SEARCH

SEARCH BY CITATION

Keywords:

  • acenocoumarol;
  • mutation;
  • phenprocoumon;
  • resistance;
  • VKORC1;
  • warfarin

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients, materials, and methods
  5. Results
  6. Discussion
  7. Addendum
  8. Acknowledgements
  9. Disclosure of Conflict of Interests
  10. References
  11. Supporting Information

Summary. Background: Vitamin K 2,3-epoxide reductase complex subunit 1 (VKORC1) is the molecular target of oral anticoagulants. Mutations in VKORC1 cause partial or total coumarin resistance. Objectives: To identify new VKORC1 oral anticoagulant (OAC) resistance (OACR) mutations and compare the severity of patient phenotypes across different mutations and prescribed OAC drugs. Patients/Methods: Six hundred and twenty-six individuals exhibiting partial or complete coumarin resistance were analyzed by VKORC1 gene sequencing and CYP2C9 haplotyping. Results: We identified 13 patients, each with a different, novel human VKORC1 heterozygous mutation associated with an OACR phenotype. These mutations result in amino acid substitutions: Ala26[RIGHTWARDS ARROW]Thr, His28[RIGHTWARDS ARROW]Gln, Asp36[RIGHTWARDS ARROW]Gly, Ser52[RIGHTWARDS ARROW]Trp, Ser56[RIGHTWARDS ARROW]Phe, Trp59[RIGHTWARDS ARROW]Leu, Trp59[RIGHTWARDS ARROW]Cys, Val66[RIGHTWARDS ARROW]Gly, Gly71[RIGHTWARDS ARROW]Ala, Asn77[RIGHTWARDS ARROW]Ser, Asn77[RIGHTWARDS ARROW]Tyr, Ile123[RIGHTWARDS ARROW]Asn, and Tyr139[RIGHTWARDS ARROW]His. Ten additional patients each had one of three previously reported VKORC1 mutations (Val29[RIGHTWARDS ARROW]Leu, Asp36[RIGHTWARDS ARROW]Tyr, and Val66[RIGHTWARDS ARROW]Met). Genotyping of frequent VKORC1 and CYP2C9 polymorphisms in these patients revealed a predominant association with combined non-VKORC1*2 and wild-type CYP2C9 haplotypes. Additionally, data for OAC dosage and the associated measured International Normalized Ratio (INR) demonstrate that OAC therapy is often discontinued by physicians, although stable therapeutic INR levels may be reached at higher OAC dosages. Bioinformatic analysis of VKORC1 homologous protein sequences indicated that most mutations cluster into protein sequence segments predicted to be localized in the lumenal loop or at the endoplasmic reticulum membrane–lumen interface. Conclusions: OACR mutations of VKORC1 predispose afflicted patients to high OAC dosage requirements, for which stable, therapeutic INRs can sometimes be attained.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients, materials, and methods
  5. Results
  6. Discussion
  7. Addendum
  8. Acknowledgements
  9. Disclosure of Conflict of Interests
  10. References
  11. Supporting Information

At present, the prevention of and therapy for thromboembolic conditions are chiefly accomplished by administration of 4-hydroxycoumarin derivatives, including warfarin, phenprocoumon, and acenocoumarol. An additional class of oral anticoagulant (OAC) drugs, represented by fluindione, is currently used in France [1]. Despite well-known drawbacks, including a narrow therapeutic dosage range and broad variation in individual drug response, OACs are among the most commonly prescribed drugs globally [2]. The mode of action of 4-hydroxycoumarins was clarified by the identification and cloning of their molecular target, warfarin-sensitive vitamin K 2,3-epoxide reductase complex subunit 1 (VKORC1; EC 1.1.4.1) [3–6]. The VKORC1 enzyme is responsible for regeneration of reduced vitamin K from vitamin K epoxide, a byproduct of γ-glutamyl carboxylation of vitamin K-dependent coagulation factors (Fig. 1) [7]. Coumarins non-competitively inhibit VKORC1-dependent reduction of vitamin K epoxide to vitamin K hydroquinone, thus blocking post-translational γ-glutamyl carboxylation of vitamin K-dependent proteins required for normal clotting function [8–11].

image

Figure 1.  Inhibitors of the vitamin K cycle directly block vitamin K 2,3-epoxide reductase complex subunit 1 (VKORC1) enzymatic activity. Glu, glutamyl residue; Gla, γ-carboxylated glutamyl residue; K, vitamin K quinone; KH2, vitamin K hydroquinone; K>O, vitamin K 2,3-epoxide. Enzymes are depicted as ovals: GGCX, γ-glutamyl carboxylase; VKORC1, vitamin K 2,3-epoxide reductase complex subunit 1. In the presence of intracellular dissolved O2, GGCX incorporates dissolved CO2 into γ-glutamyl carboxylate (Gla) groups of post-translationally modified vitamin K-dependent proteins, including blood clotting and bone homeostasis factors, and produces K>O and water as a byproduct. VKORC1 carries out both vitamin K 2,3-epoxide reductase (VKOR) and vitamin K quinone reductase (VKR) activities. The four oral anticoagulants currently prescribed for human use worldwide have direct inhibitory actions on VKORC1 (depicted as a blunt-ended bar), and thus indirectly block γ-glutamyl carboxylation by preventing recycling of intracellular K>O to KH2.

Download figure to PowerPoint

The OAC maintenance dosage was found to be dependent on multiple factors, including patient weight, sex, age, and ethnicity, as well as on polymorphisms in CYP2C9 and VKORC1 [12–15]. However, in some patients, the excessive OAC dosage requirement cannot be explained by these factors and genetic variants alone. In such cases, rare mutations in VKORC1 have been found to be associated with a substantially higher dosage requirement (operationally defined as partial OAC resistance when a stable, therapeutic International Normalized Ratio [INR] value is obtained), and sometimes even complete resistance (operationally defined for a very high dosage requirement and not reaching a stable, therapeutic INR value). Among such cases, missense mutations in VKORC1 were found to be causative for some extremely elevated OAC dosage phenotypes [4,16–19]. To date, 134 OAC-resistant patients worldwide have been reported to harbor one of 11 distinct VKORC1 missense mutations (Table S3) [1,4,16–27]. Notably, most of these patients are heterozygous for the respective mutations, with only one mutation (Asp36[RIGHTWARDS ARROW]Tyr) being found homozygously in six OAC-resistant patients [17,19,21,27].

We present here 13 novel and three previously reported OAC resistance (OACR) mutations in the protein-coding region of the VKORC1 gene identified in 23 patients (Table 1). Comparison of the OACR phenotypes revealed that, in many patients, therapeutic INR values can be achieved with substantially higher than average OAC dosages. In addition, we used bioinformatic analysis of VKORC1 homologous protein sequences, together with published results on the enzymatic function and warfarin inhibition of VKORC1 from humans and rodents, to propose a plausible mechanism for OAC action. Finally, we present a hypothesis for a molecular mechanism of OACR that is well supported by previously published in vitro studies of mutant OAC-resistent VKORC1 proteins.

Table 1.   New human vitamin K 2,3-epoxide reductase complex subunit 1 (VKORC1) oral anticoagulant resistance mutations
MutationNucleotideNovelPatient ID (sex, age)Weekly dose (mg)INRVKOR:c.−1639CYP2C9Ethnicity
  1. INR, International Normalized Ratio; P, phenprocoumon; st, stable therapeutic anticoagulation was achieved at the indicated INR value and at the indicated weekly dose; W, warfarin. Column heading ‘Novel’ refers to first published report. All mutations are heterozygous. CYP2C9 indicates haplotypes.

Ala26[RIGHTWARDS ARROW]Thrc.76G>AYes1 (M, 39)42 P (aborted)1.0GG*1*1Morrocan
His28[RIGHTWARDS ARROW]Glnc.84C>TYes2 (M, 79)69 P2.2 stGG*1*1German
Val29[RIGHTWARDS ARROW]Leuc.85G>TNo3 (M, 67)180 P2.0–3.0 stGG*1*1Swiss
Asp36[RIGHTWARDS ARROW]Tyrc.106G>TNo4 (F, 39)50 P2.2 stGG*1*1German
Asp36[RIGHTWARDS ARROW]Tyrc.106G>TNo5 (M, ?)140 W2.0–2.5 stGG*1*1Afro-American
Asp36[RIGHTWARDS ARROW]Tyrc.106G>TNo6 (M, 56)47 P2.6 stGG*1*1Russian
Asp36[RIGHTWARDS ARROW]Tyrc.106G>TNo7 (M, 71)51 P2.5 stGG*1*1Russian
Asp36[RIGHTWARDS ARROW]Tyrc.106G>TNo8 (F, 66)80 W (aborted)1.2GG*1*1Russian
Asp36[RIGHTWARDS ARROW]Tyrc.106G>TNo9 (F, 47)140 W2.5 stGG*1*1Turkish
Asp36[RIGHTWARDS ARROW]Glyc.107A>GYes10 (F, ?)140 W2.5–3.0 stGG*1*2Russian
Ser52[RIGHTWARDS ARROW]Trpc.155C>GYes11 (F, 58)69 P2.0–2.2 stGG*1*1German
Ser56[RIGHTWARDS ARROW]Phec.167C>TYes12 (F, 55)105 P (aborted)1.2GG*1*1German
Trp59[RIGHTWARDS ARROW]Leuc.176G>TYes13 (M, 55)105 P (aborted)1.2GG*1*2Iranian
Trp59[RIGHTWARDS ARROW]Cysc.177G>TYes14 (M, 42)74 P (aborted)1.0GG*1*1German
Val66[RIGHTWARDS ARROW]Metc.196G>ANo15 (M, 46)63 P (aborted)1.2GG*1*1Austrian
Val66[RIGHTWARDS ARROW]Metc.196G>ANo16 (F, 51)72 P2.3 stGG*1*1Afro-Carribean
Val66[RIGHTWARDS ARROW]Metc.196G>ANo17 (F, 56)294 W2.5 stGG*1*1Afro-Carribean
Val66[RIGHTWARDS ARROW]Glyc.197T>GYes18 (M, 51)54 P2.0 stGG*1*2German
Gly71[RIGHTWARDS ARROW]Alac.212G>CYes19 (M, 48)42 P (aborted)1.2GG*1*1German
Asn77[RIGHTWARDS ARROW]Serc.230A>GYes20 (F, 28)63 P (aborted)1.6GG*1*3German
Asn77[RIGHTWARDS ARROW]Tyrc.229A>TYes21 (M, 50)175 W; 73 P3.0 stGG*1*3German
Ile123[RIGHTWARDS ARROW]Asnc.368T>AYes22 (F, 68)147 P (aborted)1.4GG*1*1German
Tyr139[RIGHTWARDS ARROW]Hisc.415T>CYes23 (M, 63)63 P (aborted)1.0GA*1*1German

Patients, materials, and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients, materials, and methods
  5. Results
  6. Discussion
  7. Addendum
  8. Acknowledgements
  9. Disclosure of Conflict of Interests
  10. References
  11. Supporting Information

Patients

Since the initial cloning of the VKORC1 gene in 2004, 626 patients have been referred to our laboratory for genetic analysis on the basis of either elevated OAC dosage requirement or complete OACR, as subjectively assessed by the attending physician. Thus, limitations of the present study include the following: (i) patient samples for genetic analysis were referred from multiple clinicians worldwide for cases involving elevated OAC dosage requirement, but without clearly defined ascertainment criteria with respect to comprehensive patient data; and (ii) comprehensive OAC dosage and INR data were therefore only obtained retrospectively for those patients in whom OACR mutations were found. Therefore, the present study does not include population-level data, and statistical comparisons are not possible. All patients gave informed written consent prior to genetic analysis. On the basis of genetic analysis results for each patient, we subdivided the population of 626 patients with elevated dosage requirement into three subgroups: (i) patients with missense mutations found in the VKORC1 protein-coding regions (VKORC1 mutation group); (ii) patients without VKORC1 missense mutations, but with both VKORC1:−1639GG and CYP2C9*1*1 genotypes (high-dosage control group without mutations); and (iii) patients without VKORC1 missense mutations lacking either VKORC1:−1639GG or CYP2C9*1*1 genotypes (i.e. the group of patients excluded from groups 1 and 2). OAC dosage data for 77 patients from group 2 receiving phenprocoumon and reaching stable anticoagulation INR values were averaged for use as a control high-dosage threshold (HDT) reference level in the present study (Fig. 2; see also ‘Comparison of OAC-resistant patient phenotypes’ below).

image

Figure 2.  Venn diagram depicting genotypes/phenotypes among groups of patients in the present study in relation to all patients treated with oral anticoagulants. Patient numbers cited represent only 77 control patients and 23 vitamin K 2,3-epoxide reductase complex subunit 1 (VKORC1) mutation patients from the present cohort, as comprehensive VKORC1 and CYP2C9 haplotype data are not given for the majority of patients from previously published studies. HDT, high dosage threshold.

Download figure to PowerPoint

Genetic analysis

Genomic DNA was prepared from whole blood samples by standard procedures. The promoter region, three exons and flanking intronic sequences of the VKORC1 gene, as well as exons 3 and 7 of the CYP2C9 gene, were amplified by standard PCR techniques and sequenced as described previously [28]. CYP2C9 amplification was performed in the same way, except that an annealing temperature of 58 °C was applied. INR analyses were performed in laboratories at patients’ local clinics.

Comparison of OAC-resistant patient phenotypes

In order to compare the relative severities of both current and previously published OAC-resistant patient phenotypes across all reported cases of VKORC1 missense mutations (Fig. 3; for complete literature references, see Table S3), we first compiled OAC dosage ranges for patients binned by mutation type, and noted whether stable INR values were attained for each case. We operationally defined an HDT as the average weekly phenprocoumon dosage required by patients with combined non-VKORC1*2 (homozygous VKORC1:c.−1639G) and homozygous wild-type CYP2C9 (CYP2C9*1*1) alleles, but without missense mutations in VKORC1, who achieved stable anticoagulation INR values. For phenprocoumon, the HDT was calculated to be 21.1 (± 5.7 standard deviation) mg week−1 (38 male patients, age 60 ± 17 years, dose 21.4 ± 5.7 mg phenprocoumon week−1; 39 female patients, age 50 ± 20 years, dose 20.8 ± 5.8 mg phenprocoumon week−1). This dosage is in accordance with that reported by Schalekamp et al. [29] in patients with the same genotype (phenprocoumon 22.4 ± 2 mg week−1). As our cohort did not contain sufficient patients treated with warfarin or acenocoumarol for direct calculatation of average dosages, as Schalekamp et al. did for phenprocoumon, we defined HDT factors for acenocoumarol and warfarin on the basis of phenprocoumon conversion factors recently published by van Leeuwen et al. [30] (phenprocoumon to warfarin, 2.36; phenprocoumon to acenocoumarol, 1.15). Accordingly, we defined the warfarin HDT value as 49.8 mg week−1 (similar to 43.4 mg week−1 mean dosages for groups of patients with VKORC1 high-dose haplotypes reported by D’Andrea et al. [31] and by Rieder et al. [32]; comprehensively reviewed in Moyer et al. [33]) and the acenocoumarol HDT value as 24.3 mg week−1. As no phenprocoumon to fluindione conversion factor has been reported to date, we used the 19.8-mg normal reference daily dose for fluindione as reported by Lacut et al. [34] for homozygous non-VKORC1*2 patients to define the fluindione HDT as 138.6 mg week−1. Finally, we converted average weekly patient dosages to multiplicative factors of the respective drug-specific HDT, in order to allow direct comparison of anticoagulation efficacy among patients with various OACR missense VKORC1 mutations, regardless of the specific OAC drug that each patient was prescribed. Specifically, we divided the weekly OAC dosage for each VKORC1 missense mutation patient by the HDT value for the particular OAC drug taken by the patient, averaged the respective HDT factors in cases of multiple patients with each drug–mutation combination, and plotted the results in Fig. 3. As current data are sparse with regard to coverage of OACR mutational genotypes and dosages for each of the specific OAC drugs, we caution the reader not to draw inferences for any particular drug concerning superiority in treatment choice.

image

Figure 3.  Elevated oral anticoagulant (OAC) dosage requirements normalized to operationally defined drug-specific high dosage threshold (HDT) values for human vitamin K 2,3-epoxide reductase complex subunit 1 (VKORC1) missense mutations associated with OAC resistance (OACR) for 58 patients bearing OACR mutations in VKORC1. (Some patients were successively given multiple OAC drs; thus, the figure data comprise 68 drug–dosage combinations.) Bar heights indicate average values (number of patients are indicated in parentheses above bars; otherwise single patient values) for OAC dosage levels normalized to an operationally defined HDT level equal to a value of 1.0 (indicated by the horizontal black bar) for each drug corresponding to the following OAC dosages: phenprocoumon 21.1 mg week−1, acenocoumarol 24.3 mg week−1, warfarin 49.8 mg week−1, fluindione 138.6 mg week−1. Absolute dosage ranges for mutation–drug combinations with multiple patients are indicated by I-shaped bars (not error bars). All mutations are heterozygous, with the exception of Asp36Tyr+/+. All data represent dosages resulting in stable therapeutic International Normalized Ratio (INR) values, except for those indicated as bars marked with a caret (^), for which no patient reached a therapeutic INR endpoint.

Download figure to PowerPoint

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients, materials, and methods
  5. Results
  6. Discussion
  7. Addendum
  8. Acknowledgements
  9. Disclosure of Conflict of Interests
  10. References
  11. Supporting Information

Sequencing of the VKORC1 gene revealed single heterozygous coding sequence variations corresponding to single amino acid substitutions in 23 of 626 patients initially diagnosed as partially or completely OAC-resistant (Table 1). Of these, six patients exhibited an Asp36[RIGHTWARDS ARROW]Tyr substitution (c.106G>T), three a Val66[RIGHTWARDS ARROW]Met substitution (c.196G>A), and one a Val29[RIGHTWARDS ARROW]Leu substitution (c.85G>T), all of which have been previously described [4,23,24]. Among the remaining patients, 13 individual novel heterozygous mutations were detected (Table 1). Genotyping of frequent VKORC1 and CYP2C9 polymorphisms indicated a striking predominance of combined non-VKORC1*2 (Table 1, column marked VKOR:c.−1639, 22 of 23 patients) and wild-type CYP2C9 haplotypes (Table 1, column marked CYP2C9, 23 of 23 patients, including 18 homozygotes and five heterozygotes). Of the total of 157 OAC-resistant patients from published reports and this study, all were heterozygous for VKORC1 mutations, except for six with homozygous Asp36[RIGHTWARDS ARROW]Tyr mutations found in patient populations with a relatively high prevalence of Asp36[RIGHTWARDS ARROW]Tyr heterozygotes [16,17,19,21,23,27].

Thirteen VKORC1 mutation patients achieved a therapeutic INR of 2.0–3.0 with increased OAC dosages (Table 1, patients 2–7, 9–11, 16–18, and 21), whereas OAC therapy was aborted or failed entirely for the 10 remaining patients (Table 1, patients 1, 8, 12–15, 19, 20, 22, and 23), accounting for 40% of VKORC1 mutation patients in the present study. Averaged HDT factors for VKORC1 mutation patients are shown in graphical form in Fig. 3, where zero values (absent bars) for any combination of OAC drug and mutation indicate that no patients bearing that specific mutation were treated with those particular therapeutics. In Fig. 3, we use caret (^) symbols above individual bars to indicate that the values represent the highest reported dosage among all patients for the specific mutation when no patient was reported to reach a stable therapeutic INR value; that is, OAC therapy was ineffective at least up to the reported level. It is not known whether higher dosages of the respective OACs would result in stable therapeutic INR values being reached (indicative of partial OACR) or whether no elevated level of OAC would achieve a therapeutic INR (indicative of complete OACR). Thus, Fig. 3 is a comprehensive compilation of all currently available data for OAC drug dosage levels and therapy outcomes for OAC-resistant patients bearing VKORC1 missense mutations.

Alignment of the 24 known OACR mutations with human VKORC1 and with 13 vertebrate VKORC1 orthologs from sequenced genomes revealed that all OACR mutational positions are completely (nine of 16 positions, AMAS score*) or highly (five of 16, AMAS score 9; one of 16, AMAS score 7; one of 16, and AMAS score 5) conserved (Fig. S4).

Further bioinformatic analysis of 327 VKORC1 ortholog sequences (for detailed methods and analysis, see Figs S1–S4 and Tables S1–S3) resulted in a topology prediction of human VKORC1 (Fig. 4; see also Fig. S4, line labeled Putative structural features) as a polytopic membrane protein with four α-helices spanning the endoplasmic reticulum (ER) membrane, N-termini and C-termini in the cytoplasm, and a large ∼ 50-residue loop located in the ER lumen. This prediction is in excellent agreement with the topology of a recently published two-domain prokaryotic VKORC1 homolog protein structure at 3.6-Å resolution [35], in contrast to a previously reported biochemical study suggesting an alternative topology featuring an ER lumenal-exposed N-terminus and three transmembrane (TM) α-helices [36] (see Supporting Information, pg. 6 ff., for a detailed analysis). A more recent biochemical study has unequivocably established that the VKORC1 N-terminus is located in the cytoplasm, strongly supporting our topology model [37]. The positions of all currently known human VKORC1 OACR missense mutations are plotted above the human VKORC1 amino acid sequence in the protein topology model in Fig. 4. Mutations Ala26[RIGHTWARDS ARROW]Thr, Ala26[RIGHTWARDS ARROW]Pro, Leu27[RIGHTWARDS ARROW]Val, His28[RIGHTWARDS ARROW]Gln and Val29[RIGHTWARDS ARROW]Leu are predicted to be localized at the interface between the first TM helix and the ER luminal domain at positions that are completely or highly conserved across all 13 vertebrate species. Mutations Asp36[RIGHTWARDS ARROW]Tyr, Asp36[RIGHTWARDS ARROW]Gly, Val45[RIGHTWARDS ARROW]Ala, Ser52[RIGHTWARDS ARROW]Trp, Val54[RIGHTWARDS ARROW]Leu, Ser56[RIGHTWARDS ARROW]Phe, Arg58[RIGHTWARDS ARROW]Gly, Trp59[RIGHTWARDS ARROW]Leu, Trp59[RIGHTWARDS ARROW]Arg, Trp59[RIGHTWARDS ARROW]Cys, Val66[RIGHTWARDS ARROW]Met, Val66[RIGHTWARDS ARROW]Gly, Gly71[RIGHTWARDS ARROW]Ala, Asn77[RIGHTWARDS ARROW]Tyr and Asn77[RIGHTWARDS ARROW]Ser are fairly evenly distributed throughout the large loop situated in the ER lumen between the first two TM helices. Notably, this loop is predicted to contain three completely conserved residues (Cys43, Cys51, and Ser57) that are presumed to be catalytically active, as suggested by in vitro mutagenesis studies [4,38]. Of the three remaining mutations, Ile123[RIGHTWARDS ARROW]Asn and Leu128[RIGHTWARDS ARROW]Arg are predicted to lie in the third TM helix at the C-terminal end (Fig. 4, indicated as TM3), adjacent to the ER lumen, and the last mutation, Tyr139[RIGHTWARDS ARROW]His, is predicted to be in the ER lumenal half of the fourth TM helix at the putative warfarin-binding site (TYA 138–140), just one helical turn away from the redox-active CIVC catalytic motif (amino acids 132–135) [39–41]. Both Leu128[RIGHTWARDS ARROW]Arg and Tyr139[RIGHTWARDS ARROW]His represent, like the two mutations located in the first TM helix, substitutions at positions that are completely conserved across all 13 vertebrate species. The remaining Ile123[RIGHTWARDS ARROW]Asn mutation represents a non-conservative mutation at a position that is not completely conserved. Notably, all human OACR mutations of VKORC1 are situated at the lumenal side of the ER membrane, where, by analogy to the recently solved structure of a homologous prokaryotic protein [10], vitamin K substrates are expected to bind to Cys135 in the reaction center.

image

Figure 4.  A topological model of human vitamin K 2,3-epoxide reductase complex subunit 1 (VKORC1) from bioinformatic analysis of VKORC1 orthologs for 327 species, using predictions for the location of transmembrane (TM) helices. The model is consistent with recently reported X-ray structural data for a prokaryotic VKORC1 homolog protein. The VKORC1 primary sequence is indicated by amino acid single-letter codes in circles (small numbers indicate the amino acid position at every 10 residues). α-Helical TM segments are labeled TM1 to TM4. Black filled circles indicate the five completely conserved residues presumed to be catalytically active. The yellow boxed region encompasses the CIVC redox motif predicted to lie at the interface between the fourth TM helix and the endoplasmic reticulum (ER) lumen. The red boxed region surrounds residues presumed to be part of a binding site for 4-hydroxycoumarin-type oral anticoagulants. Cyan-shaded circles indicate the positions of the 17 specific VKORC1 residues that have been found to be mutated in humans with oral anticoagulant resistance.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients, materials, and methods
  5. Results
  6. Discussion
  7. Addendum
  8. Acknowledgements
  9. Disclosure of Conflict of Interests
  10. References
  11. Supporting Information

VKORC1 mutations associated with elevated OAC dosage requirements

Since the discovery of the VKORC1 gene in 2004, 11 different VKORC1 missense mutations have been reported to cause OACR in a total of 134 patients (Table S3). Of these, only Asp36[RIGHTWARDS ARROW]Tyr has been reported as a frequent sequence variant (80.2% of all published OAC-resistant cases with VKORC1 mutations), being often found in Jewish patients from Ethiopia (15% of the total population) and Europe (Ashkenazim, 4% of the total population) [33]. In our study, this mutation represents one-quarter of all patients (Table 1, six of 23 patients). With respect to OAC dosage, Asp36[RIGHTWARDS ARROW]Tyr appears to affect warfarin and phenprocoumon therapeutics in the same way. On the basis of a warfarin–phenprocoumon conversion factor [30] of 2.36, equal target INR, and VKORC1:c.−1639 GG and CYP2C9*1 genotypes, a comparable elevated dosage requirement is observed for both drugs (Fig. 3, Asp36Tyr+/−, compare the yellow and pink bars), as reported by Loebstein et al. [16] and D’Ambrosio et al. [23]. Only one Asp36[RIGHTWARDS ARROW]Tyr patient in our cohort (Table 1, patient 8) did not reach a therapeutic INR with an 80 mg week−1 (1.6 × HDT) highest attempted warfarin dosage. However, the aggregate data for heterozygous Asp36[RIGHTWARDS ARROW]Tyr patients (Fig. 3, Asp36Tyr+/−) suggest that a higher warfarin dosage of ∼ 140 mg week−1 (2.8 × HDT) would probably have been effective, as was the case for Asp36[RIGHTWARDS ARROW]Tyr patients 5 and 9 in our cohort. Interestingly, we observed another sequence variation at Asp36 (VKORC1:c.107A>G), causing a transition to Gly. For Asp36[RIGHTWARDS ARROW]Gly, the phenotype is partially resistant, requiring a highly elevated warfarin dosage, comparable to that for Asp36[RIGHTWARDS ARROW]Tyr (Fig. 3, compare pink bars).

The second most common mutation in our cohort is Val66[RIGHTWARDS ARROW]Met, which was found in three patients of the present cohort, and might represent a founder mutation in populations of Afro-Carribean origin, but the patient population studied so far is not large enough to unequivocally support this hypothesis. Similarly to patient 8 (Asp36[RIGHTWARDS ARROW]Tyr), who did not reach an effective warfarin dosage before aborting OAC therapy, patient 15 (Val66[RIGHTWARDS ARROW]Met) did not reach a therapeutic INR at 63 mg week−1 phenprocoumon (3.0 × HDT), but would probably have been treated successfully at higher dosages of either warfarin or phenprocoumon (Fig. 3, Val66[RIGHTWARDS ARROW]Met yellow and pink bars), as other patients in both the present cohort and in two previous studies were successfully treated at higher dosages (Table 1: patient 16, 72 mg week−1 phenprocoumon [3.4 × HDT]; patient 17, 294 mg week−1 warfarin [5.9 × HDT]) (Table S3: patients 3 and 5, 140–245 mg week−1 warfarin [2.8–4.9 × HDT]; patients 2, 6, and 11, 135–294 mg week−1 warfarin [2.7–5.9 × HDT]).

We found additional sequence variations at Ala26 [19], Ser52 [26], and Val66 [28] (VKORC1:c.75G>A, VKORC1:c.155C>G and VKORC1:c.197T>G, respectively) that cause transitions to Thr26, Trp52, and Gly66, respectively. We also identified two different Asn77 substitutions – Asn77[RIGHTWARDS ARROW]Tyr and Asn77[RIGHTWARDS ARROW]Ser (VKORC1:c.229A>T and VKORC1:c.230A>G) – and two new mutations were observed at Trp59 (VKORC1:c.176G>T and VKORC1:c.177G>T; Trp59[RIGHTWARDS ARROW]Leu and Trp59[RIGHTWARDS ARROW]Cys), in addition to the one first identified by Wilms et al. [18] (Trp59[RIGHTWARDS ARROW]Arg; VKORC1:c.175T>C).

As was the case for both Gly and Tyr mutations at Asp36, OAC dosage requirements were similar for patients with Met or Gly substitutions at Val66. This suggests that the mechanistic action of these mutations is probably position-specific and not dependent on the physicochemical properties of the specific amino acid side-chain. As therapy was aborted for patients with Ala26[RIGHTWARDS ARROW]Thr, Trp59[RIGHTWARDS ARROW]Leu, Trp59[RIGHTWARDS ARROW]Cys, and Asn77[RIGHTWARDS ARROW]Ser, no further conclusions can be drawn on these specific amino acid positions concerning OAC dosage, other than that these require a highly elevated phenprocoumon dosage if they are not completely resistant.

For OAC-resistant patients with VKORC1 missense mutations, we found a bias in VKORC1 and CYP2C9 haplotypes as compared with the general population. A combination of homozygous wild-type CYP2C9 and non-VKORC1*2 alleles would be expected in approximately 25% of patients with Caucasian ancestry, but was observed in 19 of 23 (82%) of patients with VKORC1 mutations in our cohort (Fig. 2). In all but four patients with OACR mutations, a homozygous c.−1639 GG genotype, which is a marker for non-VKORC1*2 haplotypes, together with a wild-type CYP2C9 genotype, was found. As the VKORC1*2 haplotype is correlated with halved VKORC1 mRNA expression, enzyme activity, and OAC dosage requirement [28], a combination of this haplotype with an OACR mutation on the same allele could mimic a normal or just slightly elevated OAC dosage phenotype, potentially masking a partial OAC-resistant phenotype. Such patients would probably evade clinical detection and genetic analysis. Similarly, homozygous CYP2C9 alleles causing reduced OAC metabolism might compensate for the increased dosage requirements of some patients with mutations in VKORC1, and result in seemingly normal dosage requirements, especially when warfarin or acenocoumarol are being used. An example of this effect is seen for four previously reported patients (Table S3, patients from lines 2, 4, 11, and 17) who are heterozygous for wild-type/non-wild-type CYP2C9 deficiency alleles in combination with the VKORC1 Leu27[RIGHTWARDS ARROW]Val or Asp36[RIGHTWARDS ARROW]Tyr OACR mutations, resulting in normal or only moderately increased warfarin dosage requirements (49–98 mg week−1; 0.98–1.97 × HDT). Such cases are likely to be overlooked in routine clinical practice.

Unfortunately, for 10 patients of our cohort, the elevated OAC dosage requirements led to discontinuation of OAC treatment. For example, in patient 20 (Table 1), although the Asn77[RIGHTWARDS ARROW]Ser mutation was diagnosed and the INR rose to 1.6, phenprocoumon therapy was aborted by the attending physician at 63 mg week−1 (3.0 × HDT). Likewise, for patients 1, 8, 12–15, 19, 22, and 23, OAC therapy was aborted, probably because of the fear of overdosing with the threat of subsequent bleeding events. The majority of these patients were subsequently treated with heparin to achieve anticoagulation. The tendency to abort OAC therapy for patients with high dosage requirements, despite approaching stable therapeutic INR values, represents a challenge for physicians treating patients with potential VKORC1 missense mutations. Therefore, we recommend that patients requiring OAC dosages > 2.0 × HDT (i.e. 42.2 mg week−1 phenprocoumon, 48.6 mg week−1 acenocoumarol, 99.6 mg week−1 warfarin, and 277.2 mg week−1 fluindione) to reach therapeutic INR levels, as well as patients who fail to reach therapeutic INR levels despite OAC doses as much as > 2.0 × HDT, should be referred for detailed genetic analysis of both CYP2C9 and VKORC1. On the basis of the data analysis in Fig. 3, at or above this recommended threshold for genetic testing referrals, most haplotype-dependent high-dosage patients (i.e. average HDT = 1.0, but also including up to two-fold higher dosages) would be excluded from referral, thus allowing targeting of costly analysis only to patients with a high likelyhood of actually having VKORC1 missense mutations. For most patients with identified VKORC1 missense mutations, OAC dosage can be incrementally increased until they either reach a stable, therapeutic INR or until they develop undersirable side-effects, in which case OAC therapy should be discontinued. Identifying OAC-resistant patients with outlying OAC dosage requirements and specific VKORC1 missense mutations could potentially lead to successful treatment of these patients, and would generate an important corpus of genotype–phenotype correlation data to aid our understanding of the molecular mechanisms of VKORC1-mediated OACR.

Effects of OACR mutations on VKORC1 structure and function

Recent studies of the enzymatic characteristics of vitamin K epoxide reductase (VKOR) have provided insights into some important differences in the function of wild-type and OAC-resistant mutant proteins. Lasseur et al. reported diminished VKOR enzymatic activity and kinetic parameters for warfarin-resistant rats with the VKORC1 Tyr139[RIGHTWARDS ARROW]Phe mutation (Km and Vmax of 77% and 67%, respectively, as compared with wild-type rats). The simultaneous reduction in both kinetic parameters results, nevertheless, in an overall enzymatic efficiency about equal to that of wild-type VKORC1 [42,43]. VKOR activity of these rats was found to be reduced relative to that of wild-type rats, and was attributed to lower mRNA levels than in the wild type. Similarly, Lasseur et al. [44] determined the enzymatic parameters for mice with complete warfarin resistance caused by a VKORC1 Trp59[RIGHTWARDS ARROW]Gly mutation and found that, again, overall VKOR activity was reduced for the mutant relative to the wild type. In this case, however, kinetic analysis revealed a more complicated situation, with two enzymatic components, one apparently with low affinity and high capacity for substrate, and greater warfarin sensitivity than the wild type, and the other with apparent high affinity and low capacity for substrate, but substantially warfarin-resistant relative to the wild type.

Earlier studies of a Welsh strain of warfarin-resistant rats – very recently shown to harbor the VKORC1 Tyr139[RIGHTWARDS ARROW]Ser mutation [45] – focused on both VKOR enzymatic parameters and reaction products [46–50]. Fasco et al. [47] reported on a reduction of VKOR enzymatic activity to 74% of that of the wild type. Furthermore, they identified (+)-3-hydroxyvitamin K as a specific byproduct, in addition to vitamin K quinone, representing up to 70% of the converted vitamin K epoxide substrate for the VKORC1 Tyr139[RIGHTWARDS ARROW]Ser mutant enzyme [48]. Interestingly, 3-hydroxyvitamin K was not produced by wild-type VKORC1. On the basis of earlier chemical modeling [51,52] and theoretical enzymatic considerations [53], Fasco et al. proposed detailed mechanisms for both VKORC1-specific reduction of vitamin K epoxide (K>O) and inhibition of VKOR activity by 4-hydroxycoumarins. Briefly summarized, K>O reduction would proceed via a substrate–thiol adduct that transiently forms between VKORC1 and K>O, resulting in opening of the epoxide oxirane ring to yield a hydroxyvitamin K intermediate species, existing in a tautomeric keto–enol equilibrium. Deprotonation of the enzyme-stabilized enol species would then result in formation of an enolate transition state (TS) that is resonance-stabilized through interactions with the enzyme. For wild-type enzyme, the enolate TS would then undergo hydroxyl elimination to yield the vitamin K quinone (K) product. It was further hypothesized that the warfarin-resistant enzyme, being less enzymatically efficient in substrate (K>O) to product (K) conversion (as experimentally demonstrated for the three OACR mutations discussed above), would have an equilibrium shifted in favor of protonation of the enolate TS state, thus effectively allowing 3-hydroxyvitamin K to be non-enzymatically formed and released from the enzyme active site before elimination of the hydroxyl group preceding the final formation of the quinone product by reductive elimination of the thiol adduct. OACs, because of their structural similarity to 2,3-substituted 1,4-naphthoquinones (vitamin K-like compounds), have long been hypothesized to achieve their inhibitory action on VKOR enzymatic activity by binding to VKORC1 and mimicking TS intermediates [50,53,54]. Thus, the mechanisms proposed by Fasco et al. imply a causal relationship between the diminished affinity of the mutant enzyme for the enolate TS intermediate, dimished VKOR activity, and diminished affinity for the 4-hydroxycoumarin inhibitor, resulting in the apparent resistance phenotype.

In order to assess a possible correlation between OACR for VKORC1 mutants and associated effects such as diminished VKOR activity, we compiled all of the available VKOR enzymatic activity data for mutant VKORC1 proteins from the literature. Table S4 summarizes the relative VKOR enzymatic activities as percentages of wild-type activity from six independent in vitro studies between 1983 and 2009 that measured VKOR activity for specific human, rat and mouse VKORC1 OACR mutations. Three of the mutant VKORC1 proteins studied were actually chimeras of point mutations from one species introduced into the cDNA from another species, so we do not discuss these further. Of the remaining results, VKOR activity data for 27 unique species-specific OACR mutations (four human, 14 rat, and nine mouse) indicate that the overwhelming majority (25 of 27) of these mutations result in diminished VKOR activity relative to the wild-type enzyme, consistent with the mechanistic model of Fasco et al., which links diminished VKOR activity to anticoagulant resistance. Included among these are the human OACR mutations Val29[RIGHTWARDS ARROW]Leu, Val45[RIGHTWARDS ARROW]Ala, Arg58[RIGHTWARDS ARROW]Gly, and Leu[RIGHTWARDS ARROW]128Arg, which suggests that OACR mutations, although widely distributed over the protein primary sequence (see Fig. 4), all result in diminished VKOR activity. Consistent with the Fasco et al. mechanistic model, we infer that OACR mutations at the ER lumenal side of VKORC1 could perturb the protein structure in such a way that the mutant enzyme is able to function, albeit with diminished VKOR activity and a lowered affinity for OACs. The major assumption is that there is a perturbation (by the mutated amino acid) of the spatial constellation of residues required for stabilizing OAC binding and/or efficient catalytic conversion of the vitamin K epoxide substrate. Given the close physical juxtaposition of the OACR mutations and the five completely conserved putative catalytic residues in our VKORC1 topological model (Fig. 4), this hypothesis appears plausible. According to our topology prediction, supported by the recently published protein structure of a prokaryotic VKORC1 homolog [35], all of the mutant residues conferring OACR are located within the lumenal loop or at the loop–membrane interface. We further infer from this line of argument that the vitamin K epoxide substrate and probably also OACs bind to VKORC1 on the ER lumenal side of the protein in close proximity to the catalytic residues.

Future perspectives for understanding OACR and patient treatment

Targeted in vitro mutational studies, together with advances in structural analyses of the VKORC1 enzyme with bound OAC inhibitors, will be required to understand the detailed molecular mechanism of OACR in humans and rodents.

The identification of VKORC1 mutations in patients with OACR phenotypes has helped to improve care for these patients by introducing novel non-standard OAC therapies, and has guided structural and functional investigations of VKORC1 enzyme function and OACR. In addition, we have presented a new diagnostic classification scheme for the severity of OACR phenotypes that allows comparison across all four OACs currently in use worldwide, taking into account the combined VKORC1:c.−1639G/CYP2C9 genotypes for defining a standard reference dosage level for the various OACs in order to aid systematic identification of OAC-resistant patients with outlying genotypes and phenotypes.

Addendum

  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients, materials, and methods
  5. Results
  6. Discussion
  7. Addendum
  8. Acknowledgements
  9. Disclosure of Conflict of Interests
  10. References
  11. Supporting Information

K. Sittinger, G. Spohn, and S. Rost: performed experiments; M. Watzka, C. Geisen, E. Seifried, C. R. Müller, and J. Oldenburg: designed the research and analyzed results; C. G. Bevans: contributed bioinformatic analysis and figure graphics; M. Watzka, C. G. Bevans, and J. Oldenburg: wrote the article; M. Watzka and C. Geisen: contributed equally to this study.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients, materials, and methods
  5. Results
  6. Discussion
  7. Addendum
  8. Acknowledgements
  9. Disclosure of Conflict of Interests
  10. References
  11. Supporting Information

We acknowledge all physicians sending patient samples to our laboratory for routine VKORC1 analysis.

Disclosure of Conflict of Interests

  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients, materials, and methods
  5. Results
  6. Discussion
  7. Addendum
  8. Acknowledgements
  9. Disclosure of Conflict of Interests
  10. References
  11. Supporting Information

The work of J. Oldenburg was supported by grants from the Deutsche Forschungsgemeinschaft (DFG – OL 100/3-1), the Bundesministerium für Bildung und Forschung – Forschungszentrum Jülich (BMBF/PTJ – 0312708E), the National Genome Research Net Cardiovascular Diseases (BMBF/DLR-01GS0424/NHK-S12T21), and Baxter Deutschland GmbH, Germany. The other authors state that they have no conflict of interest.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients, materials, and methods
  5. Results
  6. Discussion
  7. Addendum
  8. Acknowledgements
  9. Disclosure of Conflict of Interests
  10. References
  11. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients, materials, and methods
  5. Results
  6. Discussion
  7. Addendum
  8. Acknowledgements
  9. Disclosure of Conflict of Interests
  10. References
  11. Supporting Information

Fig. S1. Overview of multiple sequence alignment for 327 VKORC1 homolog proteins.

Fig. S2. Multiple prediction alignment (MPA) for VKORC1 and VKORC1-like 1 (VKORC1L1) predicted transmembrane α-helices upon which our consensus prediction is based.

Fig. S3. PolyPhobius topology predictions for VKORC1 and NST-glycosylation consensus site-tagged VKORC1.

Fig. S4. Multiple sequence alignment of VKORC1 for 13 vertebrate species showing human and rodent missense mutations responsible for OAC resistance phenotypes.

Table S1. Predicted transmembrane α-helical segments for human VKORC1.

Table S2. Predicted apparent free energy difference ΔGapp for predicted transmembrane α-helices of human VKORC1 and NST-glycosylation consensus site-tagged VKORC1 (based on enthalpic contributyions for amino acid type & position in the TM helix, helical hydrophobic moment and length).

Table S3. Previously reported human VKORC1 oral anticoagulant resistance mutations.

Table S4. Summary of published relative VKOR activities for OACR mutant VKORC1 proteins as percentages of wild-type activity.

FilenameFormatSizeDescription
JTH_4095_sm_FigS1-4-TableS1-4.doc1459KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.