Chem. Res. Toxicol. 1991,4, 349-359
349
Identification of Epoxide- and Quinone-Derived Bromobenzene Adducts to Protein Sulfur Nucleophiles Donald E. Slaughter and Robert P. Hanzlik* Department of Medicinal Chemistry, University of Kansas, Lawrence, Kansas 66045-2506 Received December 31, 1990
Bromobenzene (BB) hepatotoxicity is widely attributed to the alkylation of cellular proteins by chemically reactive metabolites, particularly BB-SP-oxide. This laboratory recently reported the first conclusive evidence that BB epoxides actually do alkylate proteins; i.e., acid hydrolysates of hepatic proteins from phenobarbital-(PB-) induced BB-treated rats contain S-(0-, S-(m-,and S-(p-bromopheny1)cysteine[Weller, P. E., and Hanzlik, R. P. (1991) Chem. Res. Toxicol. 4,17-201. However, these three compounds account for > meta, ortho). This, too, clearly implicates protein-SH alkylation by BB-2,3- and 3,4-oxides. In addition, 2,3-dimethoxy-5bromothioanisole and another unidentified isomer were observed. However, by far the major adduct (5-6’70 of total covalent binding) was 2,5-dimethoxythioanisole(Le., a debrominated adduct). When BB-d5 was administered, the latter contained mostly 3 deuterium atoms/mol. These latter results clearly show that alkylation of protein sulfur nucleophiles in vivo by quinone metabolites is 10-15 times more extensive than their alkylation by BB epoxides. After BB-d5 was administered, the bromothioanisoles and dimethoxybromothioanisoles contained 4 and 2 deuterium atoms/mol, respectively. A weighted average calculation of deuterium retention across the six major sulfur adducts agreed well with 3H/14Cretention ratios determined earlier for total liver protein covalent binding of dual-labeled [3H/14C]BB,indicating that the overall pattern of BB metabolite binding to all protein nucleophiles may closely parallel that seen here specifically for protein sulfhydryl groups. The identification of a variety of specific BB-derived adducts to protein now affords the opportunity to investigate their relative contributions to the toxicity of bromobenzene.
Introduction The hepatotoxicity of bromobenzene (BB)’ has long been correlated with its biotransformation and the concomitant covalent binding of chemically reactive metabolites to nucleophilic groups on cellular proteins (1,2).On the basis of an analysis of the substitution pattern of major urinary metabolites of BB and the correlation of changes in this pattern with changes in BB hepatotoxicity elicited by inducing agents, toxicity was ascribed mainly to its 3,4-oxidemetabolite (Table I, compound 2) (3-6).A role for quinone metabolites of BB in protein covalent binding was first suggested by Hesse et al. (7) who observed that in vitro CVB of [14C]BBto microsomal protein continued after BB consumption ceased and that inhibition of epoxide hydrolase did not increase CVB. Further evidence supporting a role for quinone metabolites of BB in covalent binding came from in vitro studies using a variety of chemical and enzymatic trapping agenh (8),and from the observation of substantial (-50%) loss of 3H relative to 14Cduring the metabolic activation and covalent binding of 3H/14C-dual-labeled BB both in vitro and in vivo (9,lO). Abbreviations: 3-MC, 3-methylcholanthrene; BB, bromobenzene; [WIBB, [U-W]bromobemne;BB-& pentadeuteriobromobenzene;PB, phenobarbital; PB/BB, phenobarbital pretreated and bromobenzene dosed; Si SPE, silica gel solid-phase extraction; fraction C1, pentane eluate from Si S P E fraction C2,5% ether/pentane eluate from Si SPE; EIMS, electron impact mass spectrum.
Scheme I. Examples of Epoxide- and Quinone-Baaed Protein Adduct Formation and Subsequent Darivatization by “Alkaline Permethylation”’
Br
Br
Br
‘The epoxide adduct elimination of water (reaction a) probably occurs before C-S cleavage (reaction b). In both cases methylation is shown as occurring after C-S cleavage, but in principle Smethylation could precede cleavage.
None of the above approaches provides direct evidence for either the structures of the adducts formed or the
reactive metabolites responsible for their formation. To this end, we recently reported the isolation of S-(bromopheny1)cysteine isomers (para >> ortho, meta) from acid hydrolysates of liver proteins of BB-treated rata, which provides unambiguous evidence for the alkylation of rat liver protein in situ by metabolically generated epoxides 1 and 2 (11). Unfortunately, this approach is unsuitable
0093-220~/91/2704-0349~02.5~/0 0 1991 American Chemical Society
350 Chem. Res. Toxicol., Vol. 4, No. 3, 1991
Slaughter and Hanzlik
Table I. Structures of Putative Reactive Metabolites of Bromobenzene, Expected Alkaline Permethylation Products, and Thioanisole Standards putative reactive metabolites alkaline permethylation products expected
bo Q &
Br
Br
SCH,
1
6
SCH,
7
2
c
CH,O
s
k
10
CH,O &H,
XH,
12
SCH,
11 &H3 SCH,
4
Br
3”% ”& 3Hc OCH,
OCH,
13
bo
17
5
for adducts of BB-derived quinones, such as 3-5, to Nacetylcysteine (or, presumably, protein-SH groups), the major obstacle being the extensive degradation of the expected S-(bromodihydroxypheny1)cysteines via a cascade of oxidation, cyclization, and polymerization reactions akin to those of melanin biosynthesis (2,12-14). To circumvent this problem, we developed an “alkaline permethylation” process for the detection of adducts of BB metabolites to protein sulfur nucleophiles as diagrammed in Scheme I. The principle involved is simply that the protein-sulfur bond is cleaved by base-induced elimination (15-17), and nucleophilic moieties (phenol and thiol groups) are stabilized and rendered nonpolar by in situ methylation. Conceivably, quinones 3-5 could each react with protein-SH groups to generate four adducts, one of which would have undergone bromine loss via addition/elimination (viz., 9, 13, 17). Alternatively, these debrominated adducts could arise from debrominated quinones. Alkaline permethylation of these S-adducts should then furnish various dimethoxythioanisoles or bromodimethoxythioanisoles (e.g., 9-20, Table I). In contrast, reaction of epoxides 1 and 2 with protein-SH groups would lead to adducts with structures analogous to known glutathione adducts and premercapturic acids derived from 1 and 2, i.e., compounds 21-24. Such structures dehydrate quite Br
Br
OH
SR 22
OCH,
15
OCH, OCH,
Br
CH,S &CH
14
@H3
OCH,
SCH,
16
0
SR
3
CH,O
9
&
21
H
CH,O
3
OH
SCH,
0 &Ha SCH,
0
Br
7
23
24
8: R = glutathhyl
b: R = CH&H(NHAc)COOH
easily under basic or acidic (e.g., protein precipitation) conditions. Thus alkaline permethylation of protein con-
OCH, OCH,
10
CH,S
OCH,
OCH,
WH,
19
20
taining BB-metabolite adducts analogous to 21-24 should give rise to 6-8 and, thereby, provide an independent assessment of the extent to which 1 and 2 alkylate proteinSH groups. In order to gain further insight into protein alkylation by BB metabolites, particularly quinones, under toxicologically relevant conditions in vivo, we subjected whole liver protein from PB/BB rats to alkaline permethylation and examined the products by RP-HPLC and capillary GC/MS for the presence of bromothioanisoles, dimethoxythioanisoles, and bromodimethoxythioanisoles (viz., Table I). Parallel experiments were carried out by using [I4C]BBand BB-d,. Authentic samples of 6-13 and 20, as well as their corresponding mercapturic acids, were synthesized to aid optimization of conditions for alkaline permethylation, HPLC, and GC/MS. Results described in this paper provide unambiguous evidence for the formation and in situ covalent binding of 1,2, and 5 to protein sulfur nucleophiles. The formation and binding of 3 and 4 or their debrominated counterparts are also clearly demonstrated. Epoxide-derived adducts (detected as 6-8) account for only 0.4% of total covalent binding (corrected for the efficiency of recovery of thioanisoles from the alkaline permethylation of mercapturic acid standards). This result is in excellent agreement with the recovery of S-(bromopheny1)cysteines(0.5% of total covalent binding) from the acid hydrolysis of PB/BB rat liver protein. Quinone-derived adducts (detected as primarily 9,13, 17, and 20) accounted for a much greater fraction of total binding (6.7%). The occurrence of major amounts of debrominated materials (9, 13, and 17) was quite unexpected, since to our knowledge no debrominated urinary or biliary metabolites of BB have been reported. When a parallel study was carried out with BB-d5, a weighted average of deuterium retention or loss in the group of six major protein S-nucleophile adducts (i.e., 6-9, 13,20) indicated 56% net retention of deuterium overall (total
Protein Adducts of Bromobenzene
%/total adducts). This figure is very close to the previous report of 50% net retention of 3H versus 14Cin the total covalent binding fraction (Le., all protein nucleophiles) resulting from the use of 3H/14C-dual-labeled BB. Thus while covalent binding of reactive BB metabolites to protein S-nucleophiles may account for only a modest fraction (ca. 743%) of total covalent binding, it may nevertheless provide a reasonable representation of the relative contribution of various reactive BB metabolites to the overall process. Experlmental Procedures Analytical Methods. Low-resolution electron impact (EIMS) or desorption/chemical ionization (DCIMS) mass spectra with ammonia as the reagent gas were taken on a Nermag R10-10b quadrupole GC/MS with a SPECTRAL 30 data system. Exact mass measurements (HRMS) were taken on a VG-ZAB-HS high-resolution MS in the peak matching mode. Full-spectrum GC/EIMS analyses were conducted on a 30-m fused silica capillmy column [DBWax bonded poly(ethy1ene glycol), J & W Scientific] linked to the Nermag instrument; run conditions accompany chromatograms. An asterisk (*) denotes the 79Br peak of a 1:l isotopic doublet of a monobromo fragment ion; relative intensities are given in parentheses. 'H and 13C NMR spectra were determined at 300 and 75.4 MHz, respectively, on a Varian XL-300 NMR spectrometer unless otherwise indicated. Chemical shifts (6) are expressed in ppm. HPLC analyses were performed on either C8 (Alltech Econosil4.6 X 250 mm 10-pm bonded silica) or phenyl (Alltech Rsil 4.6 X 250 mm 5-rm bonded silica) reverse-phase columns eluted with methanol/HzO or acetonitrile/ HzO, respectively, using a Shimadzu HPLC system (dual LC-6A pumps, an SCL-6A controller, a C-R3A Chromatopac integrator, and an SPD-6A UV detector) unless otherwise indicated. Compositions of mobile phases used accompany chromatograms. 14C radioactivity was counted on a Beckman LS-7000 liquid scintillation counter. Chemicals. 3-Bromophenyl ethyl sulfide (HPLC internal standard) was Synthesized from 3-bromothiophenol (Aldrich) and ethyl iodide (18). [WIBromobenzene was available from earlier acids work (IO,19). (Bromo-2,5-dihydroxyphenyl)mercapturic were synthesized as a mixture of the 3-, 4-, and 6-bromo isomers (20) by the addition of N-acetylcysteine to bromobenzoquinone (generated by the oxidation of 2-bromohydroquinone by AgzC03 on Celite). Analogous procedures were used to synthesize (2,3dihydroxypheny1)mercapturic acid and (5-bromo-2,3-dihydroxypheny1)mercapturic acid. Dimethoxythioanisoles 9 and 13 and bromodimethoxythioanisoles 10-12 and 20 were synthesized by the alkaline permethylation of their corresponding mercapturic acids (20). The thioanisoles are all liquids at room temperature. Their characterization data are given below. All other chemicals were obtained from commercial sources and used without further purification. GC/MS evaluation of the bromobenzene-d5 used (BB-d,; Aldrich) showed it to be 97.8% dSand 2.2% d4 (99.6 atom % D); no other deuterated species were indicated. (2,3-Dihydroxyphenyl)mercapturicAcid. Hygroscopic solid. 'H NMR (500 MHz, DMSO-d6): 6 1.83 (s, 3 H), 2.98 (dd, J = 8.9, 13.3, 1 H), 3.22 (dd, J = 4.7, 13.4, 1 H), 4.27 (ddd, J = 4.7, 8.3, 8.3, 1 H), 6.43 (t,J = 7.8, 1 H), 6.52 (td, J = 1.4, 7.8, 2 H), 8.23 (d, J = 7.6, 1 H), 9.1 (bs, 2 H). 13C NMR (125 MHz, DMSO-da): 6 22.4,33.5,51.7,114.1,119.4,120.6,121.4, 1441,145.2, 169.4, 172.1. DCIMS (NH,): m / z 272 (MH+, 5), 132 (25), 112 (30), 86 (80), 66 (45),44 (100). HRMS: m / z 271.0523; calcd for CIiHlgNO$, 271.0513. (5-Bromo-2,3-dihydroxyphenyl)mercapturic Acid. Hygroscopic solid. 'H NMR (acetone-d6): 6 1.95 (8, 3 H), 3.19 (dd, J = 8.1, 13.9, 1 H), 3.37 (dd, J = 4.5, 13.9, 1 H), 4.65 (ddd, J = 4.4, 8.1, 8.1, 1 H), 6.96 (d, J = 2.2, 1 H), 7.09 (d, J = 2.3, 1 H), 7.55 (d, J = 8.0, 1 H). 13C NMR (acetone-de): 6 22.6, 37.4, 53.1, 111.4, 118.8, 123.3, 126.9, 145.8,147.1, 170.8, 171.7. DCIMS: m/z 3501 (MH', 8), 272 (121, 147 (15), 132 (50), 130 (loo), 112 (20), 86 (20), 60 (85). HRMS: m / z 348.9615; calcd for Cl1Hl2BrNOSS, 348.9620. 2,3-Dimethoxythioanisole(13). 'H NMR (aCet"d6): 6 2.38 (s, 3 H), 3.75 (s, 3 H), 3.83 (s, 3 H), 6.76 (dd, J = 1.4, 8.0, 1 H),
Chem. Res. Toxicol., Vol. 4, No. 3, 1991 351 6.83 (dd, J = 1.4,8.3), 7.04 (t,J = 8.1). 13CNMR (acetone-de): 6 14.1, 56.1 59.7, 110.2, 117.7, 125.3, 134.3, 146.2, 153.5. EIMS: m/z 184 (M', loo), 169 (44),154 (18), 137 (17), 125 (32), 111 (42). HRMS: m/z 184.0551; calcd for CgHlzOPS,184.0558. 2,5-Dimethoxythioanisole(9). 'H NMR (acetone-de): 6 2.38 ( 8 , 3 H), 3.74 (s, 3 H), 3.78 (s, 3 H), 6.65 (dd, J = 2.9, 8.8, 1 H), 6.71 (d, J = 2.8, 1H). 13C NMR (acetone-de): 6 14.1,55.8,56.6, 109.9, 112.0, 112.8,129.7, 151.3, 155.3. EIMS: m / z 184 (M+, 100), 169 (55),154 (45), 137 (18), 123 (45). HRMS: m/z 184.0557; calcd for CgH1202S,184.0558. 5-Bromo-2,3-dimethoxythioanisole(20). 'H NMR (acetone-de): 6 2.41 (s, 3 H), 3.74 (s, 3 H), 3.87 (8, 3 H), 6.86 (d, J = 2.2, 1 H), 6.98 (d, J = 2.2, 1 H). 13C NMR (acetone-de): 6 14.1, 56.6, 59.8, 113.4, 117.5, 119.8, 136.8, 145.2, 154.0. EIMS: m / z 262* (M+, 45), 247* (22), 168 (loo), 153 (15). HRMS: m/z 261.9661; calcd for CgH1102BrS,261.9663. 3-Bromo-2,5-dimethoxythioanisole (12). 'H NMR (acetone-de): 6 2.45 (8, 3 H), 3.76 (s, 3 H), 3.81 (s, 3 H), 6.73 (d, J = 2.8, 1 H), 6.90 (d, J = 2.9,l H). EIMS: m/z 262* (M+, 70), 2471 (62), 232* (22), 168 (loo), 153 (E),125 (27). 4-Bromo-2,5-dimethoxythioanisole (1 1). 'H NMR (acetone-de): 6 2.45 (s, 3 H), 3.84 (s,3 H), 3.87 (s, 3 H), 6.89 (s, 1H), 7.11 (8, 1 H). EIMS: m/z 262* (M', loo), 247* (50), 232* (7), 168 (80), 153 (12), 125 (25). 6-Bromo-2,5-dimethoxythioanisole (10). 'H NMR (acetone-d6): 6 2.36 (s, 3 H), 3.82 (s, 3 H), 3.86 (8, 3 H), 6.99 (d, J = 9.5, 1 H), 7.05 (d, J = 9.5,l H). EIMS: m/z 262* (M', 57), 247* (20), 168 (loo), 153 (20), 125 (20). Induction a n d Dosing of Rats; Preparation of Liver Protein. Male Sprague-Dawley rats (200-300 g) were injected ip with phenobarbital (PB) (50 mg/kg, 50 mg/mL in 0.9% NaC1) for 4 days and fasted overnight (to lower hepatic glutathione content) prior to ip administration of a hepatotoxic dose of BB (2.5 M in corn oil, 2.5 mmol/kg; when used, [14C]BBwas 0.25 Ci/mol). Four hours after dosing, rats were killed by decapitation following C02 narcosis. Dissected livers were washed once in phosphate-buffered saline (0.1 M potassium phosphate, 1 mM EDTA, 1.12% KCl, pH 7.4), blotted dry, weighed, and homogenized (Polytron) in 5 mL/g of the same buffer. The resulting suspension was adjusted to 20% trichloroacetic acid (TCA). The precipitate was collected by centrifugation and washed sequentially with acetone (2X), 80% methanol/H,O (4X),acetone again (2x),and ethyl ether (2x). The protein was then dried by rotary evaporation and under high vacuum, resulting in an off-white free-flowing powder. Protein yields were 1.7 f 0.2 g of washed liver protein/200 g of rat, and the density of covalent binding was 4.2 f 2.6 pmol equiv of BB/g of protein. Alkaline Permethylation of Liver Protein. Solvent-washed protein (0.2-0.5 g) was added to a two-phase mixture of 8 mL of 16 N KOH (N, purged) and 7 mL of CH31 in a 25 X 100 mm screw-cap (Teflon-lined) glass culture tube (Corex no. 8422-A). If the reaction mixture contained radiolabel to be analyzed by HPLC nonradioactive carriers in the form of 3 pmol each of (2-, (3-, and (4-bromopheny1)mercapturicacids, (2,3- and (2,5-dihydropheny1)mercapturic acids and (3-, (4-, and (6-bromo-2,5dihydroxypheny1)mercapturic acids ([ x3 hmen] = 3 rmol; 3Br:4-Br:6-Br = 2028:52 as judged by 'H NMR) were added. Tightly capped reaction tubes were then immersed in a 120 OC oil bath to just above the meniscus and refluxed while stirring until the CH31was completely consumed (as evinced by a transition from a two-phase to a one-phase liquid system, cessation of CH31boiling and refluxing, and the appearance of a substantial white precipitate). [Caution! Considerable internal pressure develops during these reactions, apparently from the formation of dimethyl ether. Ongoing reactions should be shielded, and completed reaction mixtures should be cooled on ice and the tubes opened cautiously.] Cooled reaction mixtures were extracted with pentane (3 X 10 mL), and the extracts were dried over Na2S04. Silica Gel Solid-Phase Extraction (Si S P E ) of Thioanisoles. Pentane extracts were concentrated by fractional distillation (15-cm Vigreux column; 55-60 "C oil bath) to about 2 mL and then applied to an Si SPE cartridge (0.6 g of flash-grade silica) equilibrated in pentane. The column was eluted sequentially with 15 mL of pentane, 12 mL of 5% ether/pentane, 12 mL of ether, and 12 mL of methanol. Control experiments involving
352 Chem. Res. Toxicol., Vol. 4, No. 3, 1991 Table 11. Fractionation of Radioactive Protein Adducts and Their Alkaline Permethylation Productsa total covalent binding (n = 6), nmol equiv/mg of 4.2 f 2.6 protein pentane extract after alkaline permethylation (n = 4), 9.6 f 2.8 % of total *‘C elution after solid-phase extraction onto silica (n = 4), % of total pentane 0.3 f 0.1 5% ether/pentane 4.7 f 1.3 ether 3.1 f 0.6 methanol 0.6 f 0.1 See text for details of fractionation procedures. the application of thioanisole standards in pentane to Si SPE cartridges showed that thioanisole and the bromothioanisoles (6-8) eluted quantitatively in the pentane fraction while the brominated and nonbrominated dimethoxythioanisoles (9-13, 20) eluted quantitatively in the 5% ether/pentane fraction. The ether and methanol elutions were included in order to maximize recovery of 1%from the column, which was 85 & 17%. All nine thioanisole standards were recovered essentially quantitatively. However, Si SPE chromatography of the pentane extracts did remove a good deal of contaminating material, as noted by the amount of radioactivity separated from the thioanisole-containing fractions (see Table 11)and by the retention of most of the sample’s yellow color on the Si SPE cartridge during elution with 5 % ether/ pentane. Concentration of Si S P E Eluents for HPLC and GC/MS. Concentration of pentane or 5 % ether/pentane extracts by fractional distillation was effective only down to a volume of about 300-600 pL. For further concentration a special ‘flask” was made by heat-sealing a commercial Pasteur pipet about 1.5 cm from its large end so as to leave a sharply tapered conical rather than rounded bottom. Into this vessel a pentane concentrate (300-600 pL) was transferred by using a 1-mL glass syringe with a 6 in. X 22 guage steel needle, and the bottom 2-3 cm of the tube was immersed in an oil bath at 55-60 “C. With the air-cooled walls of the long neck of this flask acting as a reflux/fractionating column, the volume could be reduced to about 20 pL, most of which was formed by condensation of residual vapom on the inside walls upon removal from the oil bath. For HPLC analysis 3 pmol of 3-bromophenyl ethyl sulfide as chromatographic standard was added to the extract, the sample was concentrated to about 20 p L (as above), and 25 pL of methanol was added to displace the remaining pentane by azeotropic distillation in the same bath. The entire sample (20-40 pL) was then injected manually onto the HPLC column. Thioanisole standards were recovered quantitatively following these concentrating procedures. In cases where the mass chromatographic peaks of the dimethoxy- and bromodimethoxythioanisolesfrom the 5 % ether/ pentane Si SPE eluents were obscured by those of contaminating materials, low-boiling solvent was removed from the sample and replaced with methanol as described above, and the thioanisoles were further purified by HPLC (C, reverse phase, 20-70% methanol/H20 in 30 min, 1.5 mL/min, eluent from 22 to 32 min collected as a single fraction) in the absence of intemal standard. After extraction from the collected HPLC eluent with pentane, the extract was dried and concentrated for GC/MS. Identification of Recovered Thioanisoles. Recovered thioanisoles were identified tentatively by HPLC and then conclusively by GC/MS. HPLC identification was based on the coelution of radioactive peaks from [14C]BB-dosedrats with the UV peaks of standards. Identification by GC/MS was based on retention time and full-spectrum EIMS comparisons between thioanisole standards and experimentally derived materials. Mess chromatographic peaks were monitored at appropriate m/z values for their molecular ions. For retention time comparisons standards and experimental samples were run separately and peak retention times were normalized relative to those of a chromatographic internal standard (2,4-dichloronitrobenzene for 6-8; 1,2,4-trimethoxybenzene for 9-13 and 20). Quantitation of Recovered Thioanisoles. Recoveries of thioanisoles from alkaline permethylation of protein from [14C]BB-dosedrats were determined by comparing the amount
Slaughter and Hanzlik of radiolabel (corrected for counting efficiency) coeluting with added thioanisole standards to the specific radioactivity of the [“CJBB substrate. HPLC estimates of the recoveries of thioanisoles from the alkaline permethylation of mercapturic acid standards were determined by comparing integrated W ( 2 5 4 ” ) peak areas of recovered thioanisoles with those of authentic standards. All thioanisole peak areas were normalized to that of an internal standard, 3-bromophenyl ethyl sulfide. GC/MS quantitative estimates of thioanisole recoveries were determined by comparisons of integrated ion intensities (molecular ion; reconstructed ion chromatogram mode) of experimentally derived thioanisoles with those of known quantities of thioanisole standards after normalizing to those of chromatographic standards (as above). The molecular ions of compounds 6-9 and 13 either were the base peaks of their respective mass spectra or had intensities similar to those of the base peaks. The molecular ions of compounds 10-12 and 20 (m/z 262 for %r) were significantly less intense than those of their base peaks (m/z 168,M = CH3Br), but they were still substantial (240%). They were chosen over the base peaks for use in quantitation because the nearly equal intensities of their counterparts ( m / z 264) gave a measure of confidence that there was no significant contribution to their ion currents from coeluting contaminants.
Results Synthesis and Characterization of Standards. The synthesis of new thioanisole derivatives and new mercapturic acids reported herein proceeded in strict analogy to related syntheses reported previously (20). Thioanisoles were isolated by extraction and silica gel chromatography, while mercapturic acids were isolated in analytically pure form by preparative HPLC a n d lyophilization. It is interesting to note that, in t h e synthesis of (bromodimethoxypheny1)mercapturic acids via t h e nucleophilic addition of N-acetylcysteine to bromoquinones 3-5, only addition of sulfur was noted; displacement of bromine by sulfur was n o t detected under t h e conditions used. Method Development and Validation. Our principal objective was to detect, differentiate, and quantitate epoxide- and quinone-based adducts arising from the alkylation of protein nucleophiles i n vivo by reactive metabolites of bromobenzene. We focused specifically on sulfur nucleophiles because of t h e anticipated ease with which they could be separated from the protein by alkali-induced fragmentation and methylation (Scheme I). T o validate this approach, several control experiments were performed using chemically synthesized standard compounds. First, i t was ascertained that representative thioanisole derivatives were stable under the alkaline permethylation reaction conditions a n d that they could be recovered quantitatively from these mixtures by extraction with pentane. Second, i t was found t h a t t h e simple thioanisoles (e.g., 6-8) a n d t h e more polar dimethoxythioanisoles (e.g., 9-13, 20) could be separated cleanly and quantitatively by chromatography over silica gel in solid-phase extraction cartridges; t h e former eluted easily with pentane while the latter required 5% ether in pentane for elution. This simple step afforded a n expedient a n d useful class separation of epoxide- vs quinonederived thioanisole products. T h e efficiency of t h e alkaline cleavage a n d methylation reactions shown in Scheme I could not be assessed directly because of a lack of authentic protein adducts as standards. Therefore, mercapturic acids were used as surrogates for this purpose. In the presence of liver protein from control and (p-bromorats (PB induced, no BB dosing) ( 0 - , (m-, pheny1)mercapturic acids were found t o undergo conversion t o thioanisoles 6,7, a n d 8, respectively, with an average efficiency of 67 % , (2,5-dihydroxyphenyl)mercapturic acid was converted quantitatively t o 2,5-dimethoxythioanisole (91, a n d (3-, 4-, a n d 6-bromo-2,5-dihydroxy-
Chem. Res. Toxicol., Vol. 4, No. 3, 1991 353
Protein Adducts of Bromobenzene
0
10
20
40
30
50
60
Time (min.) 30.6
C' 1 .o
0.9
0.8
Relative Retention Time
I0
20
30
Time
(inin
40
5r1
hi1
)
Figure 1. C8 HPLC profiles of a pentane Si SPE eluent. UV peaks numbered in boldface are of bromothioanisoles (Table I) derived by the alkaline permethylation of nonradioactive mercapturic acid standards added to the protein sample as described under Experimental Procedures. Retention times given over peaks are in minutes. Both chromatograms are from the same run. The injected sample represents the pentane Si SPE eluent derived from 0.5 g of protein. (*) Internal standard (4-bromophenyl ethyl sulfide). Radiochromatograms: (B) pCi/fraction marked at midpoint of collection interval; (-) % methanol in HPLC mobile phase (H20).
pheny1)mercapturic acids were converted to bromo-2,5dimethoxythioanisoles 10, 11, and 12, respectively, with an average efficiency of 69%. Isolation and Quantitation of Thioanisole Derivatives from the Alkaline Permethylation of Liver Protein from Bromobenzene-Treated Rats. Liver proteins isolated from six individual phenobarbital-pretreated rats 4 h after ip administration of a hepatotoxic dose of [14C]bromobenzenecontained an average of 4.2 nmol equiv of BB residues/mg of protein (Table 11). The induction and dosing regimen was particularly effective on this occasion and resulted in notably higher average binding than reported previously (111, although there was some overlap between the two data seta. Portions of this protein (ca. 500 mg) from each of the four rats with the most covalently bound radioactivity were subjected to alkaline permethylation and the products extracted and fractionated as described above; the quantitative results are given in Table 11. The pentane extraction after alkaline permethylation removed approximately 10% of the 14Coriginally present in the protein as covalently bound residue. This is much greater than the 0.2-0.790 of total covalent binding previously identified as S-(bromopheny1)cysteine isomers in acid hydrolysates of similar protein samples (11). Fractionation of the pentane extracts over silica gel yielded a pentane fraction (hereafter called fraction Cl), a 5% ether/pentane fraction (hereafter called fraction CP), an ether fraction, and a methanol fraction (viz., Table 11). The recovery of 14C in these four fractions was 87% of that applied to the columns. As the two groups of compounds of primary interest could be cleanly separated into fractions C1 and C2, only these fractions were analyzed in detail (next two sections). However, it is worth noting that the 0.3% of bound 14Creleased into fraction C1 agrees well
20 120
I:
,Oo
206"
Peak 11'
II
lI
40
20 0 60
80
100
120
140
160
180
200
m/z Figure 2. GC/MS reconstructed ion chromatograms of bromothioanisole standards (molecular ion = 202 amu) and Si SPE pentane eluents (fraction Cl). (a) Standards as labeled; (b) eluent from rats dosed with unlabeled BB; (c) eluent from rats dosed with BB-d6 (molecular ion at 206 amu for expected d, bromothioanisoles); (d) EIMS of p-bromothioanisole standard 8 from panel a above; (e) EIMS of peak B' in panel c above. An asterisk (*) denotes the %r peak of a 1:l isotopic doublet of a monobromo fragment ion. Retention times are relative to that of a 2,4-dichloronitrobenzenestandard (ca. 16.0 min). GC/MS conditions: DBWax capillary, l-rL splitless injections, full-spectrum E1 mode, 120-250 "C at 5 OC/min, linear velocity of the carrier gas (He) = 25 cm/s. Data for panels b and c were collected at lox sensitivity. Column and instrumentation were as described under Experimental Procedures.
with the 0.2-0.790 of total binding identified earlier as S-(bromopheny1)cysteine isomers. In addition, the fact that there is much more radioactivity in fraction C2 than in fraction C1 is consistent with earlier studies of the covalent binding of 3H/14C-dual-labeled bromobenzene, which suggested that the majority of covalent binding involved metabolites at an oxidation state more like that of a quinone than that of an arene oxide (9, 10). Identification of Thioanisoles in Fraction C1. HPLC analysis of fraction C1 (Figure 1)showed a major peak of radioactivity that coeluted with p-bromothioaniaole (8) and two much smaller peaks that coeluted with the ortho and meta isomers (6 and 7); no other radioactive peaks were observed, and the overall recovery of injected 14Cwas essentially quantitative. This pattern agrees well with the distribution of S-(bromopheny1)cysteineisomers observed following the acid hydrolysis of PP/BB liver protein from [14C]BB-dosedrats (11). GC/MS analysis of fraction C1 and related standards is shown in Figure 2, The E1 mass spectra for standards 6-8 were all very similar, showing M+ and M - CH3 peaks at m/z 202* and 187* and M - HBr and M - CH3Br fragments at m/z 122 and 108, respectively. Panels a and
354 Chem. Res. Toxicol., Vol. 4, No. 3, 1991
Slaughter and Hanzlik
Table 111. Estimated Recoveries of and Deuterium Retention for Thioanisoles from Alkaline Permethylation of PB/BB Rat Liver Proteinn recovery,*nmol/g of protein deuterium retention compd from BB-do (n = 3)‘ from BB-d5 (n = 3Id atoms/mol mol % of d5 7 8 6 13 9 20
1.2 f 0.2 9 f 1 0.9 f 0.4 14 f 1 58 f 5 23 f 5
1.2 f 0.1 11 f 1 1.9 f 0.6 7 f 2 81 f 18 1.8 f 0.1
4 4 4 2.Y 2.6e 2.0 2.8
av deuterium retention1
80 80 80 54 52 40 56
a This analysis includes only those thioanisoles for which authentic thioanisole standards were available. *Recoveries were estimated from GC/MS data as described under Experimental Procedures. ‘Protein pooled from 6 rats. dDuplicate analyses for protein from 3 separate rats. Weighted average of d2 and d3 species. ’Weighted average across these 6 compounds.
0
10
20
Time (min.)
30
40
30
40
70
90
136
110 Time (min.)
16.0
IO
20 Time (min)
80
Figure 3. C8 HPLC profidea of a 5% ether/pentane Si SPE eluent (fraction C2). UV peaks numbered in boldface are of dimethoxyand bromodimethoxythioanisoles (Table I) derived from the alkaline permethylation of nonradioactive mercapturic acid standards added to the protein sample. Bromodimethoxythioanisole 20 (not included in this example) elutes just ahead of compound 12 under the conditions. Retention times given over peaks are in mintues. Both chromatograms are from the same run. The injected sample represents the 5% ether/pentane Si S P E eluent from 0.5 g of protein. (*) Internal standard (4bromophenyl ethyl sulfide). Radiochromatograms: ( 8 ) nCi fraction marked at mid retention time; (-) % methanol in HPL mobile phase (H,O). Missing data points on the radiochromatogram were zeros.
L
b in Figure 2 compare the reconstructed ion chromato-
grams at mlz 202 for standards 6-8 with that for reactions
C1 from animals dosed with unlabeled BB. Panel b indicates that all three bromothioanisole isomers were found but that para >> meta, ortho, (8 >> 7, 6), once again in agreement with the ratios of S-(bromopheny1)cysteine isomers found in the acid hydrolysates of livers from PB/BB rats. The EIMS of peak B showed an excellent match with that of 8 at all major ions. Peaks A and C, while weak, were clearly identifiable as bromothioanisole isomers by comparison of their retention times and EIMS to standards. The average recovery of compound 8 as estimated from GC/MS data was 9 nmol/g of protein (Table 111) or L 5 mol % of total thioanisole recovery. Recovery of all three bromothioanisole isomers accounted for about 10 mol % of total thioanisoles. Fraction C1 was also searched for thioanisole itself, which could conceivably arise via a debrominated metabolite of BB (see below), but none was observed. The reconstructed ion chromatogram for fraction C1 from rats dosed with BB-d, is shown in panel c of Figure 2 (now at mlz 206 corresponding to tetradeuterobromo-
100
120
Time (min.)
Figure 4. Phenyl bonded-phase HPLC profiles of 5% ether/ pentane Si S P E eluent (fraction C2). UV peaks numbered in boldface are of dimethoxy- and bromodimethoxythioanisoles (Table I) derived from the alkaline permethylation of nonradioactive mercapturic acid standards added to the protein sample. Retention times given over peaks are in minutes. Both chromatograms are from the same run. The chromatograms were developed by using a linear 10-52% CH3CN/H20 gradient over 125 min. The injected sample represents the 5% ether/pentane Si SPE eluent from 0.2 g of protein. (*) Internal standard (4bromophenyl ethyl sulfide). Radiochromatograms: (a) nCi/ fraction marked a t mid retention time; (-) % acetonitrile in HPLC mobile phase (HzO).
thioanisoles). Once again peaks corresponding to all three bromothioanisoles were seen with para >> meta, ortho; their slightly decreased retention times are consistent with their high deuterium content. Panels d and e of Figure 2 show the EIMS of standard 8 and peak B’ of panel c, respectively; the latter can be seen to contain four and only four deuterium atoms, as predicted by Scheme I. The E1 mass spectral peaks A’ and C’ in panel c were weaker but clearly confirmed their identities as 7-d4 and 6-d4, respectively. Identification of Dimethoxythioanisoles in Fraction C2. Fraction C2 from rats dosed with [14C]BBwas analyzed under two different HPLC systems: C8 reverse phase eluted with methanol/HzO and phenyl reverse phase eluted with CH3CN/Hz0. Figure 3 shows the C8 HPLC chromatograms. Almost all of the injected radioactivity eluted in two unequal peaks, an early major peak that coeluted with dimethoxythioanisolestandards 9 and 13 and a later minor peak that eluted slightly earlier than 12. Thii second peak was later found to coelute with bromodimethoxythioanisole standard 20. The overall recovery of 14Cinjected was 92 f 8% (98% for the run shown).
Chem. Res. Toxicol., Vol. 4, No.3, 1991 355
Protein Adducts of Bromobenzene
l 262'
1208
H'
I
I * ----A_. I f>
I
D' I
1.1
1.3
.
I1
I
1.5
'
I
1.7
'
I
.
1.9
120
140
160
200
180
220
240
260
1
2.1 60
Relative Retention Time
Figure 5. GC/MS reconstructed ion chromatograms of dimethoxy- (molecular ion = 184 amu) and bromodimethoxythioanisole (262 amu) standards and Si SPE 5% ether/pentane eluents (fraction C2). (a and b) Standards as labeled; (c and d) eluent from rata dosed with unlabeled BB; (e and f) eluent from rats dosed with BB-d6(molecular ions at 187 and 264 amu, respectively, for expected d3dimethoxy- and d2bromodimethoxythioanisoles). Retention times are relative to that of a 1,2,4trimethoxybenzene standard (ca. 11.8 min). Each pair of chromatograms derived from a single run. GC/MS conditions: DBWax capillary, 1-pL splitless injections, full-spectrumE1 mode, 140-250 "C at 5 OC/min, linear velocity of the carrier gas (He) = 25 cm/s. Data for panels e and f were collected at 1OX sensitivity. Column and instrumentation were as described under Experimental Procedures.
Separation of 9 and 13 was achieved by phenyl HPLC as shown in Figure 4. Here a majority of the radioactivity in fraction C2 is seen to coelute with 9, and only a minor amount with 13. A second major radioactive peak coeluted with both 12 and 20, but C8 HPLC (see above) showed the absence of 12 in fraction C2. A small amount of radioactivity elutes near 11, but as described below, no evidence for 10, 11, or 12 was ever seen during GC/MS examination of fraction C2. Results of separate GC/MS analyses of fraction C2 materials and related standards (9-13,20) are shown in Figures 5 and 6. Data were collected in the full-spectrum mode. The EIMS of dimethoxythioanisole 9 (Figure 6, panel c) shows a base peak molecular ion at m/z 184, with M - CH3, M - (CH3)2, and M - SCH, fragments, respectively, at m/z 169,154, and 137 and other fragments of uncertain origin at m/z 123 and 125; the spectrum of isomer 13 (not shown) was very similar. The EIMS of 20 (Figure 6, panel a) shows strong M+ and M - CH3 peaks a t m/z 262* and 247*, a base peak (M - CH3Br) at m/z 168, and an M - (CH3I2Brfragment at m/z 153; the mass spectra of isomers 10-12 (not shown) were very similar to that of 20. Figure 5 shows reconstructed ion chromatograms for standards 9-13 and 20 (panels a and b) and for materials in fractions C2 (panel c-0. Panels c and d are from rats dosed with unlabeled BB. The probable identities of peaks D, F, and H from panels c and d with standards 13,9, and 20, respectively, were confirmed by comparisons of their EIMS in addition to their retention times. The mass spectrum of peak E was very similar to those of dimethoxythioanisoles 9 and 13, with a molecular ion at m/z 184 and an M - CH3 peak at m/z 169 as well as other familiar fragments. Assuming the two oxygens are quinone metabolite derived, they must be 1,2 or 1,4 to each other. Since only three such dimethoxythioanisole isomers are possible (9, 13, and 171, peak E must be 17. The mass
40
20
loolPeak F' 120,
187
I
80
dl
172
126
60 40
20 00
60
80
100
120
140
160
180
m/z Figure 6. Mass spectra extracted from indicated peaks in the GC/mass chromatograms of Figure 5. (a) Bromodimethoxythioanisole 20 standard; (b) peak H' from BB-d6-dosedrat; (c) 2,5-dimethoxythioanisolestandard (9); (d) peak F' from BBd , - d d rat. An asterisk (*) denotes the 79Br peak of a 1:l isotopic doublet of a monobromo fragment ion.
spectrum of peak G showed a molecular ion at m/z 262*, a very small M - CH3 peak at m/z 247*, an M - CH3Br base peak at m/z 168, and an M - (CH3I2Brpeak a t m/z 153, which indicated it to be an isomer of 20. Peak G could thus be any of compounds 14-16,18, or 19, all bromodimethoxythioanisoles for which standards were not available. 2,5-Dimethoxythioanisole(9) (peak F) was by far the most abundant thioanisole recovered, accounting for more than 50% of the total recovery (Table 111). Among recovered thioanisoles for which standards were available (6-13,20), debrominated dimethoxythioanisolesaccounted for 68% of total recovery and the s u m of quinone-derived thioanisoles (9 + 13 + 20) accounted for 90% of the total pentane-extractable materal after alkaline permethylation. Reconstructed ion chromatograms for fraction C2 materials from rats dosed with B B d 5 are shown in Figure 5, panels e and f. The ions at m/z 187 and 264 correspond, respectively, to the 79Brmolecular ions for the expected d3-containing dimethoxythioanisoles and d2-containing bromodimethoxythioanisoles. Three peaks (D', F', and H') were detected and observed to coelute with standards 13, 9, and 20, respectively. The EIMS of peaks F' and H' (Figure 6, panels d and b, respectively), when compared to those of standards 9 and 20, not only confirmed their identities, but also that they contained primarily the expected number of deuterium atoms (i.e., 9-d3 and 20-d2, viz., Scheme I). Furthermore, the presence of deuterium in F' proves that despite the absence of bromine, this product is truly derived from the administered bromobenzene. The unexpected observation that a majority of the BBderived alkaline permethylation products were debrominated, together with the observation that peak F' also
356 Chem. Res. Toxicol., Vol. 4, No. 3, 1991 -
-2
a)
193
b,
193
')
191
x
.e
B 2!
.e u 6
2
I
I
m/z
Figure 7. Mass spectral molecular ion clusters for compound 9 (2,5dimethoxythioanimle)derived from alkaline permethylation
in NaOD, D20,and CDJ. (a) From PB/BB (unlabeled BB) rat liver protein. (b)From authentic 2,5dihydroxyphenylmercapturic acid permethylated in the presence of PB-induced control rat liver protein. (c) From a mixture of authentic (bromo-2,ti-dihydroxypheny1)mercapturicacid isomers (i.e., 3-, 4-, and &bromo). See Figure 6 for full mass spectrum of compound 9 in unlabeled form. contained 30-35 mol % d2species, led us to examine much more closely the details of the alkaline permethylation process itself. A major objective of this effort was to determine whether the loss of bromine and deuterium was occurring in vivo as part of the metabolism and binding process, or in vitro as an artifact of handling. To this end, the experiments described in the next section were carried out. Alkaline Permethylation Using Deuterated Reagents. A t the time our alkaline permethylation method was being developed no debrominated soluble metabolites of bromobenzene were known, nor was there any evidence for formation of debrominated bound residues. Once 9 was discovered to be the major alkaline permethylation product from liver protein of rats treated with BB, a closer look at the alkaline permethylation chemistry of our standards of (3-, (4-, and (6-bromo-2,5-dihydroxyphenyl)mercapturic acids was called for. To our surprise, alkaline permethylation of a mixture of these three isomers in the absence of protein yielded 9 in 25% yield (average of 6 trials) along with 10-12 in ratios proportional to the abundance of their corresponding mercapturic acids in the original mixture. While this was disturbing, two additional observations mitigate this result. First, when identical ebperiments were repeated in the presence of liver protein from control rats, 9 was formed in only 0.5 f 0.2% yield (n = 3) and the recoveries of 10-12 were correspondingly increased. The basis for this effect of added protein (like the relevance of this experiment to the alkaline permethylation of true protein adducts as opposed to "model compounds") is not clear. The second observation is that, in all of the alkaline permethylations we have conducted on liver from BB-treated rats (over 15 in all), clear evidence for the presence of 10, 11, or 12 was never observed. We feel it extremely unlikely that bromine-containing protein-S adducts of 3 would undergo complete loss of Br to form 9 exclusively. Thus, we interpret the complete absence of 10-12 from the alkaline permethylation products to indicate that 9 arises via debromination during metabolism and/or binding, and not as an artifact of alkaline permethylation. As a final attempt to support the latter view several alkaline permethylation reactions were carried out in D20/NaOD with CD31. Figure 7 shows molecular ion clusters for 9 derived from the alkaline permethylation with deut,erated reagents either of PB/BB rat liver protein (panel a), of (2,5-dihydroxyphenyl)mercapturicacid in the presence of control rat liver protein (panel b), or of a
Slaughter and Hanzlik mixture of (3-, (4-, and (6-bromo-2,5-dihydroxyphenyl)mercapturic acid isomers in the absence of protein (panel c). A comparison of panels a and b clearly shows that compound 9 derived from PB/BB rat liver protein is predominantly, if not exclusively, in the d9 (tri-CD,) form expected from the alkaline permethylation of nonbrominated S-(dihydroxyphenyl) conjugates under these conditions. Similar molecular ion clusters were seen for protein-derived 13 and 17 (data not shown). On the other hand, loss of bromine during alkaline permethylation should result in the incorporation of one additional ring deuterium atom and, hence, a dIospecies as seen in panel c. Panel c was generated in the absence of protein because the level of mercapturic acid chemical debromination dropped to
