Association of Human Cytochrome P450 1A1 (CYP1A1) and Sulfotransferase 1A1 (SULT1A1) Polymorphisms with Differential Metabolism and Cytotoxicity of Aminoflavone

Aminoflavone [AF; 5-amino-2-(4-amino-3-fluorophenyl)-6,8-difluoro-7-methyl-4H-1-benzopyran-4-one; NSC 686288] is a novel fluorinated diaminoflavone analogue that has shown antitumor activity in vitro and in vivo against a variety of human tumor types, including breast, renal, and ovarian cancers (1–4). AF is currently being evaluated in a lysine derivative prodrug form (AFP464), which is rapidly converted to AF in plasma by nonspecific plasma esterases, in phase I clinical trials at several institutions including the Karmanos Cancer Institute. It is entering phase II trials for the treatment of breast cancer. Mechanistic studies on the cytotoxicity of AF have revealed that AF induces DNA damage associated with DNA-protein cross-link, DNA single-strand breaks, and DNA replication–dependent DNA double-strand breaks (3, 5). AF also inhibits DNA synthesis and induces S-phase arrest (3). A substantial body of evidence suggests that AF requires metabolic activation by cytochrome P450 1A1 (CYP1A1) and sulfotransferase 1A1 (SULT1A1) to exert its antitumor activity (1, 2, 4, 6). Specifically, AF is metabolized by CYP1A1 to form hydroxylamines that are readily conjugated by SULT1A1, an unstable N-sulfoxy-AF metabolite that spontaneously decomposes, and highly reactive nitrenium ions that react with cellular DNA and protein, leading to cell death and apoptosis (4). AF itself also induces its own metabolism by activation of transcription of CYP1A1 and SULT1A1 through the aryl hydrocarbon receptor pathway (1, 6).

The human genes for CYP1A1 and SULT1A1 are polymorphic. Besides the wild-type (CYP1A1*1), 10 alleles have been identified for human CYP1A1 (http://www.cypalleles.ki.se/). The most common single-nucleotide polymorphisms (SNP) are CYP1A1*2C (I462V) and CYP1A1*4 (T461N). Both result in amino acid substitutions in the heme binding region of the enzyme (7). Allele frequencies of CYP1A1*2C are 2.2% to 8.9% and 19.8% in Caucasians and Japanese, respectively (8–11). CYP1A1*4 allele occurs in 2.0% to 5.7% of a Caucasian population (8–11). CYP1A1 polymorphisms have been associated with different enzyme activity (12–14) and inducibility (15), and altered susceptibility to cancer risk (16–18).

With respect to SULT1A1, besides the wild-type (SULT1A1*1), the most common SNPs are SULT1A1*2 (R213H) and SULT1A1*3 (M223V). These allele frequencies vary in different ethnic population: In Caucasians, SULT1A1*1 (frequency, 65.6%) is the most common allele followed by SULT1A1*2 (33.2%) and SULT1A1*3 (1.2%); in Chinese, SULT1A1*1 is the most predominant allele (91.4%), whereas SULT1A1*2 and SULT1A1*3 alleles are rare (8% and 0.6%, respectively); in African American, both SULT1A1*2 and SULT1A1*3 alleles are common (29.4% and 22.9%, respectively; ref. 19). The allelic variants of SULT1A1 have been associated with altered enzyme activity and thermal stability of the enzyme, with the SULT1A1*2 allele being associated with low enzyme activity and thermal stability (20, 21).

AF can be considered as a “prodrug” that requires sequential metabolic activation by CYP1A1 and SULT1A1 to exert its antitumor activity. Besides the liver, CYP1A1 and SULT1A1 are expressed in extrahepatic tissues including normal (e.g., lung) and tumor cells. Thus, functional genetic polymorphisms in these enzymes may influence the metabolic activation and cytotoxicity of AF in both normal and tumor cells, thereby contributing to interindividual or interethnic variability in clinical outcomes of efficacy and safety of patients treated with AF. The purpose of this study was to determine the functional significance of common polymorphisms of human CYP1A1 and SULT1A1 genes on the metabolism and cytotoxicity of AF. To this end, Chinese Hamster lung fibroblast V79 cells were genetically engineered to stably express human CYP1A1*1 (wild-type), CYP1A1*2C (I462V), or CYP1A1*4 (T461N) and coexpress human wild-type CYP1A1 with human SULT1A1*1 (wild-type), SULT1A1*2 (R213H), or SULT1A1*3 (M223V). The metabolism and cytotoxicity of AF were evaluated in these cellular models.

AF (NSC 686288) was provided by the National Cancer Institute (Bethesda, MD). Resorufin, 7-ethoxyresorufin, dicumarol, and DMSO were purchased from Sigma. G418 (geneticin), hygromycin B, DMEM, Opti-MEM, penicillin, streptomycin, and pcDNA3.1(+) and pcDNA3.1/Hygro(+) expression vectors were obtained from Invitrogen. QuikChange Site-Directed Mutagenesis kit was obtained from Stratagene. The wild-type human CYP1A1 cDNA was kindly provided by Dr. F.J. Gonzalez (National Cancer Institute). The wild-type human SULT1A1 cDNA and Pierce bicinchoninic acid (BCA) protein assay kit were purchased from Open Biosystems Thermo Scientific. Fetal bovine serum (FBS) was purchased from Atlanta Biologicals. Purified recombinant enzymes of CYP1A1 and SULT1A1 were obtained from BD Biosciences. 3-(4,5-Dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium inner salt (MTS) assay kit was obtained from Promega.

Chinese Hamster V79 cell line was obtained from the American Type Culture Collection. The cells were cultured in DMEM supplemented with 10% FBS, 100 units/mL penicillin, and 100 μg/mL streptomycin at 37°C in the presence of 5% CO₂. The identity of the V79 cell line was verified by the American Type Culture Collection using short tandem repeat profiling analysis. No authentication was done by us.

The wild-type human CYP1A1 cDNA was inserted into the expression vector pcDNA3.1(+) that contains G418 as selection antibiotic gene, and served as a template to generate the CYP1A1*2C and CYP1A1*4 recombinant expression vector by site-directed mutagenesis using QuikChange Site-Directed Mutagenesis kit. The entire coding region of CYP1A1, including the mutated sites, was verified by complete DNA sequence. The recombinant CYP1A1 expression vector was linearized with Scal (Invitrogen) and used for transfection of cells by electroporation. In brief, V79 cells (5 × 10⁶/400 μL Opti-MEM) were transfected with 20 μg of Scal-linearized pcDNA3.1/CYP1A1*1, pcDNA3.1/CYP1A1*2C, or pcDNA3.1/CYP1A1*4 plasmids in a 2-mm gene pulser cuvette using Gene Pulser Xcell electroporation system (200 V and 950 μF; Bio-Rad). Immediately after electroporation transfection, the cells were grown in fresh DMEM (containing 10% FBS) for 48 hours. The cells were then cultured in the presence of 1 mg/mL G418 (Geneticin) until resistant clones appeared. Multiple clones for each transfection were isolated and expanded. The levels of CYP1A1 mRNA and protein expression in each clone were determined by real-time reverse transcription-PCR (RT-PCR) and Western blot, respectively, as described below. Stable clones (denoted as V79-CYP1A1*1, V79-CYP1A1*2C, and V79-CYP1A1*4) expressing comparable CYP1A1 mRNA were selected and used for functional experiments. CYP1A1 activities in the selected clones were assessed by ethoxyresorufin-O-deethylase (EROD) assay.

The wild-type human SULT1A1 cDNA was inserted into the expression vector pcDNA3.1/Hygro(+) that contains hygromycin B as selection antibiotic gene, and served as a template to generate SULT1A1*2 and SULT1A1*3 recombinant expression vector by site-directed mutagenesis using QuikChange Site-Directed Mutagenesis kit. The entire coding region of SULT1A1, including the mutated sites, was verified by complete DNA sequence. The recombinant SULT1A1 expression vectors were linearized with SspI (Invitrogen) and used for transfection of the V79-CYP1A1*1 cell line, as described above. The cells were cultured in the presence of 1 mg/mL G418 and 200 μg/mL hygromycin B 48 hours after transfection until resistant clones appeared. Multiple clones for each transfection were isolated and expanded. Protein expressions and mRNA levels of SULT1A1 and CYP1A1 in clones were determined by Western blot and real-time RT-PCR, as described below. Stable clones (denoted as V79-CYP1A1-SULT1A1*1, V79-CYP1A1-SULT1A1*2, and V79-CYP1A1-SULT1A1*3) expressing comparable SULT1A1 and CYP1A1 mRNA were selected and used for functional experiments.

Cell lysates from the V79 cells that were transfected with the blank vector (negative control), CYP1A1 alone, or CYP1A1 plus SULT1A1 (∼40 μg protein per lane) were subjected to SDS-PAGE and transferred to nitrocellulose membranes. To detect protein expression of CYP1A1 and SULT1A1, membranes were probed with rabbit anti-CYP1A1 antibody (1:10,000; BD Biosciences) and mouse anti-SULT1A1 antibody (1:500; Sigma-Aldrich), respectively, and detected by the chemiluminescent system using horseradish peroxidase–conjugated goat anti-rabbit IgG (1:10,000; Santa Cruz Biotechnology) and donkey anti-mouse IgG (1:2,500; Santa Cruz Biotechnology) as the secondary antibody, respectively. The commercially available purified recombinant enzymes of CYP1A1 (5 μg per lane) and SULT1A1 (2 μg per lane; BD Biosciences) were used as the positive control. β-Actin was used as a loading control.

Total RNA was extracted from cell pellets using RNeasy Plus Mini kit (Qiagen). The first-strand cDNA was synthesized using qScript cDNA SuperMix kit (Quanta Biosciences). Real-time PCR was done using FastStart SYBR Green Master (Roche Applied Science) on an iCycler iQs real-time PCR detection system (Bio-Rad). Thermal cycling conditions for CYP1A1 and SULT1A1 include an initial denaturation at 95°C for 8 minutes, followed by 45 cycles of denaturation at 95°C for 20 seconds and combined annealing and extension at 60°C for 30 seconds and 72°C for 30 seconds. Primer sequences used were as follows: human CYP1A1, GCTACCTACCCAACCCTTCC (sense) and GCTCCTTGACCATCTTCTGC (antisense); human SULT1A1, AGGAGTTCATGGACCACAGC (sense) and TGAAGGTGGTCTTCCAGTCC (antisense); Chinese hamster glyceraldehyde-3-phosphate dehydrogenase (GAPDH), ATCACTGCCACCCAGAAGAC (sense) and GGATGCAGGGATGATGTTCT (antisense). Gene expression was analyzed using iQ5 Optical system software (Bio-Rad). mRNA levels of CYP1A1 or SULT1A1 were normalized to the GAPDH internal standard.

CYP1A1 enzyme activities in the CYP1A1-transfected stable clones were assessed by EROD activity, as described previously with modifications (22). Briefly, the cells (2 × 10⁵ per well) were seeded in a 96-well plate and, 24 hours after seeding, washed once with PBS and incubated in culture medium containing varying concentrations of 7-ethoxyresorufin (0.5–16 μmol/L) at 37°C for 30 minutes. Dicumarol (10 μmol/L) was also added to the assay medium to prevent further metabolism of resorufin by the cytosolic enzyme diaphorase (23). At the end of incubation, a 75-μL aliquot of cell medium was collected from each well and transferred to another 96-well plate, to which 200 μL ethanol was added into each well. The formation of resorufin was determined fluorimetrically at 530-nm excitation and 590-nm emission wavelength using a Victor X3 2030 Multilabel Reader (Perkin-Elmer). A standard curve of resorufin (1–800 μmol/L) was prepared in culture medium. The protein concentration in each cell sample was determined by BCA protein assay according to the manufacturer's protocol. The concentration of resorufin was normalized to cell protein concentration.

The functional influence of CYP1A1 polymorphisms on the metabolism of AF was determined in the V79 cells stably expressing CYP1A1*1 (wild-type), CYP1A1*2C (I462V), or CYP1A1*4 (T461N). The cells were seeded in 10-cm Petri dishes (5 × 10⁶ per dish). Twenty-four hours after seeding, the cells were incubated at 37°C in DMEM containing varying concentrations of AF (i.e., 0.5, 1, 2.5, 5, 10, and 25 μmol/L) for 6 hours; at the end of incubation, the cells were washed five times with ice-cold PBS, and cell pellet was collected and stored at −80°C until analysis. Two independent experiments were done, with duplicates in each experiment.

AF was also incubated with CYP1A1 Supersomes (BD Biosciences) as the positive control. The reaction mixtures (total volume, 0.2 mL) containing 50 μmol/L AF, 1.3 mmol/L NADP⁺, 3.3 mmol/L glucose-6-phosphate, 0.4 units/mL glucose-6-phosphate dehydrogenase, 3.3 mmol/L magnesium chloride, and 100 pmol/mL CYP1A1 enzyme in 100 mmol/L potassium phosphate buffer (pH 7.4) were incubated at 37°C for 30 minutes and terminated by adding 100 μL acetonitrile and centrifugation at 14,000 rpm at 4°C for 10 minutes. The supernatant was collected and stored at −80°C until analysis. An incubation with control Supersomes was done simultaneously with the negative control.

The functional influence of SULT1A1 polymorphisms on the metabolism of AF was determined in the V79-CYP1A1-SULT1A1*1, V79-CYP1A1-SULT1A1*2, and V79-CYP1A1-SULT1A1*3 cells. The cells were seeded in 10-cm Petri dishes (5 × 10⁶ per dish). Twenty-four hours after seeding, the cells were incubated at 37°C in DMEM containing 1 and 10 μmol/L AF for 6 hours; at the end of incubation, the cells were washed five times with ice-cold PBS, and cell pellet was collected and stored at −80°C until analysis. Three independent experiments were done, with duplicates in each experiment.

The intracellular concentrations of AF and its metabolites were determined using a validated high-performance liquid chromatography (HPLC) method. The cell pellet was resuspended in 300 μL PBS. Into an aliquot of 250 μL cell suspension, 1 mL of acetyl acetate was added, vortex mixed for 1 minute, and centrifuged at 14,000 × g and 4°C for 10 minutes. The supernatant was evaporated to dryness under nitrogen gas, and the residual was reconstituted in 100 μL of mobile phase (acetonitrile/0.5% formic acid in water, 42:58, v/v), and an aliquot of 50 μL was injected to Shimadzu LC-10ADvp HPLC system equipped with a photodiode detector (Shimadzu Corp.). AF and its metabolites were separated on an XTerra RP18 column (5 μm, 150 × 3.9 mm inside diameter; Waters) with a mobile phase consisting of acetonitrile–0.5% formic acid in water (42:58, v/v) at a flow rate of 1 mL/min. The detection wavelength was set at 350 nm, and the autosampler temperature was set at 4°C. The calibration curve for AF was constructed over the concentration range of 0.02 to 200 μmol/L. The within- and between-day precision and accuracy for the calibrators and quality control plasma samples were <15%. Because the reference standards of AF metabolites were not available, the concentrations of all metabolites were calculated from the calibration curve of AF. The protein content in each cell pellet sample was determined by BCA protein assay using human serum albumin as standards. The intracellular concentrations of AF and its metabolites were normalized to cell protein content.

The monohydroxy and dihydroxy metabolites of AF were identified using a Waters Quattro Micro triple quadrupole mass spectrometer (MS; Waters). HPLC separation was done on a Waters XTerra MS column (50 mm × 2.1 mm inside diameter) with a gradient mobile phase consisting of acetonitrile–0.5% formic acid in water at a flow rate of 0.3 mL/min. The eluent was monitored at the m/z of 321.1 (for AF), 337.1 (for monohydroxy-AF), and 353.1 (for dihydroxy-AF) under the following MS conditions: source temperature of 120°C, desolvation temperature of 350°C, cone voltage of 40 V, and capillary voltage of 300 V.

The cytotoxicity of AF in the V79 cell lines stably expressing CYP1A1 alone or coexpressing CYP1A1 and SULT1A1 was evaluated by MTS assay. The subconfluent cells were trypsinized and seeded in 96-well plates at the seeding number of 1 × 10⁴ per well. Twenty-four hours after seeding, the cells were incubated in culture medium containing AF at the concentrations of 0 (control), 0.01, 0.1, 0.5, 1, 5, and 10 μmol/L for 72 hours. At the end of treatment, drug-containing medium from all wells was removed, and cells were fed with 100 μL of drug-free medium followed by addition of 20 μL of the combined MTS/phenazine methosulfate solution. The plates were incubated at 37°C for 4 hours, after which the amount of soluble formazan produced by cellular reduction of MTS was measured spectrophotometrically at 490 nm using a Victor X3 2030 Multilabel Reader (Perkin-Elmer). The MTS assay for each cell lines was done in quadruplicate in three independent experiments.

The enzyme kinetic constants of maximum rate of reaction (V_max) and substrate affinity (K_m) for formations of hydroxylated AF metabolites were estimated by fitting the pooled rate measurement data from two independent experiments (with duplicates for each) to Michaelis-Menten equation (Eq. A). The drug concentration producing 50% inhibition of cell growth compared with the control (EC₅₀) was determined by fitting the pooled cell survival data from three independent experiments to an inhibitory effect Sigmoid E_max model (Eq. B). All fittings were done using nonlinear regression with the software WinNonlin 5.0 (Pharmsight Corp.). The estimated values of enzyme kinetic constants (V_max and K_m) and EC₅₀ were presented as mean ± SE as determined by nonlinear regression.

where S is substrate concentration, v is the rate of metabolite formation, V_max is maximum metabolite formation velocity, and K_m is the substrate concentration at which 50% of V_max is obtained.

where E represents percent cell survival under treatment relative to the control (no drug treatment), E_max is percent cell growth of the control (100%), C is the drug concentration, EC₅₀ is the drug concentration producing 50% inhibition of cell growth compared with control, and γ is the Hill coefficient.

Chinese Hamster lung fibroblast V79 cells, genetically engineered to expressed metabolizing enzymes, are widely used for determining the contributions of specific enzymes to the metabolic profile of a given xenobiotic or for examining the role of metabolism in the drug cytotoxicity (24, 25). To determine the functional influence of CYP1A1 polymorphisms on the metabolism and cytotoxicity of AF, the V79 cells were stably transfected with human CYP1A1*1 (wild-type), CYP1A1*2C (I462V), or CYP1A1*4 (T461N). To further determine the functional influence of SULT1A1 polymorphisms on the metabolic activation of AF, the V79 cells were cotransfected with human CYP1A1*1 and SULT1A1*1 (wild-type), SULT1A1*2 (R213H), or SULT1A1*3 (M223V). Real-time RT-PCR and Western blot indicated a comparable mRNA/protein expression of CYP1A1 or SULT1A1 among the selected stable clones expressing different variants of CYP1A1 or SULT1A1 (Fig. 1A and B).

EROD assay suggested that the V79-CYP1A1*2C cells exhibited a significant higher catalytic rates in terms of V_max (3,111 ± 185 pmol/min/mg protein) compared with the V79-CYP1A1*4 (V_max, 1,269 ± 918 pmol/min/mg protein) and V79-CYP1A1*1 cells (V_max, 918 ± 28 pmol/min/mg protein; Fig. 1C). Our results are in good agreement with earlier investigations in which a higher EROD activity of the CYP1A1 I462V variant was observed for the purified recombinant enzyme from Escherichia coli or baculovirus/insect expression systems (26, 27) as well as for microsomes prepared from sf9 cells coexpressing human CYP1A1 variants and cytochrome P450 reductase (28).

When incubated with the V79-CYP1A1*1, V79-CYP1A1*2C, or V79-CYP1A1*4 cells, AF produced a similar metabolite profile as that from incubation with CYP1A1 Supersomes (Fig. 2), suggesting that the metabolism of AF in these cell lines is predominantly mediated by CYP1A1. Consistent with the findings from earlier studies (29), HPLC and liquid chromatography–tandem MS (LC-MS/MS) analyses suggested that the major metabolites of AF formed in the CYP1A1 Supersomes reaction as well as in CYP1A1-expressing cells were ⁴N-OH-AF [retention time (RT), 10.0 minutes], ⁵N-OH-AF (RT, 5.9 minutes), 3-OH-AF (RT, 10.8 minutes), ⁴N,⁵N-diOH-AF (RT, 3.4 minutes), and 3,⁵N-diOH-AF (RT, 4.4 minutes; Fig. 2). Metabolic kinetic studies revealed that all CYP1A1 variants catalyzed hydroxylations of AF, yet with varying catalytic activity (Table 1; Fig. 3). Whereas there was no apparent difference in the enzyme kinetic constants of V_max and K_m between the cells expressing CYP1A1 wild-type and T461N variant, the cells expressing I462V variant exhibited significantly higher catalytic capacity in terms of the V_max for the formations of both monohydroxylated and dihydroxylated AF metabolites. The V_max for I462V for the formations of ⁴N-OH-AF, ⁵N-OH-AF, 3-OH-AF, and ⁴N,⁵N-OH-AF was estimated as 3.2-, 1.7-, 2.8-, and 2.8-fold, respectively, of those for the wild-type enzyme (Table 1; Fig. 3). With regard to the K_m, the I462V had a higher affinity for the formation of ⁵N-OH-AF (K_m, 4.2 μmol/L for I462V versus 24.7 μmol/L for wild-type) while exhibiting a higher K_m value for the formations of ⁴N-OH-AF, 3-OH-AF, and ⁴N,⁵N-OH-AF than the wild-type enzyme (Table 1). Taken together, I462V exhibited a 1.4-, 11.3-, and 1.6-fold higher intrinsic enzyme activity or catalytic efficiency (V_max/K_m) for the formations of ⁴N-OH-AF, ⁵N-OH-AF, and 3-OH-AF, respectively, than the wild-type (Table 1).

Because the sulfate metabolites of AF were unstable and not detectable under the present HPLC condition, the reductions in the intracellular concentrations of AF and hydroxylated metabolites may indicate the extent of SULT1A1-mediated sulfation in the cells coexpressing CYP1A1 and SULT1A1. As shown in Fig. 4A, after the cells were incubated with 1 μmol/L AF for 6 hours, the intracellular concentration of AF in the cells expressing CYP1A1 alone and coexpressing CYP1A1 with SULT1A1*1, SULT1A1*2, and SULT1A1*3 was 42%, 19%, 44%, and 14%, respectively, of that in the V79-vector cells (control); similarly, after the cells were incubated with 10 μmol/L AF for 6 hours, the intracellular AF concentrations in the respective cells were 64%, 27%, 64%, and 26%, respectively, of that in the control. These results suggested that the cells coexpressing CYP1A1 with SULT1A1*1 or SULT1A1*3, but not SULT1A1*2, produced significantly more overall metabolism of AF than those expressing CYP1A1 alone. The SULT1A1 allele-specific metabolism was consistently observed for the disappearance of a detectable AF hydroxylated metabolite (⁴N,⁵N-OH-AF; Fig. 4B). Taken together, these data indicated that SULT1A1.2 exhibited a lower enzyme activity in the phase II metabolism of AF, whereas SULT1A1.3 showed an equivalent activity compared with the wild-type enzyme.

The dose-response curves and IC₅₀ values of AF in the V79-vector cells (control) and in the cells stably expressing CYP1A1 alone or coexpressing CYP1A1 and SULT1A1 are shown in Fig. 5 and Table 2. The V79-vector cell line was resistant to AF. Expression of different allelic variant of CYP1A1 all conferred cell sensitivity to AF without apparent allelic-specific difference in the IC₅₀ values. Coexpressions of SULT1A1 with CYP1A1 rendered cells more sensitive to AF, yet with SULT1A1 allele-specific sensitivity: The cells expressing SULT1A1*3 exhibited the highest sensitivity, followed by SULT1A1*1, whereas SULT1A1*2 exhibited the lowest sensitivity (Fig. 5; Table 2).

AF, a clinically investigational novel anticancer agent, requires metabolic activation by CYP1A1 and SULT1A1 to exert its antitumor activities (2, 4). In the present in vitro study, we showed the functional significance of common polymorphisms within human CYP1A1 and SULT1A1 genes on the metabolism and cytotoxicity of AF. Our results showed that all common variants of CYP1A1 and SULT1A1 were actively involved in the metabolic activation of AF, but with a varying degree of activity. Most remarkably, the cells expressing CYP1A1*2C (I462V) exhibited a superior catalytic activity (mainly caused by a significantly higher V_max) in the phase I metabolism (i.e., hydroxylations) of AF, whereas cells expressing SULT1A1*2 (R213H) possessed a diminished capacity to sulfate and activate hydroxylated AF metabolites.

With regard to the functional significance of the I462V and T461V variants of CYP1A1, our findings are in striking agreement with an earlier investigation using purified recombinant CYP1A1 enzymes, in which CYP1A1.2 harboring I462V showed a significant higher catalytic efficiency (V_max/K_m), mainly caused by a significant higher V_max, toward hydroxylations of 17β-estradiol and estrone, whereas CYP1A1.4 harboring T461N exhibited activity comparable with the wild-type enzyme (13). A superior activity of the I462V variant was also observed toward a CYP1A1 probe substrate ethoxyresorufin in the present study (Fig. 2) as well as in earlier studies (27). Different inducibility has been reported among the allelic variants of CYP1A1 (30, 31). For example, CYP1A1*2A and CYP1A1*2B harboring a 3801T>C transition, resulting in a new MspI restriction endonuclease site, are associated with a highly inducible phenotype of the enzyme (32). However, the observed superior activity of the cells expressing CYP1A1*2C (I462V) for hydroxylations of AF could not be explained by differential enzyme expression or inducibility because (a) real-time RT-PCR and Western blot did not detect any apparent differences in mRNA and protein expression levels of CYP1A1 among the selected stable clones expressing CYP1A1*1, CYP1A1*2C, and CYP1A1*4 (Fig. 1) and (b) treatment of the CYP1A1-expressing V79 cells with AF did not induce CYP1A1 mRNA expression (probably due to the lack of aryl hydrocarbon receptor in V79 cells; data not shown). A more rational underlying mechanism could be variant-specific interactions of the enzyme with P450 reductase involving differential electron transfer. This interpretation is supported by a homology model of human CYP1A1 based on the CYP2C5 crystallographic template, which locates the mutated residue Val⁴⁶² on the proximal face of the heme group (adjacent to the thiolate ligand; ref. 7), a region where the interaction with P450 reductase is generally assumed to take place (33). It is plausible that the I462V amino acid substation alters the electronic state of the heme (27), thus leading to enhanced interaction of the enzyme with P450 reductase.

It is interesting to note that whereas the I462V variant exhibited a superior activity in formations of hydroxylated AF metabolites, expression of different CYP1A1 variants (i.e., wild-type, T461N, or I462V) did not confer cell differential sensitivity to AF (Fig. 5; Table 2). On the other hand, the cells coexpressing wild-type CYP1A1 with SULT1A1*1, SULT1A1*2, or SULT1A1*3 displayed SULT1A1 allele-specific metabolic activation of AF: The cells expressing SULT1A1*3 exhibited the highest sensitivity, followed by SULT1A1*1, whereas SULT1A1*2 showed the lowest sensitivity to AF (Fig. 5; Table 2). Those observations are compatible with the results from a correlation analysis of AF activity (GI₅₀) with gene expression patterns in the NCI 60 cells, which revealed a highly significant positive correlation of AF activity with SULT1A1 expression but not with CYP1A1 expression (4). These data uniformly indicated the critical role of SULT1A1 activity for AF bioactivation and cellular response. We and others consistently showed that expression of CYP1A1 conferred cell sensitivity to AF, and coexpression of SULT1A1 with CYP1A1 further rendered cell sensitive to AF (Fig. 5; Table 2), confirming the notion that AF requires sequential metabolic activations by CYP1A1 and SULT1A1 to exert its antitumor activity. Although CYP1A1 is critical and upstream from SULT1A1 for AF bioactivation, CYP1A1-catalyzed hydroxylations are not sufficient to fully activate AF, whereas further sulfation of hydroxylated AF generates highly reactive nitrenium electrophiles capable of reacting to DNA and other cellular nucleophilic macromolecules. Therefore, SULT1A1 activity plays a critical role in AF bioactivation to elicit a cytotoxic response in tumor cells.

The present study clearly showed that SULT1A1*2 (R213H) was associated with low activity, whereas SULT1A1*3 (M223V) was associated with high activity, in the metabolic activation of AF. It has been repeatedly reported that human SULT1A1*2 allele, a SNP in exon 7 of the SULT1A1 gene resulting in the substitution of a histidine for an arginine at position 213 of the translated protein, produces an enzyme with impaired catalytic activity and decreased thermal stability (20, 21). Although relatively little has been reported about the functional significance of SULT1A1*3, possibly because of its low allele frequency in Caucasians and Asians, two earlier studies consistently showed that the SULT1A1*3 variant exhibits a higher affinity for the cofactor PAPS than the SULT1A1*1 and SULT1A1*2 variants (21, 34). The crystal structure modeling of human SULT1A1 suggests that the amino acids at position 213 and 223 may interact with neighboring residues and affect cofactor or substrate binding (35). It is plausible that the R213H and M223V amino acid substitutions may result in altered enzyme stability or enzyme binding affinity and activity toward different substrates due to structural alterations in the protein.

It is noteworthy that tamoxifen, a widely used drug in the world for the treatment and prevention of estrogen receptor–positive breast cancer, undergoes a similar sequential metabolic activation by CYP2D6 and SULT1A1. Specifically, tamoxifen is metabolized predominantly by CYP2D6 to form two active metabolites, 4-hydroxytamoxifen (4-OH-TAM) and 4-hydroxy-N-desmethyltamoxifen (endoxifen), which show much greater affinity for the estrogen receptor than tamoxifen (36, 37); furthermore, 4-OH-TAM and endoxifen are metabolized by SULT1A1 to form highly reactive products, leading to DNA adducts (38, 39). There is substantial evidence that functional polymorphisms (including SNPs and gene copy number variations) of CYP2D6 and SULT1A1 are associated with tamoxifen pharmacokinetics and clinical outcomes of efficacy and adverse effects (40–44). Of note, a recent study involving 1,325 breast cancer patients receiving tamoxifen suggests that the presence of two functional CYP2D6 alleles is associated with better clinical outcomes, and the presence of nonfunctional or reduced-function alleles with worse outcomes (44). In addition, the low-activity SULT1A1*2 allele has been associated with significantly poorer overall survival in breast cancer patients receiving tamoxifen (43).

Likewise, given the crucial roles of CYP1A1 and SULT1A1 in AF bioactivation, it is likely that functional polymorphisms in human CYP1A1 and SULT1A1 genes may influence clinical outcomes of patients treated with AF. The present study clearly indicated that the CYP1A1*2C (I462V) variant exhibited superior activity in the phase I metabolism (hydroxylations) of AF, whereas the SULT1A1*2 variant conferred the lowest cellular sensitivity to AF. It would be expected that the presence of low-activity SULT1A1*2 allele may predict poor response to AF, whereas the presence of high-activity CYP1A1/SULT1A1 alleles, especially the combination of CYP1A1*2C and SULT1A1*3 or SULT1A1*1, may be beneficial to patients receiving AF. The present study provided a foundation for future clinical investigations of potential genetic biomarkers that may enable selection of patients for the greatest benefit from AF treatment. Obviously, additional polymorphisms within regions of the CYP1A1 and SULT1A1 genes other than the open reading frame, such as within the promoter or introns as well as variation in the regulation of transcription or gene copy numbers, could also have functional or clinical significance on the metabolic activation of AF. The haplotypes of the polymorphisms in CYP1A1 and SULT1A1 could be highly significant in terms of the pharmacokinetics, efficacy, and safety of AF. All of those possibilities need to be explored in the future in vitro and clinical studies.

No potential conflicts of interest were disclosed.

We thank Dr. F.J. Gonzalez (National Cancer Institute, Bethesda, MD) for providing CYP1A1 cDNA, Dr. Zengquan Yang for valuable suggestions on real-time RT-PCR experiments, and Richard Wiegand for assistance in LC-MS/MS analyses.

Grant Support: Karmanos Cancer Institute Start-up Fund.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.