Introduction

Human gamma-glutamyl carboxylase gene (GGCX), which is located on chromosome 2p12 (Kuo et al. 1995), spans about a 13-kb genomic region and consists of 15 exons (Wu et al. 1997). It encodes a 758-amino-acid protein that appears to have three transmembrane domains near its amino terminus (Wu et al. 1991). GGCX is also known as the vitamin K-dependent carboxylase, and activates a subset of proteins with calcium-binding properties, vitamin K-dependent (VKD) proteins or γ-carboxyglutamate-proteins (Gla-proteins), by catalysing them (Suttie 1985). Conversion from glutamyl residues to γ-carboxyglutamyl residues within VKD proteins allows VKD proteins to chelate calcium ions and subsequently undergo metal-dependent conformational alteration that is essential for these proteins to exert its biological functions (Furie and Furie 1988). Examples of VKD proteins are prothrombin, blood coagulation factors VII, IX, X, protein C, and protein S (FAO/WHO expert consultation on human vitamin and mineral requirements 1998). These proteins require post-translational carboxylation before they can exert their functions in hemostasis (Furie and Furie 1990).

Previous studies demonstrated that post-translational carboxylation by GGCX is impaired in the presence of a vitamin K antagonist such as warfarin, an oral-anticoagulant, which creates the anticoagulant state by interfering with the production of reduced vitamin K, an essential cofactor for the carboxylation reaction (Esmon et al. 1975; Horton and Bushwick 1999). Although most of the variability observed in patients’ response to warfarin may be attributed to genetic varations in VKORC1 (Mushiroda et al. 2006; Schwartz and Stein 2006) that functions to regenerate reduced vitamin K (Rost et al. 2004; Li et al. 2004), polymorphisms in GGCX have also been indicated to have some association with inter-individual variation in warfarin maintenance-dose requirement (Shikata et al. 2004; Chen et al. 2005; Loebstein et al. 2005; Wadelius et al. 2005; Herman et al. 2006; Kimura et al. 2007; Vecsler et al. 2006).

In this study, we report identification of 41 SNPs, seven insertion/deletion polymorphisms (indels), and a microsatellite polymorphism in an 18-kb genomic region corresponding to human GGCX by direct sequencing of DNAs from 96 Japanese individuals, and also the haplotype structure of this region. We also evaluated the association between polymorphisms in GGCX and the warfarin maintenance-dose requirements of our Japanese subjects.

Materials and methods

Population samples

Eight hundred and twenty eight Japanese individuals under warfarin therapy were recruited for the current study. All individuals had given written informed consent to participate in the study in accordance with the process approved by the Ethical Committees of the Institute of Medical Science, University of Tokyo, Japan.

SNP discovery and genotyping methods

Initially, we carried out SNP discovery by using DNA samples of 96 of the 828 above-mentioned Japanese individuals; 32 individuals from each of three categories of patient (those requiring high, medium and low warfarin maintenance doses). The mean daily warfarin dose requirements of patients from each of these categories are 6.625, 3.5, and 0.95 mg respectively.

Using the genomic sequence information (GenBank accession number, AC016753.9), we designed 20 sets of primers (Supplementary Table 1) to amplify the 18-kb genomic region from 2.4 kb upstream of the first exon to 2.5 kb downstream of the last exon of GGCX, except for the repetitive sequences screened by the Repeat-Masker program (http://www.repeatmasker.org/). We defined exon–intron boundaries of GGCX by comparison of genomic sequences with cDNA sequences (GenBank accession number, NM_000821.3) using the NCBI nucleotide analysis tool Spidey. For each of the 96 DNA samples, PCR was performed with 10 ng of genomic DNA in a total reaction volume of 20 μl. All PCRs were performed by using GeneAmp PCR system 9700 (Applied Biosystems, USA) and used Ex Taq DNA polymerase (Takara Bio Inc, Japan).

In general, we performed PCR with an initial denaturation step of 94°C for 5 min, followed by 37 cycles of denaturation at 94°C for 30 s, annealing at 60°C for 30 s, and extension at 72°C for 1 min. These cycling reactions were followed by a final extension at 72°C for 5 min. For fragments 7, 9, and 10, we performed amplification as above, but increased the extension time of the 37 cycles from 1 to 2 min. We carried out SNP discovery by direct sequencing of the PCR products by using 96-capillary 3730xl DNA Analyzer (Applied Biosystems, USA). We sequenced all amplified fragments by two pairs of sequencing primers, except for fragments 5A and 5B, which were each amplified by only a pair of primers. All SNPs were detected by the Polyphred computer program (Nickerson et al. 1997) and were confirmed by sequencing both strands of each PCR product.

Statistical analyses

We calculated and tested genotype frequencies for each marker with the standard Chi-square test of Hardy–Weinberg Equilibrium (HWE) (Weir 1996). We used only polymorphisms with minor allele frequencies of greater than 0.1 for subsequent haplotype analyses. We estimated haplotypes and corresponding haplotype frequencies using the Expectation–Maximization (EM) algorithm (Slatkin and Excoffier 1996; Kitamura et al. 2002).

Association study

We performed both the Kruskal–Wallis and Mann–Whitney U tests to evaluate association between each of the polymorphisms with the warfarin maintenance-dose requirement of the 96 individuals. For the CAA microsatellite in intron 6 of GGCX, we performed the Mann–Whitney U test to examine whether an absence or a presence of a particular allele or genotype of the CAA repeats is associated with the warfarin maintenance-dose requirement of our subjects.

We increased the number of individuals genotyped for a particular polymorphism if we detected possible association between that polymorphism with the warfarin maintenance-dose of the 96 individuals. We also genotyped more individuals for a cSNPs (rs699664) and a microsatellite (rs10654848) because previous studies (Shikata et al. 2004; Kimura et al. 2007) reported that these polymorphisms influence the warfarin dose requirement of Japanese individuals. In all cases, we PCR-amplified and direct sequenced the genomic region containing the particular polymorphism under investigation.

Results and discussion

Direct sequencing of an 18-kb genomic region corresponding to GGCX, using DNAs from 96 Japanese patients treated with warfarin, identified 49 genetic variations including 41 SNPs, seven insertion/deletion polymorphisms, and one microsatellite polymorphism. Six polymorphisms are found in the 5′ flanking region, seven in exons, 27 in introns, and nine in the 3′ flanking region of GGCX. On average, genetic variations were found in every 367 nucleotides. Figure 1 illustrates genomic organization of the 18-kb region and locations of all genetic variations, and their detailed information is summarized in Table 1.

Fig. 1
figure 1

Genomic organization of the human Gamma-glutamyl carboxylase gene (GGCX) and locations of the 49 polymorphisms detected in the 18-kb genomic region containing GGCX that we examined in this study. Tall, grey boxes illustrate exons, while short, black boxes represent introns. The numerical number on top of each grey box indicates the number of each exon

Full size image
Table 1 Characterization of 49 genetic variants in the 18-kb genomic region containing GGCX. Both 5′ and 3′ sequences flanking the genetic variants are written in lower case letters, while the variants themselves were highlighted in capital letters
Full size table

Among the 49 polymorphisms we discovered in this genomic region, 32 variations—27 SNPs and five insertion/deletions—were not deposited in the public database, and are considered to be novel. In addition, 19 polymorphisms deposited in the dbSNP database were not polymorphic in the 96 Japanese individuals we used in this study. Genotype distributions, minor allele frequencies (MAF), and P value for the Chi-square test of HWE for all loci, except for GGCX-49, are shown in Table 2.

Table 2 Genotype distributions and minor allele frequencies (MAF) of the 48 polymorphisms detected. Also shown are the P value of the Chi-square test of Hardy–Weinberg Equilibrium (HWE) of genotypes at each locus, as well as P values of the Kruskal–Wallis and Mann–Whitney U tests of the association between each polymorphism and the warfarin maintenance-dose requirements of the 96 Japanese individuals
Full size table

All insertion/deletion polymorphisms detected in this study except for GGCX-42 are biallelic (a presence or an absence of the particular inserted/deleted nucleotide(s)) but GGCX-42 was found to be a tri-allelic polymorphism of eight, nine, and ten repeats of nucleotide G. For GGCX-49, a microsatellite polymorphism at position -179 in intron 6, seven alleles and ten genotypes (Table 3) have been identified.

Table 3 Genotype distributions of GGCX-49 (Microsatellite Intron 6 -179 (CAA)n) and association between different alleles and genotypes of the microsatellite with warfarin maintenance-dose requirement of the 96 Japanese individuals
Full size table

Among the seven polymorphisms located in the exonic regions, three cause amino-acid substitutions (Pro113Leu in exon 1, Arg325Gln in exon 8, and Val460Ile in exon 10). GGCX-46, an insertion of TAAA, is located in the untranslated region of exon 15, at one of the two concensus polyA+ addition signals of GGCX. However, whether this insertion alters the function of the poly A+ addition signal of GGCX remained unknown.

Results of the Chi-square test of HWE revealed that 39 of the 41 investigated SNPs, as well as all insertion/deletions, fit the law of HWE. Two SNPs that did not follow the Hardy–Weinberg equilibrium (P ≤ 0.05), GGCX-1 and GGCX-33, are non-exonic SNPs. As summarized in Table 2, 18 polymorphisms have minor allele frequencies of greater than 0.1. With reference to the genotyping data, we eliminated six SNPs that are in absolute linkage disequilibrium (LD) with others, and used genotype information of the remaining 11 SNPs (including GGCX-42) and GGCX-49 (the microsatellite) for subsequent haplotype estimation by the EM algorithm. Twenty-one possible haplotypes of the 12 polymorphisms have been estimated, and ten of them that have frequencies of greater than 0.01 are listed in Table 4. We found that four major haplotypes (Haplotype 1–4) could cover 79% of the haplotypes in this genomic region.

Table 4 Haplotypes and corresponding frequencies as estimated by EM-algorithm (Kitamura et al. 2002) by using information of 12 polymorphisms. There are ten haplotypes with haplotype frequencies of greater than 0.01, as shown in this table
Full size table

For the association with the warfarin dosage, we screened primary results for the 96 Japanese subjects treated with warfarin by the Kruskal–Wallis and Mann–Whitney U tests (as summarized in Tables 2 and 3) and found one, GGCX-17, of the 49 polymorphisms revealed a possible association with a P value of 0.022 with the warfarin maintenance dose. In order to further evaluate the significance of this association, we genotyped an additional 732 Japanese individuals for this SNP, but no association was detected (P = 0.394).

In addition, we further investigated rs699664 G>A, because it causes a substitution of Gln for Arg at codon 325 and was reported previously by Kimura et al. (2007) to significantly relate to the warfarin maintenance dose of their 93 Japanese patients (P = 0.022). However, we genotyped 362 Japanese individuals and found no association (P = 0.636) between this cSNP and the warfarin maintenance dose. Similarly, genotyping results of 365 individuals at the CAA microsatellite of GGCX revealed no association with the warfarin maintenance dose of our patients.

GGCX, which plays an important role in the vitamin K cycle, is a candidate gene for the investigation of inter-individual differences in the warfarin maintenance dose requirement. Although warfarin has been the most frequently prescribed oral anticoagulant in prevention of thromboembolism, administration of warfarin to achieve sufficient suppression of thrombosis without causing undesired bleeding has remained a challenge for physicians, owing to a large variation in the inter-individual dose requirement (Daly and King 2003). The findings that recently known genetic and environmental factors only partially explain inter-individual variability in warfarin maintenance-dose requirements (Loebstein et al. 2005; Schwarz and Stein 2006) indicate the need to uncover other unknown genetic determinants of the response to warfarin for better administration of this oral anticoagulant.

Although we failed to identify any polymorphisms that are the genetic determinants of the warfarin maintenance dose of our Japanese subjects in the current study, the high-resolution SNP map within the 18-kb genomic region containing GGCX should serve as a useful resource for further pharmacogenetic studies for various drugs in whose pharmacological pathway GGCX may be involved.