity of the reducing molecules remains unclear, which can be either partner proteins of hVKOR (29) or small molecules abundant in the ER, such as reduced glutathione (20). The hVKOR catalysis with KO and warfarin (W) inhibition occur at different redox states. The KO reduction is carried out with R and PO states, but not in O state. In contrast, warfarin preferably inhibits the PO and O states. The inhibition of the R state is known to be weak (13, 27). Because KO and warfarin target different states, warfarin shows mixed inhibition kinetics (16).