Comparative Aflatoxin B₁ Activation and Cytotoxicity in Human Bronchial Cells Expressing Cytochromes P450 1A2 and 3A4¹

AFB₁,³ a mycotoxin produced by Aspergillus flavus and Aspergillus parasiticus, is carcinogenic in many animal models and is strongly suspected to be a human carcinogen (1). AFB₁ has been detected in respirable dusts in amounts as high as 52 ppm (2). There is some epidemiological evidence linking pulmonary exposure to AFB₁-laden grain dusts with an increase in lung tumor incidence in certain occupational settings (3). In addition, there is considerable evidence of pulmonary AFB₁ activation in animal models and human and animal tissues (4, 5, 6, 7, 8).

Although human CYPs activate AFB₁ to the electrophilic exo-AFBO, there is some ambiguity as to which isoform is most important in human liver (9, 10, 11, 12, 13, 14). CYPs 1A2 and 3A4 appear to be the most important isoforms for AFB₁ activation in human liver (15), although others (2A6 and 2B7) also activate AFB₁ in vitro (16). Both CYPs 1A2 and 3A4, in addition to their mRNA, have been detected in human lung tissues and human lung cells (17). Compared with CYP 3A4, human liver microsomal CYP 1A2 is reported to have a higher affinity toward lower AFB₁ concentrations more reflective of dietary exposures (18). However, CYP 3A4 produces exclusively the exo-AFBO, whereas 1A2 also produces equal amounts of the endo-AFBO (14). This is important because only exo-AFBO is involved in the alkylation of cellular nucleophyles such as DNA. Another complicating component of this issue is the induction of CYPs by exposure to many environmental chemicals (19).

A previous report from this laboratory has shown the potential involvement of CYP 3A4 in AFB₁ activation by human lung tissues at high substrate concentrations (8). Still other studies have shown the role of LOX in human lung AFB₁ activation as well, but very little activation has been seen at concentrations as low as 1.5 μm (7). The role of these various enzymes may be dependent on their magnitude of expression and on the AFB₁ concentration. Less is known about the enzymology of AFBO detoxification in people. In many animals, a key for the detoxification of AFBO is GST-mediated glutathione conjugation (20, 21). Although in humans the M1-1 isoform appears to be the most important isoform in vitro, M1-1, M3-3, P1-1, A1-1, and A2-2 GST isoforms can all conjugate the AFBO (20, 22).

The bronchiolar epithelium is the major site of tumor formation in the lung (23). Many cell types are recognized in the bronchiolar epithelium, but there is great variation between species in the cell types present and between the cell types of different species (24, 25, 26, 27). BEAS-2B cells are an SV40 immortalized cell line originating from normal human bronchial epithelial cells, which are progenitors of human lung cancer (28), and they have been used as a model for studying human lung cancer (23, 28). They are nontumorigenic and remain so up to very high numbers of passages (28, 29, 30). In addition, these cells have been successfully transfected with cDNA for CYPs 1A2 and 3A4 and express stable amounts of these CYPs (15, 23). The CYP-transfected BEAS-2B cell lines used in this study include B-CMV1A2 (transfected with CYP 1A2; Ref. 23) and B3A4 (transfected with CYP 3A4; Ref. 15).

Our laboratory has previously studied AFB₁ metabolism and adduct formation in airways and lung tissues from humans and animals (4, 5, 6, 8, 31, 32). The use of animal models is limited for studying the human condition because of variability between species and because human tissues are extremely variable in CYP expression and activities. Here we compare the activation and cytotoxicity of AFB₁ at low concentrations reflective of many human inhalation exposures in human lung cells that stably express CYP 1A2 and 3A4, the two principal isozymes in AFB₁ activation in human liver.

[³H]AFB₁ (30 Ci/mmol; >98% radiochemical purity) was purchased from Moravek Biochemicals, Inc. (Brea, CA) and diluted with unlabeled AFB₁ to achieve the desired concentrations. LHC-8, epinephrine, and retinoic acid were obtained from Biofluids (Rockville, MD) and used to make LHC-9. LHC Basal and BSA stock were also purchased from Biofluids. Fetal bovine serum was from Hyclone (Logan, UT). Bovine fibronectin, methoxyresorufin, nifedipine, 0.25% trypsin-EDTA, trypsin inhibitor, and Neutral Red solution were purchased from Sigma Chemical Co. (St. Louis, MO). Collagen was a product of Collagen Corp. (Fremont, CA). Resorufin was from Aldrich (Milwaukee, WI), and nitrendipine was from ICN (Irvine, CA). Quickprep Micro mRNA purification kit and First Strand cDNA Synthesis kit were purchased from Amersham (Piscataway, NJ). Riboprobe System T-7 RNA synthesis kit and Taq polymerase were from Promega (Madison, WI). DNA Now reagent was from Biogentex (Seabrook, TX). All custom oligonucleotide primers were purchased from Integrated DNA Technologies (Coralville, IA). 17α-Ethynylestradiol, 7,8-benzoflavone, and all other reagents were purchased from Aldrich.

BEAS-2B cells were cultured with LHC-9 (LHC-8 with 50 ml of 3.3 mm retinoic acid and 250 ml of 0.1% epinephrine) in Corning T₇₅ tissue culture flasks (Corning, Corning, NY) at 37°C and 5% CO₂. Cells were passed and harvested using 0.25% trypsin-EDTA and trypsin inhibitor solutions and HBSS (0.9% NaCl and 30 mm HEPES). Flasks were coated with a plate coat (added to flasks 15 min before seeding) consisting of 5 mg of bovine fibronectin, 5 ml of collagen, 50 ml of BSA stock, and 500 ml of LHC Basal.

CYP expression was quantified by competitive quantitative RT-PCR. Poly(A)⁺ RNA was isolated from B-CMV1A2, B3A4, and BEAS-2B cell types using the Quickprep Micro mRNA purification kit from Amersham. An initial range of mRNA concentrations was amplified by RT-PCR for each cell type to determine the linear range of amplification (0.5–2 ng, B-CMV1A2; 20–150 ng, B3A4; data not shown). An amount of total mRNA in this range was used for competitive quantitative RT-PCR. The primers used for RT-PCR have been described previously (33) and were specific for their target CYP isoform (Table 1). Primers were also designed to span at least one intron to exclude interference from genomic DNA contamination. A modification of the procedure of Vanden Heuvel et al. (34) was used to synthesize internal standard rcRNA for each CYP. A DNA template was constructed from the cDNA of each target CYP as filler sequence, which was approximately 60 bp shorter than the target cDNAs (using primers in Table 1). rcRNAs containing competitive primer sequences were synthesized using the Promega T-7 RNA synthesis kit. Next, a range of concentrations (0.1–0.01 ng/50 μl) of rcRNA were added to reaction mixes to compete with target CYP mRNA (within the total cellular mRNA in reaction mixes) to quantify the amount of target mRNA present. The reverse transcription step was performed with the Amersham First Strand cDNA Synthesis kit. First-strand cDNA synthesis mix was diluted to the manufacturer’s specifications, and Taq polymerase was added for amplification. The PCR was performed as described previously (35), using the following program: 95°C for 5 min; followed by 27 cycles at 95°C for 30 s, 49°C for 30 s, and 72°C for 1 min, with a 5-min extension at 72°C after the final cycle (Genemate; ISC Bioexpress, Kaysville, UT). Bands were separated electrophoretically on 3% agarose gels (7 × 7.5 × 1 cm) in TAE buffer [0.04 m Tris, 2 mm EDTA, and 0.1% acetic acid (pH = 8.5)] at 200 V for 35 min in a biological cold cabinet (4°C).

Activity of CYP 1A2 was quantitated by measuring the conversion of methoxyresorufin to resorufin in intact cells as described by Raffali et al. (36). T₇₅ flasks were seeded at 9.5 × 10⁵ cells/flask and cultured for 96 h to approximately 90% confluence. Growth media were removed and replaced with 5 ml of PBS to wash flasks. After removal of PBS, 2 μm methoxyresorufin containing TBS was added to flasks. Preliminary experiments using a range of concentrations (0–16 μm) showed that 2 μm methoxyresorufin was saturating and therefore not rate-limiting (data not shown). Incubations increasing by 3-min intervals from 0–15 min were determined to be in the linear range for methoxyresorufin conversion to resorufin at 2 μm (data not shown). The TBS-MROD solution was removed, and resorufin production was measured fluorometrically (excitation λ = 530 nm; emission λ = 584 nm; Gilford Flouro IV, San Diego, CA). Resorufin production during the 15-min interval was compared with a resorufin standard curve to quantify resorufin production. Cells were immediately harvested via trypsinization and resuspended in cold storage buffer [0.05 m Tris, 1 mm EDTA, 0.25 M sucrose, 200 μm PMSF, and 20% glycerol (pH 7.4)] before homogenization and storage at −80°C. The protein concentration of this homogenate was determined using the Bradford Protein Microassay (37) and used to calculate the total cellular protein/flask.

Cellular CYP 3A activity was quantitated by nifedipine oxidation, the conditions of which were identical to MROD except for the substrate and its concentration. T₇₅ flasks were seeded at 9.5 × 10⁵ cells/flask. After 96 h of culture, flasks were washed with 5 ml of PBS. PBS was replaced with TBS containing 45 μm nifedipine. This concentration of nifedipine was found in preliminary determinations to be within the range of saturation under these conditions (0–55 μm; data not shown). The TBS/nifedipine solution was recovered at 3-min intervals for 15 min, a time interval within the linear phase of product formation (data not shown). Oxidation of nifedipine was analyzed by reverse-phase high-performance liquid chromatography as described previously (Ref. 8; detection limit < 1 nmol). Cells were harvested immediately after TBS/nifedipine removal via trypsinization and suspended in 1 ml of cold storage buffer [0.05 m Tris, 1 mm EDTA, 0.25 m sucrose, 200 μm PMSF, and 20% glycerol (pH 7.4)] before homogenization and storage at −80°C. A 20-μl aliquot of the crude cell homogenate was used for determination of protein concentration by Bradford microassay (37). The amount of total cellular protein/flask was calculated from the results.

BEAS-2B, B3A4, and B-CMV1A2 were seeded at 9.5 × 10⁵ cells/T₇₅ flask and harvested at 90% confluence (5–6 days later) via trypsinization. The cells were then pelleted and suspended in cold homogenizing buffer [0.05 m Tris, 1 mm EDTA, 0.25 m sucrose, 0.15 m KCl, 20 μm butylated hydroxytoluene, and 200 μm PMSF (pH 7.4)], homogenized, and centrifuged at 600 × g. The supernatant was spun again at 10,000 × g for 10 min. The resulting supernatant was again centrifuged at 16,000 × g for 10 min, and the soluble fraction from this spin was centrifuged at 105,000 × g for 1 h. The 105,000 × g supernatant (cytosol) was stored at −80°C for later use.

GST activity was quantified in the cytosol from each cell type by measuring the rate of conjugation of GST-mediated glutathione to 1 mm CDNB spectrophotometrically (λ = 340 nm; Model DU 640; Beckman, Fullerton, CA) for 10 min at 25°C as described previously (38).

AFB₁ cytotoxicity toward BEAS-2B, B-CMV1A2, and B3A4 cells was determined by the Neutral Red uptake assay (39). For this assay, 96-well plates were seeded at a density of 2.5 × 10⁴ cells/well (in 0.18 ml medium/well). Each plate had a blank and control column. After 24 h of growth, cells were dosed with AFB₁, which was serially diluted from the high-dose concentration in subsequent microplate columns, for 24 h (AFB₁) or 48 h (7,8-benzoflavone and 17α-ethynylestradiol). After the incubation period, the wells were washed with PBS. Neutral Red media [LHC-9 with Neutral Red (5 mg/ml)] was added, and cells were then incubated for 4 h. Neutral red media were removed, and wells were again washed with PBS before 100 μl of elution buffer (50% ethanol and 1% acetic acid) were added to each well. Plates were shaken for 10 min and read at λ = 540 nm to measure dye uptake and at λ = 650 nm for background subtraction. The measured Neutral Red dye uptake was used to calculate the percentage of inhibition of dye uptake as described previously (40) where percentage of inhibition of dye uptake = [1 − (A₅₄₀test −A₆₅₀test/A₅₄₀control − A₆₅₀control)] × 100. A₅₄₀test is the spectral absorbance at 540 nm of a test (AFB₁) dose group, and A₆₅₀test is the spectral absorbance of the same test group at 650 nm. A₅₄₀control is the 540 nm absorbance of the control (no AFB₁) wells, and A₆₅₀test is the 650 nm absorbance of the control group.

AFB₁-DNA adduct formation was quantified as described previously (23). [³H]AFB₁ was diluted with cold AFB₁, and the specific activity of the stock solution used was 62 μCi/μg AFB₁, as determined by liquid scintillation and spectrophotometry (λ = 360 nm; ε = 21,800 A/mol AFB₁). Cells were seeded at a density of 9.5 × 10⁵ cells/T₇₅ flask and cultured 96 h before exposure to [³H]AFB₁. Time course studies were performed by exposing cells to 1.5 μm [³H]AFB₁ for 2, 4, 8, 16, or 24 h before harvest. Once the optimal incubation time for adduct formation was determined (8 h), subsequent experiments used a range (0.015–15 μm) of AFB₁ concentrations. Cultures were harvested via trypsinization, and DNA was isolated from cellular pellets using DNA Now reagent. From the isolated DNA samples 46 μg of genomic DNA were brought to a volume of 1 ml in H₂O and added to 5 ml of scintillation mixture in a 7-ml scintillation vial. The activity of each vial was measured by scintillation counting (Model LS3801; Beckman).

Cytotoxicity plots (percentage of inhibition versus μm AFB₁) were fit using an empirical three-parameter Hill equation model (41, 42, 43):

where R is the measured response (percentage of inhibition), K_0.5 is the AFB₁ concentration yielding half of the maximal response (R_max), and n is the Hill exponent, which is a measure of cooperativity. An n = 1 represents a linear response at low concentrations, an n > 1 represents a sublinear (sigmoidal) response indicating cooperativity, and an n < 1 represents a supralinear response. The parameters for this equation were calculated using Sigma Plot (Jandel Scientific, San Rafael, CA). Groups were compared for differences using one-way ANOVA (Sigma Chemical Co. Stat Software). A P < 0.05 was judged significant.

Using competitive quantitative RT-PCR, we were able to measure the amounts of mRNA specific to each CYP isoform expressed in these cell types. RT-PCR cDNA detection fragments for CYP 1A2 and 3A4 were generated from B-CMV1A2 and B3A4 cells, respectively, using forward and reverse 1A2 or 3A4 primers. DNA templates for recombinant RNA internal standards were constructed using 1A2 or 3A4 rcRNA forward and 1A2 or 3A4 rcRNA reverse primers, and part of each target gene was used as filler sequence. Finally, rcRNAs were synthesized from the DNA templates using the Riboprobe System T-7 RNA synthesis kit. Thus, rcRNAs were as similar as possible to target mRNA in sequence and differed in length by only approximately 60 bp so that PCR products would sufficiently separate by gel electrophoresis.

Fig. 1 clearly shows the predicted PCR products of 241 and 301 bp or 256 and 323 bp from CYP 1A2 and 3A4 rcRNA/mRNA, respectively, generated in this protocol. In each case, the band intensities of products from the rcRNA (241 and 256 bp) decreased as the concentration of spiked rcRNA decreased. Likewise, the band intensities representing PCR products from CYP 1A2 and 3A4 target mRNA (301 and 323 bp, respectively) increased as the PCR product from the rcRNA decreased; the equivalency point between these two products occurred at approximately 0.5 pg of spiked rcRNA competitor. This amount, therefore, represented the amount of target mRNA present in 2.5 and 75 ng of B-CMV1A2 and B3A4 total mRNA, respectively. Using this value, we calculated that B-CMV1A2 cells expressed approximately 5 times more of their respective CYP mRNA than the B3A4 cells (0.37 ± 0.1 amol CYP 1A2 mRNA/ng total mRNA versus 0.07 ± 0.02 amol CYP 3A4 mRNA/ng total mRNA, respectively; Fig. 1, A and B). No CYP 3A4 mRNA was detected in cells other than B3A4, and no CYP 1A2 mRNA was detected in cells other than B-CMV1A2 (data not shown). Furthermore, neither CYP 1A2 nor 3A4 mRNA was detected in BEAS-2B cells (data not shown).

Cellular CYP expression was also assessed by measuring catalytic activities of specific enzyme markers. As expected, activities of MROD and nifedipine oxidase, prototype activities for CYP 1A2 and 3A, respectively, were mainly found in cells that were transfected with cDNA coding for those isozymes. For example, only B-CMV1A2 cells expressed MROD activity, whereas nifedipine oxidase was most significant in B3A4 cells, although small amounts of activity of this enzyme were detected in all cell types (Table 2).

All cell types retained GST activity, as determined by conjugation to CDNB (Table 3). In addition, CDNB conjugation was not significantly different between the cell types.

Whereas there were consistent AFB₁ concentration-dependent decreases in cell survival, there were profound differences with regard to their sensitivity to AFB₁. B-CMV1A2 cells were by far the most susceptible cell type to AFB₁. B3A4 cells were approximately 70-fold more sensitive to AFB₁ than BEAS-2B cells, whereas B-CMV1A2 cells were roughly 45-fold more sensitive to AFB₁ than cells expressing 3A4 and approximately 3000-fold as sensitive as BEAS-2B cells (Table 4). In each case, regression lines from the percentage of inhibition of dye uptake versus AFB₁ concentration curves had good fit for B-CMV1A2 as judged by their r² values (Fig. 2, A−C). The amount of CYP mRNA expressed by each cell type, as determined in competitive RT-PCR, was then used to calculate a ratio of percentage of inhibition:CYP mRNA expressed as a representation of cytotoxicity generated per molecule of CYP mRNA expressed (Fig. 3; low range data are emphasized in the inset plot). Thus, on a per molecule mRNA basis, CYP 1A2 was approximately 6-fold more efficient than CYP 3A4 at producing cytotoxic metabolites (in their respective cells) at low (0.25 μm) concentrations (Table 4), and CYP 3A4 was approximately 5-fold more efficient in this regard at high (10 μm) AFB₁ concentrations (Table 4).

Formation of AFB₁-DNA adducts is the result of CYP-mediated activation of AFB₁ to the reactive AFBO that binds to nuclear DNA. In agreement with the cytotoxicity results, there was a significant difference in the ability of the three cell types to form AFB₁-DNA adducts. A preliminary experiment showed that for all cell types, an 8-h incubation time was optimal for AFB₁-DNA adduct formation (Fig. 4), a duration that was used in subsequent experiments examining adduct formation over a range of AFB₁ concentrations. AFB₁-DNA adduct formation was barely detectable in BEAS-2B cells incubated with 0–15 μm AFB₁ (Fig. 5). At all concentrations of AFB₁ examined, B-CMV1A2 cells bioactivated considerably more AFB₁ to DNA-binding species than did B3A4 cells, as evidenced by higher amounts of AFB₁-DNA adducts in B-CMV1A2 cells than in B3A4 cells (Fig. 5; data at low concentrations were emphasized in the inset plot). This discrepancy between the bioactivation of the two cell types was not as great when corrected for the amount of CYP mRNA expression (Fig. 6; data at low concentrations were emphasized in the inset plot). In fact, at higher concentrations of AFB₁ (i.e., 8–15 μm), B3A4 cells were more efficient than B-CMV1A2 cells in adduct formation when the amount of CYP mRNA expressed was taken into consideration.

Predictably, the CYP 1A2 inhibitor 7,8-benzoflavone inhibited the formation of AFB₁-DNA adducts in B-CMV1A2 cells in a concentration-dependent manner (Fig. 7,A), whereas adducts in B3A4 cells were inhibited by 17α-ethynylestradiol, a CYP 3A4 inhibitor (Fig. 7 B). The observed decreases in AFB₁ bioactivation caused by 7,8-benzoflavone and 17α-ethynylestradiol were not due to cytotoxicity of these agents at the nanomolar concentrations used in the inhibition assay at the same exposure duration (8 h). These inhibitors were cytotoxic, but only at concentrations several orders of magnitude higher than that found to inhibit AFB₁-DNA adduct formation, even when cultured for 48 h (data not shown). At these higher concentrations, 7,8-benzoflavone was significantly more toxic than 17α-ethynylestradiol. A decline in cytotoxicity was observed at higher concentrations (60–120 μm) of 7,8-benzoflavone, resulting from poor solubility of the compound.

The biotransformation of AFB₁ is a complex process wherein a multitude of CYP isoforms produce several oxidized metabolites of this mycotoxin. In human liver, CYP 1A2 and 3A4 appear to be principal isoforms in AFB₁ bioactivation. However, in human lung, the question of which of these isoforms may predominate in bioactivation of AFB₁ has not been established. Previous attempts to correlate CYP isoform specificity with respect to AFB₁ bioactivation in human lung have often been complicated by interindividual variability and poor stability of phase I enzymes in human lung tissue samples (8). The use of human lung cells that stably express these isoforms potentially overcomes these limitations (23).

Although specific AFB₁-DNA adducts were not analyzed in this study, it is possible that the profile of individual adducts is different between the two cell types. For example, in a similar study examining DNA adduct formation and mutations in CYP-expressing human liver cells exposed to AFB₁ in vitro, Macé et al. (9) found that, as in the present study, CYP 1A2-expressing human liver cells were significantly more sensitive to the cytotoxic effects of AFB₁ than 3A4-expresing cells. However, the 3A4-expressing cells produced a greater amount of the AFB₁-N⁷-Gua and ring-opened AFB₁-N⁷-Fapyr adducts compared with those that express CYP 1A2, which also produced some aflatoxin M₁-based DNA adducts (9). Determination of specific AFB₁-DNA adducts in these lung cells will be the subject of future investigations.

Competitive RT-PCR was used to measure differences in expression of CYP 1A2 and 3A4 mRNA in the human lung cell types used in this study. This method is highly specific to the isoform being measured (33) and is highly quantitative (34). Importantly, previous studies have shown that 1A2 and 3A4 mRNA levels correlate strongly with their corresponding protein expression and activity (44, 45, 46).

Here we report that whereas both CYP 1A2 and 3A4 are capable of bioactivating AFB₁, as determined by the ability of cells to produce cytotoxic and genotoxic intermediates, the former isoform appears to be a more predominant contributor at low substrate concentrations. This finding is in agreement with Gallagher et al. (47), who demonstrated that CYP 1A2-expressing microsomes likewise predominated in AFB₁ activation in human liver compared with those expressing CYP 3A4 at low AFB₁ concentrations relevant to many “real world” exposures.

As expected, the transformed cells exhibited CYP activity toward appropriate prototype substrates. For example, the introduction of CYP 3A4 cDNA increased nifedipine oxidase activity, a marker for CYP 3A, 15-fold compared with control cells. Likewise, MROD activity, a marker for CYP 1A2, was detected only in the CYP 1A2-expressing B-CMV1A2 cells. Expression of CYP mRNA was specific to transfected cells. In nontransfected BEAS-2B cells, we detected neither CYP 1A2 nor 3A4 mRNA. This is strongly supported by data measuring cytotoxicity, DNA binding, and isoform-specific activity assays. This result is also consistent with a previous report that BEAS-2B cells did not express CYPs 1A2 and 3A4 (48). The lack of CYP expression in this cell line may be a result of immortalization by SV40 large T antigen. It is also possible that the progenitor normal cell(s) for this cell line, which survived SV40 infection to become immortalized, was not a CYP-expressing cell type such as a Clara cell.

All cell types retained GST activity, the key phase II enzyme with respect to AFB₁ detoxification. It is also noteworthy that there were no significant differences in GST activity between cell types, suggesting that differences in susceptibility and DNA binding are not due to differences in GST activity but to CYP-mediated activation.

Cytotoxicity is dependent on the enzymatic generation of cytotoxic intermediates. As observed previously (23), introduction of CYP 1A2 and 3A4 cDNA into BEAS-2B cells had a profound effect on the sensitivity of cells in culture toward AFB₁. Of the three cell types used in this study, B-CMV1A2 cells were by far the most sensitive to AFB₁. B-CMV1A2 cells were 45-fold more susceptible than B3A4 cells and 3000-fold more susceptible than BEAS-2B cells. However, B3A4 cells were 70-fold more susceptible than the nontransformed BEAS-2B cells. When cytotoxicity was corrected for cellular CYP mRNA expression, CYP 1A2 was approximately 6-fold more efficient than CYP 3A4 in generating cytotoxic intermediates of AFB₁ at low AFB₁ concentrations (Table 4). Our data showing intact B-CMV1A2 cells to be more efficient at bioactivating AFB₁ at low concentrations compared with B3A4 cells supports earlier findings showing that CYP 1A2-expressing microsomes have a higher affinity toward low AFB₁ concentrations than do those expressing CYP 3A4 (10, 18, 49). The Hill constant generated from the AFB₁-mediated cytotoxicity curve in B3A4 cells indicated positive cooperativity (Fig. 2 B). This observation seems to support previous evidence of two cooperative AFB₁-binding sites in recombinant human liver 3A4 (18, 48).

Formation of AFB₁-DNA adducts, which likewise are dependent on CYP-mediated bioactivation of AFB₁, reflected the comparative susceptibility of these cells to the cytotoxicity of AFB₁. When expressed as a ratio of AFB₁-DNA adducts formed/CYP mRNA expressed, B-CMV1A2 cells were more efficient at activating at low AFB₁ concentrations (0–1.5 μm), whereas B3A4 cells were more efficient at higher concentrations (8–15 μm; Fig. 6). Interestingly, the concentration at which a cross-over of the relative efficiencies of 1A2- and 3A4-expressing cells to mediate AFB₁-DNA adduct formation (∼1.5 μm) occurred was near the point at which a cross-over in the efficiency to generate cytotoxicity also occurred (1–2 μm; Fig. 3).

The profound and specific effect of CYP inhibitors on AFB₁-DNA adduct formation suggests that activation of AFB₁ is primarily due to CYP activity rather than activity by other possible AFB₁-activating enzymes in these cells, such as LOX. LOX has a role in AFB₁ activation in human lung cytosol, but only at relatively high (50 μm) AFB₁ concentrations (7). It may also indicate that CYP-mediated AFB₁ activation may predominate at much lower concentrations.

Whereas CYP 3A4 is commonly isolated in human lung (8, 17, 50, 51), there are conflicting reports regarding whether CYP 1A2 is constitutively expressed. One group of investigators (17) found both CYP 1A2 protein and mRNA in human airway tissues, whereas others did not (8, 50, 52, 53), indicating that this isoform may not be universally expressed or that this protein is particularly unstable. In any event, the expression of this particular isoform is likely to be an important factor in potential risk to inhaled AFB₁.

Any assessment of risk to inhaled AFB₁ should take into account the relative amounts of CYP expression. It is feasible that under conditions where appropriate CYPs are expressed in lung, inhalation of AFB₁ may result in increased risk to lung cancer in exposed persons. Our data indicate that intact human lung cells expressing AFB₁-bioactivating CYP isoforms, especially 1A2, are extremely susceptible to AFB₁, even at low concentrations. It is also important to consider that localized AFB₁ concentrations within human lung cells exposed to AFB₁-laden dusts are largely unknown, making it difficult to accurately predict the in vivo roles of CYPs 1A2 and 3A4 for AFB₁ activation in the lung. Additional studies on the relative amounts of pulmonary CYP expression and the range of individual variation of expression will shed more light on the relative roles of these isoforms in vivo.