CYP2A6 Substrate Specificity

A very specific reaction catalyzed by CYP2A6 is 7-hydroxylation of the naturally occurring plant compound coumarin, and a coumarin phenotyping test has therefore been developed to estimate an individual's CYP2A6 in vivo activity (Rautio et al., 1992). CYP2A6 is also one of the major enzymes involved in human nicotine metabolism (see below) and the major enzyme responsible for the metabolism of the platelet-activating factor receptor antagonist SM-12502 [(+)-cis-3,5-dimethyl-2-(3-pyridyl)thiazolidin-4-one hydrochloride] and the neuroprotective drug chlormethiazole. It also contributes to the metabolism of some other pharmaceuticals including methoxyflurane, halothane, valproic acid, and disulfiram, and can activate a number of precarcinogens e.g., aflatoxin B1, 1,3-butadiene, 2,6-dichlorobenzonitrile and the nitrosoamines NNK [4-(methylnitrosoamino)-1-(3-pyridyl)-1-butanone], NNAL [4-(methylnitrosoamino)-1-(3-pyridyl)-1-butanol], NDEA [N-nitrosodiethylamine], and NNN [N′-nitrosonornicotine] (see Oscarson et al., 1998;Pelkonen et al., 2000 and references therein).

Interindividual Variability in CYP2A6 Activity

Early studies using human liver microsomes revealed a pronounced interindividual variability in CYP2A6 levels and activity, with some livers completely lacking the enzyme (Miles et al., 1990; Yun et al., 1991; Camus et al., 1993; Shimada et al., 1996). Coumarin phenotyping studies carried out in European and Middle East populations (Finnish, British, Turkish, and Jordanian) have also shown important interindividual variability in CYP2A6 activity (Cholerton et al., 1992;Rautio et al., 1992; Iscan et al., 1994; Hadidi et al., 1998). Very few poor metabolizers (PMs) have, however, been described in these populations (Hadidi et al., 1997; Oscarson et al., 1998). In contrast, the CYP2A6 PM phenotype is much more prevalent in Asian populations where a small percentage of the population lack active enzyme (see below). A generally decreased nicotine clearance has also been reported in African-Americans (Benowitz et al., 1999), and it was suggested that decreased CYP2A6 activity in combination with decreasedN-glucuronidation may contribute to this. As discussed below, a lot of the CYP2A6 variability can be attributed to polymorphisms in the CYP2A6 gene, but CYP2A6 activity is also modified by certain drugs and environmental factors. Thus, CYP2A6-dependent coumarin 7-hydroxylation is induced in vivo by several antiepileptic agents (Sotaniemi et al., 1995) and decreased CYP2A6 activity has been demonstrated in subjects with severe alcohol-induced liver cirrhosis (Sotaniemi et al., 1995) and viral hepatitis A (Pasanen et al., 1997).

Genetic Polymorphism of the CYP2A6 Gene

The CYP2A6 gene spans a region of approximately 6 kilobase pairs, contains 9 exons and has been mapped to the long arm of chromosome 19 (between 19q12 and 19q13.2) (Miles et al., 1989). It is located within a 350-kilobase pair gene cluster together with the CYP2A7 and CYP2A13 genes, twoCYP2A7 pseudogenes, as well as genes in the CYP2Band CYP2F subfamilies (Hoffman et al., 1995). The CYP2A7 enzyme has been shown to not incorporate heme and is therefore inactive (Yamano et al., 1990; Ding et al., 1995). A full-length CYP2A13 cDNA was only recently cloned and expressed (Su et al., 2000),2 which revealed an enzyme active in NNK, coumarin, nicotine, and cotinine metabolism. CYP2A13 seems, however, to be expressed mainly in nasal mucosa, which limits its importance for total body clearance of CYP2A substrates.

Some of the interindividual variability in expression can be attributed to genetic polymorphisms in the CYP2A6 gene, and two variant alleles (CYP2A6*23and CYP2A6*3) were described several years ago (Yamano et al., 1990; Fernandez-Salguero et al., 1995). The CYP2A6*2allele encodes an enzyme with a L160H substitution, which is inactive because it does not incorporate heme (Yamano et al., 1990). It has also been suggested that the L160H substitution results in an enzyme with altered regiospecificity that catalyzes coumarin 3-hydroxylation instead of the usual 7-hydroxylation (Hadidi et al., 1997). A more likely explanation for this in vivo finding could, however, be a shunting of coumarin to the 3-hydroxylation pathway catalyzed by CYP1A1, CYP1A2, CYP2E1, and CYP3A4 (Zhuo et al., 1999) in the absence of CYP2A6-dependent 7-hydroxylation. The CYP2A6*3 allele was initially reported as a hybrid allele created through multiple gene conversions with the inactive CYP2A7 gene (Fernandez-Salguero et al., 1995) resulting in an allele where exons 3, 6, and 8 are of CYP2A7 origin. This allele is also considered to be inactive, although this has never been demonstrated. In a PM for coumarin 7-hydroxylation we detected an additional inactive allele, CYP2A6*5, where a highly conserved Gly-479 has been replaced with valine (Oscarson et al., 1999a). When expressed in yeast, this substitution resulted in a very unstable enzyme devoid of catalytic activity, which is consistent with the PM phenotype.

Using the original genotyping method described to detect theCYP2A6*2 and CYP2A6*3 alleles, we noticed a poor correlation between the CYP2A6 genotype and phenotype determined in vivo using the probe drug coumarin. In particular, we found an individual who displayed coumarin 7-hydroxylation activity in vivo, although the subject was initially determined to be homozygous for the inactive CYP2A6*2 allele. We therefore developed a more specific method for the detection of these alleles, which gave a much better correlation with the phenotype (Oscarson et al., 1998). Using this method, we detected a CYP2A6*2 allele frequency of 1 to 3% in European populations, in contrast to the 15 to 17% previously reported (Fernandez-Salguero et al., 1995). Furthermore, we failed to detect any true CYP2A6*3 alleles (Oscarson et al., 1999a); a finding that has been supported by several other groups (Chen et al., 1999; Kitagawa et al., 1999; Sabol and Hamer, 1999). ACYP2A7 gene conversion in the 3′-flanking region of theCYP2A6 gene (CYP2A6*1B) that abolishes the binding site for one of the primers used in the earlier method is the likely explanation for most of these misclassifications (Oscarson et al., 1999a).

In Asian populations, earlier reports suggested that the frequency of the CYP2A6 PM phenotype was much higher than in Caucasians and an in vitro study using microsomes from Japanese subjects demonstrated that eight of 30 livers had very low or no CYP2A6 immunoreactivity and activity (Shimada et al., 1996). Studies by Nunoya and coworkers (1998)suggested that a partial or whole deletion of the CYP2A6gene could contribute to the PM phenotype in Japanese individuals. We studied this phenomenon using Southern blotting, PCR, and DNA sequencing, and described the structure of a CYP2A locus where the whole CYP2A6 gene has been deleted, resulting in abolished CYP2A6-dependent metabolism (Oscarson et al., 1999b). The origin of this locus appears to be an unequal crossover event between the 3′-flanking regions of the highly similar CYP2A6 andCYP2A7 genes (Nunoya et al., 1999; Oscarson et al., 1999b), resulting in a deletion allele (CYP2A6*4A) and an allele with two CYP2A6 genes in tandem (Rao et al., 2000) as the two reciprocal outcomes. In a Spanish individual, we also identified a slightly different type of CYP2A6 deletion allele (CYP2A6*4D), where the crossover region is instead located in intron 8 or exon 9 (Oscarson et al., 1999a). A PCR-based method for the detection of the CYP2A6*4 alleles was developed, and the allele frequency was found to be 15% in Chinese subjects, but only 1.0% in Finns and 0.5% in Spaniards (Oscarson et al., 1999b). It can be concluded that the CYP2A6*4 allele is one of the major defective alleles contributing to the PM phenotype in Asian populations.

A schematic overview of the different CYP2A6 alleles currently described is shown in Fig. 1, and Table 1 summarizes the allele frequencies in European and Asian populations. Taken together, it can be observed that there are multiple alleles present in the population, which are the result of unequal crossover events and/or gene conversions between the CYP2A6 and CYP2A7 genes, and care must therefore be taken when designing genotyping methods to avoid similar genotype misinterpretations such as those that occurred for the CYP2A6*3 allele. The high-throughput methods that are now becoming available for rapid screening of novel polymorphisms will likely reveal a number of additional polymorphisms in theCYP2A6 gene that may modulate CYP2A6 levels and activity.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Schematic representation of theCYP2A6 alleles described thus far.

Shaded and unshaded regions represent CYP2A7 andCYP2A6 sequences, respectively. Seehttp://www.imm.ki.se/CYPalleles for details and for an updated list ofCYP2A6 alleles.

Implications of the CYP2A6 Polymorphism for Nicotine Metabolism and Smoking Behavior

In humans, 70 to 80% of the nicotine dose isC-oxidized by cytochrome P450 to a nicotine Δ^1′(5′)-iminium ion, which is further metabolized to cotinine by a cytosolic aldehyde oxidase. Nicotine is, however, also N-oxidized by flavine monooxygenases to nicotine N′-oxide. Cotinine is subsequently converted to a number of hydroxylated products includingtrans-3′-hydroxycotinine, 5′-hydroxycotinine, and norcotinine (Fig. 2). In addition, nicotine, cotinine, and trans-3′-hydroxycotinine are glucuronidated (see Benowitz and Jacob, 1997 for a review on nicotine metabolism). In vitro studies using human liver microsomes and recombinant P450s have shown that CYP2A6 is the most important P450 responsible for the C-oxidation of nicotine (Cashman et al., 1992; Nakajima et al., 1996b; Messina et al., 1997), and the nicotine clearance was decreased in liver microsomes from individuals heterozygous for the CYP2A6*2 allele (Inoue et al., 2000). Other P450s such as CYP2B6 can, however, also catalyze theC-oxidation of nicotine but with a lower affinity (Yamazaki et al., 1999). This is relevant especially in individuals who lack active CYP2A6 enzyme. CYP2A6 is also involved in the subsequent oxidation of cotinine to several different products (Nakajima et al., 1996a; Murphy et al., 1999), although the relative importance of different enzymes in catalyzing these reactions has not yet been fully elucidated. An in vivo study further stressed the importance of CYP2A6 in nicotine metabolism, as individuals homozygous for aCYP2A6 gene deletion displayed only 15% of urinary cotinine levels compared with individuals carrying at least one activeCYP2A6 gene after smoking the same number of cigarettes (Kitagawa et al., 1999). Earlier reports suggesting a major contribution of CYP2D6 (debrisoquine hydroxylase) to the in vivo metabolism of nicotine have, however, not been reproduced (Benowitz et al., 1996), thereby questioning the role of this enzyme in nicotine metabolism.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Phase I metabolism of nicotine in humans.

Compiled from Benowitz and Jacob, 1997; Murphy et al., 1999 and references therein. FMO, flavine monooxygenase.

Because of the substantial involvement of CYP2A6 in nicotine elimination, it has been proposed that the CYP2A6 polymorphism is a major determinant of an individual's smoking behavior. It has also been suggested that CYP2A6 inhibitors can be used as a new approach to treat tobacco dependence (Sellers et al., 2000). Pianezza and coworkers (1998) showed that a lower number of individuals carrying theCYP2A6*2 and CYP2A6*3 alleles were found in a tobacco-dependent group as compared to a never tobacco-dependent group, and that smokers carrying an inactive CYP2A6 allele smoked fewer cigarettes. These findings have, however, been questioned. First, the original nonspecific genotyping method (Fernandez-Salguero et al., 1995) was used, and a total allele frequency of 6 to 10% was found for the CYP2A6*2 and CYP2A6*3 alleles. As discussed above, this genotyping method leads to a significant overestimation of the CYP2A6*2 allele frequency, and it also scores many individuals as positive for the CYP2A6*3 allele, an allele which does not seem to exist. Second, several groups have failed to reproduce the correlation between CYP2A6 genotype and smoking behavior using both the original as well as more specific genotyping methods (London et al., 1999; Sabol and Hamer, 1999;Loriot et al., 2001; Tiihonen et al., 2000). It would be interesting to conduct this kind of study in Asian populations where the frequency of inactive alleles is considerably higher and the impact on interindividual variability in nicotine metabolism should be much greater. Furthermore, it would be beneficial to phenotype the individuals with coumarin as an indicator of in vivo CYP2A6 activity, since there are probably still a number of CYP2A6 alleles that have not yet been identified. This would clarify the predictive value of CYP2A6 genotyping for the alleles known presently. It should also be remembered that nicotine addiction and smoking behavior are complex multifactorial processes that involve many different molecular targets. It is therefore not surprising that a number of other candidate genes have been proposed as modulators of smoking behavior, including genes involved in dopamine biosynthesis and metabolism, as well as dopamine receptors, nicotinic acetylcholinergic receptors, and other enzymes involved in nicotine metabolism (Rossing, 1998).

Implications of the CYP2A6 Polymorphism for Lung Cancer Susceptibility

Most human cancers are the result of an interaction between the environment and the individual's genetic predisposition (Perera, 1997). In the case of lung cancer, the individual's pattern of enzymes, which activate and inactivate the precarcinogens inhaled in cigarette smoke, could modulate the risk of cancer development. TheCYP2A6 genotype is one of the polymorphisms that has been suggested to modulate this risk, but as the progression of a normal cell into a tumor is a complex multistep process this is probably modulated by polymorphisms in a number of different genes. Theoretically, the absence of CYP2A6 enzyme could reduce the risk of lung cancer in two ways. First, if the hypothesis raised by Pianezza and coworkers (1998) is correct, individuals lacking CYP2A6 activity would not become smokers or at least smoke fewer cigarettes. Second, as a number of precarcinogens found in tobacco smoke such as NNK, NDEA, and NNN are metabolically activated by CYP2A6, a lack of this enzyme would be protective also in this respect. In support of this hypothesis, a Japanese case-control study involving 492 lung cancer patients and 402 healthy controls showed that subjects with aCYP2A6 gene deletion (CYP2A6*4) were at reduced risk for lung cancer (odds ratio = 0.25) (Miyamoto et al., 1999). In a French case-control study involving 301 male lung cancer patients and 310 matched healthy hospital controls, there was, however, no significant differences in the distribution of either theCYP2A6*2 or CYP2A6*4 alleles (Loriot et al., 2001). Although studies in European populations are limited by the low allele frequencies of inactive alleles, these findings indicate that the relationship between the CYP2A6 polymorphism and lung cancer is more complex and that many additional studies are needed to elucidate if there is any involvement of CYP2A6 in determining an individual's susceptibility to lung cancer.

Conclusions and Future Directions

Recent studies of the CYP2A6 polymorphism has revealed several novel alleles, and reliable genotyping methods are now available for the prediction of CYP2A6 PMs. Although several newCYP2A6 polymorphisms might be described in the future, CYP2A6 phenotyping studies using coumarin suggest that the most common inactive alleles, at least in Caucasian populations, have been found. It is, however, likely that there are additional alleles that modulate CYP2A6 activity and thereby nicotine metabolism, and additional studies are needed to understand the mechanisms behind the tremendous variability within the extensive metabolizer group. The involvement of the CYP2A6 polymorphism concerning interindividual differences in smoking behavior and risk of lung cancer is still very unclear. Several well designed studies are needed to fully clarify this important issue. Furthermore, it remains to be determined how critical the CYP2A6 polymorphism is in determining an individual's disposition of drugs that are primarily metabolized by this enzyme.

• Download figure
• Open in new tab
• Download powerpoint

Mikael Oscarson received the Ph.D. degree in molecular genetics from Karolinska Institutet, Stockholm, in 1999. His thesis was “Genetic polymorphism of human drug metabolising enzymes. Structural and Functional studies”. He is currently completing the M.D. degree at Karolinska Institutet.  He is with the Division of Molecular Toxicology at the Institute of Environmental Medicine of Karolinska Institutet. Dr. Oscarson serves as the webmaster for the Human Cytochrome P450 (CYP) Allele home page and has served on the CYP Allele Nomenclature Committee since 1990. He has coauthored 12 research papers and 6 reviews or book chapters.