Activation of Polycyclic Aromatic Hydrocarbon trans -Dihydrodiol Proximate Carcinogens by Human Aldo-keto Reductase (AKR1C) Enzymes and Their Functional Overexpression in Human Lung Carcinoma (A549) Cells*

Polycyclic aromatic hydrocarbons (PAH) are environmental pollutants and suspected human lung carcinogens. In patients with non-small cell lung carcinoma, differential display shows that aldo-keto reductase (AKR1C) transcripts are dramatically overexpressed. However, whether AKR1C isoforms contribute to the carcinogenic process and oxidize potent PAH trans -di-hydrodiols (proximate carcinogens) to reactive and redox active o -quinones is unknown; nor is it known whether these reactions occur in human lungs. We now show that four homogeneous human recombinant aldo-keto reductases (AKR1C1–AKR1C4) are regioselective and oxidize only the relevant non-K region trans -dihy-drodiols. However, these enzymes are not stereo-selec-tive, since they oxidized 100% of these racemic substrates. The highest utilization ratios ( V max / K m ) were observed for some of the most potent proximate carcinogens known ( e.g. 7,12-dimethylbenz[ a ]anthracene-3,4-diol (DMBA-3,4-diol) and benzo[ g lysed by hypotonic shock in distilled water. The lysate was homogenized in a glass homogenizer and microcentrifuged for 5 min. Aliquots of the lysate were then added to 1.0-ml systems containing 100 m M potassium phosphate buffer (pH 7.0), 2.3 m M NAD (cid:2) , 1 m M 1-acenaphthenol, and 4% acetonitrile. The enzyme reaction was monitored by following the change in pyridine nucleotide absorbance at 340 nm for 15 min at 25 °C. Isolated proteins from each cell line were boiled for 5 min and were separated by SDS-PAGE and electrotransferred to nitrocellulose filters. Filters were incubated with polyclonal rabbit anti-AKR1C9 antiserum (71536) at a 1:1000 dilution (34). Immunoblots were developed by in- cubation with the goat anti-rabbit IgG-horseradish peroxidase conjugate using ECL Western blot detection reagent.

PAHs 1 are ubiquitous environmental pollutants and are tobacco carcinogens implicated in the causation of human lung cancer. PAHs are metabolically activated to exert their deleterious effects. Three principal pathways have been proposed for PAH activation and are shown for the representative compound BP (Fig. 1).
The first pathway involves the formation of radical cations catalyzed by P450 peroxidases (1). Radical cations form N-7 guanine-depurinating DNA adducts, a process that can lead to G to T transversions in ras (2,3).
In the second pathway, PAHs are activated by members of the CYP superfamily to form an arene oxide on the terminal benzo-ring; subsequent hydrolysis by epoxide hydrolase results in the formation of non-K region trans-dihydrodiols, which are potent proximate carcinogens (4). The trans-dihydrodiols can then undergo a secondary epoxidation to form bay region antior syn-diol epoxides, which are potent mutagens and tumorigens (4). Preferential formation of N 2 -dGuo stable adducts can lead to G to T transversions via error-prone trans-lesional synthesis (5). G to T transversions are common mutations observed in K-ras and p53 2 in human lung cancer (6).
Recently, differential display has shown the overexpression of human AKR1C isoforms in non-small cell lung carcinoma. High expression was observed in 384 patients and was a prognostic indicator of poor disease outcome (17). Because AKR1C1 is induced in ethacrynic acid-resistant human colon cells (18), it was suggested that overexpression of AKR1C may predispose patients to resistance to cancer chemotherapeutic agents. However, AKR1C isoforms have not been shown to metabolize these drugs.
In this study, our first goal was to determine the substrate specificity of all four human AKR1C isozymes toward a variety of PAH trans-dihydrodiols including DMBA-3,4-diol, 5MC-7,8diol, and B[g]C-11,12-diol. Each of these trans-dihydrodiols are precursors of diol-epoxides, which are potent rodent lung carcinogens (19 -21). Demonstration that these trans-dihydrodiols could be converted to o-quinones would suggest that human AKRs contribute to the further activation of these proximate carcinogens. We demonstrate that DMBA-3,4-diol is enzymatically oxidized to DMBA-3,4-dione, which was trapped as its mono-and bis-thioether conjugates. Our second goal was to determine whether AKR1C isoforms are functionally active in lung cells. We used array analysis to identify the human lung adenocarcinoma cell line A549 as a rich source of AKR1C isoforms and demonstrate that this cell line is also capable of converting DMBA-3,4-diol to DMBA-3,4-dione. The preferential oxidation of potent proximate carcinogenic trans-dihydrodiols by AKR1C isoforms in human lung cells suggests that the AKR pathway plays a role in PAH-induced lung carcinogenesis. The demonstration that the highly reactive DMBA-3,4-dione forms bis-thioether conjugates by sequential 1,6-and 1,4-Michael addition raises the possibility that bis-conjugates may form with other cellular nucleophiles and that these may have unique end organ toxicity.

EXPERIMENTAL PROCEDURES
Chemicals and Reagents-Cell culture media and reagents were obtained from Invitrogen. NADP ϩ and NAD ϩ nucleotide cofactors were obtained from Roche Molecular Biochemicals. 1-Acenaphthenol was obtained from Sigma. All solvents were HPLC grade, and all other chemicals used were of the highest grade available. Recombinant human AKR1C1-AKR1C4 proteins were overexpressed in Escherichia coli and purified to homogeneity as previously described (16). Their representative specific activities were as follows: AKR1C1, 2.1 mol/min/mg; AKR1C2, 2.5 mol/min/mg; AKR1C3, 2.8 mol/min/mg; AKR1C4, 0.21 mol/min/mg. Specific activities were measured using 100 mM potassium phosphate (pH 7.0), 2.3 mM NAD ϩ , and 1 mM 1-acenaphthenol (AKR1C1-3) or 75 M androsterone (AKR1C4).
Synthesis of PAH Metabolites-NP-1,2-diol was synthesized by reducing naphthalene-1,2-dione with sodium borohydride in ethanol (22). Other racemic trans-dihydrodiols were synthesized according to the following methods cited: BA-3,4-diol (23) (32). Note that all PAHs are potentially hazardous and should be handled in accordance with National Institutes of Health guidelines for the laboratory use of chemical carcinogens.
Spectrophotometric Assays-The initial velocities of enzymatic oxidation of each trans-dihydrodiol substrate were determined spectrophotometrically using 2.3 mM NADP ϩ as cofactor in 1.0 ml of 50 mM AMPSO buffer, pH 9.0, at 25°C. The trans-dihydrodiol substrates were dissolved in Me 2 SO, and the final concentration of organic solvent in the assay was 8%. Reactions were monitored by following the increase in absorbance of the reduced pyridine nucleotide at 340 nm on a Beckman DU 640 spectrophotometer. The specific activity for 1-acenaphthenol oxidation in the presence of 8% Me 2 SO was reduced by 50% for AKR1C1, 20% for AKR1C2, 22% for AKR1C3, and 65% for AKR1C4; therefore initial velocity data were corrected for the level of inhibition observed.
Preparation of the 2-Mercaptoethanol Conjugate(s) of 7,12-Dimethylbenz[a]anthracene-3,4-dione-Potassium phosphate buffer (50 mM, pH 7.0) containing 5 mM 2-mercaptoethanol and 25 M DMBA-3,4-dione in 8% Me 2 SO was incubated for 20 h at 37°C with stirring. The reaction was terminated by extraction with ethyl acetate. Organic extracts were combined and dried over sodium sulfate, and the organic solvent was removed under reduced pressure. The resulting solid was redissolved in acetonitrile and was analyzed by LC/MS.
Trapping of the Product of the AKR1C4-catalyzed Oxidation of DMBA-3,4-diol-To characterize the product of DMBA-3,4-diol oxidation catalyzed by AKR1C4, incubations were conducted in 50-ml systems containing 25 M DMBA-3,4-diol, 2.3 mM NADP ϩ , 50 mM potassium phosphate buffer (pH 7.0), and 5 mM 2-mercaptoethanol as trapping agents in the presence of 8% Me 2 SO. Following the addition of the purified enzyme (ϳ1 mg), the reaction was incubated for 20 h at 37°C and then terminated by extraction of the reaction mixture with ethyl acetate (two 50-ml aliquots). The organic extracts were combined and dried over sodium sulfate, and the organic solvent was removed under reduced pressure. The resulting solid was redissolved in acetonitrile and was analyzed by LC/MS. MS and MS/MS of the reaction product was compared with that obtained for the synthetically prepared DMBA-3,4-dione thioether conjugates.
Mass Spectrometric Analysis-Mass spectrometric data were acquired on a Finnigan LCQ ion trap mass spectrometer (ThermoQuest, San Jose, CA) equipped with a Finnigan electrospray ionization source.

TABLE I Oxidation of structurally diverse PAH trans-dihydrodiols by human AKR1Cs
Enzymatic reactions were in 50 mM AMPSO (pH 9) with 2.3 mM NADP ϩ . The concentrations of PAH diol substrates used were as follows: The mass spectrometer was operated in the positive ion mode. On-line chromatography was performed using a Waters Alliance 2690 HPLC system (Waters Corp., Milford, MA). A YMC C 18 ODS-AQ column was used at a flow rate of 0.9 ml/min. Solvent A was 5 mM ammonium acetate in water containing 0.01% trifluoroacetic acid, and solvent B was 5 mM ammonium acetate in methanol containing 0.01% trifluoroacetic acid with the gradient conditions as follows: 30% B at 0 min, 30% B at 5 min, 100% B at 16 min, 100% B at 24 min, and 30% B at 26 min. Cell Culture-The A549 human lung carcinoma cell line was obtained from the American Type Culture Collection (ATCC number CCL-185) and maintained in F-12K nutrient mixture (Kaighn's modification) with 10% heat-inactivated fetal bovine serum, 1% L-glutamate, and 1% OPI solution. HepG2 hepatoma cells were maintained in Eagle's minimal essential medium supplemented with 10% heat-inactivated fetal bovine serum, 1% L-glutamate, and 100 units/ml penicillin/streptomycin solution. H441 and H358 human bronchoalveolar cells were maintained in RPMI 1640 medium with 10% heat-inactivated fetal bovine serum, 1% L-glutamate, and 100 units/ml penicillin/streptomycin, whereas H358-AKR1C2 transfectants were grown in the same conditions as the parental cell line in the presence of 0.4 mg/ml G418 to maintain selection. Cells were incubated at 37°C in a humidified atmosphere containing 5% CO 2 and were passaged every 4 days at 1:10 dilution.
Multiple Tissue Expression Array-A dot blot containing poly(A) ϩ RNA from multiple human tissues (Human Multiple Expression Array; CLONTECH; catalog no. 7776-1) was hybridized to a randomly primed cDNA probe containing the entire open reading frame of AKR1C2 (XhoI/XbaI (ϳ1 kb)) fragment from pET16b-AKR1C2 (16). Random priming was achieved with radiolabeled [␣-32 P]dCTP, and a final specific activity greater than 10 9 cpm/g of DNA was attained. Hybridization was performed at 65°C for 6 h, and blots were washed with 2ϫ standard saline citrate plus 1% SDS at 60°C for 40 min. The blots were subjected to autoradiography at Ϫ70°C.
RNA Isolation and Northern Analysis-Cellular RNA was isolated using TRIZOL reagent. Total RNA (20 g) was separated by electrophoresis on 1% agarose/formaldehyde gels and transferred overnight to a Duralon-UV membrane (Stratagene). Membranes were prehybridized in ExpressHyb hybridization solution (CLONTECH) at 65°C for 30 min. After prehybridization, membranes were hybridized to 10 7 dpm of [␣-32 P]dCTP probe corresponding to the entire open reading frame of AKR1C2 cDNA that was labeled by random priming to a specific activity of 10 9 cpm/g of DNA. Hybridization was performed at 65°C for 6 h. After hybridization, the blot was subjected to two washes with 2ϫ SSC plus 1% SDS at 60°C for 20 min each. The blot was then exposed to x-ray film at Ϫ80°C overnight. For purposes of normalization, the blot was stripped and reprobed with a 1-kb fragment of ␤-actin labeled by random priming as described above.
Isoform-specific Reverse Transcription PCR of AKR1C mRNA in A549 Cells-Total RNA (1 g) from A549 cells was reverse transcribed into cDNA and was subjected to 25 cycles of reverse transcription PCR amplification. The PCR program included denaturation at 95°C for 2.5 min and then, for each cycle, a 30-s denaturation step at 95°C, a 2.5-min annealing step at 60°C, and a 7-min extension step at 72°C. The AKR1C isoform-specific primers used have been reported elsewhere (33). These amplimers yielded a PCR product of 500 bp, whereas the amplimers for GAPDH gave a product of 400 bp, which was used as a positive control.
Expression of AKR1Cs in Cell Lysates-Confluent cells (1 ϫ 10 7 ) were washed with 3 ml of phosphate-buffered saline containing 1 mM EDTA and harvested by centrifugation at 3000 rpm for 10 min. The cells were lysed by hypotonic shock in distilled water. The lysate was homogenized in a glass homogenizer and microcentrifuged for 5 min. Aliquots of the lysate were then added to 1.0-ml systems containing 100 mM potassium phosphate buffer (pH 7.0), 2.3 mM NAD ϩ , 1 mM 1-acenaphthenol, and 4% acetonitrile. The enzyme reaction was monitored by following the change in pyridine nucleotide absorbance at 340 nm for 15 min at 25°C.
Isolated proteins from each cell line were boiled for 5 min and were separated by SDS-PAGE and electrotransferred to nitrocellulose filters. Filters were incubated with polyclonal rabbit anti-AKR1C9 antiserum (71536) at a 1:1000 dilution (34). Immunoblots were developed by incubation with the goat anti-rabbit IgG-horseradish peroxidase conjugate using ECL Western blot detection reagent.

RESULTS
trans-Dihydrodiol Oxidation Catalyzed by Human AKR1C1-AKR1C4 -To determine the preference of AKR1C enzymes to oxidize potent PAH trans-dihydrodiol proximate carcinogens, a structural series of diols ranging in ring number and arrangement from naphthalene-1,2-diol to benzo[g]chrysene-11,12-diol was examined as potential substrates. In this substrate series, it is the presence of a non-K region trans-dihydrodiol, a bay region, and ultimately a methylated bay region or fjord region that dictates carcinogenicity of the resultant diol-epoxide (Fig.  3). Of the various trans-dihydrodiols examined, DMBA-3,4-diol and B[g]C-11,12-diol represent potent methylated bay region and fjord region proximate carcinogens, respectively (Fig. 2). Due to the limited solubility of PAH trans-dihydrodiols, direct determination of K m and V max values were not possible. Instead, the initial velocity of trans-dihydrodiol oxidation was measured at very low substrate concentrations relative to K m in the presence of saturating concentrations of NADP ϩ . The Michaelis-Menten equation in this case simplifies to v/[S] ϭ V max /K m , providing a direct estimation of utilization ratio. Specific activities and utilization ratios are reported for the various PAH trans-dihydrodiols examined with NADP ϩ as coenzyme at pH 9.0 (Table I).
In the case of naphthalene, which is a noncarcinogen, the following rank order of utilization ratios was observed for the oxidation of naphthalene-1,2-diol: AKR1C1 Ͼ AKR1C2 Ͼ AKR1C4 Ͼ AKR1C3. In the case of benzo[a]pyrene, a prominent tobacco carcinogen, the K region dihydrodiol, benzo-[a]pyrene-4,5-diol, was not a substrate for any of the AKRs tested, whereas the non-K region dihydrodiol, BP-7,8-diol, was oxidized by all of the AKRs. In this instance, AKR1C1 and AKR1C2 gave the highest utilization ratios. By contrast, AKR1C1 gave modest utilization ratios for the oxidation of other trans-dihydrodiols that contained more than two rings. The highest ratios were seen for 5-MC-7,8-diol, benz[a]anthracene-3,4-diol, and DMBA-3,4-diol.
AKR1C4 is a liver-specific AKR (33) that displayed the highest utilization ratios among the 1C members for many of the PAH trans-dihydrodiols tested. AKR1C4 had the highest utilization ratio for the most potent proximate carcinogens (e.g. DMBA-3,4-diol and the fjord region B[g]C-11,12-diol). Thus, a common trend observed across the human AKR1C isoforms was preferential oxidation of trans-dihydrodiols of potent proximate carcinogens and, in particular, DMBA-3,4-diol and B[g]C-11,12-diol.
Determination of the End Point of PAH trans-Dihydrodiol Oxidation Catalyzed by Human AKR1C Enzymes-To determine whether the AKR1C isoforms exhibit any stereochemical preference for the oxidation of the racemic PAH trans-dihydrodiols substrates, discontinuous HPLC-based enzyme assays were performed. This is an important consideration, since it is ultimately the (Ϫ)-R,R-trans-dihydrodiols that are metabolically formed. In these assays, the disappearance of PAH sub- strate over time was monitored in the presence of enzyme and cofactor. The nmol of trans-dihydrodiol consumed were calculated from appropriate standard curves, which related nmol of the authentic standard to peak area. Enzymatic reactions were conducted until no further change in peak area was observed. Data are shown for the AKR1C2-or AKR1C4-mediated oxidation of benz[a]anthracene-3,4-diol, DMBA-3,4-diol, 5MC-7,8diol, and B[g]C-11,12-diol (Fig. 4). In each instance, AKR1C2 and AKR1C4 enzymes consumed 100% of the trans-dihydrodiol substrate. The ability of the two isoforms (AKR1C2 and AKR1C4) with the highest utilization ratios for the oxidation of DMBA-3,4-diol to consume both of the enantiomers of the racemic substrate indicates that they have the capacity to oxidize either enantiomer formed during the incubation of DMBA with human liver microsomes.
Characterization of the Product of AKR1C4-mediated Oxidation of the Potent Proximate Carcinogen DMBA-3,4-diolo-Quinones are highly reactive Michael acceptors that have the propensity to form conjugates and/or DNA adducts via 1,4-Michael addition. Of the quinones anticipated to be formed by AKRs, DMBA-3,4-dione is among the most reactive; estimates of its bimolecular rate constant for GSH addition are on the order of 2.0 ϫ 10 6 min Ϫ1 M Ϫ1 (35). To determine whether DMBA-3,4-diol is oxidized by AKR1C isoforms to DMBA-3,4dione, an enzymatic reaction was performed in the presence of 5 mM 2-mercaptoethanol to act as a trapping agent. For these studies, we elected to use recombinant AKR1C4, which has a robust utilization ratio for the oxidation of DMBA-3,4-diol.
It was found that purified recombinant human AKR1C4 catalyzed the oxidation of DMBA-3,4-diol to yield two major peaks on HPLC (Fig. 5D). To verify that the peaks were derived from DMBA-3,4-dione, the o-quinone was reacted with an equivalent concentration of 2-mercaptoethanol. The synthetically prepared conjugates were found to have identical retention times on RP-HPLC to the peaks isolated from the enzymatic reactions (Fig. 5, B and D).
LC/MS was performed to further identify the peaks isolated by HPLC. This showed that the enzymatically formed products and the thioether conjugates prepared synthetically gave iden-tical molecular ions [MH ϩ ] and fragment ions (Fig. 6) 1 H NMR supports the assignment of the mono-thioether conjugate of DMBA-3,4-dione as the 1,6-Michael addition product. The NMR spectrum showed that this conjugate lacked the CH-2 vinylic proton (6.32 ppm), which is present in the synthetically prepared DMBA-3,4-dione as a double-doublet. Therefore, the first Michael addition can only be at C-2, yielding the 1,6-Michael addition product as drawn.
Expression of AKR1C Isoforms in Human Lung Cells-The observation that AKR1C enzymes efficiently oxidize DMBA-3,4-diol to DMBA-3,4-dione in vitro raised the question of whether there was functional expression of this activity in human lung, a site of PAH carcinogenesis.
Human multiple tissue expression analysis was performed in which 96 human tissues or cell lines were screened for AKR1C expression. In this array, the loading of the RNA transcript was normalized against eight housekeeping genes including ubiquitin. Data showed that AKR1C transcripts are abundantly expressed in human liver. Of the cell lines analyzed, the A549 lung carcinoma cell line showed the highest level of expression (Fig. 7A). When the array was stripped and reprobed with the cDNA for ubiquitin, equal loading of the blots was confirmed (data not shown). The presence of AKR1C transcripts in A549 lung cancer cells supports the earlier detection of the overexpression of these transcripts in non-small cell lung carcinoma but, importantly, identifies a cell line in which the functional expression of AKR1C can be assessed.

Detection of AKR1C Transcripts in Human Lung Carcinoma
Cells (A549)-To confirm the presence of human AKR transcripts in the A549 lung carcinoma cell line, a Northern blot analysis was performed. Human HepG2 cells are known to express AKR1C1-AKR1C4; therefore, mRNA from this cell line was used as a positive control. H441 and H358 cells (human bronchoalveolar cells of epithelial origin) are a null environment for human AKR expression and were used as a negative control. H358-AKR1C2 transfectants were used as a positive control for the AKR1C2 probe used in the hybridization. This probe was expected to detect all of the isoforms (AKR1C1-AKR1C4) due to their high sequence identity (Ͼ86%). As anticipated, AKR1C transcripts were detected in HepG2 cells and H358-AKR1C2 transfectants (Fig. 7, lanes 1 and 5). Moreover, H441 and H358 cells had no signal showing the absence of the transcript (Fig. 7, lanes 3 and 4). By contrast, the A549 lung carcinoma cell line had the highest level of AKR1C transcripts as seen from the blot, when normalized against the GAPDH loading control (Fig. 7, lane 2).
Using our recently established isoform-specific RT-PCR protocol (33), we next identified which AKR1C isoforms were present in the A549 cell line. Isoform-specific primers designed to yield 500-bp fragments for each transcript were used to amplify the mRNA species present in this cell line. AKR1C1, AKR1C2, and AKR1C3 were detected as 500-bp PCR products, whereas AKR1C4 was not found (Fig. 8). The absence of AKR1C4 is consistent with previous findings, which indicates that this is a liver-specific enzyme (33,36). GAPDH primers were used to generate a 400-bp fragment as a positive control.
Functional Expression of AKR1C in A549 Cells-To confirm the presence of AKR proteins in the A549 lung carcinoma cell line, Western blot analysis was performed. Polyclonal AKR1C9 antibody was used to detect AKR1C1-3 proteins. HepG2 cells known to contain AKR proteins were used as positive controls, whereas H358 cells known to be a null environment for AKRs were used as negative controls. A 37-kDa band corresponding to AKR1C proteins was seen in lanes containing A549 and HepG2 cell lysates, confirming the presence of AKR proteins in those cells. H358 cell lysates did not show a band at 37 kDa as was expected (Fig. 9A).
To verify that the AKR1C proteins expressed in the A549 lung carcinoma cell line were functional, enzyme activity was measured using a standard assay in which the NAD ϩ -dependent oxidation of 1-acenaphthenol was measured. The specific activity for 1-acenaphthenol oxidation was 50 and 35 times greater in A549 and HepG2 cell lysates, respectively, than that observed in H358 and H441 cell lysates. The H358-AKR1C2 stable transfectants had a specific activity that was 10 times greater than the null cells (Fig. 9B). Thus, the level of AKR1C enzyme activity in A549 cells exceeds that in any other cell line measured to date.
Demonstration That A549 Lung Carcinoma Cells Convert DMBA-3,4-diol to DMBA-3,4-dione-To determine whether the AKR1C activity in A549 cells was sufficient to convert DMBA-3,4-diol to DMBA-3,4-dione, A549 cell lysates (2.0 ϫ 10 7 cells) were incubated with DMBA-3,4-diol in the presence of NADP ϩ . The reaction was incubated for 20 h and monitored by RP-HPLC. After extraction with ethyl acetate, two product peaks identical in terms of HPLC retention times to those observed in the reaction with AKR1C4 were observed. LC/MS analysis gave identical molecular ions and fragment ions to those observed with the authentic synthetic standards, m/z ϭ 439 ([MH ϩ ],

DISCUSSION
This is the first complete study demonstrating the ability of four homogeneous recombinant human AKR1C enzymes to oxidize a wide variety of PAH trans-dihydrodiols. We show that AKR1C isoforms preferentially oxidized PAH trans-dihydrodiols derived from potent carcinogenic hydrocarbons (e.g. DMBA). Further, this demonstrates that DMBA-3,4-diol is oxidized to the corresponding DMBA-3,4-dione. This ortho-quinone represents one of the most electrophilic and redox-active quinones that can be derived from the metabolic activation of PAH by AKRs. Bimolecular rate constants for the addition of GSH are typically 2.0 ϫ 10 6 min Ϫ1 M Ϫ1 , and rates of O 2 consumption under redox-cycling conditions indicate that the quinone is cycled rapidly (13,35,37). This observation is coupled with the finding that human multiple tissue expression array analysis demonstrates that of 96 tissues and cell lines examined, the human lung carcinoma cell line A549 showed one of the highest expression levels of AKR1C enzymes. The ability of A549 cell lysates to mirror the conversion of DMBA-3,4-diol to DMBA-3,4-dione suggests that the AKR pathway may contribute to PAH activation in human lung.
AKR1C enzymes were found to oxidize a wide variety of PAH trans-dihydrodiols of increasing ring number and methylation. All four AKRs examined were regiospecific; they oxidized non-K region trans-dihydrodiols but were inactive on K region trans-dihydrodiols. We have previously documented the ability of rat AKR1C9 to oxidize benzo[a]pyrene-7,8-diol and not benzo[a]pyrene-4,5-diol (7). Of the non-K region trans-dihydrodiols oxidized by the human enzymes, there was a distinct preference for the oxidation of methylated non-K region trans-dihydrodiols, and one of the most preferred substrates was DMBA-3,4-diol. This bay region methylated diol is one of the most potent proximate carcinogens known. In addition, AKR1C4 showed a distinct preference for both DMBA-3,4-diol and the fjord region trans-dihydrodiol B[g]C-11,12-diol.
In previous work, we showed that benzo[a]pyrene-7,8-dione was the product of human AKR1C oxidation of the corresponding BP-7,8-diol by measuring the co-chromatography of o-quinone-glycine conjugates. However, the product of DMBA-3,4diol oxidation catalyzed by any AKR had not been previously identified. The presumptive product, DMBA-3,4-dione, has a bimolecular rate constant for the addition of GSH that exceeds that observed for the addition of N-acetyl-L-cysteine to anti-BPDE, a diol-epoxide, by 3 orders of magnitude (38). This high bimolecular rate constant poses a challenge to the effective trapping of DMBA-3,4-dione. This o-quinone is also highly redox-active. In this study, the DMBA-3,4-dione was successfully trapped as thiol ether conjugates using 5 mM 2-mercaptoethanol. The enzymatic products were identical to the synthetically prepared standards when analyzed by LC/MS. LC/MS analysis was revealing, since it identified two major product peaks. In previous reactions using rat AKR1C9 and either naphthalene-1,2-diol or benzo[a]pyrene-7,8-diol as substrates, only one product was trapped with 2-mercaptoethanol, and this corresponded to the 1,4-Michael addition product of the fully oxidized o-quinone. By contrast, the products obtained from the AKR1C4 oxidation of DMBA-3,4-diol and subsequently trapped with thiol correspond to a mono-thioether conjugate that arises from 1,6-Michael addition and a bisthioether conjugate formed by 1,4-Michael addition to the mono-thioether conjugate. It is proposed that the first Michael addition occurs via 1,6-addition through the ring system, reminiscent of the Michael addition to quinone-methides, and may occur due to the steric hindrance of the bay region methyl group (Fig. 11). NMR assignments support this sequential addition, since the mono-thioether conjugate lacks the vinylic proton at C-2.
The ability to form a novel bis-thioether conjugate may explain why previous attempts to chemically synthesize a simple thioether conjugate of DMBA-3,4-dione failed. Importantly, the detection of the bis-conjugate raises the issue of whether other cellular nucleophiles could form similar bis-conjugates (e.g. GSH). Bis-and trisglutathionyl ether conjugates of a variety of polyphenols have been implicated in hematotoxicity and nephrotoxicity, raising the prospect that bis-thioether conjugates of DMBA-3,4-dione may have their own unique end organ toxicity (39).
The ability of human AKR1C isoforms to efficiently oxidize DMBA-3,4-diol with high utilization ratios was an impetus to determine whether these enzymes were highly expressed in PAH target tissues. Analysis of a multiple human tissue expression array provided the clue that AKR1C isozymes were highly expressed in the human lung carcinoma cell line, A549. The high expression of AKR1C isoforms in this cell line was validated by Northern blot analysis, RT-PCR, Western blot analysis, and functional assay for enzyme activity. By contrast, the enzyme was not expressed in human broncholaveolar cells (H358 or H441). The high expression of AKR1C isoforms observed in lung carcinoma cell line implicates the AKR pathway in the metabolic activation of PAH in human lungs.
AKR1C transcripts were shown to be overexpressed in human non-small cell lung carcinoma using the technique of differential display, but the enzymes were not shown to be functionally active (17). This earlier study found that overexpression of AKR1C isoforms was a prognostic marker of poor disease outcome. The findings were interpreted as being related to resistance to chemotherapeutic agents, yet the detoxification of cancer chemotherapeutic agents catalyzed by AKR1C has never been demonstrated.
To further support a possible role of this enzyme in the metabolic activation of PAH in lungs, we determined whether A549 cell lysates were capable of catalyzing the same transformation observed with recombinant AKR1C4 in vitro. LC/MS of the metabolites of DMBA-3,4-diol isolated from A549 cell lysates showed that identical products to those formed by the recombinant enzyme could be detected. Thus, lung carcinoma cells can produce DMBA-3,4-dione as a result of AKR-mediated trans-dihydrodiol oxidation.
As discussed, DMBA-3,4-dione is also highly redox-active and will produce significant amounts of ROS in the presence of reducing equivalents. One measure of ROS production is the formation of 8-oxo-dGuo. Importantly, PAH target tissues exposed to the parent hydrocarbon, DMBA, do show either a significant amount of 8-oxo-dGuo formation (40) or etheno-adducts derived from decomposition of lipid hydroperoxides (41). Thus, the AKR pathway described provides one route to these oxidatively derived DNA adducts, in DMBA-exposed sites.
When these data are taken together, we propose the following. Tobacco smoke is causally related to lung cancer, and PAHs are tobacco carcinogens. Exposure to PAH induces CYP1A1/CYP1B1 in human lung cells (42,43), and AKR1C1 is induced by PAH and ROS (44). This concerted induction would increase the flux of trans-dihydrodiols to redox-active o-quinones and enhance the production of ROS. This would lead to a further overexpression of AKR1C isozymes. AKR1C isozymes will on the one hand provide a defense mechanism against the harmful effects of ROS, since they are efficient catalysts of 4-hydroxy-2-nonenal reduction, a decomposition product of lipid hydroperoxides (45). On the other hand, induced AKR1C isozymes can divert PAH trans-dihydrodiols from CYPs to the deleterious o-quinones. If the o-quinones and the ROS they generate are not eliminated, these have the potential to cause covalent and oxidative DNA lesions, respectively, increasing the mutational load of PAH-exposed lung cells. We have recently shown that AKR-derived PAH o-quinones will cause change-in-function mutations in p53 via G to T transversions and that these mutations result from ROS (46). Together, these events may contribute to PAH-induced lung carcinogenesis.