Chem. Res. Toxicol. 1994, 7, 177-184
177
Covalent Binding of Benzoquinone to Reduced Ribonuclease. Adduct Structures and Stoichiometry Robert P. Hanzlik,' Shawn P. Harriman, and M a r y
M. Frauenhoff
Department of Medicinal Chemistry, University of Kansas, Lawrence, Kansas 66045-2506 Received September 7,1993"
As a model for the reaction of chemically reactive quinone metabolites with cellular proteins in vivo, the reactions of benzoquinone (BQ; 1-64 mol/mol of protein) with bovine pancreatic ribonuclease A (RNase), reduced RNase, S4amidomethylated) reduced RNase, and reduced guanidinated RNase were investigated. The reaction stoichiometry and products were characterized by means of HPLC, UV-vis spectrophotometry, electrospray mass spectrometry, amino acid analysis, alkaline permethylation analysis, and measurement of the covalent binding of [ W I B Q to protein. Native RNase and S-(amidomethylated) reduced RNase show no reaction with BQ over 60 min a t pH 7.4-9.6,whereas reduced RNase, which has 8 free S H groups/mol, reacts rapidly with exactly 24 molecules of BQ, of which 8 become covalently bound to proteinS H groups while 16 are reduced to hydroquinone. Half of the latter is formed via BQ oxidation of the initially formed S-(2,5-dihydroxyphenyl)cysteinemoieties. Michael addition of a second protein nucleophile to each resulting S-(2,5-quinonyl)cysteinemoiety, followed by reoxidation of that addition product by BQ, generates the second group of 8 molecules of HQ and results in cross-linking. Reduced guanidinated RNase, in which most of the lysines are blocked by guanidination with O-methylisourea, also reacts rapidly with BQ, but only ca. 16 equiv are consumed; of these, 8 become covalently bound t o protein-SH groups while the others are reduced to HQ. Thus, even though the lysine residues in native RNase and S-(amidomethylated) reduced RNase do not react with BQ, they may react with (2,5-quinonyl)-S-proteinmoieties. These results indicate that, under toxicologically relevant conditions (Le., low concentrations, pH 7.4,short reaction times), BQ is a highly selective electrophile which reacts much faster with sulfhydryl groups than with any other protein nucleophiles. Covalent modification of proteins by small molecules is an exceedingly important process. From a physiological perspective events such as glycosylation, isoprenylation, and phosphorylation modulate the cellular function or activity of many proteins. From a biochemical perspective covalent modification of proteins by affinity labeling reagents is important for the identification of active site residues and for enzyme inhibition and drug design. In recent years the covalent modification of cellular proteins by chemically reactive metabolites has also received considerable attention (1-3). Such events can initiate allergic sensitization and are frequently correlated on a cellular level with the toxic effects of xenobiotic chemicals. Measurement of reactive metabolites covalently bound to readily accessible proteins such as albumin or hemoglobin also provides a means for assessing time-integrated human exposure to xenobiotics in the workplace (i.e., biomonitoring) (4). Bromobenzene,a simple aromatic compound with little chemical or pharmacological reactivity, undergoes biotransformation via chemically reactive intermediates, some of which react covalently with cellular proteins (5,6).Since the latter event is strongly correlated with cellular injury, bromobenzenehas become an important model compound for studying reactive metabolite-induced tissue injury, particularly in liver (7)but also in kidney (8)and lung (9). Brodie et al. first proposed that bromobenzene 3,4-oxide, the apparent precursor of the major water-soluble me-
* Address correspondence to this author. Phone: 913-864-3750; FAX: 913-864-5326. Abstract published in Aduance ACS Abstracts, February 1, 1994. 0893-228x/94/27Q7-0177$04.50/0
tabolites of bromobenzene,was also the reactive metabolite responsible for protein covalent binding (10). Later in vitro studies by Hesse et al. suggested that quinone metabolites of bromobenzene could also be responsible for ita metabolism-dependent protein covalent binding (11). The first protein adduct of a bromobenzene metabolite to be characterized structurally was S-@-bromophenyl)cysteine (121,which clearly derives from the alkylation of a protein sulfhydryl group by bromobenzene 3,4-oxide in a process analogous to mercapturic acid biosynthesis, except that the latter is catalyzed by glutathione transferase while the former is spontaneous. Surprisingly, however, S-@-bromophenyl)cysteine, along with minor amounts of its ortho and meta isomers, accounted for only 1-2 % of the total protein binding of bromobenzene to rat liver protein. More recently, analysis of covalent bromobenzene residues in rat liver protein by alkaline permethylation confirmed the above findings but also revealed a quantitatively much greater contribution from arylation of protein S-nucleophiles by quinonemetabolites of bromobenzene (13). Of the 10 or so different protein S-adducts identified after release by alkaline permethylation (14), ca. 60% of the total consisted of a single compound,2,5-dimethoxythioanisole, undoubtedlyformed from (2,5-dihydroxyphenyl)-S-protein moieties via baseinduced C(p)-S cleavage and methylation of the 0-and S-nucleophiles in the fragment. In principle, such proteinbound residues could arise by Michael addition of proteinSH groups to 1,4-benzoquinone, or by an additionelimination reaction between 2-bromo-1,4-benzoquinone and protein-SH groups. However, since 1,Cbenzoquinone 0 1994 American Chemical Society
178 Chem. Res. Toxicol., Vol. 7, No. 2, 1994
is a known metabolite of bromobenzene (16), the former route seems more likely. Many quinones are known to be cytotoxic (17,181. In some cases this toxicity is attributed to the occurrence of redox cycling, in which the quinone essentially catalyzes the transfer of reducing equivalents to molecular oxygen, leading to the formation of various reactive oxygen species capable of damaging cellular constituents, but such redox cycling is probably not important in the cases of benzoquinone or 2-bromo-1,4-benzoquinone(17, 19,201. Another general mechanism underlying quinone toxicity is the electrophilic arylation of protein nucleophiles (17,18, 21). In most cases the evidence for protein arylation by quinones is based on radiolabel binding studies and/or the ability of some quinones to impart permanent color to proteins; in very few cases has structural elucidation of quinone-protein adducts been provided (22). Because benzoquinoneis a common metabolite of a number of drugs and chemicals (15, 23,24), we undertook a study of its reaction with bovine pancreatic ribonuclease A (RNase)l and several of its well-characterized derivatives, as model proteins containing a representative array of amino acids (25). In this paper we report our findings on the stoichiometry of the reaction and the structures of the adducts produced.
Materials and Methods Instrumentation. GUMS analyses were performed using a Hewlett-Packard system [HP5890 series I1 gas chromatograph, HP5971A mass-selective detector, and HP59940A MS ChemStation (HP-UX series) controller] equipped with a 30-m fused silica capillary column (DB-5, 5% phenyl/methyl silicone, J & W Scientific); the temperature was programmed to rise at 2 OC min-1 from 60 to 110 OC and then at 10 OC min-l to 250 OC with a hold time of 5 min (14). HPLC analyseswere performed using a Shimadzu system (dual LC-6A pumps, SCL-6A controller, and an SPD-6A variablewavelength detector). HPLC method A utilized a C-18 column (Alltech Econosil, 10 pm, 4.6 X 250 mm) eluted with methanol/ water (2278) at 1mL/min. HPLC method B utilized a C-8column (Alltech Econosil, 10 wm, 4.6 X 250 mm) eluted with methanol/ water (2080) containing 0.1 % trifluoroacetic acid at 1mL/min. Amino acid analyses were performed by the University of Kansas Biochemical Research Service Laboratory using a Beckman automated amino acid analyzer with postcolumn derivatization by ninhydrin; System Gold software was used for data analysis. Protein samples were oxidized with performic acid (26) and hydrolyzed (6 N HC1,12 h, 110 OC, invacuo) prior to analysis. Control experiments showed that authentic N-acetyl-S-(2,5dihydroxypheny1)cysteine (16)did not yield cysteic acid or any other interfering peaks. Electrospray mass spectra were collected using a Sciex APII11 mass spectrometer (PE/Sciex, Thornhill, Ontario), equipped with a pneumatically assisted electrospray ion source. Samples were dissolved in 50% (v/v) aqueous methanol containing 2% acetic acid and solutionsinfused using a Harvard Model 22 syringe pump (Harvard Apparatus, South Natick, MA) at a rate of 2 pL/min into the ion spray needle, which was maintained at 5400 V. The orifice voltage was 80 V. A mass range of m/z 300-2400 was scanned at a rate of 21 s/scan. Successive spectra were summed until an acceptable signal/noise ratio was obtained (usually after 1-4 min). 1 Abbreviations: BQ,1,ri-benzoquinone; Cbz, benzyloxycarbonyl;DTT, dithiothreitol;ESMS,electroepraymass spectrometry;GdnHC1,pu8f’dine hydrochloride; GR, guanidmted ribonuclease; HQ, hydroqumone; HQMA,hydroquinonemercapturicacid;QMA,quinone mercapturic acid; rGR,reducedguanidinatd ribonuclease;rR, reduced ribonuclease;RNase, ribonuclease A; TNBS, trinitrobenzenesulfonicacid.
Hanzlik et al. Chemicals a n d Reagents. All operations were performed at room temperature unless otherwise indicated. Ribonuclease A (type IIA), dithiothreitol, 5,5’-dithiobis(2-nitrobenzoicacid) (Ellman’s reagent), 0-methylisourea sulfate, and picrylsulfonic acid (TNBS) were obtained from Sigma. Guanidine hydrochloride (GdnHC1) was Ultrapure grade from ICN. Methanol and acetonitrile were HPLC grade from Fisher Scientific. Hydroquinone, benzoquinone,2,4-dichloronitrobenene,p-nitrobenzyl alcohol, and iodoacetamide were from Aldrich. Water was purified through a Millipore MilliQ system before use. 2,5Dimethoxythioanisole, N-acetyl-S-(2,5-dihydroxyphenyl)cysteine [i.e., (2,5-dihydroxyphenyl)mercapturicacid], @-chloropheny1)mercapturicacid, and p-chlorothioanisolewere available from previous studies in our laboratory (16). [14ClBenzoquinonewas synthesized from [l4C1aniline as follows. A solution of NazCr20,.2H~O (210 mg, 0.74 mmol) in 5 mL of water was added to a stirred solution of [“Claniline (0.62 mmol, 1.35 mCi) in 15 mL of 4 M H2SO4 and stirred for 5 min, after which 2-propanol (1mL) was added. After 30-min stirring zinc powder (3 g) was added and stirring continued overnight. The reaction product ([14C]hydroquinone) was extracted with ethyl acetate, the extracts were dried over anhydrous MgSO4 and concentrated, and the resulting light brown solid was purified by silica flash chromatographyeluting with 20-30 % ethyl acetate in hexane: yield 28 mg (41%), specific activity 1.74 Ci/mol, radiochemical purity >97% as judged by HPLC. [14C]Hydroquinone was stored as a 60.5 mM solution in methanol at -15 OC. [l4C1Benzoquinonewas prepared as needed by oxidationof [14C]hydroquinone. A 0.75-mL aliquot of the latter was dried by rotary evaporation in a screw-cap culture tube and the residue redissolved in 2% H2SO4. A small magnetic stir bar, NaClOa (6.5mg), and V205 (1mg) were added, after which the tube was capped and stirred for 3 h in the dark. Extraction with methylenechloride and evaporation yielded crude [l4C1benzoquinone,which was dissolved in acetonitrile and purified by HPLC method A yield 2 mg (40%),radiochemical purity >98% as judged by HPLC (225 nm). Preparation of Reduced RNase. Reduced RNase (rR) was prepared by dissolving RNase A (10 mg) and dithiothreitol (DTT, 18mg, 20-fold excess per disulfidemoiety)in 1mL of 4 M GdnHCl in 84 mM sodium phosphate buffer (pH 7.4 before addition of GdnHC1). The solution was then purged with nitrogen for 5 min, sealed, and allowed to stand in the dark for 4 h. The GdnHCl and excess DTT were removed on a 1.5 X 17 cm column of Sephadex G-25 equilibrated and eluted with 0.1 M acetic acid (1-2 mL/min). After 5 mL had run through the column, 1-mL fractions were collected. The absorbance of each fraction at 280 nm was determined, and the major protein-containing fractions were pooled and generallyused immediately,although they could be lyophilizedand reconstituted after storage with no loss of free sulfhydryl groups. Preparation of Reduced Guanidinated RNase (rGR). RNase (5 mg) was guanidinated by treatment with 0-methylisourea sulfate (0.5 mmol in 1.0 mL of distilled water adjusted to pH 10.5 by titration with 4 M NaOH) for 48 h at room temperature (27). Excess 0-methylisourea was removed using a Sephadex G-25 column as described above. Guanidinated RNase (GR) prepared this way showed 0.9 free NH2 group/mol as determined with picrylsulfonic acid (28)using an extinction coefficient of 12000 M-l cm-l (28-30). In a parallel control experiment, native RNase, which has 11 free NH2 groups (10 lysine c-NH2 plus one a-NHz), showed 7.1 free amine groups/ mol. Reduced guanidinated RNase (rGR) was prepared as described above for reduction of native RNase. Determination of Protein Concentration. Protein concentrations were measured photometrically at 280 nm. Native RNase is reported to have a molar absorptivity of 9800 M-1 cm-l at pH 6.5 (31). For native RNAse (as supplied by Sigma) we determined a molar absorptivity of 9470 M-l cm-l in water. For native RNase obtained from Sigma, desalted on a (3-25 column, lyophilized, weighted, and then redissolved in water, we deter
Chem. Res. Toxicol., Vol. 7, No. 2, 1994 179
Protein Binding of Benzoquinone Scheme 1
Step 2
OH
SteD 4
rmas($)x
p hslHQL
bm 0
0
0 ( Ap = alkaline permethylation )
0
mined a molar absorptivity (280 nm) of 9380 M-l cm-1; we used the latter value for solutions of both native and reduced RNAse. Determination of Protein Sulfhydryl Groups. The sulfhydryl content of rR was determined using Ellman's reagent (32).Aliquots of protein solution (0.9 mL) were added to 0.1 mL of Ellman's reagent (10 mM in sodium phosphate buffer, 0.1 mM, pH 8.0), and after 10 min A412was measured vs a proteinfree blank. The concentration of sulfhydryl groups was calculated using a molar extinction coefficient of 13 600 M-l cm-1 (32). Amidomethylationof Reduced RNase. Native RNase (20 mg) and DTT (36 mg) were dissolved in 1mL of 84 mM sodium phosphate buffer containing 4 M GdnHCl (pH 7.4 prior to addition of GdnHCl, pH 6.3 after), purged with nitrogen for 5 min, and allowed to stand in the dark for 4 h. This solution was then cooled to 0-3 "C, iodoacetamide (0.111 g) was added, and the solution was stirred for 60 min in the dark a t 0-3 "C followed by passage over Sephadex G-25 and collection of the protein fraction as described above. Reaction of Benzoquinonewith Proteins. In general 5-40pL aliquots of freshly prepared solutions of BQ in MeCN (3.65 or 36.5 mM) were added via a microliter syringe to nitrogenpurged aqueous solutions of protein (18-72 nmol in a volume of 1.0 mL) to attain the desired BQ/protein molar ratios. In all experiments the total volume of BQ solution added was less than 5 % than of the aqueous protein solution; further details are given in the table footnotes and figure legends. Native RNase is quite soluble in simple aqueous buffers, but rR is not (unless the pH is kept below ca. 5.0). To solubilize rR in pH 7.4 phosphate buffer, GdnHCl was added to 2 mM; this caused the pH to drop to 6.3 but did not impede the reaction with BQ. In some later experiments solutions of rR in 0.1 M HOAc (obtained directly from the G-25 column used to remove DTT and GdnHCl after reduction) were treated directly with MeCN solutions of BQ. Under these conditions the reaction of BQ with rR was also rapid, but as discussedunder Results and Discussion,the overallreaction course differed in some minor respects from that in GdnHCl at pH 6.3. Determination of Benzoquinone and Hydroquinone. After addition of BQ to rR, HQ formed and unreacted BQ remaining were determined by HPLC method B after adding p-nitrobenzyl alcohol as an internal standard; detection was by absorbance a t 227 nm. Chromatograms were recorded and integrated using a Rainin Dynamax system. The order of elution was HQ (2.5 min), BQ (3.5 min), and p-nitrobenzyl alcohol standard (8.6 min). The concentrations of HQ and BQ were determined from calibration plots of peak area ratios vs mole ratios of known mixtures. Alkaline Permethylationof BQ-TreatedProtein. Alkaline permethylation is a process that converts S-(2,5-dihydroxypheny1)cysteine moieties in proteins or mercapturic acids into 2,5-dimethoxythioanisole(compound 9 in Scheme 1)which can
10
20 30 40 mol BQ added I mol rR
50
60
70
Figure 1. Covalent binding of benzoquinone to reduced RNase. Varying aliquots of BQ in MeCN (3.65 or 36.5 mM; up to 32 pL total) were added to constant aliquots of rR (18.3 nmol) in 1.0 mL of nitrogen-purged buffer (42 mM sodium phosphate, pH 6.3, containing 2 M GdnHCl), and each reaction mixture was kept sealed under a Nz atmosphere in the dark. After 60 min the reaction mixture was extracted 5 times with 2 volumes of ethyl acetate, after which a 0.2-mL aliquot of the remaining aqueous (protein-containing)phase was removed for scintillation counting. The number of BQ equivalents covalently bound per mole of protein was determined from the specific activity of the starting BQ and the initial concentration of protein. The points are experimental; the lines are drawn to show a slope of 1from 0 to 8 mol of BQ added and a limiting value of 8 mol bound/mol of protein. be determined quantitatively by GC/MS; the entire procedure has been described in detail (14). In the work described herein, was a known quantity of N-acetyl-S-(p-chloropheny1)cysteine added to each protein sample prior to alkaline permethylation. This internal standard is converted top-chlorothioanisole whose quantitation provides an in situ measure of the efficiency of the alkaline permethyaltion process in that particular sample. After the thioanisole reaction products were extracted for GUMS analysis,a known quantity of 2,4-dichloronitrobenzenewas added as an internal standard for absolute quantitation of both p-chlorothioanisole and 2,5-dimethoxythioanisole.
Results and Discussion Reaction of BQ with Native, Reduced, and Amidomethylated Reduced RNase. Native RNase contains 4 disulfide bonds and presents one terminal amine group plus the side chains from 10 lysines, 4 histidines, 6 tyrosines, and 4 methionines as potential targets for electrophiles. Despite this array of nucleophilic sites, exposure of native RNase to 64 mol of [l4CIBQ/mol of protein for 1h resulted in no significant incorporation of radioactivity into the protein. In contrast, when reduced RNase (containing 8 mol of sulfhydryl groups/mol) was treated with up to 8 mol of [W]BQ/mol of protein, virtually all of the radioactivity became covalently bound within 10-15 min, but further addition of BQ resulted in very little additional covalent binding (Figure 1). In another set of experiments native, reduced, and S-amidomethylated reduced RNases were each dissolved in buffers at pH 7.4,8.1,9.0, or 9.6, and 3 mol of BQ/mol of protein was added to each reaction. After 60 min the concentration of unreacted BQ was determined by HPLC. In the reactions with rR no BQ remained free in solution while in the other two cases there was no decrease in BQ concentration, even at the higher pH values where nucleophilic side chains such as histidine and lysine would be substantially deprotonated; in no case was any hydroquinone observed. Lysine side chains in proteins are known t o be reactive toward a,@-unsaturatedaldehydes and reactive carbonyl
180 Chem. Res. Toxicol., Vol. 7,No. 2, 1994
'0
10
20
30 40 50 mol BQ added / mol rR
60
Hanzlik et al.
70
Figure 2. Effect of benzoquinone on free sulfhydryl groups in reduced RNase. As described for Figure 1, varying aliquots of [WIBQ in MeCN were added to rR in 1.0 mL of nitrogen-purged buffer, and each reaction mixture was kept sealed under a Nz atmosphere in the dark for 60 min, after which a 0.9-mL aliquot was removed and combined with 10 mL of Ellman's reagent (10 mM in 0.1 M phosphate buffer, pH 8.0) and its A412 determined after 10 min. The number (mol) of SH groups/mol of protein was calculated after subtracting the absorbance of protein-free blanks. The line is drawn to have a slope of -1; the points are experimental.
compounds such as malondialdehyde ((33), and 2,5hexanedione (35). Using SYBYL software, we examined the crystal structure of RNase (36) using coordinates obtained from the Brookhaven Protein Data Bank (file 3RN3) (37) and ascertained that nearly all of the lysines in RNase are on the surface or in the active site. Furthermore, these residues are known to react readily with electrophilic reagents such as O-methylisourea (27) or methyl acetimidate (29). Hence their apparent lack of reactivity toward BQ was investigated more directly by combining a 9-fold excess of Na-Cbz-lysine with 0.1 mM BQ in buffered GdnHCl solution and observing the BQ chromophore spectrophotometrically, but no change was observed over a 2-h period. In another experiment NaCbz-lysine was added to a mixture of BQ and quinonemercapturic acid (QMA; generated at pH 7.4 by adding BQ to HQMA as shown in eq l),but after 1h HPLC analysis
pJL :H
QXL -= H OIQMA)
H
BQ
HQ
a 250
300
350
450
400
Wavelength (nm)
Figure 3. Spectrophotometric titration of reduced RNase with benzoquinone. Reduced RNase (20 pM in GdnHCl buffer) in a quartz cuvette was titrated with BQ in MeCN as described for Figure 1except that the reaction was conducted under air. After each addition the solution was mixed and the spectrum recorded immediately; in some cases repeat scans were made a t later times, but these spectra were the same as those obtained immediately after addition of BQ. A protein-free base-line scan is shown in curve a, and a spectrum or rR alone is shown in curve b. Under these conditions h , for BQ and HQ are 248 and 291 nm, respectively. Curves e l represent spectra obtained after addition of 1, 3,6,8, 10, 12, 15,18, 24, and 64 mol of BQ/mol of protein, respectively. The inset shows a plot of AAm8 vs the equivalents of BQ added. 35. '
30.
0 mol HQ formed / mol rR mol BO remaining
25. 20.
(1)
(QMA)
indicated that no reaction had taken place. These results indicate that on this time scale neither BQ nor QMA is reactive toward lysine side chains. Figure 2 shows that titration of rR with BQ leads to a stoichiometric loss of free sulfhydryl groups detectable with Ellman's reagent. Concomitantly a new chromophore (Ama = 305 nm) which is not extractable into organic solvents appears in solutions of rR titrated with
