Introduction

UDP-glucuronosyltransferases (UGTs) are considered among the most relevant phase II enzymes of drug metabolism. The reaction catalyzed by UGTs involves the addition of a glucuronic acid moiety to xenobiotics. Thus, UGT-mediated glucuronidation is a major path for the human body’s elimination not only of about 35 % of all drugs metabolized by phase II enzymes, but also of foreign chemical removal for most drugs, dietary substances, toxins and endogenous substances [1–3].

The human UGT is a large gene superfamily, including the UGT1A locus on chromosome 2q37.1, which encodes at least nine functional proteins such as UGT1A4 [4]. UGT1A4 is known to be the primary enzyme that catalyzes N-glucuronidation mostly of primary, secondary and aromatic amines, including clinically-important drugs such as phenytoin, lamotrigine, olanzapine, tamoxifen, trifluoperazine, tacrolimus and amitriptyline [5–7]. In addition, UGT1A4 shows O-glucuronidation activity towards steroidal compounds and phytochemicals [8]. Nakajima et al. [9] have demonstrated that UGT1A1 and UGT1A4 polymorphisms might be a cause of the interindividual differences in O-glucuronidation of phenytoin in human liver microsomes. Ehmer et al. [10] identified the UGT1A4 P24T (c.70C>A) and L48V (c.142T>G) variants that exhibited reduced glucuronidation activities with substrate specificity. These natural variations of the UGT1A4 gene show a differential hepatic metabolic activity toward mutagenic amines and endogenous steroids [10]. Recently, the variability of the clinical outcome in epileptic patients on monotherapy or polytherapy was found to be related to a decreased in the serum concentration of the antiepileptic drug, lamotrigine, due to the L48V polymorphism of UGT1A4 [11]. In addition, data obtained in human liver microsomes suggest that the L48V polymorphism also contributes significantly to interindividual variability in olanzapine metabolism [12]. Zhou et al. [13] developed an in vitro evaluation with three recombinant UGT1A4 enzymes (UGT1A4 wild-type, UGT1A4 P24T and UGT1A4 L48V) demonstrating that the presence of these two last variants may lead to interindividual variations in lamotrigine metabolism in vivo.

The functional implications for the genetic polymorphisms in UGT1A4 giving rise to the P24T and L48 V substitutions have been documented in several investigations [9–16]. In an association study, the L48 V change was studied in patients with schizophrenia suggesting that this SNP is associated with lower average plasma concentrations of olanzapine [12, 16]. Both, P24T and L48 V polymorphisms affect the UGT1A4 in vivo metabolism activity and may be good candidates for antiepileptic clinical response variation and as genetic causes of lamotrigine-induced adverse events [11, 16].

The synonymous UGT1A4*1b (C157C; c.471C>T) variant has been reported with a high frequency in Caucasians (20.9 % in French-Canadians and 15 % in Germans) [1, 10]. The C157C, P24T and L48V variants correspond to UGT1A4*1b, *2 and *3 alleles, respectively [17].

The frequency of the genetic polymorphisms in UGT1A4 has been reported for several populations revealing important differences among them [1, 10, 12, 17–23]; however there is no information regarding these polymorphisms in Mexican Mestizos (MM) and Spaniards (SP). Therefore, the main goal of this study was to determine the genotype and allele frequencies of UGT1A4*1b, UGT1A4*2 and UGT1A4*3 in a sample of MM and SP healthy volunteers.

Subjects and methods

Subjects

Participants were unrelated healthy volunteers of Mexican Mestizo ethnicity from Mexico City and surrounding areas (MMs; n = 248; 62 % female; age: 35 ± 16 years, range: 19–65 years) and of Caucasian Spaniard origin from Extremadura (SPs; n = 318; 51 % female; age: 43 ± 17 years, range: 18–70 years). The subjects were informed about the aims of the study and gave their written informed consent prior their participation. The study was performed according to the Helsinki Declaration and it was approved by the National Institute of Neurology and Neurosurgery (INNN) Internal Review Board (Mexico City, Mexico) and the Ethical Committee of the Extremadura University Hospital (Badajoz, Spain).

UGT1A4 genotyping

Genomic DNA was obtained from peripheral blood leukocytes by standard procedures. PCR–RFLP or real time PCR (RT–PCR) were employed to genotype the different allelic variants *1b, *2 and *3 of the gene UGT1A4 (MIM 606429). Nomenclature of the three gene variations analyzed was according to the official page of UGT nomenclature [24] The SNPs 70C>A; P24T (rs6755571) and 471C>T; C157C (rs2011404) were genotyped with commercially available TaqMan validated SNP assays (C_25957120_10 and C_8739217_10) from Applied Biosystems (Applied Biosystems, Carlsbad, Massachusetts, USA). Amplification was done in accordance with manufacturer′s instructions in an AB7300 real time PCR instrument. An allelic discrimination plot was used to identify individual genotypes (v2.0 software, Applied Biosystems). In order to scan the allelic variant 142T>G; L48V (rs2011425) (1A4*3), PCR–RFLP with the enzyme Stu I was performed with published primers [10, 16].

Subsequently, each allelic variant was confirmed by direct sequencing in five percent of the samples. Oligonucleotide primers and sequencing conditions were those published by Saeki et al. [15]. Purified amplicons were sequenced with Big Dye Terminator v3.1 Cycle Sequencing Kit and separated in a genetic analyzer AB3130 (Applied Biosystems, Carlsbad, Massachusetts, USA). All sequences were analyzed with Sequencing Analysis v5.3 and every allelic variant was confirmed by sequencing the product of an independent PCR on both strands and verified in different public databases. Nomenclature and positions of the diverse identified polymorphisms are relative to the adenine (+1) of the ATG of the reference sequence of NM_007120.2 → NP_009051.1.

Statistical analysis

The genotype and allele frequencies were assessed for Hardy–Weinberg equilibrium, differences in allele frequencies between the two studied populations and others were calculated by a Chi square test or Fisher′s exact test. A p value <0.05 was considered statistically significant. Statystical analyses were performed using GraphPad Prism 5.00 (GraphPad Software, Inc., San Diego; CA, USA). Linkage disequilibrium (LD) and haplotype analysis were performed using the Haploview software version 4.0 [25]. Haplotypes were inferred from individual genotypes using the software PHASE v2.1 [26]. Conditions employed for the haplotype inference procedure using 3 biallelic SNPs consisted in 100 iterations and 5 thinning interval steps through the Markov chain.

Results

The UGT1A4 genotype frequencies corresponded to those predicted by the Hardy–Weinberg law. The most common genotypes in SPs and MMs were wt/*1b and wt/*3, respectively. Interestingly the genotype UGT1A4*2/*2 was not identified in MMs and only in two SPs, being the least frequent genotype (Table 1). The allelic frequencies of UGT1A4*2 and UGT1A4*1b (Table 2) were higher (p < 0.05) in SP individuals (6.9 % and 17.4 %) than in MMs (3.4 % and 7.6 %), respectively. In contrast, the allelic frequency of UGT1A4*3 was higher (p < 0.05) in MMs than SPs. In Table 3, the allele frequencies found in this study are compared with those previously reported in other populations.

Table 1 Frequencies of UGT1A4 genotypes among Mexican and Spanish healthy volunteers
Full size table
Table 2 Frequencies of UGT1A4 alleles among Mexican-Mestizo (n = 248) and Spanish (n = 318) healthy volunteers
Full size table
Table 3 UGT1A4 allelic frequencies in the Mexican Mestizo and Spanish populations compared with other population studies
Full size table

Haplotypes were inferred from the genotype data (Fig. 1). A total of seven and six different haplotypes were observed in MM and SP populations, respectively. However, the same four haplotypes above 5 % were observed in both populations but with different frequencies. In SPs, the most common haplotype was the wild-type (wt) CTC (68.0 %), and only three more haplotypes were found above 5 %: CTT (16.6 %), CGC (7.8 %) and ATC (6.5 %). The frequency of these haplotypes and the corresponding SNPs suggests that the C>T mutation in position 3 occurred first and the other two happened later and independently (Fig. 1a). In MMs, the most common haplotype was also the wt CTC (75.7 %), but only two other haplotypes were found above 5 %: CGC (12.8 %) and CTT (8.0 %) (Fig. 1b). The haplotype ATC (3.2 %) was the fourth most common but with a frequency <5 %. All these haplotypes correspond to those where the three SNPs are found independently in the same wild-type background.

Fig. 1
figure 1

Haplotype structure of UGT1A4 allelic variants studied produced by Haploview software. (a) Spanish population; (b) Mexican population. (B = UGT1A4*2; C = UGT1A4*3b; A = UGT1A4*1b), relative distances B–C = 72 nt, C–A = 392 nt, B–A = 401 nt. (nt = nucleotide). D′ values 0–1 are shown. White boxes: D′ < 1 and LOD < 2; blue boxes: D′ = 1 and LOD < 2; red boxes: D′ = 1 and LOD > 2. LOD = log odds score for linkage disequilibrium. D′ = 1 indicates that the two SNPs have not been separated by recombination. LOD > 2 indicates strong evidence of LD. Lower D′ values indicate some evidence of recombination

Full size image

Discussion

To our knowledge, this is the first study that examined the allele frequencies of UGT1A4*1b, *2 and *3 in MM and SP individuals, showing significant differences in these three alleles between the two populations. Estimation of interethnic differences in relevant genetic polymorphisms is interesting for the clinical implementation of pharmacogenetics in order to optimize the drug treatments for a given population with regard to the ethnic background. This is one of the main goals to achieve by the Ibero-Latino-American network of Pharmacogenetics and Pharmacogenomics (RIBEF).

The synonymous C157C variant (UGT1A4*1b) was included in the present study because it showed high heterozygosis in an initial pre-screening. The UGT1A4*1b allele frequencies were different between MMs and SPs (Table 2). This is in agreement with the 1000 Genomes Project data, where a higher allele frequency of UGT1A4*1b was observed for Europeans (18 %) vs. admixed Americans (13 %). Indeed, this allele is very rare in Asian (1 %) and African (4 %) populations [23].

UGT1A4*2 frequency was two times higher in SPs than in MMs. In contrast, UGT1A4*3 was lower in SPs than in MMs. These allele frequencies also presented interethnic differences with other published populations (Table 3). UGT1A4*2 and UGT1A4*3 frequencies in SPs were similar to those reported in other Caucasian populations [1, 10, 19]. However, in MMs, the frequency of UGT1A4*2 was lower and UGT1A4*3 was higher than in several Caucasian populations. The frequencies of UGT1A4*2 in SPs and MMs were higher than those observed in Asian populations [15, 18, 21], whereas the UGT1A4*3 frequencies were lower in SPs but similar in MMs compared to Asians. In addition, the allele frequency of UGT1A4*2 in Jordanians was similar in SPs but higher than in MMs, while UGT1A4*3 in Jordanians was two and three times lower than in SPs and MMs, respectively [22]. In Africans UGT1A4*2 is absent [23], being higher in both SPs and MMs, whereas UGT1A4*3 frequency was similar to both SPs and MMs observed frequencies.

In general, contrary to the relative homogeneity observed in SPs within Caucasians for UGT1A4*2 and UGT1A4*3, MMs exhibited differences when compared to Caucasians [10, 19]. While the UGT1A4*2 frequencies were found to be similar in SPs and Jordanians, UGT1A4*3 frequencies were similar among SPs, MMs and Africans and between MMs and Asians. Thus, the distribution of UGT1A4*2 and UGT1A4*3 in the MM population might be influenced by its admixture characteristics. Genetic admixture and diversity estimations in MM have been characterized, revealing a 69 % of Amerindian, 26 % of European, which comes mainly from Spain (Castilla, Andalucía, and Extremadura), and 5 % of African contribution [27].

The three analyzed SNPs variants in this genetic region indicated positive LD for UGT1A4*2 and UGT1A4*3 for the SP population. However, these variants are not in LD in the MM population, suggesting that these mutations have arisen independently in the same genetic background.

The substitutions P24T and L48 V have been observed to be in LD in Germans, but not in the French-Canadian population [10, 17]. The LD pattern observed in the French-Canadian population was similar to that in the MM. In contrast, Haploview analysis indicated positive LD between UGT1A4*2 and UGT1A4*3 variants for the SP population, in a manner similar to Germans. These allelic variants have been associated with reduced glucuronidation activities [10]. The discrepancies in LD pattern between SPs and MMs could be attributed to ethnic differences and additional studies must clarify whether this non-random association of these alleles may contribute to differences in the glucuronidation of UGT1A4.

Present data confirm that the three analyzed UGT1A4 variants are highly polymorphic in the studied MM and SP populations. Functional variants of UGT1A4 have been shown to exhibit differences in glucuronidation capacity that may contribute significantly to interindividual variability in drug metabolism [11, 16]. Taking into account that UGT1A4 metabolizes many therapeutic drugs widely used in SPs and MMs, it is plausible to suggest that the observed genetic differences in both populations might partially explain differences in drug responses among them. In the light of these findings, we could hypothesize that drugs metabolized by UGT1A4 may show different profile of efficacy/adverse effects in MM and SP populations. Therefore, further research on different ethnic groups in America is warranted.