Evidence That Warfarin Anticoagulant Action Involves T w o Distinct Reductase Activities*

The dithiothreitol-dependent vitamin K and vitamin K 2,3-epoxide hepatic microsomal reductase activities of warfarin-susceptible and warfarin-resistant rats were compared to gain insight into the role(s) of these activities in vitamin K metabolism and function, In microsomes from resistant rats, 3- to 4-fold more warfarin was required to produce 50% inhibition (I6o) of vitamin K reduction to vitamin K hydroquinone than in microsomes from susceptible rats. For the reduction of vitamin K 2,3-epoxide to vitamin K a 6-fold higher warfarin concentration was required. In microsomes from resistant rats, the ISo warfarin concentration re- quired to inhibit y-carboxylation of microsomal precursor protein was also 4-fold higher with vitamin K as substrate and was 6-fold higher with the epoxide as substrate than in microsomes from susceptible rats. Collectively, these data suggest that the vitamin K reductase contributes to the metabolism of vitamin K in intact rats and that warfarin inhibition of both the vitamin K and vitamin K 2,3-epoxide reductases is in- volved in its anticoagulant effect. Vitamin K is required for the post-translational formation of y-carboxyglutamyl residues from Glu residues in a limited number of proteins including clotting Factors 11, VII, IX, and X (1). The 02-dependent microsomal carboxylase responsible for this carboxylation event requires the

The dithiothreitol-dependent vitamin K and vitamin K 2,3-epoxide hepatic microsomal reductase activities of warfarin-susceptible and warfarin-resistant rats were compared to gain insight into the role(s) of these activities in vitamin K metabolism and function, In microsomes from resistant rats, 3-to 4-fold more warfarin was required to produce 50% inhibition (I6o) of vitamin K reduction to vitamin K hydroquinone than in microsomes from susceptible rats. For the reduction of vitamin K 2,3-epoxide to vitamin K a 6-fold higher warfarin concentration was required. In microsomes from resistant rats, the ISo warfarin concentration required to inhibit y-carboxylation of microsomal precursor protein was also 4-fold higher with vitamin K as substrate and was 6-fold higher with the epoxide as substrate than in microsomes from susceptible rats. Collectively, these data suggest that the vitamin K reductase contributes to the metabolism of vitamin K in intact rats and that warfarin inhibition of both the vitamin K and vitamin K 2,3-epoxide reductases is involved in its anticoagulant effect.
Vitamin K is required for the post-translational formation of y-carboxyglutamyl residues from Glu residues in a limited number of proteins including clotting Factors 11, VII, IX, and X (1). The 02-dependent microsomal carboxylase responsible for this carboxylation event requires the reduced form of vitamin K, vitamin KHz,' as a cofactor (2), and current evidence suggests that vitamin K 2,3-epoxide may be a product of this reaction (3). Vitamin KHz formation from vitamin K may be catalyzed by DT-diaphorase (EC 1.6.99.2) (4, 5) and/ or an uncharacterized microsomal reductase which can utilize * This research was supported in part by grants from the United States Public Health Service, Department of Health and Human Services and by The College of Agricultural and Life Sciences and the School of Pharmacy, University of Wisconsin-Madison. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ' The abbreviations used are: vitamin KH1, vitamin K hydroquinone; HPLC, high performance liquid chromatography; Ix,, concentration at which 50% inhibition occurs. dithiothreitol as cofactor in vitro (6).
Warfarin and many other 4-hydroxycoumarin drugs are potent inhibitors of vitamin K-dependent protein formation and have consequently found extensive use in clinical medicine and as rodenticides. Warfarin causes an accumulation of vitamin K 2,3-epoxide in liver and strongly inhibits a microsomal vitamin K 2,3-epoxide reductase (7).
The discovery of rats resistant to the anticoagulant effects of warfarin (8) has made available a valuable tool to study the action of this drug. Metabolism of warfarin or vitamin K is apparently not altered in these animals (9)(10)(11), although differences in warfarin binding to an unidentified protein in the microsomes have been demonstrated (12,13). The vitamin K 2,3-epoxide reductase of warfarin-resistant rats is much less sensitive to inhibition by warfarin than is the reductase of susceptible rats (14), but the DT-diaphorase, vitamin K-dependent carboxylase, and vitamin K epoxidase activities are not (15,16). These data support the conclusion that the genetic alteration which confers resistance is highly selective and affects only vitamin K 2,3-epoxide reductase (6,15,16).
The microsomal dithiothreitol-dependent, warfarin-sensitive vitamin K reductase described by Whitlon et al. (6) has not been investigated as thoroughly as other enzymes of the vitamin K metabolic cycle, partially because of difficulties associated with quantification of the vitamin KH, product of the reaction. Recently, an assay has been developed which quantifies microsomal vitamin KHz formation with either vitamin K or its 2,3-epoxide as substrate (17). We report here two separate lines of evidence which suggest that the anticoagulant action of warfarin is due to inhibition of the reduction of vitamin K as well as of vitamin K 2,3-epoxide.

EXPERIMENTAL PROCEDURES
Warfarin-susceptible Holtzman male rats (Holtzman Company, Madison, WI) and warfarin-resistant male rats (from a breeding colony maintained at the University of Wisconsin-Madison) were all fed Purina Lab Chow. Resistant rats were provided 1 mg of menadione sodium bisulfite/liter of drinking water. The rats were fasted overnight prior to decapitation. All further operations were performed at 5 "C.
The livers excised from five susceptible or five resistant rats were pooled, rinsed with 0.02 M Tris-HC1, 0.25 M sucrose buffer (pH 7.4), minced, and homogenized with a Potter-Elvehjem homogenizer in two volumes (by weight) of the same buffer. The cellular debris, nuclei, and mitochondria were removed by centrifugation a t 10, OOO X g for 10 min and the microsomes were sedimented from the supernatant solution by centrifugation at 105,000 X g for 60 min. The microsomes were resuspended by homogenization in the Tris-sucrose buffer, and aliquots were frozen at -80 "C until use. Microsomal protein concentrations were determined by the method of Bradford (18), with the use of Bio-Rad reagents and bovine serum albumin as standard.
Vitamin K2 (Sigma) was used without further purification. Vitamin K 2,3-epoxide was prepared by hydrogen peroxide oxidation of vitamin K and was purified by HPLC (19). Vitamin KH2 was prepared by dithiothreitol reduction of vitamin K at pH 8.5 (17), and warfarin (Calbiochem) sodium salt was prepared by the method of West et al. (20). Aqueous Emulgen 911 (Kao Atlas, Tokyo, Japan) solutions (9:L v/v) of vitamin K or vitamin K 2,3-epoxide (20 mg/ml) were prepared as described previously (17) and were diluted with water for use in metabolism studies.
Reaction mixtures contained, in order of addition: 1.35 ml of 200 mM Tris-HCI, 0.15 M KC1 buffer (pH 7.4) with or without warfarin Vitamin K, was used throughout this work. sodium salt; 0.6 ml of microsomes (10 mg of protein/ml); and 0.01 ml of vitamin K or vitamin K 2,3-epoxide (2 mg/ml). Each mixture was incubated at 25 "C with gentle agitation for 1 min, after which the reaction was initiated by addition of 0.04 ml of dithiothreitol solution (100 m). After incubation at 25 "C for an additional 5 min, vitamin substrates and metabolites were extracted with 4 ml of isopropanolhexane (1:l) by vortex mixing for 20 s. After brief, low speed centrifugation, a 1.5-ml aliquot of the solvent phase was evaporated to dryness at 30 "C under oxygen-free nitrogen, and the residue was redissolved in 0.2 ml of isopropanol. The vitamin components were separated by HPLC as described (17), except that the column was a Waters RCM C,S (5 mm, inner diameter), solvent A was water:acetonitrile:isopropanol (54:l) and solvent B was acetonitri1e:isopropa-no1 (41). Compounds were detected at 254 nm and were quantitated by integrated area comparison with external standards. Variations of this method are described in the figure legends.
Vitamin K deficiency was induced in rats, and liver microsomal preparations were obtained as previously described (16). Vitamin Kdependent carboxylation of endogenous proteins in suspensions of microsomes was measured in triplicate incubations by using the enzyme and substrate concentrations detailed above plus NaHI4C03 (25 pCi/d, 58 Ci/mmol; Amersham/Searle). The reactions were initiated by adding vitamin K or its epoxide, dissolved in ethanol, and were stopped after 10 min. The trichloroacetic acid-insoluble material was prepared for liquid scintillation counting as described previously (21).

RESULTS AND DISCUSSION
Warfarin was a potent inhibitor of the conversion of vitamin K to vitamin KHz in microsomal preparations from susceptible rats. The I, for this reaction was 1 p~ warfarin, compared to 3 to 4 p~ for resistant rat microsomes (Fig. 1A). Warfarin also inhibited the conversion of vitamin K 2,3-epoxide to vitamin K and vitamin KHz more effectively in susceptible than in resistant rat microsomes (Fig. 1B); the 1 5 0 warfarin concentrations were approximately 2 and 12 p~, respectively. These findings clearly demonstrate that (a) warfarin not only acts at the epoxide reduction step but can also inhibit reduction of vitamin K and ( b ) both reactions are less sensitive to warfarin in the resistant rat.
Numerous in vitro studies have indicated that vitamin KHz formation is a physiologically relevant step in vitamin Kdependent carboxylation (1). With respect to inhibition by warfarin, the vitamin K-supported carboxylation activity in microsomes from resistant rats required a 4-fold higher 1% warfarin concentration than did microsomes from susceptible rats ( Fig. 2A); for the vitamin K epoxide-dependent reaction, the difference in sensitivity was about 6-fold (Fig. 2B). These  The 15,, warfarin concentrations for vitamin K reduction, however, were approximately 12-fold lower than those for vitamin K-dependent carboxylation (1 uersus 12 p~ in susceptible and 3-4 uer-sus 47 p~ in resistant rat microsomes). A possible explanation for this difference is that in the presence of excess vitamin K substrate, the KHz formation rate is greatly in excess of that required for optimum activity of the carboxylase. The higher warfarin concentrations required for 50% inhibition in this assay reflect that difference. When vitamin K epoxide is used as substrate, two reduction steps are involved in generating the active hydroquinone form of the vitamin and its formation rate may be limiting for the carboxylase. Indeed, relative to vitamin K 2,3-epoxide, vitamin K promoted the incorporation of much higher concentrations of I4CO2 into microsomal precursor proteins of either rat strain ( Fig. 2 legend). Whatever the cause of the differences in the I, warfarin concentrations in the two assay methods, the data of both experiments support the concept that both vitamin K and vitamin K 2,3,-epoxide reduction are warfarin-sensitive and altered in the resistant rat.
With microsomes from both rat strains, metabolism of vitamin K epoxide to vitamin K and vitamin KHz as a function of time (Fig. 3) demonstrated that the only product formed during the initial 2 min was vitamin K. Thereafter, vitamin K formation attained a steady state rate, and the concentration of vitamin KHz, the second metabolite of the reaction, increased. These data are consistent with those obtained for microsomes of Wistar rats (22) and suggest that the reaction sequence is vitamin K epoxide to vitamin K to vitamin KHz, where vitamin K must attain a minimum concentration before it can serve as a substrate for further reduction. After this concentration is achieved, the vitamin KHz formation rate becomes equal to the vitamin K formation rate, preventing further accumulation of vitamin K. Vitamin K epoxide metabolism to vitamin KHz is therefore an ordered process in which vitamin K must interact at a site either within the same enzyme or on another enzyme prior to its reduction to hydroquinone.
Microsomes of susceptible and resistant rats had similar profiies of vitamin K 2,3-epoxide metabolism to vitamins K and KH, over the time span investigated, although higher metabolite concentrations were produced by microsomes of susceptible rats at the 20-pM concentration of vitamin K epoxide used. Warfarin slows the rate of vitamin KHz formation both by decreasing the concentation of available vitamin K and by inhibiting the conversion of vitamin K to vitamin KH,. The combined effect produces a pronounced decrease in vitamin KHz formation (Fig. 3A).
In microsomes from resistant rats, 3 PM warfarin did not inhibit vitamin K formation from the epoxide as completely as did 1 p~ warfarin in microsomes from susceptible rats. Thus, the microsomes of resistant rats have higher concentrations of vitamin KH,. Since the conversion of vitamin K to KH, in these microsomes is also less sensitive to inhibition by warfarin, quantities of vitamin KHz are further increased, relative to those in microsomes of susceptible rats. Thus, after 15 min of reaction, 1 PM warfarin in microsomes of susceptible rats produced an 84% reduction of vitamin KHz formation, while 3 warfarin in microsomes of resistant rats reduced vitamin KHz f o~a t i o n by only 26%. If the assumption is made that both the vitamin K and vitamin K 2,3-epoxide reductase activities are normally required to maintain tissue vitamin KHz concentrations, warfarin insensitivity at both sites will confer greater resistance than would be predicted from the extent of inhibition of either reaction. Indeed, relative to susceptible rats, resistant rats are 50 t.o 200 times less sensitive to warfarin.
Previous knowledge regarding the importance of vitamin K 2,3-epoxide reductase as the physiologically important site of action of warfarin has been derived from comparisons of vitamin K metabolism in susceptible and resistant rats. The evidence presented here clearly indicates that vitamin KHz arises from vitamin K 2,3-epoxide through a vitamin K intermediate and that this second reduction step is both warfarin-sensitive and altered in the resistant rat. The physiologic importance of the NADH-dependent reductases of microsomes cannot be determined from these data, but both dithiothreitol-dependent steps are clearly involved in the action of warfarin as an inhibitor of vitamin K-dependent reactions.
The nature of the physiologic reductant replaced by dithiothreitol in these reactions is not known. Neither do these data clearly indicate whether both reductive steps are catalyzed by the same site on a single enzyme, by separate sites on a single enzyme, or by two enzymes. The low probability of a genetic alteration affecting two enzymes might argue against the last possibility unless there is some protein-protein interaction between these enzymes. A clear understanding of the factors which mediate vitamin K and vitamin K 2,3epoxide reduction will probably come only from efforts to purify these membrane-bound activities.