202
Chem. Res. Toxicol. 1992,5, 202-210
Oxidation of Aflatoxlns and Sterigmatocystin by Human Liver Microsomes: Significance of Aflatoxin Q, as a Detoxication Product of Aflatoxin 6, Kevin D. Raney,? Tsutomu Shimada,*J Dong-Hyun Kim,*!''John D. Groopman,l Thomas M. Harris,? and F. Peter Guengerich*v* Departments of Chemistry and Biochemistry and Center in Molecular Toxicology, Vanderbilt University School of Medicine, Nashville, Tennessee 37232, and Department of Environmental Health Sciences, Division of Environmental Chemistry, Johns Hopkins University School of Hygiene and Public Health, Baltimore, Maryland 21205 Received October 24, 1991
Aflatoxin Q18,g-oxide was synthesized and found to yield lower levels of NT-guanyl adducts than obtained from aflatoxin B18,g-oxide when mixed with calf thymus DNA or Salmonella typhimurium TA 98 cells. However, when S. typhimurium TA 98 was treated with the (analogous) epoxides of aflatoxin B1,aflatoxin G1,aflatoxin Q1, or sterigmatocystin, the ratios of revertants to NT-guanyl DNA adducts were similar. Aflatoxin Q1and aflatoxin B18,g-oxide (trapped here as the glutathione conjugate) are the major oxidative products formed from aflatoxin B1at all substrate concentrations in human liver microsomes, and cytochrome P-450 (P-450) 3A4 appears to be the dominant enzyme involved in both oxidations, as judged by studies involving correlation of activities in different liver samples, chemical inhibition, immunoinhibition, and reconstitution with purified hepatic and yeast recombinant P-450 3A4. Aflatoxin Q1is not appreciably oxidized in human liver microsomes and is not very genotoxic. The postulated formation of both aflatoxin Q1and aflatoxin 8,g-oxide from aflatoxin B1can be rationalized by a model in which P-450 3A4 binds the substrate in either of two different configurations. This is further demonstrated by the dichotomous effect of 7,8-benzoflavone-this flavone stimulates 8,9-epoxidation while inhibiting the 3a-hydroxylation reaction to form aflatoxin Q1.Thus, the 3a-hydroxylation of aflatoxin B1to aflatoxin Q1is viewed as a potentially significant detoxication pathway, Aflatoxins and closely-related dihydrofuran mycotoxins such as sterigmatocystin are of concern in many parts of the world because of their widespread occurrence in foodstuffs and their abilities to cause hepatic and other cancers in some animal models (1-3). Even with consideration given to hepatitis B virus, a substantial fraction of human liver cancer can be attributed to AFB,' exposure (1). AFB, is considered to be the most toxic and carcinogenic of these dihydrofurans (1)-the compound is not genotoxic itself, but all indications point to a critical role of epoxidation at the 8,9-vinyl bond to yield an electrophile which reacts with DNA (10). The only primary covalent nucleic acid adduct identified to date is that formed by reaction at the N7-guanyl position (11). A dramatic correlation between levels of this adduct and tumor incidence has been demonstrated among different experimental animal models (12) and may well be applicable to humans. The wide variation in tumorigenicity of AFB, and related compounds among animal models poses questions related to the metabolism of these compounds, and a number of issues must be addressed in vitro before further *Author to whom correspondence should be addressed at the Department of Biochemistry and Center in Molecular Toxicology, Vanderbilt University, Nashville, Tennessee 37232. +Departmentof Chemistry and Center in Molecular Toxicology, Vanderbilt University. Department of Biochemistry and Center in Molecular Toxicology, Vanderbilt University. %Presentaddress: Osaka Prefectural Institute of Public Health, Nakamichi, Higashinari-ku, Osaka 537, Japan. 11 Present address: Korea Institute of Science and Technology, P.O. Box 131, Chongyangni, Seoul, South Korea. Johns Hopkina University School of Hygiene and Public Health. f
0893-228x/92/2705-0202%03.00/o
progress can be made. Among these are comparison of the relative abilities of the different dihydrofuran epoxides to yield N-guanyl DNA adducts and genotoxic damage, the nature of the oxidation and conjugation products of the AFB, products formed in human liver (particularly at lower substrate concentrations relevant to in vivo exposure situations), which of the P-450s and GSH $-transferases are dominant in the formation of the various products, and whether certain dihydrofurans (that are either found along with AFB, or formed from it) are oxidized to genotoxic products. In this report we describe the characterization of the oxidation of AFB, to ita 8,9-oxide and AFQ, by human liver P-450 3A4 and the lack of genotoxicity of AFQ,, in the context of ita relationship to other structures (AFB,, AFG1, and sterigmatocystin).
Experimental Sectlon Enzyme Preparations and Antibodies. Human liver samplea were obtained from organ donors through Tennessee Donor Services (Nashville, TN) and stored at -80 O C ; microsomal fractions were prepared as described elsewhere (13,14). Several of the liver samples used here have been utilized in other studies (15-18). P-450 3A4 (P-45OW) was purified to homogeneity from human liver sample HL 115 by modifications of procedures described elsewhere (6). Recombinant P-450 3A4 was produced in Saccharomyces cereuisiae and partially purified as described elsewhere (9). Antibodies were raised in rabbits against human Abbreviations: AF, aflatoxin (used for the individual compounds AFB1, AFG1, AFM1, AFQ1, and AFB& P-450, liver microsomal cytochrome P-450; GSH, (reduced) glutathione; HPLC, high-performance liquid chromatography. For reference to the nomenclature of P-45Os see ref 4; see ref 5 for discussion of human P-450s. P-450 3A4 appears to be the major P-450 3A family enzyme expreseed in human liver and has been extensively characterized as reported elsewhere (6-9). Q 1992 American Chemical Societv
Chem. Res. Toxicol., Vol. 5, No. 2, 1992 203
Aflatoxin QIFormation and Reactions Chart I. Structures of Aflatoxins and Sterigmatocystin
9
AFLATOXIN B1
II
0
I1
\
9 0
la
0
AFLATOXIN Q1
0
@
9
0
,/'
AFLATOXIN G1
@ ; OH I
\
3s
OCHl
'O
6 5'
OCH,
STERIGMATOCYSTIN
P-450 3A4 (6),P-450 1A2 (19,201, and P-45OW (a 2C enzyme) (21). Liver cytosol was prepared from female B6C3F1 mice (20 g, Charles River Laboratories, Wilmington, MA) and used as a source of GSH S-transferase in experiments where the AFB, 8,g-oxide-GSH conjugate was prepared or trapped. Spectroscopy. lH NMR spectra were obtained on a Bruker (Billerica, MA) instrument, either AM-400 or -300. 4,4-Dimethyl-4-silapentane-1-sulfonic acid and tetramethylsilane were used as internal standards. W spectra were recorded on a Varian Cary 2030 spectrophotometer (Varian, Walnut Creek, CA). Fast atom bombardment mass spectra (+ and - modes) were obtained with a VG 70-250 instrument (VG, Manchester, U.K.) using a probe and standard Ion-Tech saddle field gun producing xenon atoms of &kV energy. Glycerol, thioglycerol, and 3-nitrobenzyl alcohol were used as matrices. Chemicals. AFB,,AFG,, and sterigmatocystin were purchased from Aldrich Chemical Co. (Milwaukee, WI). AFQ, was either isolated by HPLC (vide infra) from preparative incubations of AFB, and NADPH-fortified human liver microsomes or synthesized by oxidation of AFB, as described by BDchi et al. (22): both samples showed the appropriate UV, NMR, and mass spectral properties. (See Chart I for structures.) Synthesis of Epoxides. Epoxides of AFB1, AFG1, AFQ,, and sterigmatocystin were prepared according to the procedure of Baertschi et al. (11). Preparations were usually on a 1-to 2-mg scale. Aflatoxins and sterigmatocystin were dried from benzene under a stream of N2 before oxidation and dissolved in distilled CHzClzto a concentration of -10 mg mL-'; 1.5 equiv of freshly prepared dimethyldioxirane t0.05 M dried over anhydrous MgSOl (23)] was then added to the mycotoxin. After 20 min at room temperature, the solvent was removed under a stream of Nz and the resulting epoxide (obtained in nearly quantitative yield) was dissolved in deuterated solvent for 'H NMR analysis. (Solutions of epoxides in acetone can be stored at -20 OC for more than 6 months if kept dry. For extended storage periods, epoxides were stored neat under argon at -20 OC). The 'H NMR spectrum of AFBl 8,g-oxide has already been reported elsewhere (11). The 'H NMR spectral assignments of the other three epoxides used in this report were made by comparison with dihydrofuran precursors and AFB, 8,g-oxide. AFG, 9,10-oxide: 'H NMR ( a ~ e t o n e - ~ H 6 3.51 ~ ) (d oft, 2 H, J = 2.3,6.0 Hz,H-4), 4.04 (s,3 H, 5-OCHJ, 4.04 (H-10 overlapped by S-OCHs), 4.43 (t, 2 H, J = 6.0 Hz, H-3), 4.55 (d, 1H, J = 5.8 Hz, H-loa), 5.49 (d, 1 H, J = 1.5 Hz, H-9),6.20 (d, 1H, J = 5.8 Hz, H-7a), 6.64 (s, 1 H, H-6). AFQ, 8,9-oxide: 'H NMR ( a c e t ~ n e - ~ H 6 2.38 ~ ) (d of d, 1 H, J = 18.5, 1.5 Hz, H-2), 2.89 (d of d, 1H, J = 18.5, 6.8 Hz, H-29, 4.03 (d, 1 H, J = 1.7 Hz, H-9), 4.14 (9, 3 H, 4-OCH3), 4.46 (d, 1 H, J = 4.3 Hz, 3-OH), 4.57 (d, 1 H, J = 5.8 Hz, H-ga), 5.50 (d, 1 H, J = 1.7 Hz, H-8), 5.69 (m, 1 H, H-3), 6.22 (d, 1H, J = 5.8 Hz, H-6a), 6.72 (8, 1 H, H-5). Sterigmatocystin 1,Z-oxide: 'H NMR ( a ~ e t o n e - ~ H 6 3.97 ~) (8, 3 H, 6-OCHJ, 4.05 (d, 1 H, J = 1.7 Hz, H-1), 4.55 (d, 1 H, J = 5.8 Hz, H-12~),5.36 (d, 1 H, J = 1.7 Hz, H-2), 6.15 (d, 1 H, J = 5.8 Hz, H-3a), 6.45 (8, 1 H, H-5), 6.75 (d of d, 1 H, J = 1.0, 8.3
Hz, H - l l ) , 6.91 (d of d, 1 H, J = 1.0, 8.4 Hz, H-9), 7.54 (t, 1 H, J = 8.3 Hz, H-lo), 13.18 (9, 1 H, 8-OH). Preparation a n d Analysis of N7-Guanyl Adducts. The preparation of AFB,, AFG,, and sterigmatocystin W-guanyl adduct standards was described previously (24). &(W-Guanyl)-9hydroxy-8,9-dihydro-AFQ,was prepared in a similar manner by reaction of AFQ, 8,g-oxide with calf thymus DNA and hydrolysis of the resulting cationic adduct; purification was by reverse-phase HPLC (11): 'H NMR [(C2H3)zSO]6 2.25 (d, 1 H, J = 18.4 Hz, H-2), 2.85 (d of d, 1 H, J = 18.4, 6.6 Hz, H-2'), 3.89 (8, 3 H, 4-OCHJ, 4.16 (d, 1H, J = 6.0 Hz, H-ga), 5.20 (8, 1 H, H-9), 5.48 (m, 1 H, H-3), 6.09 (bs,2 H, guanine NHJ, 6.23 (s, 1H, H-8), 6.62 (e, 1H, H-5), 6.85 (d, 1H, J = 5.9 Hz, H-6a), 7.3 (8, 1H, guanine H-8), 9.86 (s, 1 H, guanine H-1); fast atom bombardment mass spectrum, m/z 496 (MH+, 25% relative abundance). Analysis of this AFQ, N7-guanyl adduct at varying DNA concentrations was performed using an Alltech octadecylsilane HPLC column (4.6 x 250 mm, 5 pm, Alltech, Deerfield, IL)with isocratic elution conditions consisting of 90% HzO and 10% CH&N (v/v); the t R for the AFQ, N7-guanyl adduct was 4.5 min. W-Guanyl adducts were quantified as previously described (24, 25). Chemical Synthesis of Glutathione Conjugates. GSH (200 mg, 0.65 mmol) was stirred in 2 mL of CH30H, and 50 mg of sodium (2.2 mmol) was added. The GSH quickly dissolved, and 2.5 mg of AFB, 8,g-oxide (7.6 pmol) in 0.75 mL of CH2Clzwas added to the solution. The solution immediately became yellow and after 1min was neutralized with -1.5 mL of 1M CH3COZH. The solvent was removed in vacuo, and the residue was dissolved in 4 mL of H 2 0 and filtered through a 0.2-pm nylon filter before purification by HPLC. The entire sample was loaded onto an Alltech semipreparative octadecylsilane column (10 X 250 111111, 5 pm) via the pump head of the system. A linear gradient of 100% H20 to 50% of a mixture of CH3CN/CH30H/H20,1:1:2 (v/v/v), in 25 min was then run at a flow rate of 2.5 mL min-' to elute the AFB, 8,g-oxide-GSH conjugate which had concentrated on the column (tR18.8 min). The area of the AFB, 8,9-oxide-GSH conjugate peak (A,) accounted for 70% of all absorbing eluents. The collected fraction was dried using a Speed-Vac concentrator (Savant, Farmingdale, NY),and the residue was lyophilized twice from 2Hz0before obtaining 'H NMR (in 2H20). A similar procedure was used for synthesizing the GSH conjugates of the epoxides of AFG, and sterigmatocystin. The gradient used to purify these conjugates was applied over 50 min (same solvents). The retention time for the AFG, 9,lO-oxideGSH conjugate was 18 min, and the yield was 20%. The sterigmatocystin 1,a-oxide-GSH conjugate (monitored at 254 nm) eluted at 24.4 min and was obtained in a 50% yield. Chemical synthesis of the AFQ, 8,9-oxide-GSH conjugate was not successful using this approach. The compounds prepared using this approach were indistinguishable (by HPLC and NMR and UV spectra) from those prepared by enzymatic synthesis (vide infra). Enzymatic Synthesis of GSH Conjugates. Mouse liver cytosol (10 mg of protein mL-') was prepared in 50 mM sodium phosphate buffer (pH 7.4) containing 5 mM GSH. AF&,8,g-oxide (75 pg, 0.30 pmol in 500 pL of acetone) was added in 5 aliquota over 5 min to 10 mL of mouse liver cytosol, with stirring. Two milliliters of 0.4 M HCOzH was then added to precipitate protein. The mixture was centrifuged (3000g, 10 min), and the resulting supematant was filtered (0.4-pm nylon filter) before loading the entire sample onto an Alltech semipreparative octadecylsilyl column (10 X 250 mm, 5 pm) via the pump head of the system. A linear gradient of 0--40% CH3CN in HzO was applied over 40 min. The AFQ, 8,9-oxideGSH conjugate eluted at 14 min. The CH&N in the collected material was fractionally distilled under reduced pressure using a rotary evaporator, and the remaining solution was taken to dryness by lyophilization. The resulting yellow residue was dissolved in 2 H 2 0'H NMR assignments were made by comparing the resulting spectrum with that previously reported for the AFBl 8,9-oxide-GSH conjugate (24). Attempts to obtain mass spectral data by fast atom bombardment techniques and by HPLC-mess spectrometry in the thermospray mode were unsuccessful. 8 - ( S - G l u t a t h i o n y l ) - 9 - h y d r o x y - 9 , 1 ~ ~ ~'H d ~NMR ~Bl: (2HzO)6 2.08 (m, 2 H, Glu j3), 2.47 (t, 2 H, J = 8.0 Hz, Glu -y), 2.59 (m, 1 H, partially exchanged, H-2), 2.82 (d of d, 1 H, J = 14.2, 8.6 Hz, CYSOb), 3.16 (d of d, 1 H, J = 14.3, 5.8 Hz, CYSpa),
Raney et al.
204 Chem. Res. Toxicol., Vol. 5, No. 2, 1992 3.30 (m, AFQ, 8,9-oxide (Table II). The numbers of mutants and N7-guanyl levels were well correlated with each other in these and repeated experiments with both S. typhimurium TA 98 and TA 100. Oxidation of AFB, to AFQ, and AFB, 8,s-OxideGSH Conjugate. Previous work showed that AFQ, was a major oxidative product of AFB, in human liver microsomes (26,27). Since enzymatically derived AFB, 8,9-oxide can be efficiently trapped as a GSH conjugate in the presence of purified GSH 5'-transferase or mouse liver cytosol, this end point was used as a marker of the capability of microsomes to form the 8,g-oxide. HPLC analysis showed that AFQ, and AFBl 8,g-oxide (measured as the GSH conjugate) were the major oxidation products detected when many different human liver microsomal preparations were incubated with NADPH at high or low concentrations of AFB, (Figure l),consistent with the observations of Moss and Neal (28) and Ramsdell and Eaton (27). Further studies were done at the relatively low concentration of 2.5 pM AFB,,at which the sensitivity for detection was still adequate. When different human liver microsomal preparations were compared, rates of formation of the AFB, 8,9oxide-GSH conjugate were highly correlated with rates of nifedipine oxidation, a marker of P-450 3A4 (15, 29) (Figure 2). Rates of AFQ, formation were also highly correlated with those of nifedipine oxidation (Figure 2B), a result reported anecdotally (29), and of AFB, 8,9oxide-GSH conjugate formation (Figure 2C). Anti-P-450 3A4 was highly effective at inhibiting the formation of both AFQ, and AFB, 8,9-oxide-GSH conjugate from AFBl in human liver microsomes (Figure 3). A preimmune serum or antibodies raised against human P-45om (a P-450 2C protein) or P-450 1A2 were ineffective inhibitors of the reactions. Preincubation of human liver microsomes with either of the selective P-450 3A inhibitors
Aflatoxin
Chem. Res. Toxicol., Vol. 5, No. 2, 1992 205
Q1Formation and Reactions
Figure 1. Formation of major products in human liver microsomes as a function of concentration of the substrate AFB1. Human liver microsomes (3.7 mg of protein mL-'; A, sample HL 105; B, sample HL 107; C, sample HL 110) were incubated with varying concentrations of AFB,, an NADPH-generating system, 0.1 M sodium phosphate buffer (pH 7.4), 5 mM GSH, and 3 mg of mouse liver cytosolic protein mL-' (as a source of GSH S-transferase). Incubations proceeded for 15 min a t 37 "C, and AFQ, (0)and the AFB, 8,9-oxide-GSH conjugate ( 0 )were estimated by HPLC as described under Experimental Section.
40-
-
0
0
1
2
3
4
5
8
Nifedlpine oxldatlon, nmol/mln/mg protein
0
2
4
8
8
1
0
1
2
AFBl 8,9-oxlde:GSH conjugate, pmoi/min/mg protein
Nlfedipine oxldatbn, nmol/mlnlmg protein
Figure 2. Correlations involving aflatoxin oxidation products in various human liver samples. Rates of nifedipine oxidation (6) and conversion of AFBloxidation to AFQ, and the AFB, 8,9-oxiddSH conjugate were estimated as described under Experimental Section and Figure 1. The AFB, concentration in these incubations was 4 pM; samples included HL 102, HL 105, HL 106, HL 107, HL 110, and HL 125 (part C only). Correlation coefficients ( r ) in parts A, B, and C were 0.99, 0.98, and 0.94, respectively. Table 111. Inhibition of Human Liver Microsomal 3-Hydroxylation of AFB, rate of AFQl concentration formation (% of inhibitoP (LLM) unhibited rate) troleandomycin 20 15 40 12 gestodene 7,s-benzoflavone 20 58 2 104 quinidine diethyldithiocarbamate 100 88 sulfaphenazole 100 84
Human liver microsomal preparation HL 110 was used in this study, and the AFBl concentration was 4 pM. The uninhibited rate of AFQ,formation was 65 pmol min-' (mg of microsomal protein)-'. Results are presented as means of duplicate experiments (which differed by
