Aflatoxin B1 (AFB1), a mycotoxin produced mainly by Aspergillus flavus and Aspergillus parasitcus, is a potent carcinogen in humans and animals.1, 2 Mutation-induced inactivation of p53 tumor suppressor gene and/or activation of K-ras oncogene has been proposed as the major molecular mechanism in AFB1-induced cancers.3, 4, 5, 6 The worldwide human exposure to AFB1, particularly in developing countries, remains to be a serious public health problem. Since AFB1 is a major etiological agent of human liver cancer, extensive studies have been focused on dietary exposure to AFB1 and AFB1-induced liver cancer.7, 8 However, there are epidemiological evidences to suggest that human respiratory tract is also a target for AFB1 carcinogenicity. Occupational exposure to inhaled AFB1, particularly in the form of contaminated grain dusts, is associated with increased respiratory cancers.2, 8, 9, 10, 11, 12 The reported highest amount of AFB1 in respirable grain dusts was 52 ppm.13 Human lung is also at risk of cancer for dietary AFB1 exposure.14 AFB1 also induces lung tumors in laboratory animals.15, 16, 17
Metabolic activation is required for AFB1 to exert its carcinogenicity and toxicity (Fig. 1). The major carcinogenic and mutagenic metabolites of AFB1 are AFB1-8,9-epoxide and AFM1-8,9-epoxide, although the latter is relatively less active in the Ames mutagenesis test.18 In contrast, the formation of AFP1 and AFQ1 metabolites is considered as pathways of detoxicification. Cytochrome P450 (CYP) enzymes are critical for the metabolic activation of AFB1, particularly at low substrate concentrations that are relevant to human environmental exposure.19, 20, 21 Although some non-CYP enzymes such as lipoxygenases and prostaglandin H synthase were reported to be able to metabolize AFB1, their activities are much lower than those of the AFB1-activating CYP enzymes and are usually only active at higher AFB1 concentrations.22, 23, 24, 25, 26 Among human CYPs, CYP1A2 and CYP3A4 are the major AFB1-metabolizing enzymes in human liver. In particular, CYP1A2 is believed to play a predominant role in the metabolic activation of AFB1in vivo.19, 20, 21 Human lung is also able to activate AFB1, and the cultured human lung bronchus and tracheobronchial explants can activate AFB1 to form AFB1-N7-guanine adducts.12, 27, 28, 29 However, the key CYP enzymes in human lung for efficient activation of AFB1 have not been identified, as CYP1A2 is almost exclusively expressed in human liver.30
CYP2A13 is one of the three members in human CYP2A gene family. The other two members are CYP2A6 and CYP2A7. Heterologously expressed CYP2A7 protein showed no catalytic activity.31, 32 CYP2A6 is the major coumarin 7-hydroxylase and nicotine C-oxidase in man,31, 33 but is also able to metabolically activate AFB1.34 CYP2A6 is mainly expressed in human liver, while CYP2A13 is predominantly expressed in human respiratory tract, with the highest level in the nasal mucosa, followed by the lung and trachea.35 Previously, we demonstrated that CYP2A13 is the most efficient human CYP enzyme in the metabolic activation of a major tobacco-specific carcinogen, 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone (NNK).35 We have recently shown that CYP2A13 is also active in the metabolism of nicotine and cotinine, with a catalytic efficiency higher than that of CYP2A6.36 The present study identified CYP2A13 as an efficient catalyst for the metabolic activation of AFB1, a human carcinogen with lung as a suspected target.
Material and methods
AFB1, AFM1, AFP1, AFQ1, NADP+, glucose 6-phosphate, and glucose-6-phosphate dehydrogenase were purchased from Sigma–Aldrich (St. Louis, MO). AFB1 (98% purity) was further purified through HPLC with a reverse-phase Prep-C18 Scalar column (Agilent, Foster City, CA). [3H]-AFB1 (99% purity) was purchased from Moravek Biochemicals (Bres, CA). [3H]-AFB1 was passed through a Sep-Pak C18 cartridge (Waters, Milford, MA) to remove unbound free tritium. Authentic standards of AFB1 metabolites, AFB1-diol and AFM1-diol, were made according to procedures previously described.37, 38 The Cell Titer 96 Aqueous nonradioactive cell proliferation assay kit was purchased from Promega (Madison, WI). BaculoDirect baculovirus expression system, Flp-In CHO (Chinese hamster ovary) cells, pcDNA5/FRT, pOG44, Lipofectamine™ 2000, Ham's F-12 nutrient mixture, fetal bovine serum and hygromycin B were purchased from Invitrogen (Carlsbad, CA). Baculovirus-expressed human CYP1A2 protein and a monoclonal anti-human CYP2A6 antibody that cross-reacts with human CYP2A13 were obtained from BD Genetest (Woburn, MA). Triethylammonium formate (HPLC grade) was purchased from Regis Technologies (Morton Grove, IL).
Expression of human CYP2A13 and CYP2A6 proteins by baculovirus/insect cell system
All the approaches and the methods used to produce CYP2A6 and CYP2A13 proteins, including the CYP2A13 Val117 and Arg372 mutants, have been described in detail previously.39
P450 content determination and immunoblot analysis
P450 content in the membrane preparation (microsomes) of the infected SF9 cells was determined as described.40 Microsomal proteins were diluted to 1 mg protein/ml with 0.1 M phosphate buffer and aliquoted into two 1 cm optical path cuvettes. After recording the base-line, the sample cuvette was bubbled with CO for 60 sec. Using a UV/visible spectrophotometer (Shimazu UV 160u, Japan), the difference in absorption spectrum was measured by scanning from 500 to 400 nm, immediately after adding dithionite to the sample cuvette, with the other cuvette as a reference.
Immunoblotting to determine the protein level of CYP2A13 or CYP2A6 in the microsomes was carried out as described.39 Microsomal proteins were separated by SDS-polyacrylamide gel electrophoresis and transferred to a nitrocellulose sheet. The nitrocellulose sheet was then incubated with the anti-CYP2A6 antibody (BD Gentest), which cross-reacts with CYP2A13 with the same binding efficiency, and followed by the binding of a secondary antibody (goat anti-mouse IgG) conjugated with horseradish peroxidase. The immunoblot was visualized by ECL detection according to the manufacturer's protocol (Amersham Bioscience, Piscataway, NJ).
AFB1 metabolism and HPLC analysis
The incubation protocol used for AFB1 metabolism was modified from a reported method.19 The incubation mixture consisted of CYP enzymes (50 pmol), NADPH-CYP oxidoreductase (P450/reductase = 10 pmol:30 units), 15 or 150 μM AFB1 (each spiked with 0.5% of [3H]-AFB1), an NADPH-generating system (5 mM glucose 6-phosphate, 0.5 unit/ml glucose-6-phosphate dehydrogenase, 1 mM NADP+), and a buffer system containing 190 mM sucrose, 60 mM potassium phosphate, 80 mM Tris, 15 mM NaCl, 5 mM KCl and 4 mM MgCl2 (pH 7.6). Total volume of incubation solution was 240 μl. The reaction was carried out at 37°C for 10–30 min and was terminated by addition of ice-cold methanol for precipitation of proteins and extraction of AFB1 metabolites. Human CYP1A2 protein obtained from the same baculovirus expression system was used as a positive control. For the negative control, methanol was added before initiation of the reaction. The reaction mixture was stored at −20°C overnight, and then centrifuged at 10,000g for 5 min prior to HPLC analysis.
HPLC analysis was conducted on an Agilent Model 1100 system (Agilent Technologies), which was equipped with a diode-array detector connected in series with a fluorescence detector (excitation, 366 nm and emission, 436 nm) and a C18 5-μm (4 × 150 cm2) Zarbox analytical column. Chromatographic separation of AFB1 metabolites was obtained by a 5–25% ethanol linear gradient in water generated over a 25 min period, followed by isocratic elution with 25% ethanol in water, at a flow rate of 1 ml/min. The mobile phase was buffered with 5 mM triethylammonium formate (pH 3.0), and the column temperature was maintained at 45°C. Authentic standards of AFB1 metabolites (M1, P1, Q1, AFB1-diol and AFM1-diol) were used to establish the chromatographic retention times. The correspondent integrated peaks were quantitated with the regression formulae obtained from standard curves for each metabolite. In addition, HPLC fractions were collected, and the radioactivity associated with labeled AFB1 was counted in a Beckman LS6500 liquid scintillation counter for verification of the known AFB1 metabolites and identification of unknown AFB1 metabolites.
Construction of expression vector and transfection of CHO cells
The wide-type CYP2A13 and CYP2A6 cDNAs as well as the mutant CYP2A13 cDNAs were cloned into a pcDNA5/FRT vector. They were used to cotransfect Flp-In CHO cells with pOG44 by Lipofectamine 2000.41 After re-seeding and cultivating for 24 hr, the cells were split and grown for 2–3 weeks in the presence of hygromycin B (800 μg/ml) for selection. Individual hygromycin B-resistant stable transfectants were isolated and subcultured in the presence of hygromycin B (500 μg/ml). Control cells were transfected only with pcDNA5 vector. Expression of CYP2A13 and CYP2A6 proteins in the stable transfectant cells was confirmed by immunoblot analysis as previously described.
AFB1-induced cytotoxicity was determined by a modified MTS assay as described previously.41 CHO cells stably expressing a particular CYP enzyme (CHO-2A6, CHO-2A13, CHO-2A13Val117 and CHO-2A13Arg372) were plated into a 24-well plate (1×105 cells per well) with F12 complete medium containing 200 μg/ml hygromycin B. They were grown under 95% humidity and 5% CO2 at 37°C overnight to allow for attachment. The cells were then treated with AFB1 (dissolved in DMSO) at different concentrations (ranging from 1 nM to 10 μM) for 24 or 48 hr. Flp-In CHO cells transfected with pcDNA5 vector alone (containing no CYP cDNA insert) were used as a negative control. After AFB1 treatment, the medium in each well was removed and 0.5 ml of MTS mixture was added. The cells were incubated with MTS mixture at 37°C in dark for 30 min. The plate was then read at 490 nm wavelength by a μQuant plate reader (Bio-Tek Instrument, Winooski, VT). In each group, the viability of cells incubated with DMSO only was set at 100%. The LC50 was calculated to compare AFB1-induced cytotoxicity among CHO cells expressing different CYP enzymes.
Catalytic activity of CYP2A13 in metabolizing AFB1
Our study clearly showed that CYP2A13 is an efficient enzyme in metabolizing AFB1 (Fig. 2 and Table I). There was no detectable formation of AFB1 metabolites in the negative controls (methanol-preinactivated CYP2A13 or 1A2), while incubation of AFB1 with CYP2A13 protein resulted in the generation of multiple AFB1 metabolites, including the carcinogenic/toxic epoxides (detected as diols). At both 15 and 150 μM AFB1 concentrations, the activity of CYP2A13 was as high as that of CYP1A2 in the formation of AFM1-8,9-epoxide and was approximately one-third of CYP1A2 in the formation of AFB1-8,9-epoxide (Table I). CYP2A13 also catalyzed the formation of other known and unknown AFB1 metabolites, including AFM1, AFQ1 and AFP1, with the activities comparable to CYP1A2 (Fig. 2 and Table I). It is of interest to note that at 150 μM AFB1 the formation of AFB1-8,9-epoxide by CYP2A13 and CYP1A2 was significantly lower than at 15 μM AFB1, while the formation of AFM1-8,9-epoxide and most of the other nonepoxide metabolites was much higher. A possible explanation is that high AFB1 concentration or the production of some metabolites may selectively inhibit the metabolic pathway for AFB1-epoxide formation. Although CYP2A6 was previously reported to be able to activate AFB1,34 there was no detectable AFB1-metabolizing activity of CYP2A6 at 15 μM substrate concentration. At 150 μM AFB1, CYP2A6 catalyzed the formation of AFM1, AFQ1 and AFP1 but not the epoxide formation (Table I).
Table I. Comparative Activity of Human CYP1A2, 2A6, 2A13 and 2A13 Mutants in AFB1 Metabolism
AFB1 (15 μM)
AFB1 (150 μM)
The activity is expressed as pmol/min/pmol P450. The results are the average of duplicated determinations with <10% of assay variations. ND, nondetectable.
AFB1-induced toxicity in CHO cells expressing CYP2A13 and CYP2A6
To further demonstrate the role of human CYP2A13 in AFB1 metabolic activation, we established the Flp-In CHO cell lines with stable expression of human CYP2A13 or CYP2A6 and used them to determine the metabolism-mediated cytotoxicity of AFB1. Immunoblotting demonstrated that the protein expression levels of CYP2A13 (including its mutants) and CYP2A6 were the same (Fig. 3). This is consistent with the working principle of the Flp-In cell system, which ensures the same level expression of the transgenes.42 Up to 10 μM AFB1, there was no significant cell death after 24 or 48 hr in the CHO cells transfected with vector alone (without CYP cDNA). However, there was a striking difference (∼800-fold) in cell viability between the CHO cells expressing CYP2A13 and CYP2A6 (Figs. 4 and 5). The LC50 values of AFB1 in CHO-2A6 cells were >100 μM and 39 μM for 24 hr and 48 hr treatment, respectively; but were only 190 nM for 24 hr and 50 nM for 48 hr in CHO-2A13 cells (Fig. 5). This result is consistent with the difference in AFB1-metabolizing activities between CYP2A13 and CYP2A6 (Table I).
Alteration of AFB1-metabolizing activities in CYP2A13 mutants
We have previously generated a series of CYP2A13 mutants in which the amino acid residues of CYP2A13 were replaced with the residues of CYP2A6 at the corresponding positions.39 In the present study, 2 of these mutants, Ala117Val and His372Arg, were assayed for their AFB1-metabolizing activities. Both mutants showed a significant alternation in the formation of the 8,9-epoxides of AFB1 and AFM1. The Ala117Val mutant is more active than the wild-type CYP2A13 in producing AFM1-8,9-epoxide and AFB1-8,9-epoxide, while the His372Arg mutant has completely lost the activity in AFB1 epoxidation (Table I). Consistent with these metabolic activity changes, CHO cells expressing Ala117Val mutant showed a reduced viability after AFB1 treatment, while the CHO cells expressing His372Arg mutant were as resistant as the negative control cells to AFB1 toxicity, with an increased LC50 of >100 μM (Figs. 4 and 5).
Metabolic activation of carcinogens in their target tissues is critical for tissue-specific tumor production by the carcinogens. Although epidemiological studies have suggested a positive link between AFB1 inhalation exposure and human lung cancers, the molecular mechanisms and the key CYP enzymes in human lung that are responsible for the metabolic activation of AFB1in situ are unknown. Our results clearly demonstrate that CYP2A13, which is predominantly expressed in human respiratory tract, is efficient in AFB1 metabolic activation, particularly at low substrate concentrations. This finding provides a positive support to the etiological role of AFB1 in human lung carcinogenesis.
Human occupational exposure to inhaled aflatoxins, particularly in the form of airborne grain dusts, has been reported worldwide.8, 9, 10, 11, 12, 13 AFB1 in respirable grain dusts could reach as high as 52 ppm in feed factories.13 During unloading of ships, AFB1 was found in bilge at levels as high as 300 ng/m3.43 Autrup et al. found that 7 of 45 blood samples collected from workers in animal feed processing plant in Denmark were positive for AFB1, with an average daily intake through inhalation of 64 ng/kg of body weight.44 Ghosh et al. found that airborne aflatoxins in rice- and maize-processing plants in India were always higher in the respirable dust samples (<7 μm) than in total dust samples. Concentrations of total airborne aflatoxins in the respirable dusts were 26 pg/m3 in the rice mill, 800 pg/m3 in a maize-processing plant and 816 pg/m3 in an oil mill, respectively.45 A study in Thailand found that levels of aflatoxin airborne dust generated during handling of animal feed were 1.55–6.25 ng/m3 as compared to 0.99 ng/m3 in the control air samples. The exposed workers had altered lactate dehydrogenase activity and tumor necrosis factor-α level in plasma.46 The role of occupational exposure to inhaled AFB1 in human lung carcinogenesis needs to be further studied.
The participation of several human CYP enzymes, including CYP1A2, CYP3A4 and CYP2A6, in metabolizing AFB1 to its carcinogenic and toxic epoxides has been well documented by previous studies.19, 20, 21, 47 In particular, CYP1A2 has been demonstrated as the principle enzyme responsible for AFB1 activation in human liver in vivo at low substrate concentrations that are relevant to human environmental exposure.21 CYP1A2, CYP3A4 and CYP2A6 are all predominantly expressed in human liver. While CYP3A4 and CYP2A6 are also expressed in human lung,12, 48, 49, 50 most reported studies failed to detect constitutive expression of CYP1A2 in human lung.12, 48, 49, 51, 52 For AFB1-8,9-epoxide formation, CYP1A2 showed Michaelis–Menten kinetics with Km value of 41 μM, whereas CYP3A4 displayed a sigmoidal curve with Km value of 140–180 μM.21 The cells or the microsomes expressing CYP1A2 have a higher affinity toward low AFB1 concentrations than do those expressing CYP3A4.19, 20, 21 There was no reported enzyme kinetics of CYP2A6 for AFB1 metabolism, and the activation of AFB1 by CYP2A6 was mainly demonstrated by mutagenesis assay.34 Results from our metabolism and cytotoxicity studies clearly show that CYP2A13 is much more active than CYP2A6 in the metabolic activation of AFB1. Further studies to compare the enzyme kinetics of these human CYP enzymes in AFB1 metabolism are warranted.
CYP2A13 and CYP2A6 share a 93.5% identity in the amino acid sequence, and both proteins are composed of 494 amino acid residues. However, there is a large difference between CYP2A13 and CYP2A6 in their activities in metabolizing coumarin and NNK. CYP2A13 has a much higher activity in NNK metabolism, while its activity in metabolizing coumarin is only one-tenth of CYP2A6.35 Such a significant difference in catalytic activity was also observed in the present study for AFB1 epoxidation. Our previously results demonstrate that amino acid residues Ala117 and His372 are critical for the activity difference between CYP2A13 and CYP2A6 in coumarin 7-hydroxylation and NNK α-hydroxylation.39, 53 Substitution of these amino acid residues in CYP2A13 with the residues at corresponding position of CYP2A6 resulted in a drastic increase in coumarin 7-hydroxylation activity and a significant decrease in NNK α-hydroxylation activity. The present study demonstrates that the activity of CYP2A13 in the formation of the 8,9-epoxides of AFB1 and AFM1 was also significantly altered in the Ala117Val and His372Arg mutants. Interestingly, the substitutions resulted in a totally opposite effect on AFB1 epoxidation. The substitution of Ala117 caused an increase in AFB1 expoxidation, while the substitution of His372 abolished the epoxidation activity. Consistent with this metabolic activity change, the CHO cells expressing these 2 mutants showed the corresponding changes in their responses to AFB1 toxicity. A detailed molecular analysis of the interactions between these CYP2A13 amino acid residues and AFB1 would help us elucidate the structural basis of CYP2A13 for its high efficiency in AFB1 epoxidation. We have recently established a homology-based CYP2A13 protein model39, 53 and will dock AFB1 into the active site of CYP2A13 to explore the role of these amino acid substitutions in AFB1 epoxidation.
It is known that AFB1-8,9-epoxides can have 2 stereoisomeric forms. Human CYP1A2 catalyzes the conversion of AFB1 to both exo- and endo-AFB1-8,9-epoxides, while human CYP3A4 only produces the exo-epoxide.54 The production of these 2 stereoisomeric expoxides of AFB1 can be monitored indirectly by their conjugation with glutathione in the presence of glutathione-S-transferase, followed by the separation with chiral HPLC.55 AFM1-8,9-epoxides may also exist as the endo- and exo-forms but there were no reported studies on their existence and preferential production by different human CYPs. Since there is a significant difference between the exo- and endo-forms of AFB1-8,9-epoxide in DNA binding and mutagenicity,56, 57 it would be of interest to determine in future studies the identity of the AFB1-8,9-epoxides and AFM1-8,9-epoxides produced by CYP2A13.
Before 2000, CYP2A13 was predicted as a nonfunctional protein based on the observation that the features of its amino acid sequence resemble the nonfunctional CYP2A7 and CYP2A6*2 (or CYP2A6 v1).31, 32 As a result, there was little research interest on CYP2A13. However, our recent study demonstrated that CYP2A13 is the most efficient human CYP enzyme in the metabolic activation of a major tobacco-specific carcinogen NNK. Compared to CYP2A6, which was believed to be the major enzyme for NNK activation, the Km value of CYP2A13 in generating the carcinogenic metabolites (the keto aldehyde and keto alcohol) was ∼10 times lower (˜10 μM for CYP2A13 versus ˜100 μM for CYP2A6) and the catalytic efficiency (Vmax/Km) was ˜30 times higher.35 Since cigarette smoking is the single most important etiological factor of human lung cancer, and NNK, which also induces lung and nasal tumors in animals, is a highly suspected human lung carcinogen, our finding suggests a significant role of CYP2A13 in human lung carcinogenesis related to cigarette smoking. The present study further increases the importance of CYP2A13 in human lung carcinogenesis by identifying its high efficiency in the metabolic activation of AFB1.
Our study was supported by the NIH grants ES10048 to J-Y Hong, CA90997 to J-S Wang, and NIHS center grant ES05022. We thank Dr. X. Ding (Wadsworth Center, Albany, NY) for the gift of CYP2A13 cDNA, Dr. A.Y. Lu (Rutgers University, Piscataway, NJ) for helpful discussions and Dr. A.-N.T. Kong (Rutgers University, Piscataway, NJ) for permission to use his plate-reader.