VKORC1 mutations detected in patients resistant to vitamin K antagonists are not all associated with a resistant VKOR activity

Authors


Virginie Lattard, USC 1233 INRA – Vetagro Sup, 69280 Marcy l’Etoile, France.
Tel.: +33 4 78 44 24 11; fax: +33 4 78 87 05 16.
Email: v.lattard@vetagro-sup.fr

Abstract

Summary.  Background: The VKORC1 gene codes for the VKORC1 enzyme, which is responsible for the reduction of vitamin K epoxide into vitamin K. VKORC1 enzyme is the target of vitamin K antagonists (VKA). Twenty-eight rare single mutations in the VKORC1 coding sequence have been reported from resistant patients receiving unusually high doses of VKA to achieve therapeutic anticoagulation.

Objectives: It has been suggested that these mutations are responsible for the resistant phenotype, while biochemical consequences of these mutations on the VKORC1 enzyme have not yet been evaluated. Therefore, the aim of this study was to investigate the causality of the VKORC1 mutations in the resistance phenotype.

Methods: Wild-type VKORC1 and its spontaneous mutants were expressed in Pichia pastoris and susceptibility to VKA was assessed by the in vitro determination of kinetic and inhibition constants.

Results and Conclusions: The in vitro analysis revealed that six mutations only (A26P, A41S, V54L, H68Y, I123N and Y139H) were associated with increase in Ki, suggesting their involvement in the resistance phenotype observed in patients. A41S and H68Y led to selective resistance, respectively, to indane-1,3-dione and 4-hydroxycoumarine derivatives. The other mutations did not increase the Ki. Furthermore, 10 mutations (S52L, S52W, W59L, W59R, V66M, V66G, G71A, N77S, N77T and L128R) led to an almost complete loss of activity. These results suggest the existence of other resistance mechanisms.

Introduction

Vitamin K antagonists (VKAs) are widely prescribed in the prevention and treatment of thromboembolic disorders. Warfarin, a derivative of 4-hydroxycoumarin, is the most used VKA worldwide and is prescribed to 2 million new patients in the USA each year. The number of dispensed outpatient prescriptions for warfarin increased by 45%, from 21 million in 1998 to nearly 31 million in 2004 [1]. In France, fluindione, an indane-1,3-dione derivative, is traditionally prescribed in more than 80% of patients. Thrombotic and hemorrhagic accidents associated with the use of these drugs represent the first cause of iatrogenic accidents, according to the EMIR study in 2007 (Adverse Effects of Drugs, Incidence and Risk) from AFSSAPS (Agence Française de Sécurité des Produits de Santé). Roughly one in five patients are hospitalized for bleeding within 6 months of starting the drugs and lethal cases account for 0.6% of the treated patients [1]. VKAs are difficult to use because of a narrow therapeutic index and a wide variability of dosage necessary to achieve stable anticoagulation. This variability is partly due to basic physiological parameters such as age and bodyweight, but also to co-occurring disorders, and drug and food interactions. Nonetheless, 30 to 50% of the dosage variability might be explained by genetic polymorphisms in CYP2C9 involved in VKA metabolism and by genetic polymorphisms in the promoter of VKORC1 [2–6]. Even if genotyping of VKORC1 and CYP2C9 has been suggested as a prerequisite to optimize dosage adjustment with VKAs [7,8], the correlation between genotype and dose requirement is complicated. Indeed, many allele combinations of CYP2C9 and VKORC1 are possible, resulting in the need for various doses of VKA.

The VKORC1 encodes the vitamin K epoxide reductase (VKORC1) [9,10]. VKAs exert their anticoagulant activity by inhibiting this enzyme in a non-competitive manner [11]. The function of VKORC1 is to regenerate vitamin K quinone and hydroquinone (K and KH2) from vitamin K 2,3-epoxide (K > O), a byproduct of the vitamin K-dependent gamma carboxylation reaction (Fig. 1) [12]. Inhibition of VKORC1 by VKAs limits the amount of KH2 available for the carboxylation reaction and results in partially carboxylated vitamin K-dependent blood clotting factors. The VKORC1 is a 163 amino acid integral membrane protein that contains a C132XXC135 redox motif located in the fourth transmembrane domain [13,14].

Figure 1.

 The vitamin K cycle. GGCX, gamma-glutamyl-carboxylase; NQO1, NADPH quinone oxidoreductase 1; VKOR, vitamin K epoxide reductase.

More than 26 missense mutations of the VKORC1 have been found to be associated with a substantially higher dosage requirement that cannot be explained by demographic or environmental factors (Table 1). These mutations are located either in the transmembrane domains TM1, TM3 and TM4, or in the large intraluminal loop between TM1 and TM2 [15]. Detected in patients with total or partial resistance to VKAs, these mutations are considered as linked to this phenotype [10,16]. Nonetheless, biochemical and structural consequences of these mutations of VKORC1 still remain to be elucidated and additional study is warranted to know if genetic abnormalities of VKORC1 can explain VKA resistance. Therefore in order to characterize the relationship between genetic variation in VKORC1 and the VKA resistance phenotype, we conducted a systematic analysis of the functionality of the VKORC1 mutated enzymes expressed as membrane-bound proteins in Pichia pastoris.

Table 1.   Genetic variations in the coding sequence of human VKORC1 detected in patients requiring a high dose of a vitamin K antagonist
MutationPositionNumber of patientsResistance toDose (mg day−1)Stable anticoagulationRef
  1. High-dosage threshold values : phenprocoumone (P) 3.0 mg day−1, acenocoumarol (A) 3.5 mg day−1, warfarin (W) 7.1 mg day−1, fluindione (F) 19.8 mg day−1, for mean age patients [23].

  2. TM, transmembrane domain; LL, luminal loop; het, heterozygous state; hom, homozygous state; ∼, depending on patient.

A26PTM11 (het)W20No[21]
F100No
A26TTM11 (het)W6No[23]
L27VTM11 (NI)W7Yes[31]
F60No
H28QTM11 (het)P10Yes[23]
V29LTM12 (het)W14?[10,23]
P26Yes
A34PLL1 (het)W27Yes[28]
D36YLL16 (het/hom)W14 ± 6[3,21–23,32]
F45No
A7No
P10 ± 6Yes
D36GLL1 (het)W20Yes[23]
A41SLL1 (het)W16?[2]
V45ALL1 (het)W45No[10]
S52LLL1 (het)P9No[33]
S52WLL1 (het)P10Yes[23]
V54LLL2 (het)W21 ± 16[21,22]
F60No
A8No
S56FLL1 (het)P15No[23]
R58GLL1 (het)W34?[10]
W59RLL3 (het)A9 ± 2No[33,34]
P9 ± 0No
W59CLL1 (het)P11No[23]
W59LLL1 (het)P15No[23]
V66MLL18 (het)W30 ± 7Yes[21–23,28,29]
P10 ± 1
V66GLL1 (het)P8Yes[23]
H68YLL1 (het)[24]
G71ALL1 (het)P6No[23]
N77SLL1 (het)P9No[23]
N77YLL1 (het)W25Yes[23]
I123NTM31 (het)P21No[23]
L128RTM316 (het)W44 ± 5No[4,10,21,22]
F80No
A13 ± 3No
P30No
Y139HTM41 (het)P9No[23]

Material and methods

Chemicals

K1 (Phylloquinone) was converted to K > O according to Tishler et al. [17]. Purity was estimated by LC/MS and was higher than 99%.

Plasmid preparation

The coding sequence corresponding to the human VKORC1 fused with a c-myc tag in its 3′-extremity was optimized for heterologous expression in yeast (i.e. by exchanging rare codons in the target gene for codons that are more frequently used in yeast) and synthesized by GenScript (Piscataway, NJ, USA). The synthesized nucleotide sequence included EcoRI and XbaI restriction sites at its 5′- and 3′-extremities, respectively. This nucleotide sequence was subcloned into pPICZ-B (Invitrogen, Cergy Pontoise, France) and sequenced on both strands. Construction of amino acid substituted mutants of VKORC1 was carried out using pPICZ-VKORC1 as a template with the QuikChange site-directed mutagenesis kit (Agilent Technologies, Massy, France). Mutants were systematically checked by sequencing, and the various mutants were individually expressed in P. pastoris.

Heterologous expression and subcellular fractionation of recombinant yeast cells

Heterologous expressions of VKORC1 proteins in Pichia pastoris and microsome preparations were performed as described previously [18].

Immunoblot analysis

Expression level quantification of VKORC1 proteins was determined by Western blotting. Microsomal proteins were separated on 12% SDS-polyacrylamide gel electrophoresis, transferred onto Immobilon-P membranes and probed with anti-c-myc antibodies (Invitrogen). The resulting immunocomplexes were visualized using alkaline phosphatase-conjugated anti-mouse immunoglobulins as secondary antibodies and a BCIP/NBT solution. Quantification of the stained bands was performed by densitometry using the Scion Image software. The relative intensity (RI) of the signal was correlated with the quantity of microsomal proteins, as shown in Fig. 2(A), and the relationship was linear from 0 to 10 μg for microsomal proteins containing WT-VKORC1 (Fig. 2B) or from 0 to 50 μg for microsomal proteins containing mutated VKORC1 that was weakly expressed.

Figure 2.

 Semi-quantitative analysis of hVKORC1 mutant by Western blotting. (A) Immunoblot analysis of microsomal proteins (lanes 1 to 5, respectively, 2.5, 5 10, 20 and 30 μg) of recombinant yeast expressing WT-VKORC1. (B) Relationship between the amount of microsomal proteins containing WT-VKORC1 and the relative intensity of the signal detected. (C) Semi-quantitative analysis of VKORC1-A26P in the microsomal fraction. From lanes 1 to 4, the membrane fractions of recombinant yeast expressing WT-VKORC1 (lanes 1 to 4, respectively, 1.25, 2.5, 5 and 10 μg) were loaded; from lanes 5 to 8, the membrane fractions of recombinant yeast expressing VKORC1-A26P (lane 5s to 8, respctively 10, 5, 2.5, 1.25 μg) were loaded.

To evaluate the expression level of VKORC1 proteins, the expression of wild-type VKORC1 (WT-VKORC1) was designated as the basal expression. For the quantification of all the mutants, the same unique pool of yeast microsomes containing WT-VKORC was used. Therefore, its expression factor was by definition 1. The expression level of the mutated VKORC1 was evaluated by comparison with the expression of WT-VKORC1. For this purpose, various amounts (from 0 to 10 μg) of microsomal proteins containing WT-VKORC1 and various amounts (depending on the expression level) of microsomal proteins containing one of the mutated VKORC1 were analyzed on the same Western blot (Fig. 2C). Two linear relations (RI = a × quantity of microsomes loaded) were obtained, the first one for microsomes containing WT-VKORC1 (characterized by a specific slope aWT), the second one for microsomes containing the mutated VKORC1 (characterized by a slope amut). Ratio amut/aWT allowed us to determine the expression factor characterizing the expression level of the mutated VKORC1 in the microsomal fraction compared with the expression level of the WT-VKORC1.

Vitamin K epoxide reductase activity (VKOR) assays and kinetics

Analysis of VKOR activity and reaction products by liquid chromatography-mass spectrometry was performed as described previously [18]. Km, Vmax and Ki values were obtained from at least three separate experiments performed on three different batches of protein. The estimation of Km and Vmax values was achieved by the incubation of at least nine different concentrations of K > O (from 0.003 to 0.2 mM) to the standard reaction. Incubations were performed in duplicate. Data were fitted by nonlinear regression to the Michaelis-Menten model using the R-fit program [19].

In order to evaluate the inhibiting effect of warfarin, acenocoumarol or fluindione (Fig. 3) on VKOR activity, Ki were determined after addition of various concentrations of anticoagulant to the standard reaction in the presence of increasing amounts of K > O (from 0.003 to 0.2 mM) using anticoagulant concentrations from about 0.05 to 20 × Ki. Data were fitted by nonlinear regression to the non-competitive inhibition model using the R-fit program [19].

Figure 3.

 Chemical structures of the VKA compounds used for the inhibition of the wild-type and mutated hVKORC1 expressed in Pichia Pastoris. (A) warfarin; (B) acenocoumarol; (C) fluindione.

Results

Expression of VKORC1 and its mutants

To assess the functional properties of the WT-VKORC1 and its 25 spontaneous mutants, these proteins were overexpressed as c-myc-fused proteins in P. pastoris. Depending on the mutation, the levels of expression of the recombinant protein in the microsomal fraction were different. For hVKORC1-V29L, absence of expression by the yeast was noticed. This absence of expression prevented more characterization of this mutated protein. All other mutated proteins were efficiently expressed in Pichia pastoris with the same expected molecular mass of approximately 20-kDa and the expression factor was comprised between 0.1 and 2 compared with the expression of WT-VKORC1.

The ability of each membrane protein to catalyze the reduction of K > O to K was determined. Fig. 4 compares the VKOR activity of 25 mutated VKORC1 proteins after normalization of their expression at 200 μM of K > O. Ten mutations (S52L, S52W, W59L, W59R, V66M, V66G, G71A, N77S, N77T and L128R) led to an almost complete loss (> 98%) of VKOR activity towards K > O.

Figure 4.

 Specific activity of WT-VKORC1 and its mutants. Enzyme activity was evaluated in the presence of 200 μM of KOX and 0.25 to 2 g L−1 of microsomal proteins containing membrane WT or mutated VKORC1. Values are expressed as % of activity of WT-VKORC1. Each data point represents the mean ± SD of three individual determinations and is representative of an experiment performed on two different batches of protein. *P < 0.02 compared with VKORC1.

Consequences of the VKORC1 mutations on the VKOR activity

Among the 25 mutations characterized in this study, six mutations led to an increased Km value. For mutations located in the transmembrane domains, only one mutation (i.e. A26P) located in TM1 led to an increase in the Km value (3-fold) (Table 2). For mutations located in the luminal loop, the Km of five mutations (D36G, A41S, V54L, R58G and W59C) were significantly higher than that of the WT. These increases varied between 2-fold for D36G and 9-fold for W59C (Table 3).

Table 2.   Apparent kinetic parameters towards K > O of the mutated VKORC1 proteins with mutation located in transmembrane domains
VKORC1 K m (μM) V max (pmol min−1mg−1 of total protein) V max/Km (nL min−1 mg−1 total protein)(% compared with WT)
  1. V max values determined at saturating concentration of K > O were evaluated after normalization of the VKORC1 expression level by immunoquantification. Each data point represents the mean ± SD of three individual determinations. *P < 0.02 compared with wild type (WT).

WT19.8 ± 4.59.4 ± 1.6476100
TM1
 A26P57.4 ± 10.1*3.4 ± 1.25912
 A26T18.7 ± 1.412.3 ± 2.7657138
 L27V22.8 ± 2.98.3 ± 0.629963
 H28Q29.8 ± 4.69.1 ± 1.630664
TM3
 I123N27.0 ± 2.114.2 ± 1.6522110
 L128RNDNDND< 2
TM4
 Y139H9.2 ± 3.05.5 ± 2.3598126
Table 3.   Apparent kinetic parameters towards K > O of the mutated VKORC1 proteins with mutation in the luminal loop
VKORC1 K m (μM) V max (pmol min−1mg−1 of total protein) V max/Km (nL min−1 mg−1 total protein)(% compared with WT)
  1. V max values determined at saturating concentration of K > O were evaluated after normalization of the VKORC1 expression level by immunoquantification. Each data point represents the mean ± SD of three individual determinations. *P < 0.02 compared with wild type (WT).

WT19.8 ± 7.79.4 ± 1.6476100
D36Y23.6 ± 0.211.4 ± 4.0483101
D36G43.8 ± 0.2*2.9 ± 0.1*6514
A41S65.9 ± 5.4*41.1 ± 0.7*623131
V45A26.9 ± 2.31.8 ± 0.4*8117
S52LNDNDND< 2
S52WNDNDND< 2
V54L102.5 ± 28.6*2.86 ± 0.1*286
S56F23.2 ± 6.246.5 ± 5.1*2004421
R58G71.0 ± 10.9*4.4 ± 1.36213
W59RNDNDND< 2
W59C179.7 ± 12.5*20.45 ± 7.411324
W59LNDNDND< 1
V66MNDNDND< 2
V66GNDNDND< 1
H68Y16.9 ± 2.810.4 ± 2.6618130
G71ANDNDND< 2
N77SNDNDND< 2
N77YNDNDND< 2

Among the 25 mutations, five mutations, all located in the luminal loop, led to modification of Vmax values (Tables 2 and 3). Vmax values of D36G, V45A and V54L were 3, 5 and 3.5-fold lower than that of the WT-VKORC1, respectively. Two mutations (A41S and S56F) located also in the luminal loop led to a 5-fold increase in the Vmax value.

Consequences of the VKORC1 mutations on the susceptibility to VKA

To better understand involvement of the VKORC1 mutations in the VKA resistance phenotype, Ki of three different VKAs were determined. VKAs inhibited the VKOR activity in a non-competitive manner for all the mutants we analysed (data not shown). Ki values obtained for the mutated VKORC1 are reported in Tables 4 and 5 for mutations located, respectively, in the transmembrane domains or luminal loop. Among the 25 mutations detected in patients clinically resistant to VKA, only six mutations led to increase in Ki. Four mutations (i.e. A26P, I123N, Y139H and V54L) led to resistance to both 4-hydroxycoumarine or indane-1,3-dione derivatives. The A41S mutation led to an increase in the Ki only towards fluindione (×9) and H68Y, only towards 4-hydroxycoumarine derivatives (×4 and ×6 for warfarin and acenocoumarol, respectively).

Table 4. Ki values of the mutated VKORC1 proteins with mutation located in transmembrane domains towards various vitamin K antagonists
VKORC1Warfarin (μM)Acenocoumarol (μM)Fluindione (μM)
  1. Inhibition parameters were assessed using anticoagulant concentrations from about 0.05 to 20 × Ki.

WT1.65 ± 0.790.33 ± 0.180.25 ± 0.14
TM1
 A26P18.43 ± 5.82*2.93 ± 1.25*> 5
 A26T2.13 ± 0.560.22 ± 0.020.43 ± 0.32
 L27V1.83 ± 0.620.52 ± 0.330.54 ± 0.08
 H28Q0.65 ± 0.420.26 ± 0.070.44 ± 0.30
TM3
 I123N4.01 ± 1.01*1.23 ± 0.33*1.54 ± 0.45*
TM4
 Y139H5.91 ± 1.77*0.80 ± 0.241.01 ± 0.13*
Table 5. Ki values of the mutated VKORC1 proteins with mutation in the luminal loop towards various vitamin K antagonists
VKORC1Warfarin (μM)Acenocoumarol (μM)Fluindione (μM)
  1. Inhibition parameters were assessed using anticoagulant concentrations from about 0.05 to 20 × Ki.

WT1.65 ± 0.790.33 ± 0.180.25 ± 0.14
D36Y1.82 ± 0.700.99 ± 0.650.35 ± 0.08
D36G0.74 ± 0.250.13 ± 0.030.95 ± 0.32
A41S1.78 ± 0.020.44 ± 0.073.04 ± 0.45*
V45A1.10 ± 0.040.26 ± 0.130.17 ± 0.03
V54L7.95 ± 1.32*3.90 ± 0.71*4.81 ± 0.68*
S56F1.05 ± 0.820.46 ± 0.170.42 ± 0.25
R58G1.50 ± 0.360.30 ± 0.060.95 ± 0.29
W59C1.16 ± 0.200.40 ± 0.140.18 ± 0.05
H68Y6.21 ± 0.85*2.04 ± 0.98*0.65 ± 0.18

Discussion

Spontaneous VKORC1 mutations were all described in patients with total or partial resistance to VKAs that cannot be explained by demographic or environmental factors (Table 1). Therefore, an association between phenotype and genotype was immediately suggested, but never demonstrated. Therefore in this study, we expressed 25 spontaneous mutants of VKORC1 and determined the functional consequences of the mutation on the recycling of vitamin K by the determination of Km and Vmax and on the susceptibility to VKA by the determination of Ki. Recombinant VKORC1 proteins were expressed as membrane bound proteins in Pichia pastoris as previously performed to characterize spontaneous rVKORC1 mutations [18]. This expression system allowed us to obtain recombinant rVKORC1 with catalytic properties (i.e. Km and Ki) similar to those of the native proteins [18]. On the other hand, these catalytic parameters determined from yeast or liver microsomes were demonstrated to correctly reflect the phenotype observed [11,18,20] in healthy animals of the same age and the same sex, in the presence of a standardized food. Consequently, the determination of the properties of VKORC1 proteins will be useful in the interpretation of the clinical data in humans.

All the spontaneous mutations of the VKORC1 gene were described in patients with moderate to severe resistance to VKAs. Surprisingly, these mutations are widely spread in the VKORC1 sequence and it seems unlikely that all of these residues are involved in VKA interaction. In our study, only six mutations (A26P, A41S, V54L, H68Y, I123N and Y139H) led to a significant increase in the Ki towards VKA (Fig. 5). The A26P mutation was described only from one heterozygote patient who was treated with up to 20 mg day−1 warfarin or 100 mg day−1 fluindione [21] without reaching a stable anticoagulation. Two patients heterozygous for the V54L mutation were described [21,22]. The stable anticoagulation was not reached for one of the patients despite the higher dose (×3) and the changes of VKA used [21], while for the other patient, a 5-fold dose of warfarin led to a stable anticoagulation [22]. I123N and Y139H mutations were recently reported also from heterozygote patients treated with a high dose of phenprocoumon, a derivative of the 4-hydroxycoumarine (i.e. ×7 and ×3 for I123N and Y139H, respectively) [23] without reaching stable anticoagulation. In our study, A26P, V54L, I123N and Y139H mutations led to 11-, 5-, 3- and 4-fold increases in the Ki towards warfarin, respectively, suggesting the involvement of these mutations in the phenotype observed in the heterozygote patients carrying these mutations. Unfortunately, it is difficult to evaluate the relationship between the increase in the Ki and the increase in the dose of VKA necessary to reach stable anticoagulation. Indeed, despite the increase of the dose of VKA, a stable anticoagulation was not reached for most of the patients who were carriers of one of these four mutations, except for one patient described by Harrington et al. [22]. For this patient stable anticoagulation was reached with 35 mg day−1 warfarin (i.e. 5-fold the usual dose) [22].

Figure 5.

 Comparison of the inhibition effect of warfarin, acenocoumarol and fluinidione using recombinant yeast microsomes expressing mutated VKORC1, comparative to WT-VKORC1.

In our study, Ki for warfarin increased 5-fold, which is consistent with the increase in the dose reported by Harrington et al., suggesting the importance of knowing the mutation and the increase in the Ki induced by the mutation for treatment based on VKAs. For A41S and H68Y mutations also associated with increase in Ki, no or few clinical data are available in the related literature [2,24] and comparison of our results with the clinical phenotype is thus impossible. Surprisingly, both mutations do not seem to lead to global resistance to VKA. A41S mutation leads to an increase in the Ki only towards fluindione, an indane-1,3-dione derivative. H68Y mutation leads to an increase in the Ki only towards 4-hydroxycoumarine derivatives. These results suggest that patients who are carriers of the A41S mutation must be treated with a normal dose of 4-hydroxycoumarine derivatives to obtain stable anticoagulation, while patients who are carriers of H68Y mutations must be treated with a normal dose of fluindione. The demonstration of the causality of A26P, A41S, V54L, H68Y, I123N and Y139H point mutations in the resistance phenotype should be taken into account to direct the choice of the molecule and the dose to be used.

Among these six mutations, only the consequences of the Y139H mutation were predictable. Indeed, Y139 was proposed to be involved in the binding of VKAs [25] and the substitutions of Y139 by F, C or S lead to severe resistance to VKAs in rats [16,18,20]. The I123 located in TM3 [23], in front of the proposed T138-Y139-A140 binding site of VKA [25], could be involved in the binding of VKA by hydrophobic interaction. But it is more likely that its replacement by an N could modify the positioning of the VKA in the binding site. The A41, V54 and H68 are located in the luminal loop, suggesting the involvement of this luminal loop in the stabilization of VKAs. The involvement of the A26 in the binding of VKAs can be excluded because the substitution of this amino acid residue by T does not modify the Ki values in humans and rats [26]. P substitution at position 26 could drastically change the structure of the protein and therefore would be responsible for the increase in Ki and Km.

In our study, the 19 VKORC1 mutations do not lead to an increase in Ki, while these mutations were described in patients resistant to VKAs. Among these 19 mutations, nine mutated proteins are able to catalyze the reduction of K > O to K (i.e. A26T, L27V, H28Q, D36Y, D36G, V45A, S56F, R58G and W59C). Partial characterization of V45A and R58G was previously reported [10] and Ki determined herein are strictly coherent with this partial characterization. For these nine mutations, various hypotheses should be considered. (i) VKORC1 mutation is not causative for the resistance phenotype and the cause of the resistance was not found in the patient who is a carrier of this mutation. (ii) VKOR activity is not mediated by a single protein, but by a multienzyme complex and our model, too simplified, does not mimic certain mechanisms of resistance such as VKORC1-calumenin interaction described as a possible mechanism of resistance [27].

Among the 19 mutations, 10 mutated proteins (i.e. S52L, S52W, W59L, W59R, V66M, V66G, G71A, N77S, N77T and L128R) are almost not able to catalyze the reduction of K > O to K and three mutated proteins (i.e. D36G, V45A and R58G) catalyze the reduction of K > O to K with an enzymatic efficiency (Vmax/Km) at least 85% lower than that of the WT. These mutations were all detected in heterozygote patients, suggesting that VKOR activity supported by a unique valid allele is sufficient to maintain a normal coagulation and thus a sufficient pool of gamma-carboxylated clotting factors II, VII, IX and X. Indeed, Harrington et al. showed that the V66M mutation in heterozygotes is not associated with an increase in the INR (International Normalized Ratio), an increase in the PIVKA (protein induced by vitamin K absence) or an increase in the plasma vitamin K > O [28]. These ‘inactivating’ mutations were all detected in patients requiring a high dose of VKA. Nevertheless, the link between an ‘inactivating’VKORC1 mutation and a clinical resistance to VKA is difficult to explain. These results might suggest the existence of other mechanisms of resistance still not identified in patients who are carriers of these mutations.

The VKORC1 mutations impairing partially or totally the VKOR activity are essentially located in a 42-aa segment in the luminal loop, suggesting its crucial role in the functioning of the enzyme. These mutations could modify the conformation of the VKORC1 or the internal electron transfer pathway. Indeed, it was suggested that the luminal loop containing C43 and C51 participates in the electron transfer from the redox partner to C132-135 of the CXXC active site [13,29,30]. In our experimental conditions, the redox partner is mimicked by the dithiotreitol molecule. Several studies described that dithiotreitol can directly reduce the active site of VKORC1 and hence activate the protein for catalysis even when C43 and 51of VKORC1 are mutated [14,29]. Nevertheless, in our experimental conditions in the presence of VKORC1 protein located inside the membrane without any detergent, the substitution of C43 or 51 by A leads to a loss of activity (data not shown), similarly to what was observed by Rost et al. [30]. This result suggests that in our experimental conditions, the reduction of C43-51 by dithiotreitol might precede the reduction of C132-135. The characterization of the VKOR activity catalyzed by the recombinant hVKORC1 expressed in the microsomal fraction of yeasts thus could allow us to explore the reduction of vitamin K > O to vitamin K, but also the internal electron transfer pathway from C43-51 to C132-135. In this case, because many mutations of the luminal loop lead to modification of the velocity of the VKORC1 contrary to the mutations located in the TM domains, the internal electron transfer could be the rate-limiting step of the VKORC1 cycle. It is also interesting to note that many mutations of the luminal loop also lead to an increase in the Km values, suggesting the involvement of the luminal loop in the substrate interaction. According to the 3-D structure of the bacterial homologue of VKORC1 from Synechococcus sp., this luminal loop would form a lid on the four-helix bundle where K > O is located, shielding the substrate from the periplasmic space [13]. The mutations leading to a drastic or complete loss of activity of the enzyme are almost all located in this luminal loop. Because the luminal loop is likely to be involved in the internal electron transfer and in the substrate interaction by shielding the substrate from the periplasmic space, we can suggest that ‘inactivating’ mutations disrupt either the internal electron transfer or the substrate binding. Further studies will be necessary in order to elucidate this question.

Addendum

A. Hodroge, B. Matagrin, C. Moreau, I. Fourel, A. Hammed and V. Lattard performed experiments; A. Hodroge, E. Benoit and V. Lattard designed the research, analyzed results and wrote the paper.

Acknowledgements

This work was supported by grants from Agence Nationale pour la Recherche (RODENT 2009-CESA-008-03) and by DGER. The authors thank V. Siguret (Service d’Hématologie Biologique – Hôpital Européen Georges Pompidou) for helpful discussion.

Disclosure of Conflict of Interests

The authors declare no competing financial interests.

Ancillary