Phenotypic polymorphism and gender-related differences of CYP1A2 activity in a Chinese population

Abstract

Aims

To investigate the distribution characteristics of CYP1A2 in a Chinese population, and to examine gender-related differences in CYP1A2 activity.

Methods

Two hundred and twenty-nine healthy subjects, 120 men and 109 women, were enrolled in this study. CYP1A2 activity was measured by plasma paraxanthine/caffeine (1,7X/1,3,7X) ratio 6 h after administration of 300 mg caffeine. The concentrations of paraxanthine and caffeine in plasma were detected by h.p.l.c.

Results

A 16-fold variation of CYP1A2 activity (range 0.09 to 1.46) was shown in this study. The coefficient of variation (CV%) of CYP1A2 activity was 62.9%. Non-normal distribution of CYP1A2 activity was indicated by the Shapiro-Wilk test (P < 0.001). Probit plots of CYP1A2 activity revealed a bimodal distribution with breakpoint of 1,7X/1,3,7X ratio of 0.12. The percentage of poor metabolizers (PMs) was 5.24% (95% CI: 2.35%~8.13%) in this Chinese population. Residual analysis of the data also supported bimodality (P < 0.01). The CYP1A2 activity of men was higher than that of women (median: 0.33 vs 0.23, P < 0.001). A probit plot of CYP1A2 activity in men was shifted to the left compared with that in women. Based on phenotype, the gender-related difference was observed in extensive metabolizers (EMs) (P < 0.001), but not in PMs (P > 0.1). In addition, there was no sex-related difference in the incidence of PMs (P > 0.1).

Conclusions

There is a phenotypic polymorphism in CYP1A2 activity in this Chinese population, and CYP1A2 activity is higher in men than that in women.

Introduction

CYP1A2 is an important member of the cytochrome P450 superfamily [1]. The coding gene CYP1A2, which locates on chromosome 15 in humans, contains 6 introns and 7 exons, and the gene length is about 7.8 kb. CYP1A2 is mainly expressed in human liver and the protein content of CYP1A2 contributes 13% of the total CYP protein in liver [2]. CYP1A2 activity can be used to monitor the alteration of liver function in clinical practice [3]. CYP1A2 is involved in the metabolism of many clinical drugs, environmental toxins and endogenous substrates [4, 5]. It may play a critical role in the activation of a wide range of procarcinogens to genotoxic intermediates or ultimate carcinogens [5]. Individual variation in CYP1A2 is relevant to drug efficacy, adverse reactions and susceptibility to certain carcinoma [5–7].

In vivo activity of CYP1A2 can be measured by administration of probe drugs such as caffeine, phenacetin and theophylline [5, 6, 8]. Presently, caffeine is the most commonly used probe because of its low toxicity and good acceptance [8]. Nine enzymes including CYP1A1, CYP1A2, CYP2A6, CYP2D6, CYP2E1, CYP3A4, CYP3A5, NAT2 and XO (xanthine oxidase) are involved in the metabolism of caffeine and at least 14 metabolites are formed by demethylation and hydroxylation [8]. Of all the metabolic pathways, caffeine 3-demethylation is the most prominent reaction accounting for 80% of systemic caffeine clearance [8, 9]. Since caffeine 3-demethylation is almost entirely mediated by CYP1A2, and the contribution of the other CYPs in this reaction is negligible, both systemic clearance and 3-demethylation partial metabolic clearance of caffeine can serve as a marker of CYP1A2 activity or as a convenient standard for evaluating CYP1A2 activity.

A number of different parameters based on caffeine metabolism have been selected as markers of CYP1A2, which include respiratory excretion of ¹³CO₂ or ¹⁴CO₂ following a test dose of caffeine labelled at the 3-methyl group by isotope ¹³C or ¹⁴C, urinary caffeine metabolic ratios (UCMRs) and plasma caffeine metabolic ratios (PCMRs) [8, 10]. These parameters are calculated by various formulae based on the metabolism of caffeine and its metabolites. All may, to some extent, be used to reflect the activity of CYP1A2. However, previous studies have shown that not only the results in the same population measured by different caffeine metabolic parameters, but also those in different populations measured by the same caffeine metabolic parameters are inconsistent [10–18]. Various investigators have reevaluated which caffeine metabolic parameters are the most suitable and reliable markers of CYP1A2 [8, 19, 20].

Although the caffeine breath test has been validated, it is still a relatively impracticable approach for large scale population studies [8]. Several UCMRs have been proposed to reflect CYP1A2 activity, but their validation remains controversial. Notarianni and coworkers carried out a mathematical comparison with five UCMRs to measure CYP1A2 activity in 230 healthy subjects and found that the five ratios reflected at least three different entities [11]. The urinary ratio of (AAMU or AFMU+1X+1 U)/1,7 U [(5-acetylamino-6-amino-3-methyluracil or 5-acetylamino-6-formylamino-3- methyluracil+1-methyl-xanthine+1-methyluric acid)/ 1,7-dimethyluric acid] was claimed as a promising candidate for measurements of CYP1A2 activity compared with other UCMRs [8, 13, 14]. In contrast, Butler and her coworkers thought the 4~5 h urinary ratio of (1,7X+1,7 U)/1,3,7X [(1, 7-dimethylxanthine+1, 7-dimethyluric acid)/caffeine] after dosing caffeine was more accurate in reflecting CYP1A2 activity by comparison with the previously proposed ratios (AAMU+ 1X+1 U)/1,7 U (AFMU+1X+1 U)/1,7 U and 1,7X/1,3,7X [21].

UCMRs, based on either secondary or tertiary metabolites, are not ideal indices of CYP1A2 activity because these metabolites are not formed exclusively by CYP1A2. Furthermore, these ratios might vary substantially with factors such as urinary flow, interethnic differences in renal function and different sampling protocols [8, 10]. Thus, another ‘simple and reliable’ CYP1A2 activity marker, 1,7X/1,3,7X ratio in plasma or in saliva 5–7 h postdosing caffeine was proposed and validated by taking the systemic clearance of caffeine as a [10]. Later, Fuhr and coworkers further confirmed that the 1,7X/1,3,7X ratio at 6 h postdosing appeared to be the most advantageous parameter, after they assessed the correlation of 1,7X/1,3,7X ratio in plasma and in saliva with the frequently used ratio (AAMU+1X+1 U)/1,7 U in urine. That has also been supported theoretically by others [20].

In previous reports, up to 200-fold interindividual differences in CYP1A2 activity were found [12, 21, 22]. The large variability of CYP1A2 activity implies genetic polymorphism. However, there is controversy as to the exact distribution of this activity, which has been described as unimodal [12, 13, 16–18], bimodal or trimodal in various population studies [10, 17, 21, 23–25]. CYP1A2 activity can be influenced by environmental factors. As reported, CYP1A2 activity may be increased by exposure to cigarette smoking [26], charbroiled food [16], caffeine-containing drinks [16, 27], and heavy exercise [16], while the inhibitory effects may appear after administration of some drugs such as oral contraceptives [28], quinolones [29], furafylline [30], and fluvoxamine [31]. In females, CYP1A2 activity varies with menstrual cycle and pregnant status [32]. For gender-related differences in CYP1A2, results from different reports are divergent [12, 15, 18, 24].

The present study was designed to investigate the distribution characteristics of CYP1A2 in Chinese population using the plasma 1,7X/1,3,7X ratio at 6 h after administration of caffeine as a CYP1A2 activity index and to test the hypothesis of gender-related differences in CYP1A2 activity.

Methods

Subjects

This study was approved by the Ethics Committee of Hunan Medical University. Two hundred and twenty-nine healthy Chinese (Han) living in Changsha, 120 men and 109 women, participated in this study after they gave informed consent. All participants were medical students. Age (mean±s.d.) of them was 20±1 years (range 18–23 years). The average weight of men was 60.4±7.4 kg (range 45–87 kg), and that of women was 50.3±4.7 kg (range 45–87 kg). Each subject was in good health on the basis of medical history, physical examination and laboratory evaluation. No individual had taken any medications for at least 2 weeks before the study. None was a habitual user of caffeine (from coffee, tea or soft drinks) or a regular drinker of alcohol. Smokers and contraceptives users were also excluded from this study since smoking and contraceptives induce and inhibit CYP1A2 activity, respectively.

Chemicals

Paraxanthine (1,7X), caffeine (1,3,7X) and β-hydroxyethyl-theophylline (used as internal standard, IS) were purchased from Sigma Chemical Co. (St Louis, Mo.). Methanol and acetonitrile were of h.p.l.c. grade and other reagents were of AR grade unless indicated.

Experimental protocol

Subjects were required to avoid methylxanthine- containing foods and beverages for at least 48 h before the study and during the entire study. Each subject ingested 300 mg anhydrous caffeine as a single oral dose with a glass of water (about 150 ml) in the morning of the study. A 3 ml blood was drawn at 6 h postdosing, and plasma was isolated and stored at −20° C until assayed.

Analytical procedures

1,7X and 1,3,7X in plasma were measured by h.p.l.c [33]. A 300 μl aliquot of thawed plasma was mixed with 100 μl of IS (100 μm ) and 300 mg of ammonium sulphate. The sample was extracted with 5 ml chloroform and isopropanol (9:1). After being vortexed for 1 min, centrifuged at 1500 g for 5 min, the organic phase was evaporated to dryness at 45–50° C under nitrogen. The residue was dissolved in 100 μl mobile phase, and a 20 μl aliquot was injected on to the column. The h.p.l.c. system consisted of a Hewlett-Packard 1050 series, a reversed-phase column (Spherisorb ODS-2, 250 mm×4 mm ID, 5 μm particle size). The solvent used for elution was acetic acid 0.05% (A), acetonitrile (B) and methanol (C). Typical conditions for elution are 81.5% A, 8.5% B and 10% C for 0–5 min, 81.5–72% A, 8.5–18% B and 10% C for 5–13 min. A linear gradient was used for all solvent changes. The flow rate was 0.7 ml min⁻¹. The detection wavelength was 282 nm.

Data analysis

The data were illustrated by probit plots [16, 21], and were examined for normality of distribution by the Shapiro-Wilk test of normality. The possibility of bimodality was further assessed by residual analysis developed by Endrenyi et al. [34]. Median or mean values of CYP1A2 activity in men and women were compared using the Wilcoxon test or t-test. The Chi-square test was applied to compare the frequency of CYP1A2 phenotypes between subgroups. A P value of less than 0.05 was considered to be statistically significant. The software SPSS 7.0 was used for statistical analysis.

Results

H.p.l.c

1,7X, IS and 1,3,7X were eluted rapidly with a complete resolution and sharp symmetrical peaks. Run time for the substances of interest was 13 min. 1,7X and 1,3,7X were quantified using calibration curves obtained from blank plasma spiked with pure standards. The calibration curves for 1,7X and 1,3,7X were linear within the range from 1 to 100 μm and from 1 to 200 μm, respectively, with correlation coefficients of 0.99 and 0.99. Relative recovery for both 1,7X and 1,3,7X was greater than 96% at 5, 25 and 100 μm. Both intra-and interday variation coefficients were less than 10%. The detection limit was 0.1 μm.

CYP1A2 activity

The molar plasma concentration ratio of 1,7X/1,3,7X was selected as the CYP1A2 activity index. The median CYP1A2 activity was 0.27 in the 229 subjects. A 16-fold variation of CYP1A2 activity was observed with a range of from 0.09 to 1.46. The coefficient of variation (CV%) of CYP1A2 activity in this population was 62.9%. The CYP1A2 activity was not normally distributed as assessed by the Shapiro-Wilk test of normality (W = 0.8015, P < 0.001). Probit transformation of log (1,7X/1,3,7X) in the subjects resulted in clear bimodal probit plots with a breakpoint at a 1,7X/1,3,7X ratio near 0.12 (Figure 1). This suggested that the subjects could be grouped into poor metabolizers (PMs) or extensive metabolizers (EMs) according to their of 1,7X/1,3,7X ratios. The percentage of PMs was 5.24% (95% CI: 2.35~8.13). On the basis of probit plots, the residual analysis of ‘linear model fitness’ offered further statistical evidence of bimodality (F_2,225=28.26, P < 0.01) [34]. When sorted by gender, bimodality was seen in men (F_2,116=12.64, P < 0.01), as well as in women (F_2,105=8.98, P < 0.01).

Figure 1.

Probit plot and frequency distribution of CYP1A2 activity index in 229 Chinese subjects.

Gender-related differences

The median value of 1,7X/1,3,7X of men (0.33) was higher than that of women (0.23) (Z = 5.27, P < 0.001). Grouped by phenotype, men had higher CYP1A2 activity than women in EMs (t = 3.60, d.f.=215, P < 0.001), but there were no significant gender related differences in PMs (t = 1.67, d.f.=10, P > 0.1) (Table 1). The effect of weight on CYP1A2 activity was assessed by correlation analysis, because the average weight of men was more than that of women (t = 12.21, d.f.=227, P < 0.0001). The results showed that there was a negative correlation between CYP1A2 activity and weight in men (r_s=−0.38, d.f.=118, P < 0.01) but there was no correlation in women (r_s=0.05, d.f.=107, P > 0.5). These results suggested that the observed gender differences in CYP1A2 activity were not due to the confounding effect of weight of subjects. Probit plots of CYP1A2 activity in men shifted right more obviously compared with women (Figure 2). However, there were no sex-related differences in the number of PMs (χ²=1.13, d.f.=1, P > 0.25).

Table 1.

Gender differences of CYP1A2 activity (mean±s.d.) in the phenotyped subjects.

Figure 2.

Probit plots of CYP1A2 activity index in men (○) and women (•).

Discussion

Considerable interindividual variability in CYP1A2 activity with 16-fold variation among the subjects was observed in this study. This is consistent with previous findings showing 15- to 40-fold variations in CYP1A2 mRNA level [35] and 67-fold interindividual variation in CYP1A2 protein content in human liver [36]. Similar variation in level of hepatic microsomal CYP1A2 has also been shown by several in vitro studies [2]. Furthermore, 6-to 200-fold variations of CYP1A2 activity have been reported in in vivo studies with different CYP1A2 markers, derived mainly from urinary metabolites of caffeine [13, 21, 22, 24]. The large interindividual differences in CYP1A2 may result from genetic variation and environmental influences. All of the subjects in this study were medical students from the same university and exposed to similar inducers or inhibitors from the environment and diet. Hence, the interindividual variations of CYP1A2 activity observed in this study may reflect mainly by genetic variation. The striking variability of CYP1A2 raises the possibility of genetic polymorphism.

The bimodal distribution characteristic of CYP1A2 activity in this study suggested the existence of poor and extensive CYP1A2 phenotypes in the mainland of China. Butler and her coworkers have studied CYP1A2 activity with (1,7X+1,7 U)/1,3,7X in subjects from China, Italy and Arkansas. They found a trimodal distribution of CYP1A2 activity among nonsmokers in each of the three populations [21]. The trimodal distribution was also observed in subjects from Western Australia [23]. Nakajima and his coworkers found a bimodal distribution in Japanese with the same ratio of (1,7X+1,7 U)/1,3,7X, and they suggested that the poor phenotype of CYP1A2 was inherited as an autosomal recessive trait [24]. When CYP1A2 activity was measured by the ratio of 1,7X/1,3,7X in plasma, which was used in present study, the data complied with a bimodal distribution in both smoking and nonsmoking groups [10]. Nevertheless, a normal or lognormal distribution of CYP1A2 activity was also observed in several other reports, where (AFMU+1X+1 U)/1,7 U or (AAMU+1X+1 U)/1,7 U was employed as a marker of CYP1A2 activity [12, 16, 18].

The different findings in the above-mentioned studies may result partially from the different CYP1A2 activity markers [5, 10, 13]. Another possible explanation for the divergent findings could be the composition of the population studied. For example, smoking can increase CYP1A2 activity and shift the breakpoint in the histogram of probit-plots [24]. It is likely that the analysis of such a mixed population results in the masking of a possible polymorphism of CYP1A2 activity. Additionally, it may be due to differences in data processing [39, 40]. Genetic analysis at the molecular level will be required to provide definite evidence that CYP1A2 exhibits a genetic polymorphism [34]. The molecular basis of phenotypic polymorphism can often be ascribed to a mutation or deletion in the genes themselves. Although it has been proposed that the observed phenotypic polymorphism of CYP1A2 has genetic basis, limited studies have not identified specific structural gene variations, which can explain the alteration in the expression of CYP1A2 [5, 24, 42]. Recently, a novel rare point mutation of C₂₈₆₈→G in exon 2, which causes a Phe₂₁ to Leu change, was detected in a Chinese (Han) population living in Taiwan. The significance of this mutation remains unknown [43]. However, numerous cis-acting elements and trans-acting factors could participate in regulating the expression of the human CYP1A2 gene. Early studies found two XREs (xenobiotic-responsive-element-like sequences), termed X1 and X2, in the CYP1A2 5′-flanking region, and the interaction of X1 with the AhR complex was thought to be required for full transcriptional activation of CYP1A2 gene by 3-MC (3-methylcholanthrene) [44]. Recently, two functional polymorphisms were found to relate to the inducibility of CYP1A2 by smoking in intron 1 and 5′-flanking region of human CYP1A2 gene, respectively [45, 46]. Another study by Chung and coworkers demonstrated that there were three protein binding sites located at the 259-bp fragment (bp-2353 to bp-2094) of CYP1A2. These sites contain activator protein-1, nuclear factor-E1.7, and one-half hepatic nuclear factor-1 (HNF-1) binding consensus. Of the cis-acting factors, HNF-1 could contribute to the liver-specific expression of human CYP1A2 [47, 48]. Raffalli-Mathieu and coworkers identified that two nuclear proteins of 37-KDa and 46-KDa probably involved in the regulation of the CYP1A2 gene expression [49]. In addition, there exists an ARE (antioxidant responsive element) at −1555 bp to −1545 bp of CYP1A2 gene, but this region of DNA is not involved in 3-MC initiated induction of CYP1A2 gene [44]. Nevertheless, whether the regulatory sequences and factors are polymorphic, and whether their polymorphism constitutes the molecular basis for phenotypic polymorphism of CYP1A2, remains to be explored.

The frequency of PMs in Chinese was 5.24% (95% CI: 3.35%~8.13%) in this study. Ignoring the differences in CYP1A2 activity markers used, the incidence of PMs is not significantly different (χ²=0.40, df = 1, P > 0.5) when compared with the percentage of PMs (10%) from 30 nonsmokers in an overseas Chinese population [21]. It also approximates to the percentage (5%) of PMs in Australians [23] but this value is much lower than that observed (14.1%, n = 205) in Japanese (χ²=10.03, df = 1, P < 0.01) [24]. These findings suggest the possibility of interethnic differences in the frequency of PMs [25].

The decreased CYP1A2 activity in women [15, 16], reported previously, was also observed in the present study. The differences in CYP1A2 activity between men and women who were neither pregnant nor did taking oral contraceptives, did not reach statistical significance in previous reports [12, 18], as was the case in PMs in this study. This may have been due to the small population or due to more marked interindividual variation than the gender-related difference. Another explanation is that the plasma ratio used in this study may more specifically reflect CYP1A2 activity than the ratios of urinary caffeine metabo1ites [19]. The CYP1A2 gene is located in chromosome 15. The frequency of PMs is approximately equal between genders in this study, and the PM subjects inherit this characteristic as Mendelian autosomal recessive trait [24]. These findings indicate that the sex-linked differences in CYP1A2 activity might not result from differences in genotypes. For such a difference, the possibility that sex-related hormonal status modifies gene expression can not be excluded. This is due mainly to findings that women, who had higher levels of progesterone and oestrogen from taking oral contraceptives [28] in the luteal phase of menstrual cycle, or in pregnant status, exhibited decreased CYP1A2 activity. Higher concentrations of oestrogen were consistently associated with relatively lower rates of caffeine elimination [28, 32], which depends mainly on CYP1A2 activity. Additionally, an in vitro study demonstrated that oestrogen (oestradiol) could inhibit the CYP1A2-catalysed O-demethylation of 7-ethoxyresorufin (EROD) [50]. This suggested that oestrogen may also affect CYP1A2 activity.

Since CYP1A2 is responsible for the metabolic activation of a number of promutagens and procarcinogens, and for the biotransformation of many of therapeutic medications, the polymorphism in CYP1A2 and gender differences may influence individual susceptibility to chemical-induced cancers and to adverse effects of particular drugs. Further investigations are required to clarify the mechanisms underlying the genetic polymorphism and the gender differences in human CYP1A2.