Naturally Occurring Coumarins Inhibit Human Cytochromes P450 and

Chem. Res. Toxicol. 2003, 16, 415-422

415

Naturally Occurring Coumarins Inhibit Human Cytochromes P450 and Block Benzo[a]pyrene and 7,12-Dimethylbenz[a]anthracene DNA Adduct Formation in MCF-7 Cells Heather E. Kleiner, Melissa J. Reed, and John DiGiovanni* Department of Carcinogenesis, Science Park-Research Division, The University of Texas M.D. Anderson Cancer Center, P. O. Box 389, Smithville, Texas 78957 Received October 2, 2002

Naturally occurring coumarins (NOCs) inhibit polycyclic aromatic hydrocarbon-induced skin tumor initiation in mice by blocking cytochrome P450 (P450)-mediated bioactivation of benzo[a]pyrene (B[a]P) and 7,12-dimethylbenz[a]anthracene (DMBA). Bergamottin selectively inhibits tumor initiation by B[a]P, whereas imperatorin and isopimpinellin inhibit tumor initiation with both carcinogens. The goals of the current study were to examine the ability of NOCs to inhibit human P450s in vitro and to establish whether NOCs, which are anticarcinogenic in mice, can block carcinogen bioactivation in cultured human cells. For the initial experiments, incubations containing 5 µM P450, P450 substrate, an NADPH generating system, and NOCs were used to determine the concentrations of each inhibitor that blocked 50% of P450 activity (IC50). These results confirmed that NOCs are capable of inhibiting multiple human P450s and that they exhibit selectivity for certain isoforms of human P450s. In subsequent experiments, we examined the effects of bergamottin, imperatorin, and isopimpinellin on DMBA and B[a]P DNA adduct formation in the human breast MCF-7 adenocarcinoma cell line. Coincubation of cells with the three different NOCs significantly inhibited DMBA DNA adduct formation by 29-82% at doses ranging from 2 to 10 µM and significantly inhibited B[a]P DNA adduct formation by 37-80% at doses ranging from 20 to 80 µM. HPLC analysis of the DNA hydrolysates demonstrated that inhibition of DNA adducts corresponded to inhibition of the major B[a]P and DMBA diol-epoxide-derived adducts. Although bergamottin was not effective at blocking DMBA bioactivation in the mouse skin model, it was similar in effectiveness to imperatorin and isopimpinellin in MCF-7 cells. These results demonstrate that NOCs, which are present in citrus fruits and other components of the human diet, are capable of inhibiting carcinogen metabolizing enzymes and blocking bioactivation of both B[a]P and DMBA in MCF-7 cells.

Introduction 1

We have previously demonstrated that NOCs, to which humans are routinely exposed in the diet, block P450-mediated 7-EROD and PROD activities in mouse liver microsomal incubations (1). Additional analyses have shown that some of these compounds (e.g., coriandrin) are mechanism-based inactivators of P450s (2). P450 1A1 and other members of the P450 superfamily are known to bioactivate procarcinogens such as PAH, and their inhibition is one strategy for the chemoprevention of cancer. To this end, we (3, 4) and others (5-7) have shown that NOCs possess anticarcinogenic properties in mice. For example, certain linear furanocoumarins (e.g., bergamottin, imperatorin, isopimpinellin) and a simple coumarin (osthruthin) were found to inhibit * To whom correspondence should be addressed. Tel: (512)237-9414. Fax: (512)237-3439. E-mail: [email protected]. 1 Abbreviations: AHR, aryl hydrocarbon receptor; B[a]P, benzo[a]pyrene; 7,8-BF, 7,8-benzoflavone; BPDE, benzo[a]pyrene-diol-epoxide; dAdo, deoxyadenosine; P450, cytochrome P450; dGuo, deoxyguanosine; DMBA, 7,12-dimethylbenz[a]anthracene; DMBADE, 7,12-dimethylbenz[a]anthracene-diol-epoxide; 1-EP, 1-ethynylpyrene; ERR, estrogen receptor R; EROD, ethoxyresorufin O-deethylase; NOCs, naturally occurring coumarins; PAH, polycyclic aromatic hydrocarbon; PROD, pentoxyresorufin O-dealkylase; TCDD, 2,3,7,8-tetrachlorodibenzo-pdioxin; TPA, 12-O-tetradecanoylphorbol-13-acetate.

epidermal DNA adduct formation and skin tumor initiation by B[a]P and/or DMBA. In addition, imperatorin blocked complete carcinogenesis by DMBA (3). In the mouse skin tumorigenesis system, significant differences were observed in the inhibitory effects of bergamottin and 7,8-BF as compared to imperatorin and isopimpinellin. In this regard, bergamottin was an effective inhibitor of B[a]P, but not DMBA, skin tumor initiation in SENCAR mice (3). In contrast, 7,8-BF was an effective inhibitor of DMBA skin tumor initiation but not B[a]P (8-10). We recently reported that a likely explanation for these differences was related to their differential effects on murine P450 1A1 and 1B1 (11). Whereas imperatorin and isopimpinellin appeared to inhibit both of these murine P450s, bergamottin appeared to be very selective for inhibiting P450 1A1, with little effect on P450 1B1. In contrast, 7,8-BF appeared to be very selective for inhibiting P450 1B1 as compared to P450 1A1. Species differences in P450 activities as well as the effects of P450 inhibitors are well-known (12-16). For example, P450-mediated coumarin 7-hydroxylase, tolbutamide hydroxylase, and testosterone 6β-hydroxylase activities were significantly different in human liver microsomal incubations as compared to horse, dog, and

10.1021/tx025636d CCC: $25.00 © 2003 American Chemical Society Published on Web 02/25/2003

416

Chem. Res. Toxicol., Vol. 16, No. 3, 2003

cat (17). Furthermore, furafylline did not inhibit phenacetin O-deethylase activity in cat and troleandomycin did not inhibit testosterone 6β-hydroxylase activity in horse (17). Also, species differences in B[a]P metabolism were observed in genetically engineered V79 Chinese hamster cells that stably express rat or human P450 1A1 (16). Human P450 1A1 metabolized B[a]P in the 7,8,9,10-position, forming B[a]P-7,8-diol, 9,10-epoxide (BPDE), the ultimate carcinogen (16). In contrast, rat P450 1A1 metabolized B[a]P in the 4,5-position as well as the 7,8,9,10-position. As a consequence, the human P450 1A1 V79 cells were more susceptible to B[a]P cytotoxicity than the rat (16). Also, the regioselectivity of DMBA metabolism in vitro was different in reconstituted purified recombinant human P450 1B1 as compared to mouse P450 1B1 (18). In this regard, phenols, DMBA-3,4-diol, and 10,11diol were the major metabolites formed by mouse P450 1B1, which represented 40, 22, and 21% of the total metabolites, respectively. In contrast, phenols, DMBA 5,6-diol, and 8,9-diol were the major metabolites formed by human P450 1B1 (18). Studies with human microsomal P450s and mutagenesis assays have implicated P450s 1A1 and 1B1 in the first oxidation step of PAHs, but the second bioactivation step leading to the diol epoxides may be catalyzed by P450s 1A1, 1B1, 3A4, 1A2, 2B6, and 2C9 (reviewed in ref 19). Because coumarins are present in the human diet, it is important to compare the effects of NOCs on human P450 activity. In the current study, we have examined the abilities of NOCs (bergamottin, imperatorin, isopimpinellin) and several established P450 inhibitors (7,8-BF and 1-EP) to block activities of cDNA-expressed human P450s in vitro. We have also evaluated the abilities of these compounds to block B[a]P and DMBA DNA adduct formation in the human breast MCF-7 adenocarcinoma cell line. The results demonstrate that NOCs can effectively inhibit several human P450s and block PAH DNA adduct formation in a human tumor cell line. In addition, the data demonstrate that some differences exist between mouse and human P450s in terms of the selectivity of inhibition by these compounds.

Experimental Procedures Caution: B[a]P and DMBA are carcinogenic and mutagenic and should be handled with extreme caution using the guidelines for carcinogenic materials developed by the National Cancer Institute. In addition, several of the linear furanocoumarins are phototoxic and should be handled with extreme care. Materials. B[a]P, DMBA, and 7,8-BF (R-naphthoflavone) were purchased from Aldrich Chemical Co. [3H]B[a]P and [3H]DMBA (specific activity (SA) 59-72 Ci/mmol) were obtained from Amersham Co. (Arlington Heights, IL) and diluted with unlabeled B[a]P or DMBA to a SA of 5 Ci/mmol. Typical batch analysis by tritium NMR spectroscopy shows the distribution of the label as follows (although batch variation may occur): for [G-3H]B[a]P, position 1, 8.4%; 2, 6%; 3, 8.3%; 4, 11%; 5, 9%; 6, 10.1%; 7, 8.1%; 8, 9.1%; 9, 10.3%; 10, 3.7%; 11, 3.7%; 12, 12.3%. For [G-3H]DMBA, distribution of label was as follows: 7-methyl position, 6%; 12-methyl position, 1.5%; 9,10-position, 27.1%; 2,3,5-positions, 43%; position 4, 17.6%; position 8, 4.9%; and, 1,6,11-positions, nil. 1-EP was a generous gift from William L. Alworth (Tulane University). Bergamottin, imperatorin, and isopimpinellin were purchased from Indofine Chemical Co. (Somerville, NJ). DNase I (bovine pancreas, EC 3.1.4.1), 7,8BF, snake venom phosphodiesterase (Crotalus atrox, EC 3.1.4.1), and alkaline phosphatase (Escherichia coli, type III, EC 3.1.3.1) were supplied by Sigma Chemical Co. (St. Louis, MO). Sephadex

Kleiner et al. LH-20 was obtained from Pharmacia, Inc. (Piscataway, NJ). Sep-pak (C-18) cartridges were supplied by Waters Corp. (Milford, MA). [3H]B[a]P and [3H]DMBA DNA adduct markers were constructed as described previously (11, 20). PAH and furanocoumarins are light sensitive. All studies were performed under subdued lighting. P450 Inhibition Studies. Human P450s + P450 reductase (2C9, 2B6, 2A6 with cytochrome b5) supersomes were purchased from Gentest Corp. (Woburn, MA). For all assays, an NADPH generating system was preincubated in 0.05 M Tris buffer (pH 7.5) with substrate ( inhibitors at 37 °C for 5 min in a total volume of 1 mL. The reaction was initiated with 5 pmol of human P450s. The conditions for EROD assays (P450 1A1, 1A2, and 1B1) were as follows: 5 nmol of ethoxyresorufin was used as substrate; P450 1A1, 1A2, and 1B1 were incubated for 10, 20, and 20 min, respectively; reactions were terminated using 2.5 mL of ice-cold methanol, protein centrifuged out, and supernatants analyzed by spectrofluorometer at an excitation wavelength of 550 and an emission of 580 nm, as previously described (21). For the coumarin 7-hydroxylase assay (P450 2A6), 5 nmol of coumarin was used as substrate; samples were incubated for 30 min and terminated using 0.2 mL of 20% trichloroacetic acid, from which 0.1 mL was added to 1.9 mL of 0.1 M Tris buffer (pH 9.0), and the product (7-hydroxycoumarin) was analyzed by a spectrofluorometer at an excitation of 368 nm and an emission of 456 nm. For the 7-ethoxy-4-trifluoromethyl coumarin deethylase assay (P450 2B6), 50 nmol of substrate was used, samples were incubated for 15 min and terminated using 0.2 mL of 20% trichloroacetic acid, from which 0.1 mL was added to 1.9 mL of 0.1 M Tris buffer (pH 9.0), and the product (7-hydroxy-4trifluoromethylcoumarin) was analyzed by a spectrofluorometer at an excitation of 410 nm and an emission of 510 nm. For the diclofenac 4-hydroxylase assay (P450 2C9), 50 nmol of diclofenac was used, samples were incubated for 30 min and terminated using 94% acetonitrile/4% acetic acid. Samples were analyzed by HPLC as described in the Gentest product bulletin. For all assays, boiled control values were subtracted from experimental values. Experimental values were extrapolated against authentic product standard curves using linear regression analysis. IC50 values were determined by interpolation as the concentration of inhibitor that inhibits 50% of the enzyme activity in the control (solvent onlysno inhibitor). With the exception of the 2C9 study, which used absolute ethanol, substrates and products were dissolved in DMSO (1.1%). Culture and Treatment of Cells. MCF-7 cells were purchased from ATCC (Manassas, VA, item No. HTB-22). Cells were cultured in Dulbecco’s Modified Eagle Medium (DMEM), containing l-glutamine, glucose, and pyridoxine (Gibco, Grand Island, NY, item no. 11965-084), supplemented with 100 units of penicillin G/mL, 69 µM streptomycin sulfate, 0.27 µM amphotericin B, 5% fetal bovine serum (Atlanta Biologicals, Norcross, GA), 0.1 mM nonessential amino acids (Gibco), and 1.5 µM insulin in a humidified atmosphere at 5% CO2 at 37 °C. Cells were passaged using 5% trypsin/0.03% EDTA in PBS at ∼80% confluence. For experiments, MCF-7 cells (passages 6-14) were seeded in polystyrene tissue culture plates at a density of 1 × 106 cells/mL. At ∼70-80% confluence, cells were treated with [3H]DMBA or [3H]B[a]P (5 Ci/mmol) at a final media concentration of 2 µM. Cells were cotreated with acetone or the test compounds dissolved in acetone. Total acetone concentration in media was 0.75% (v/v). For DNA adduct analysis, at 24 h after treatment, media was removed, cells were lysed with 0.75 M guanidine isothiocyanate, and the DNA was subsequently isolated as described (22). DMBA DNA adducts were analyzed as described below. Analysis of B[a]P and DMBA DNA Adducts. DNA was isolated by phenol extraction and RNase A digestion as described previously (23). Purity of DNA was determined spectrophotometrically, and DNA content was quantitated spectro-

Inhibition of Human P450s

Chem. Res. Toxicol., Vol. 16, No. 3, 2003 417

Table 1. IC50 Values (µM) for Selected NOCs, 1-EP, and 7,8-BF for Inhibition of CDNA-Expressed Human P450sa P450b

7,8-BF

1-EP

bergamottin

imperatorin

isopimpinellin

1A1 1B1 1A2 2A6 2B6 2C9 3A4

0.186 ( 0.012 ( 0.002 0.042 ( 0.017 >500 >500 185 (211, 159) Ndc

1.1 (0.71, 1.4) 0.08 (0.03, 0.13) 0.11 (0.10, 0.13) >500 8.26 (5.3, 11.2) >500 Nd

0.37 ( 0.04 1.52 ( 0.35 0.88 ( 0.19 3.45 (3.35, 3.55) 0.88 (1.00, 0.75) 3.08 (3.35, 2.82) 0.57 (0.28, 0.85)

2.76 ( 0.64 0.71 ( 0.17 0.38 ( 0.12 11.7 (10, 13.3) 0.32 (0.4, 0.25) 111 (133, 89.1) 0.53 (0.62, 0.45)

3.67 ( 0.49 3.72 ( 0.69 0.86 ( 0.18 27.6 (25.4, 29.9) 34.1 (11.9, 56.2) >500 0.20 (0.08, 0.32)

0.037d

a IC b 50 values (µM) were determined by interpolation. Control activities (pmol/min/pmol P450) for 1A1, 1B1, 1A2, 2A6, 2B6, 2C9, and 3A4 were 15.9 ( 1.3, 2.77 ( 0.47, 0.99 ( 0.28, 13.8 ( 0.7, 4.6 ( 0.2, 1.0 ( 0.16, and 63.7 ( 8.6, respectively (means ( SE). c Nd, not done. d Values represent means ( SE (n ) 3) or means (range in parentheses, n ) 2).

photometrically using absorbance at 260 nm. To quantitate total DNA adducts, DNA was hydrolyzed with DNase I and aliquots were analyzed by liquid scintillation spectroscopy. The SA of DNA binding was represented as pmol DMBA bound per mg of DNA. To analyze individual DNA adducts, the DNase I DNA hydrolysates were sequentially hydrolyzed with snake venom phosphodiesterase and alkaline phosphatase as described previously (24) and purified through Sep-pak (C-18) cartridges (22). HPLC analysis of B[a]P and DMBA DNA adducts was conducted by a method previously described (3, 22), using radiomatic detection (Flo-one beta, Packard Instrument Co., Meriden, CT), as described previously (11). By comparing retention times of individual peaks with the synthetic standards (described in ref 11) and with previous identification of adducts (25), the three major DMBA DNA adduct peaks were identified as anti-DMBA-diol epoxide-dGuo (anti-DMBADE-dGuo) (peak I), syn-DMBADE-dAdo (peak II), and anti-DMBADE-dAdo (peak III). Minor peaks included peak a, syn-DMBADE-dGuo; peak b, anti-DMBADE-dGuo; peak f, syn-DMBADE-dAdo; and peak g, syn-DMBADE-dAdo. The major adduct resulting from B[a]P treatment was anti-B[a]P diol-epoxide-dGuo (anti-BPDE-dGuo). Analysis of B[a]P and DMBA Metabolites. Metabolites were identified as previously described (11). Briefly, media were extracted twice with 2 vol of ethyl acetate:acetone (1:1, v/v) and the organic layers were dried under nitrogen. The residue was resuspended in methanol and injected onto HPLC-UV with radiomatic detection. Metabolites and parent compounds were identified by comparison of retention times with authentic standards. Analysis of P450 Activity and Protein Levels in MCF-7 Cells. Cells were treated with acetone, B[a]P, or DMBA (2 µM) as described above. For EROD assays, cells were also cotreated with acetone (0.5%, v/v), bergamottin (20× equimolar), imperatorin (20×), isopimpinellin (40×), or 7,8-BF (1×). At 24 h after treatment, media were discarded, ice-cold isolation buffer (0.05 M Tris buffer (pH 7.5), containing 0.5 M sucrose, 5 mg/mL aprotinin, 5 mg/mL leupeptin, 1 mM EDTA, and 0.1 mM phenylmethylsulfonyl fluoride) was added to the dishes, and cells were removed using a cell scraper. Cells were homogenized using a Polytron for 20 s. EROD assays were performed using 0.4 mg/mL protein for 60 min following a method previously described (21). Western blot analyses were performed using 50 µg of cell lysate separated electrophoretically (26) on a 10% acrylamide gel with 3% acrylamide stacking gel and transferred (27) to PVDF, probed with 1:1000 dilutions of antirat P450 1A1 (with cross-reactivity to P450 1A2), antihuman P450 1B1 (Gentest, Woburn MA), or anti-β-actin antibodies (Santa Cruz Biotechnology, Santa Cruz, CA) by an enhanced chemiluminescence method previously described (28). Statistical Analysis. Data were analyzed by ANOVA followed by Fisher protected least significant difference (PLSD) test. Fisher PLSD tests were performed on a Macintosh computer with Statview 5.0 software (Altura Software, Inc.).

Results Inhibitory Activities of Selected NOCs Against Specific Human Cytochrome(s) P450 Involved in the Metabolic Activation of B[a]P and DMBA. We

have performed studies using human cDNA-expressed P450 microsomal systems (Gentest, Woburn, MA) to determine the concentration of each inhibitor that blocks 50% of the enzyme activity in vitro (IC50 values). We used the synthetic P450 inhibitors 7,8-BF and 1-EP (29) for comparison. The following assays were performed to evaluate the effects of NOCs on P450 activities (in parentheses): EROD (1A1, 1B1, and 1A2), coumarin 7-hydroxylase (2A6), 7-ethoxy-4-trifluoromethylcoumarin dealkylase (2B6), diclofenac 4-hydroxylase (2C9), and testosterone 6β-hydroxylase (3A4). The IC50 values are summarized in Table 1. All three linear furanocoumarins were effective inhibitors of P450s 1A1, 1B1, 1A2, and 3A4, with IC50 values ranging from 0.3 to 3 µM. Bergamottin and imperatorin also inhibited P450 2B6 at IC50 values