Chem. Res. Toxicol. 1992,5, 779-786
779
Fluoranthene Metabolism: Human and Rat Liver Microsomes Display Different Stereoselective Formation of the trans-2,3- Dihydrodiolf Billy W. Day,' Yousif Sahali,O Deborah A. Hutchins,ll Michael Wildschutte,l Roberta Pastorelli,%Thanh T. Nguyen,A Stephen Naylor,v Paul L. Skipper,o
John S. Wishnok,O and Steven R. Tannenbaum*lO
Departments of Environmental & Occupational Health and Pharmaceutical Sciences, University of Pittsburgh, 260 Kappa Drive, Pittsburgh, Pennsylvania 15238,and Department of Chemistry, Division of Toxicology, Massachusetts Institute of Technology, Room 56-309,77 Massachusetts Avenue, Cambridge, Massachusetts 02139 Received May 15,1992 The metabolism of the environmental carcinogen fluoroanthene by human liver microsomes was compared to that by liver microsomes from rats treated with Aroclor 1254. Although the human-derived system gave primarily one product, similar metabolites were noted from each system. Enantiomers of the major metabolic product, in both cases the tram-2,3-dihydrodiol, were separated by chiral stationary-phase chromatography. Absolute configurations were assigned by application of the benzoate exciton chirality rules to the CD spectra of the 4- (dimethy1amino)benzoyl esters. Liver microsomes from Aroclor 1254-treated rats produced the R,R enantiomer of the diol in 75-78% enantiomeric excess, while human liver microsomes produced this enantiomer in only 6-12% excess. The activities of these enantiomers were compared in Salmonella typhimurium strain TM677 mutagenicity assays employing the 9OOOg supernatant of Aroclor 1254-induced rat liver homogenates. Both the syn- and anti-2,3dihydrodiol 1,lOb-epoxides, which had only been inferred to be metabolites in previous studies, were isolated from the microsomal incubations by preparative reverse-phase HPLC. The evident exceptional aqueous stabilities of these diol epoxides were further examined by half-life determination experiments. Their tetrahydrotetrol hydrolysis products were also noted in the metabolite HPLC profiles. The structures of the tetrahydrotetrols were confirmed by total synthesis.
Introduction Fluoranthene (1)is a tumorigenic (11,mutagenic (2,3), and cocarcinogenic (4,5)polycyclic aromatic hydrocarbon (PAH)' that is present in many sources of environmental and occupational exposure to petroleum products or the products of incomplete combustion. These include mainstream and sidestream tobacco smoke (6), industrial sources such as creosote (7, 8),and human diet (9,101. Levels of 1 in these samples are often higher than those of the most studied PAH carcinogen, benzo[alpyrene. t Taken in part from the doctoraldiasertation of D.A.H., Massachueetts Institute of Technology, 1987. * Author to whom correspondence should be addressed. f University of Pittsburgh. 8 Present address: Miles Inc., Agriculture Division, Environmental Research Section, Stillwell, KS. 11 Present address: The Procter and Gamble Co., Paper Products Division, Cincinnati, OH. 1 Present address: WCI-Unwelt Technik, GmbH, Wennigsen, Germany. # Present address: Istituto di Richerche Farmacologiche Mario Negri, Milan, Italy. A Present address: Universal Foods Corp., Milwaukee, WI. Present address: DepartmentofBiochemistryandMolecular Biology, Mayo Clinic, Rochester, MN. 0 Massachusetts Institute of Technology. Abbreviations: CD, circular dichroism; DAD, diode array detector; DBN, l,bdiazabicyclo[4.3.0Inon-bene; ee, enantiomeric excess; EI, electron ionization;HPLC, high-performanceliquid chromatography;GCMS, gas chromatography-mass spectrometry; GC-MSD, gas chromatograph-mass selective detector; NICI, negative ion chemical ionization; PAH, polycyclic aromatic hydrocarbon(s); PCI, positive chemical ionization; S9,9OOOg homogenate supernatant; THF, tetrahydrofuran.
Previous investigations of the metabolic activation of 1 by rat liver microsomes (11,12),perifused rat liver cells (13),and rats in vivo (14) have shown that trans-2,3dihydroxy-2,3-dihydrofluoranthene(2) is the major metabolite formed (see Scheme I for numbering system used for metabolites). This diol has been shown to be mutagenic to bacteria, both directly and in the presence of a metabolic activating system (111,and is the critical intermediate metabolite in further oxidations to give, presumably,both the anti-2,3-trans-dihydroxyand syn-2,3-trans-dihydroxy 1,lOb-epoxides 3 and 4. This latter assumption has been based on the observation that macromolecular adducts, shown to form with hemoglobin (15)and DNA (16)in vivo after dosing rats with 1,as well as after in vitro adduction of DNA via rat liver microsomal activation of 1 (17), have structures which contain the 1,2,3-trihydroxy-l,2,3,10btetrahydrofluoranthen-lob-ylresidue. In the case of DNA, these adducts have structures which further strongly suggest that they arise largely from reaction with the bayregion anti-diol epoxide 3. Compound 3 has been proposed to be the ultimate form of 1through which ita carcinogenic effects are mediated (11,17). Studies on the metabolism of 1by human tissues have not been reported to date. Since humans are exposed to a wide variety of environmental chemicals, we wished to compare metabolism of this PAH by human liver microsomes with easily obtainable liver microsomes from highly induced animals. We have made an attempt to develop useful methods for the chromatographicsepara0 1992 American Chemical Society
780 Chem. Res. Toxicol., Vol. 5, No. 6,1992
Day et al.
Scheme I. Metabolic P a t h w a y s f o r P r i m a r y and Secondary Oxidation of Fluoranthene 1 by Liver Microsomes from H u m a n s a n d from Aroclor 1254-Treated RatsP two phenolic dihydrodiols
1-phenol
3-phenol ,to
OH
\
2
B-Phenol
2,3-dione
4
-
\
0 Absolute stereochemistryindicated is of the major products produced by the rodent system. Heavy arrows indicate paths of metabolism noted from the human systems.
tions of metabolites of 1 after its incubation with both human and Aroclor 1254-induced rat liver microsomes. During these experiments, we noted previouslyunreported metabolites of 1. We herein report a further characterization of the metabolism of 1,including isolation directly from microsomal systems of both the diol epoxides 3 and 4 and their hydrolysis products 5-7, determination of the diol epoxides' half-lives in aqueous medium, and conformationalanalysis of the tetrahydrotetrols 5-7 by 'H-NMR. We have also isolated 2 from the incubations of rat and human liver microsomes with 1and have determined these diols' absolute configurations in both cases by the exciton chirality method. The relative mutagenicities and toxicities of the enantiomers of 2 were also compared in a bacterial assay. Materials and Methods Instrumentation and Chromatography. A Jasco-500 spectropolarimeter and 1-cm path length cells were used for circular dichroism (CD) measurements. Mass spectral analyses were performed using a HewlettPackard 5987A GC-MS with a standard EI/CI source or a 5971A GC-MSD with both E1 and PCI sources. The experimental parameters for the 5987A have been outlined previously (18)and were adapted for use on the MSD. Samples were introduced either by direct insertion probe or by GC after splitless injection and separated on 12-, 15-,or 30-m 0.25" i.d. capillary columns of 0 % , 5%, or 50% phenylsiliconized methyl silicone (HewlettPackard, Palo Alto, CA, and J&W Scientific, Rancho Cordova, CA). Trimethylsilyl ethers, formed on-column with (trimethylsily1)imidazole in pyridine, or methyl ethers, formed with diazomethane or on-column with trimethylanilinium hydroxide, were the derivatives used in GC-MS structure determinations. '€3-NMR spectra were obtained on Bruker WM 250-MHz or Varian VXR 500-MHzinstruments at the Massachusetts Institute of Technology Chemistry Department, or on an in-house man-
ufactured 500-MHz instrument at the Massachusetts Institute of Technology Francis Bitter National Magnet Laboratory. HPLC analyses were performed with a Hewlett-Packard HP 1090 liquid chromatograph equipped with a Hewlett Packard 1040diode array detector. Novapak C18 3.9- X 150-mmcolumns from Waters (Milford, MA) were used for analysis of metabolites with the following mobile phase gradient at a flow rate of 0.7 mL/min: 0% B to 100% B in 40 min; A = 9:l H&CH,OH; B = CH30H. One-minute fractions were collected. A Nucleosil C18 (10-mm particle size, 4- X 250-mm) column from Machery/ Nagel (Rainin Instrument Co., Woburn, MA) was used in the analysis of diol epoxide hydrolysis experiments with the following mobile phase gradients at a flow rate of 1mL/min: 15% to 65% B in 25 min; 15-min isocratic at 30% B; 30% to 35% B in 5 min; 10 mM (NHdzHPOd, pH 7.951A = 9 1 [50 mM NH~OZCH, CH30H; B = CH30H. Enantiomer resolution was performed using a 250- X 4.6-mm Pirkle Ionic 1-A 5-rm spherical aminopropyl silanized silica HPLC column with (R)-(-)-N-(B$-dinitrobenzoy1)-a-phenylglycineas the ionically bound stationary phase (RegisChemicalCo.,MortonGrove, IL). Themobilephase was 18:2:1 hexanes-EtOH-CH&N employed at aflow rate of 1.8 mL/min. Chemicals. Caution: Fluoranthene, its dihydrodiol, and its diol epoxides should be regarded as potential carcinogens, and any skin or inhalational contact with them should be avoided. Diazomethane, dioxane, and acrylonitrile are volatile toxic substances and should be used only in a fume hood. Compound 1 was obtained from the Midwest Research Institute (MRI, Kansas City, KS). Dihydrodiol 2 was either synthesized as described below or purchased from MRI and purified by HPLC. Purities were >99% as determined by HPLC and/or GC-MS prior to use. Glucose 6-phosphate, NADP+,and glucose-6-phosphate dehydrogenase were obtained from Sigma Chemical Co. (St. Louis, MO). Synthetic precursors were obtained from Aldrich Chemical Co. (Milwaukee, WI). Diol epoxides 3 and 4 were synthesized by previously described methods (19-22) with modifications as outlined below. Briefly, an outline of the syntheses, with previously unreported methods of preparations of intermediates given in detail, is as follows. All
Human and Rat Fluoranthene Metabolism
Chem. Res. Toxicol., Vol. 5, No. 6,1992 781
intermediates yielded NMR spectra which were consistent with their structures. 2 J 9-Fluorenylcarboxylicacid was esterified with diazomethane, alkylated with acrylonitrilein KOH/monoglyme,and then treated with aqueous acid to give 3-(9-fluorenyl)propanoicacid. 3-0x01,2,3,1Ob-tetrahydrofluoranthenewas prepared by heating a E mixture of 3-(9-fluorenyl)propanoicacid (2.856 g, 12 mmol) and 0 phenolic dihydrodlols PBr3 (0.752 mL, 2.166 g, 8 mmol) in dry ClCHzCHzCl (50 mL) lla a non-vicinal tetrol tetrol non-vicinal 2,3dione N at 40-50 OC, for 2-3 h, followed by stirring at room temperature &phenol m overnight. Anhydrous FeC13 (3.895 g, 24 mmol) was added along 6 5 1. 0 1. & & 3-phenol 3-phenol with a wash of ClCHzCHzCl(25mL) and stirred for 1h a t room C L A -. m temperature. The resulting Lewis complex was poured onto and hydrolyzed with 28.5% HC1 (40 mL). The resulting brown solution was extracted with CHC13 (3 X 40 mL). The combined organic phases were washed with 1N HCl(90 mL), 50 % saturated NaHC03 (2 X 90 mL), and HzO (90 mL), dried (MgSOd),filtered, and concentrated onto Celite. The ketone was obtained in 77 % yield after purification by flash chromatography over Si02 (4:l hexane-EtOAc). 3-Hydroxy-1,2,3,10b-tetrahydrofluoranthene was obtained by reduction of the %oxo compound with NaBH4 in CH30H. The alcohol was dehydrated with p-toluenesulfonic acid in dry C6H6 and converted to trans-2,3-bis(benzoyloxy)1,2,3,1Ob-tetrahydrofluorantheneby Prevost oxidation. The tetrahydro dibenzoate was oxidized to tra~s-2,3-bis(benzoyloxy)10 20 30 40 2,3-dihydrofluoranthene with o-chloranil in dioxane, or, alterTime (min) natively, by bromination of the 10b position with N-bromosucFigure 1. C-18 reverse-phase HPLC chromatograms of liver cinimide in a C c 4 solution containing benzoyl peroxide, followed microsomal metabolism of fluoranthene 1. Upper: Aroclor 1254by dehydrobromination with DBN in THF. Benzoyl protecting induced rat liver microsomes; lower: typical human liver migroups were removed with NaOCH3 in THF to give 2. crosomal pattern. The slight differences in retention times The anti-diol epoxide 3 was prepared from 2 by epoxidation between the two chromatograms are due to use of different HPLC with 3-chloroperbenzoic acid. The syn-diol epoxide 4 was columns. prepared from 2 by conversion to its bromohydrin with aqueous N-bromoacetamide, followed by oxirane ring formation via glucose 6-phosphate, 0.5 unit/mL glucose-6-phosphate dehydehydrobromination with potassium tert-butoxide. drogenase, 0.5-7.5 mg/mL microsomal protein, and 125-1000 Hydrolyses of diol epoxides 3 and 4 (0.5 mg) were performed nmol of 1 or 2 previously dissolved in acetone (95 95 versus saline (pH 7.4) containing 10 mM MgC12,0.7 mM NADP+, 5 mM
1
1, 6 1
Day et al.
782 Chem. Res. Toxicol., Vol. 5, No. 6, 1992 5690%). Compound 2 was the major metabolite from enzymatic activation of 1 with both microsomal systems. The identity of this metabolite was confirmed by comparison to synthetic standards of 2 via HPLC retention times, UV absorption spectra, and GC-MS characteristics. It was necessary to exercise care during the collection of chromatographic fractions corresponding to this metabolite, as a minor metabolite, the 7,8-dihydrodiol 8, only had a slightly smaller retention volume than 2 and often nearly coelutedwith2. Dihydrodiol8 was alsoasubstantial contaminant of the commercial preparations of 2 that we purchased during this study. The chromatographic conditions developed for metabolites of 1 allowed for separation and characterization of compounds more polar than 2 and 8. The rat microsomal incubations produced, upon C-18 HPLC/DAD-UV analysis, 13 peaks attributable to metabolites. Metabolite structures were proven by UV and/ or MS analysis, or by coelution with synthetic standards. Only six of these have been reported previously. Previously identified metabolites of 1 (2, 8, 2,3-dione, 8-,1-, and 3-phenols) were characterized from the metabolizing systems by comparing their UV and mass spectra to those in the literature (11-14). The 1-and 3-phenols coeluted in our chromatographic system. We also noted eight metabolites of 1 in the rat liver microsomal profile that had not previously been reported. Five of these new metabolites were positively identified in this study by matching UV spectra and coelution with synthetic standards (the diol epoxides 3 and 4) and also by MS analysis (5-7 and a nonvicinal tetrahydrotetrol). The vicinal tetrahydrotetrols were identified by coelution with standards whose stereochemistries were assigned by 1H-NMR analysis (vide infra). It should be noted that, although the range of concentrations of 1 we used was similar to that of earlier studies, the incubation period was up to 5 times as long. The compound which eluted intermediate to the two syn-diol epoxide-derived tetrols 5 and 6 is believed to be a tetrahydrotetrol on the basis of its retention volume and its UV spectrum (upper trace, Figure 2). The electron ionization mass spectral data from its per(trimethylsily1) derivative [380 (M - (TMS0)2, loo), 291 [M - (TMS0)3, 15],290(M-(TMS0)2-TMSOH, 161,202 [M-(TMSO)*, 281, 200 [M - (TMS0)2 - (TMSOHh, 1211 support the assignment of a nonvicinal arrangement of hydroxyl groups, because the vicinal tetrols 5-7 all fragment by a retro-Diels-Alder mechanism to give a base peak at m/z 354. Incubation of 2 with rat liver microsomes showed that this compound did not arise from further oxidations of the 2,3-dihydrodiol. Two other metabolites that probably are dihydrotriols were seen in the HPLC profiles of rat liver microsomal activation of 1. Their UV spectra have very similar band shapes and energy levels to that of the 2,3-dihydrodiol2, with each spectrum exhibiting a slight red shift in certain of the maxima (Figure 2). These metabolites were always formed in lesser amounts than were the diol epoxides by the rat liver microsomes (data not shown). It thus appears that these triols are phenolic dihydrodiols, arising from 2, most likely via epoxidation of a nonvicinal site on the dihydrodiol, followed by rearrangement of the resulting nonvicinal diol epoxide to yield a trihydroxy dihydro compound. We were, however, unable to obtain meaningful MS data from these compounds, which leaves their
al 0
c m 9 L
0 u)
9
a
‘7 240
280 320 360 Wavelength (nm)
400
Figure 2. UV absorptionspectraof three of the new metabolites of 1 noted in this study compared to the spectrum of the 2,3dihydrodiol2(lower). From the top and designatedby retention time (see Figure 1, upper trace): 16.6 min, putative nonvicinal tetrahydrotetrol;18.2 and 22.0 min, putative phenolic dihydrotriols. The retention time of 2 was 27.6 min. structural assignments as somewhat speculative. In the rat liver microsomal incubations, diol epoxides 3 and 4 represented 17% of the total metabolites a t low substrate concentrations and 9 % at the highest concentration of 1 used. These values are based on summation of the diol epoxides and their tetrahydrotetrol hydrolysis products 5-7. All 11 of the human liver microsome samples used in this study gave dihydrodiol 2 as the major metabolic product, accounting for 7Ck95% of the metabolites in a given sample. The human samples, however, showed considerable variability in both the extent of metabolism as well as the occurrence of the other specific metabolites formed from 1. For example, all but one of the human liver microsomal preparations tested produced the antidiol epoxide 3 and its tetrahydrotetrol hydrolysis product 7, but only one also gave the syn-diol epoxide isomer 4, and that in such low yield that we were unable to determine if its tetrahydrotetrol hydrolysis products 5 and 6 were also produced. Only one human liver microsomal sample that we analyzed in this study produced fluoranthene-2,3-dione. Interestingly, this liver sample was the one which did not yield diol epoxides. This is of some potential importance, as it has been reported that this quinone is a potent directacting mutagen in bacterial assays (11). A few of the human liver microsomes produced 8 and the 8-phenol in very low yield. All of the human samples produced the putative dihydrotriols which originated from the 2,3dihydrodiol. These were invariably the most abundant of the double oxidized metabolites and were formed in varying abundance relative to each other, dependingon the specific liver sample used.
Chem. Res. Toxicol., Vol. 5, No. 6, 1992 783
Human and Rat Fluoranthene Metabolism Hydrolysis of 3 and 4; Assignment of Stereochemistry of 5-7. The hydrolysis of the syn-diol epoxide 4 yielded two tetrahydrotetrols. At pH 1-6, the ratio of the two was consistently 5347. At pH 1-4, all 4 had hydrolyzed within 0.5 h, while ita t1/2 at pH 5 was calculated to be 26 min. The t1/2 at pH 6 was considerably longer (10.2 h). At pH 12,compound4 disappearedmore slowly than other pH and yielded mostly products other than tetrols, which were formed but comprised only a few percent of the total. Formation of the 1-ketone rearrangement product at pH 12 is not excluded by the results. The stereochemistries of the two tetrols arising from hydrolysis of 4 were assigned through the following interpretation of their 'H-NMR spectra, which was aided by inspection of molecular models. It is assumed that opening of the epoxide ring occurs only by attack at ClOb, as indicated by previous studies of other diol epoxides (26), and, therefore, only two stereoisomers are possible. In both, the relative stereochemistry of the 1, 2, and 3 positions is predetermined and is trans, trans. The Cl&&lob bond must adopt an equatorial orientation with respect to the saturated ring as a result of the limited ability of the cyclopentadiene ring to pucker, thereby causing the hydroxyl group at Glob to exist in a largely axial position. The axial c l o b hydroxyl in turn dictates that C2H in the stereoisomer resulting from cis addition of H20 resides in a quasi-equatorial position and that C2H in the stereoisomer resulting from trans addition of H2O resides in a quasi-axial position. C2H can be unambiguously identified by the multiplicity (dd) of ita NMR signal. It resonates at markedly different chemical shifts in the two stereoisomers,and ita strong upfield shift in one isomer identifies that isomer as 5, the one arising from trans addition of H20, because in this isomer the quasi-axial orientation of C2H places it under the influence of the shielding cone of one of the aromatic rings. The resonances at 6 4.278 and 4.936 in the spectrum of 6 can be assigned to C1H and C2H, respectively, on the bmis of the magnitude of the coupling constants (J = 1.5 and 7.2 Hz). Molecular models indicate that the dihedral angle between C1H and C2H is somewhat less than 120' while C2H and C3H are nearly antiparallel. Thus, C3H should exhibit the larger coupling constant. The CIH-C~H and C2H-C3H coupling constants observed for 6 do not differ by enough to permit assignment of resonances in the same way. In 6, though, it is clear that C1H is forced far out of the plane of the aromatic rings and toward the shielding cone of one ring in the same manner as the C2H in tetrol5. Thus, C1H can be assigned to the resonance observed at 6 4.374. The anti-diol epoxide 3 yielded only one tetrol(7) upon hydrolysis at each pH tested and gave no evidence of ketone formation. In the acidic range of pH 1-4, 3 hydrolyzed completely within 0.5 h. The product was not stable for any extended period of time at pH 3 or lower. At pH 5, 3 hydrolyzed to the extent of 88%within 1h, while at pH 6, the half-life was 3.8 h. Under basic conditions (pH 12) the half-life of 3 was on the order of days. The stereochemistry of 7 was determined to include 1,lOb-trans functionality by analysis of its 1H-NMR spectrum as above for 5 and 6. The doublet of doublets observed for C2H was not shifted to high field, indicating that C2H occupies a quasi-equatorial position and therefore C2H and the Glob hydroxyl group are trans to each other. The relatively greater coupling constant associated with
I 0
I
10
20
Time (min) Figure 3. Resolution of the enantiomers of 2 from human liver microsomes (upper), chemical synthesis (middle), and liver microsomes from Aroclor 1254-treated rata (lower) by chiral stationary-phase chromatography (see Materials and Methods section for details).
the resonance at 6 4.404 indicates that this resonance correspondsto C1H since the dihedral angle between C1H and C2H is nearly zero. Absolute Configuration of Metabolically Generated 2. Upon epoxidation of C2C3 of 1, two enantiomeric oxiranes are formed. As reported in an earlier study (111, this intermediate epoxide was not isolatable from the incubation mixtures and thus either is unstable or is rapidly converted to dihydrodiols by epoxide hydrolase. Hence, we could not directly determine the enantioselectivity of fluoranthene 2,3-oxide formation. The absolute configurations of the diols produced by the epoxide hydrolmecatalyzed attack of H2O at the oxirane carbons of fluoranthene 2,3-oxide, though, are completely controlled by the stereochemistry of the epoxide. We thus resolved the enantiomers of 2 by chiral column chromatography and determined their absolute configurations by the exciton chirality method through circular dichroism spectroscopy. Subjecting synthetic racemic 2 to chromatography on (R)-(-)-N(3,5-dinitrobenzoyl)-a-phenylglycineionically bound to aminopropyl silanized silica resulted in resolution of the two enantiomers (Figure 3). The C-18 chromatographic fractions containing metabolically formed 2 from both human and rat microsomal incubationswere analyzed by applying the same chiral chromatography resolution conditions. Liver microsomes from Aroclor 1254-treated rata produced 2 enantioselectively in 7 5 7 8 % ee. The major enantiomer was more retained by the chiral column. The human liver microsomes we used in this study, on the other hand, produced the later eluting isomer in only 6-12% ee, depending on the individual liver sample examined. The circular dichroism (CD) curves of the resolved enantiomers of 2 (not shown) were very weak (
