Identification of Three Protein Targets for Reactive Metabolites of

To identify protein targets for bromobenzene metabolites in cytosol, we incubated ... cytosolic proteins followed by phosphor imaging analysis reveale...
2 downloads 0 Views 177KB Size
Download PDF
1326

Chem. Res. Toxicol. 2000, 13, 1326-1335

Identification of Three Protein Targets for Reactive Metabolites of Bromobenzene in Rat Liver Cytosol Yakov M. Koen, Todd D. Williams, and Robert P. Hanzlik* Department of Medicinal Chemistry, University of Kansas, Lawrence, Kansas 66045-7582 Received August 2, 2000

The hepatotoxicity of bromobenzene and many other simple organic chemicals is believed to be associated with covalent binding of chemically reactive metabolites to cellular proteins. Recently, a rat liver microsomal esterase was shown to be targeted by bromobenzene metabolites formed in vitro [Rombach, E. M., and Hanzlik, R. P. (1998) Chem. Res. Toxicol. 11, 178-184]. To identify protein targets for bromobenzene metabolites in cytosol, we incubated liver microsomes and glutathione-depleted liver cytosol from phenobarbital-treated rats with [14C]bromobenzene in vitro. In a separate experiment, we intraperitoneally injected a hepatotoxic dose of [14C]bromobenzene to phenobarbital-treated rats. The cytosol fractions from both experiments were recovered and analyzed for protein-bound radioactivity. Under the conditions that were used, 2.6 and 3.9 nmolar equiv of bromobenzene/mg of cytosolic protein was bound in vitro and in vivo, respectively. Denaturing polyacrylamide gel electrophoresis of these cytosolic proteins followed by phosphor imaging analysis revealed several radiolabeled protein bands over a broad molecular mass range, the patterns observed in vitro and in vivo being generally similar to each other. Cytosolic proteins labeled in vitro were separated by ion exchange chromatography and electrophoresis, and three major radioactive bands with estimated molecular masses of ca. 14, 25, and 30 kDa were in-gel digested with trypsin, followed by on-line HPLC electrospray ionization mass spectrometry of the resulting peptide mixtures. For the three protein bands, the observed peptide masses were found to match the predicted tryptic fragments of liver fatty acid binding protein, glutathione transferase subunit A1, and carbonic anhydrase isoform III, respectively, with 83, 45, and 59% coverage of the corresponding complete sequences. The possible relationship of the adduction of these proteins to the toxicological outcome is discussed.

Introduction Liver injury caused by drugs and chemicals can take a number of different forms, including some which involve extensive cell death (1). For example, some simple aromatic compounds such as acetaminophen [N-acetylp-aminophenol (APAP)]1 and bromobenzene cause a classical pattern of liver injury known as centrilobular necrosis. Tissue injury caused by these compounds is strongly correlated with the rate and extent of their biotransformation and with the covalent binding of reactive metabolites to cellular proteins. Because this covalent binding correlates quantitatively, temporally, and histologically with the degree of cellular injury, and because no injury occurs in the absence of covalent binding, it is widely believed to have a causative role in the process (2, 3). Our laboratory has been interested in elucidating the chemical details of the process by which hepatocytes are injured by bromobenzene metabolites. Unlike acetami* To whom correspondence should be addressed: Department of Medicinal Chemistry, University of Kansas, Lawrence, KS 66045-7582. Fax: (785) 864-5326. E-mail: [email protected]. 1 Abbreviations: APAP, acetaminophen; CA, carbonic anhydrase; ESI, electrospray ionization; GST, glutathione transferase; IEF, isoelectric focusing; iLBP, intracellular lipid binding protein; LFABP, liver fatty acid binding protein; MALDI-TOF MS, matrix-assisted laser desorption ionization time-of-flight mass spectrometry; NAPQI, Nacetyl-p-benzoquinonimine; PVDF, polyvinylidene difluoride; SDS, sodium dodecyl sulfate; TCA, trichloroacetic acid.

nophen, which appears to give rise to a single major reactive metabolite (NAPQI), bromobenzene gives rise to a number of chemically reactive (and potentially toxic) metabolites, including bromobenzene 2,3-oxide, bromobenzene 3,4-oxide, and several quinones and bromoquinones derived via further oxidative metabolism of initially formed bromophenol metabolites; both epoxide and quinone metabolites of bromobenzene participate in protein covalent binding in vitro and in vivo (4, 5). However, since p-bromophenol, which is a major metabolite of bromobenzene in vivo, also undergoes metabolic activation and covalent binding but does not cause hepatotoxicity, it is apparent that not all covalent binding events are equally harmful to the cell (6). A similar situation exists with APAP in that its meta isomer (N-acetyl-m-aminophenol) is known to undergo extensive metabolic activation and covalent binding without causing detectable hepatotoxicity (7-10). Furthermore, ample evidence exists, indicating that covalent binding is not a completely random process involving all proteins to an equal degree; rather, selectivity is clearly apparent in that some minor proteins become more extensively labeled than would be predicted merely by their relative abundance (3, 11-16). These realizations have shifted the focus of research away from questions about the relative toxicity of one metabolite versus another to questions about the identity of the proteins which become modified, the role of these target proteins in the overall biology of the cell under

10.1021/tx000165l CCC: $19.00 © 2000 American Chemical Society Published on Web 11/10/2000

Cytosolic Targets of Bromobenzene Metabolites

normal circumstances, and the consequences of covalent modification on the ability of these proteins to perform their normal cellular functions. Over the past few years, more than 20 protein targets for APAP metabolites have been identified (12, 17, 18). Many of these proteins are enzymes of intermediary metabolism, and the activities of some of these enzymes are lower in liver tissue from animals treated with toxic doses of APAP than in tissues from control animals. However, it is frequently difficult in these cases to discern whether the lowered activities are a cause, or an effect, of the injury to the liver. Another important question which arises is whether injury stemming from covalent binding actually is derived from a few highly specific molecular lesions to critical targets which disrupt specific vital cellular functions or from a nonspecific combination of many different biochemical insults to many different target proteins. A corrolary to this is the question of whether centrilobular necrosis caused by different agents (e.g., APAP vs bromobenzene) may involve some common protein targets. Until now, only two protein targets of bromobenzene metabolites have been identified, and both are isozymes of an esterase previously identified as a target protein for reactive metabolite of the fluorinated inhalation anesthetic halothane (14). In this paper, we present evidence showing that under both in vivo and in vitro conditions, [14C]bromobenzene metabolites bind to substantially the same set of rat liver cytosolic proteins. In addition, we show that three of the major cytosolic target proteins are a fatty acid binding protein, an isozyme of carbonic anhydrase, and a subunit of glutathione transferase. Finally, we consider the possible relevance of these targets to the cytotoxicity of chemically reactive protein-modifying metabolites.

Experimental Procedures Materials. [14C]Bromobenzene (5.17 Ci/mol) was prepared in our laboratory from [14C]aniline by a sequence of bromination, diazotization, and reduction as described previously (19) and was stored as a solution in pentane at -20 °C. Immediately before being used, pentane was removed by fractional distillation, and bromobenzene was dissolved in acetonitrile (for in vitro incubations) or in corn oil (for ip injections). Sequencing grade trypsin was obtained from Boehringer Mannheim (Indianapolis, IN). TPCK-treated trypsin, type XIII, and dithiothreitol were obtained from Sigma (St. Louis, MO). 4-Vinylpyridine was obtained from Sigma and was distilled and stored under N2 at -20 °C until it was used. Electrophoresis supplies were obtained from Bio-Rad Laboratories (Hercules, CA); ion exchange chromatographic supports were from Pharmacia Biotech (Uppsala, Sweden), and HPLC grade solvents were from Fisher (Fair Lawn, NJ). Mega-Pure water (resistivity of >17 MΩ/cm) was used for the preparation of all buffers. Treatment of Animals and Preparation of Subcellular Fractions. Male Sprague-Dawley rats (150-180 g, Charles River Laboratories, Wilmington, MA) were housed in a temperature- and humidity-controlled room with a 12 h light/dark cycle and ad libitum access to food and water. After being acclimated for g3 days, animals were given three daily ip injections of sodium phenobarbital (80 mg/kg) in 0.9% saline (1.0 mL/kg). After the third dose, food was withheld overnight, and the next morning the rats were killed (for preparation of cytosol for in vitro binding experiments) or treated with [14C]bromobenzene (2 mmol/kg ip) in corn oil (1 mL/kg) and killed 4 h later by decapitation under CO2 narcosis. Livers were removed and homogenized in ice-cold 50 mM potassium phosphate buffer (pH 7.4) containing 0.15 M KCl, 5 mM EDTA, and 1 mM phenylmethanesulfonyl fluoride (4 mL/g of tissue). The homo-

Chem. Res. Toxicol., Vol. 13, No. 12, 2000 1327 genate was successively centrifuged at 3000g (10 min), 12000g (20 min), and 100000g (60 min). The final 100000g supernatant (cytosol) and pellet (microsomes) were collected and processed separately as follows. The cytosol from rats not treated with [14C]bromobenzene was clarified by recentrifugation at 100000g and stored at -70 °C until it was used for incubations. Cytosol from animals treated with [14C]bromobenzene (hereafter called cytosol 2) was clarified identically and dialyzed against 20 mM potassium phosphate (pH 7.4) containing 0.5 mM dithiothreitol (4 × 60 volumes), to remove reversibly bound radioactivity, and stored at -70 °C. The microsomal fraction was washed by homogenization in 0.1 M sodium pyrophosphate buffer (pH 8.2, 1.3 mL/g of tissue), followed by centrifugation at 100000g. The resulting pellet was resuspended in homogenization buffer, followed by centrifugation at 100000g. The final microsomal pellet was resuspended in 100 mM potassium phosphate buffer (pH 7.4) containing 0.5 mM EDTA and 20% glycerol and stored at -70 °C. Incubations. Incubations were carried out in 25 mL glassstoppered Erlenmeyer flasks for 60 min at 37 °C in a shaking water bath. To facilitate in vitro binding of bromobenzene metabolites to cytosolic proteins, nonprotein thiols (GSH) were removed from cytosol by gel filtration of aliquots (2 mL, 45 mg of protein) on a 1.5 cm × 12 cm Sephadex G25 column, immediately before the incubation. The removal of GSH was monitored using Elman’s reagent. The incubation mixture contained 100 mM potassium phosphate buffer (pH 7.4), 30 mg of GSH-depleted cytosolic protein, 30 mg of microsomal protein, 15 µmol of [14C]bromobenzene (delivered in 100 µL of acetonitrile), and a NADPH-generating system consisting of 30 µmol of NADP (added as three 10 µmol portions at 20 min intervals), 300 µmol of glucose 6-phosphate, and 10 units of glucose-6phosphate dehydrogenase, all in a final volume of 10 mL. Incubations were terminated by placing the incubation flask in an ice bath, and the cytosol was separated from microsomes by centrifugation at 100000g for 60 min. The recovered cytosol was supplemented with 1 mM dithiothreitol, and reversibly bound radioactivity was removed by dialysis against 3 × 40 volumes of 20 mM potassium phosphate (pH 7.4) containing 0.5 mM dithiothreitol. The removal of radioactivity was monitored by scintillation counting of aliquots of dialysates. The dialyzed cytosol (hereafter called cytosol 1) was stored at -70 °C until it was used for protein separation. Protein Separation by Ion Exchange Chromatography. An aliquot of dialyzed cytosol 1 (8 mL) was loaded onto a 1.5 cm × 12 cm DEAE-Sephacel column equilibrated with 20 mM potassium phosphate (pH 7.4) containing 0.5 mM DTT (buffer A). The column was washed with buffer A, and 3 mL eluate fractions were collected until the radioactivity of the eluate was less than twice the background level. Then the column was successively washed with buffer A, containing 0.1, 0.2, 0.3, and 0.5 M NaCl, and 5 mL eluate fractions were collected. Protein elution was monitored by measuring A280, and the amount of radioactivity of eluate fractions was determined by scintillation counting. Individual fractions were analyzed by SDS-PAGE (as described below) using silver staining, and those exhibiting identical protein patterns and appreciable amounts of radioactivity were combined into four pools, DE1-DE4 (fractions 5-11, 46 and 47, 49, and 60 and 61, respectively). Radiolabeled proteins in the DE1 pool (14 mL) were further separated by cation exchange chromatography on a 1.5 cm × 15 cm CM Sepharose column equilibrated with 20 mM potassium phosphate (pH 6.6) containing 0.5 mM DTT (buffer B). Proteins were eluted by washing the column successively with buffer B, containing 0, 0.05, 0.1, 0.15, 0.2, 0.3, and 0.5 M NaCl. Individual eluate fractions (5 mL) were combined into five pools, CM1CM5 (fractions 5-8, 9, 14-16, 26 and 27, and 29 and 30, respectively). Analysis of Protein Fractions. To determine the amount of radioactivity covalently bound to protein, aliquots of the subcellular fractions were precipitated with 10% TCA, and protein precipitates were successively washed with acetone,

1328

Chem. Res. Toxicol., Vol. 13, No. 12, 2000

methanol and water (80:20 v/v, three times), acetone, and diethyl ether, dried (rotovap), dissolved in 1 M NaOH, and neutralized with 1 M HCl, after which the amounts of radioactivity was measured by scintillation counting and protein were determined by the Bradford assay using a standard kit (BioRad). To locate radiolabeled proteins, aliquots of cytosol or column eluate fractions were separated by SDS-PAGE or twodimensional electrophoresis, followed by transblotting to PVDF membranes as described below. Blots were subjected to phosphor imaging analysis for 14C using a Molecular Dynamics storage phosphor screen, scanning unit, and software. MALDI-TOF MS Analysis of Intact Proteins. To analyze the protein composition of eluate pool CM1, protein was precipitated with 10% TCA, washed with acetone, and subjected to MALDI-TOF mass spectrometry (Biotechnology Core Facility, Kansas State University, Manhattan, KS). Samples were combined with sinapinic acid (matrix) and renin substrate tetradecapeptide (internal standard for mass calibration, MH+ ) 1760.0 Da) and analyzed with a Finnigan Lasermat instrument of linear configuration. Data were smoothed and centroided using LASERMAT peak detection software (five shots per spectrum). Electrophoresis. All electrophoretic separations were performed using a Mini-PROTEAN II cell equipped with a 1000/ 500 Power Supply (Bio-Rad Laboratories). Proteins in dialyzed cytosols 1 and 2 or in column eluate fractions were separated by SDS-PAGE according to Laemmli et al. (20) or by TricineSDS-PAGE according to the method of Scha¨gger and von Jagow (21). Proteins were precipitated with acetone, air-dried, and redissolved in appropriate electrophoresis sample buffer prior to application on the gel. Precast 4 to 20% gradient minigels (Bio-Rad) or laboratory-made nongradient separating gels with acrylamide concentrations in the range of 7.5-16.5% (3% Bis) and 4% stacking gel were used. For analytical purposes, the gels were run at 150-200 V and 60 mA/gel for 45-60 min (SDSPAGE) or at 90 V for 3-4 h (Tricine-SDS-PAGE), after which the proteins were visualized using either a silver staining kit (Bio-Rad) or 0.125% Coomassie brilliant blue R250 in a 50% methanol/10% acetic acid mixture or electrophoretically transferred to PVDF Trans-blot membranes (0.2 µm; Bio-Rad), followed by staining the blots with 0.4% Coomassie brilliant blue R250 in an 80% methanol/5% acetic acid mixture. Corresponding solvent mixtures were used for destaining gels and blots. For semipreparative electrophoretic separations used in digestion experiments, running conditions were optimized for the best separation of one particular protein band of interest at a time, and progress of the run was monitored by migration of prestained molecular mass standards (Bio-Rad). The run was continued until the band of interest was expected to have reached the lowest quarter of the gel. The gels were stained with 0.2% Coomassie in a 20% methanol/0.5% acetic acid mixture for 30 min and destained with 30% methanol for 5-7 min or until the protein bands appeared. Two-dimensional gel electrophoresis was performed using tube minigels (0.1 cm × 5.4 cm) with nonlinear pH 3 to 10 gradients for isoelectric focusing. The pH gradients were formed using mixtures of ampholytes (Bio-lyte 3/10 and 5/7; Bio-Rad), and runs were performed according to the manufacturer’s recommendations. SDS-PAGE or Tricine-SDS-PAGE was used for separations in the second dimension. In-Gel Digestion. Aliquots of eluate pool CM1 or CM3 (30 µg of protein each) were separated by SDS-PAGE (separation of the 25, 30, and 50 kDa radiolabeled proteins) or Tricine-SDSPAGE (separation of the 14 kDa protein) as described above, and the Coomassie-stained protein bands of interest from two to three replicate gels were excised and in-gel digested with trypsin, according to the method of Rosenfeld et al. (22) as modified by Shevchenko et al. (23) and Moritz et al. (24). Briefly, the excised gel pieces were washed twice with a 200 mM NH4HCO3/50% CH3CN mixture for 30 min at 30 °C and dehydrated with acetonitrile. The proteins were then reduced with 10 mM DTT in 100 mM NH4HCO3 (56 °C, 60 min), followed

Koen et al. by alkylation of free sulfhydryls with 2% 4-vinylpyridine in the same buffer (45 min at room temperature in the dark). After alkylation, the gel pieces were washed with 100 mM NH4HCO3, dehydrated with CH3CN (the last two steps were repeated), and vacuum- or air-dried. Then 10-20 µL of 100 mM NH4HCO3, containing 5 mM CaCl2 and 12 ng of trypsin/µL, was added; the gels were incubated on ice until they were completely rehydrated, after which a small excess of the same buffer containing no trypsin was added, and incubation was continued at 37 °C overnight. The incubations were terminated by placing the incubation tubes on ice, and the resulting peptide mixtures were extracted from the gels with 60% CH3CN and 0.1% TFA, vacuum-dried, and stored at -20 °C until they were analyzed. As a control, protein-free gel pieces of approximately the same size were processed identically. HPLC/Mass Spectrometry of Tryptic Digests. The peptide mixtures were separated on a 0.32 mm × 50 mm microbore Zorbax C18 column (300 Å, 3.5 µm) using a Micro-Tech Scientific chromatograph operating at a flow rate of 10 µL/min. Buffer A consisted of 2% MeOH, 98% ddH2O, and 0.25% formic acid, and buffer B was 98% MeOH, 2% H2O, and 0.25% formic acid. The segmented linear gradient was from 0 to 20% buffer B over 5 min followed by 20 to 80% buffer B over 60 min. The column effluent was monitored at 214 nm with a UV/vis 200 (Linear Scientific) detector (250 nL cell, 2 mm path) and passed into the electrospray ionization source. ESI mass spectra were acquired on an AUTOSPEC-Q sector instrument of EBEqQ configuration (VG Analytical, Manchester, U.K.) equipped with the Mark III ESI source. This version has the “pepper pot” counter electrode and hexapole transfer optics. The instrument was operated at an acceleration potential of 4 kV with the ESI needle at 7.5 kV and the counter electrode at 5 kV; the remaining lens voltages were optimized for maximum sensitivity of the peptide gramicidin S. The instrument was tuned to a resolving power of 1800 (10% valley) and scanned from 300 to 2500 amu at 10 s/decade. Magnet scan calibration was performed with a CsI solution. Database Searching. The measured peptide masses were compared to masses of theoretical tryptic fragments of known proteins by searching the NCBInr, SWISS-PROT, and Genpept data banks using the program MS-Fit (available at http:// prospector.ucsf.edu). For each protein, many searches were conducted in a reiterative manner using varied search parameters. On the basis of the SDS-PAGE and MALDI-TOF MS results (see below), the molecular mass intervals entered for the 14, 25, and 30 kDa protein bands were 14-15, 25-26, and 29-30 kDa, respectively. For each of these proteins, however, we also used a very wide molecular mass interval of 0-100 kDa for searches, with no restriction with respect to a source species, but these searches offered no new good matches. In all searches, S-pyridylethylation of cysteine residues and various modifications, including N-terminal acetylation, were considered. Initial searches were performed with relatively high values for the peptide mass tolerance (up to 0.3%) and the number of missed cleavages allowed (up to six), and with a small value for the minimum number of peptides required to match (three peptides). This “soft” approach was aimed at preventing us from missing possible candidates and sometimes resulted in large series of proteins whose theoretical tryptic fragment masses matched experimental peptide masses. Thus, for example, increasing the minimum number of peptide matches required from three to six dramatically reduced the number of suggested matches that were found (e.g., from 28 to 3 in the case of the 25 kDa protein; see below). Final searches of the complete databases were then performed with allowed mass tolerances in the range of 0.03-0.05% and with not less than six peptides required to match, and the highest-scoring matches found were considered further. After obtaining a database-derived list of expected tryptic fragments of tentatively identified protein candidates (including those peptides containing up to six uncleaved sites), we went back to the mass spectra to search for possible additional mass matches.

Cytosolic Targets of Bromobenzene Metabolites

Figure 1. SDS-PAGE and phosphor imaging analysis of cytosols 1 (in vitro) and 2 (in vivo). Aliquots of both samples containing equivalent amounts of protein (50 µg) were separated by a 12% SDS-PAGE followed by electroblotting to a PVDF membrane. The membrane was stained with Coomassie R250 and exposed to a Molecular Dynamics phosphor imaging plate. Panel A shows the protein patterns on the PVDF membrane and panel B the results of phosphor imaging of the same membrane: lane 1, cytosol 1 (in vitro); lane 2, cytosol 2 (in vivo); and S, molecular mass standards.

Chem. Res. Toxicol., Vol. 13, No. 12, 2000 1329

Figure 2. Separation of in vitro 14C-labeled cytosolic proteins (cytosol 1) by DEAE-Sephacel anion exchange chromatography: radioactivity (disintegrations per minute per 200 µL) and protein (A280) elution profiles. A 1.5 cm × 12 cm column was loaded with 8 mL of dialyzed cytosol containing 4.9 mg/mL protein with a specific activity of 0.013 µCi/mg. The column was washed with 20 mM potassium phosphate (pH 7.4) containing increasing concentrations of NaCl (shown by arrows); 3 mL fractions were collected before the NaCl gradient started and 5 mL fractions thereafter. Fractions 5-11, 46 and 47, 49, and 60 and 61 were combined to form pools DE1-DE4, respectively.

Results and Discussion In Vitro and in Vivo Covalent Binding of Bromobenzene Metabolites to Cytosolic Proteins. The cytosolic proteins labeled in vitro (cytosol 1) and in vivo (cytosol 2) were found to contain 2.6 and 3.9 nmolar equiv of bromobenzene/mg of protein, respectively. To detect radiolabeled proteins, both dialyzed cytosolic fractions were analyzed by SDS-PAGE followed by transblotting and phosphor imaging. Representative results are shown in Figure 1. Several radioactive protein bands with variable labeling densities are seen over a wide range of molecular masses. Although the protein patterns of cytosol 1 versus cytosol 2 appear quite similar to each other and to those of their respective untreated control cytosols (data not shown), the relative label density of individual bands varies noticeably between the two preparations. Thus, in cytosol 1 there are two major radioactive bands with estimated molecular masses of ∼15 and ∼65 kDa, respectively, and a few less-abundant radioactive bands at ∼25-30 and ∼50 kDa. In contrast, in cytosol 2 the most abundant radioactive bands are seen near 15, 30, and 50 kDa while the 65 kDa band contains much less radioactivity than cytosol 1. It is noteworthy, however, that in both cytosol fractions protein labeling is relatively selective, and that for the most part the major radiolabeled bands have the same apparent molecular masses despite apparent differences in their relative abundances and/or labeling densities. Fractionation and Analysis of Cytosolic Proteins Adducted in Vitro. The in vitro-radiolabeled cytosol was fractionated by DEAE-Sephacel anion exchange chromatography using low ionic-strength potassium phosphate buffer (20 mM) followed by a discontinuous NaCl gradient for elution (Figure 2). Two major and four minor protein peaks can be seen on the UV chromatogram. A majority of the radioactivity is associated with the first major protein peak that eluted with flow-through equilibration buffer. Two additional radioactive peaks coeluted with small protein peaks in 0.2 and 0.3 M NaCl,

Figure 3. Analysis of proteins in pools DE1-DE4 by SDSPAGE and phosphor imaging. Aliquots of DE1-DE4 (∼1000 dpm each) were precipitated with acetone, and the proteins were separated by SDS-PAGE (4 to 20% gradient) followed by electroblotting to a PVDF membrane (A) and phosphor imaging of that membrane (B): lanes 1-4, DEAE pools 1-4, respectively.

respectively. A major UV-absorbing peak eluted with 0.2 M NaCl actually contained only a small amount of protein (as visualized by Coomassie staining). Since only a trace of radioactivity was found in that fraction, it was not analyzed further. On the basis of the radioactivity and A280 elution profiles, fractions 5-11, 46 and 47, 49, and 60 and 61 were combined into four pools (called DE1-DE4, respectively); these pools contained 46, 4.7, 1.5, and 1.8% of the radioactivity applied to the column, respectively. The remainder of the radioactivity originally applied to the column was distributed at low levels throughout other eluate fractions. The pooled DEAE eluates were analyzed by SDSPAGE followed by transblotting and phosphor imaging (Figure 3). The results show that the protein-bound radioactivity is distributed mainly between two DE pools. Pool DE1 is enriched in three of the major radioactive cytosolic bands with estimated molecular masses of ∼15, ∼25-30, and ∼50 kDa, respectively, and pool DE2 contains only one radioactive band at 65 kDa. Pool DE3, which represents the overlap between a small radioactive and a large nonradioactive protein peak eluted with 0.2 M NaCl (Figure 2), also contained some radioactivity in the 65 kDa range. No distinct radioactive bands were

1330

Chem. Res. Toxicol., Vol. 13, No. 12, 2000

Figure 4. Separation of the DE1 pool by CM Sepharose cation exchange chromatography: radioactivity and protein elution profiles. An aliquot of pool DE1 (14 mL) containing 0.5 mg/mL protein (specific activity of 0.022 µCi/mg of protein) was dialyzed against 20 mM potassium phosphate (pH 6.6) and separated on a 1.5 cm × 12 cm CM Sepharose column eluted with the same buffer containing increasing concentrations of NaCl (shown by arrows). The eluate was collected in 5 mL fractions, and fractions 5-8, 9, 14-16, 26 and 27, and 29 and 30 were combined to form pools CM1-CM5, respectively.

Figure 5. SDS-PAGE and phosphor imaging analysis of pooled CM eluate fractions. Aliquots of CM1, CM3, and CM5 (∼1000 dpm each) and CM2 and CM4 (∼500 dpm each) were precipitated with acetone and the proteins separated by SDSPAGE (12%) followed by electroblotting to a PVDF membrane (A) and phosphor imaging of that membrane (B): lanes 1-5, CM pools 1-5, respectively; and S, molecular mass standards.

found in pool DE4, although in some blots a trace of radiolabel could be seen at ∼80 kDa. Since the majority of the radioactivity in the DEAE eluates was associated with the flow-through and buffer wash fractions, the pool DE1 material was further fractionated by CM Sepharose cation exchange chromatography using 20 mM potassium phosphate buffer followed by elution with a discontinuous NaCl gradient (from 0.05 to 0.5 M). As seen in Figure 4, there are two protein peaks in the flow-through and buffer wash and radioactivity elutes with these peaks. An additional radiolabeled peak coeluted with protein in 0.05 M NaCl; two small protein peaks eluted with 0.1 and 0.15 M NaCl, but they contained negligible amounts of radioactivity. Increasing the NaCl concentration further failed to elute any additional protein or radioactivity (not shown). On the basis of the elution profiles, fractions 5-8, 9, 14-16, 26 and 27, and 29 and 30 were combined into five pools (called CM1-CM5, respectively); these pools contained 35.5, 5.3, 15.3, 2.7, and 8.8% of the radioactivity applied to the column, respectively. The pooled CM eluates were analyzed by SDS-PAGE followed by transblotting and phosphor imaging. As can be seen in Figure 5, pool CM1 contains three major protein bands with apparent molecular mass values of 14, 25, and 30 kDa, and each of these bands is definitely

Koen et al.

radioactive. There is also a detectable amount of radioactivity associated with a minor protein band at ∼52 kDa. Pool CM3 material separates into three distinct protein bands and two or three very close or partially overlapping bands at ∼25-28 kDa. Only the major band at 52 kDa and the group of poorly resolved bands at 25-28 kDa contain detectable amounts of radioactivity. Pool CM5 contains two major protein bands, of which only the one at ∼28 kDa is slightly radioactive. Pools CM2 and CM4 contain one and two protein bands, respectively, in the 30 kDa range, but because of their extremely low protein content, we did not pursue them further. Since pool CM1 contained the greatest quantity of radiolabeled cytosolic proteins, and since those proteins could be easily separated by SDS-PAGE, further experiments were aimed at identification of the major radioactive proteins in this fraction from the in vitro-radiolabeled cytosol. As shown above, there are at least three distinct radiolabeled protein bands in this fraction. No further separation of additional bands was achieved by varying the SDS-PAGE conditions, suggesting that the bands that were seen were electrophoretically homogeneous or nearly so. To examine this point further, fraction CM1 was analyzed by two-dimensional electrophoresis (not shown) using mixtures of ampholytes with pH ranges of 3-10 and 5-7 for isoelectric focusing. Proteins were precipitated with acetone prior to loading on the IEF tube gel. After two-dimensional separation, the proteins were stained with silver or electroblotted onto a PVDF membrane followed by phosphor imaging. The transfer of proteins from the two-dimensional gel to the PVDF membrane was complete on the basis of observation of the gels stained with Coomassie after blotting. Thus, the pattern seen on the blots reflected the actual distribution of the proteins. It was found that during IEF the 30 kDa protein moved into the neutral-pH area of the gel, resulting in one protein spot in the second dimension. The presence of radioactivity in this spot was confirmed by subsequent electroblotting followed by phosphor imaging of the blot. Unfortunately, using standard twodimensional techniques, we failed to obtain consistent results for the 14 or 25 kDa proteins. The former, being a small protein, usually moved close to the dye front on SDS-PAGE, which resulted in a line rather than a spot on two-dimensional gels. Curiously, the 25 kDa protein did not enter the IEF gel from either a basic or an acidic solution, presumably because of poor resolubilization of this protein. In an attempt to enhance the resolution of lowmolecular mass proteins in fraction CM1, we used the Tricine-SDS-PAGE technique. The results of electrophoresis under these conditions, followed by electroblotting and phosphor imaging, are presented in Figure 6. A major 14 kDa protein band still is seen on the blot, but a small additional band at ∼13 kDa is also clearly seen (Figure 6A). Phosphor imaging of this blot (Figure 6B) shows a broad area of radioactivity covering the area of both protein bands (13 and 14 kDa), with the lower portion of the spot being much more dense, indicating that the minor 13 kDa protein has a much higher labeling density than the major 14 kDa protein. To verify this perception of the relative distribution of radioactivity between these two proteins, the blot was next cut in half (so as to physically separate the two protein bands) and the two pieces were re-exposed for phosphor imaging.

Cytosolic Targets of Bromobenzene Metabolites

Figure 6. Analysis of the pool CM1 protein by Tricine-SDSPAGE and phosphor imaging. Aliquots of CM1 containing different amounts of protein were separated using a TricineSDS-PAGE (16.5%) protocol essentially according to the method of Scha¨gger and Jagow (see Experimental Procedures for details). Separated proteins were electroblotted to a PVDF membrane (A), and the membrane was exposed for phosphor imaging (B): lanes 1-3, 50, 20, and 10 µg of CM1 protein/lane, respectively; and S, molecular mass standards.

Figure 7. MALDI-TOF mass spectrum of eluate fraction CM1. See the text for details.

This reconfirmed the presence of radiolabel in both proteins, with the labeling density of the 13 kDa protein being obviously much greater than that of the 14 kDa protein (data not shown). Further analysis of the CM1 fraction of the in vitroradiolabeled cytosolic proteins was performed using MALDI-TOF mass spectrometry. The results are shown in Figure 7. Several peaks are seen in the mass spectrum, of which four fall into the molecular mass range of interest (6.5-30 kDa). Three of the peaks have m/z values of 25 876, 29 788, and 14 423, which agree well with the estimated molecular masses of corresponding radioactive protein bands seen on SDS-PAGE (Figure 5). These data are also in good agrement with the calculated molecular masses of the three proteins with which they were subsequently identified (see below). The other two peaks have m/z values of 10 030 and 12 889, similar to the estimated molecular masses of two other protein bands seen on Tricine-SDS-PAGE (Figure 6). The absence of any additional MS peaks in the 6.5-30 kDa range suggests that five corresponding protein bands on SDSPAGE represent substantially pure single proteins. Identification of Radiolabeled Cytosolic Proteins. With a sequence of anion and cation exchange chromatography followed by SDS-PAGE, we isolated

Chem. Res. Toxicol., Vol. 13, No. 12, 2000 1331

four radioactive cytosolic proteins in an apparently pure form. For further identification, the 14, 25, 30, and 50 kDa radiolabeled proteins in fraction CM1 were isolated by excision of the appropriate bands from blots of SDSPAGE gels and the blot segments were submitted for N-terminal Edman sequencing. Disappointingly, these proteins were found to be inert to Edman degradation. Since controls showed that adequate amounts of protein were present, we interpreted these results to indicate that these proteins all had blocked N-termini, either naturally or because of reaction with acrylamide during SDS-PAGE (25-27). Because of the very small amounts that were available, we did not attempt to analyze the 13 kDa protein by the Edman method. To gain information about the structure and identity of the 14, 25, 30, and 50 kDa proteins, we resorted to the technique of peptide mapping by mass spectrometry. Proteins from the CM1 fraction of cytosol 1 were separated by either standard or Tricine-SDS-PAGE with conditions being varied to optimize resolution in the molecular mass range of interest. Separated bands (stained with Coomassie) were excised from the gel, ingel reduced with dithiothreitol, alkylated with 2-vinylpyridine, and digested with trypsin. The digests were extracted from the gel and analyzed by HPLC/ESI-MS. As a control, segments of gel containing no protein were processed and analyzed identically. HPLC chromatograms of digests usually contained a large number of peaks detected by total ion current. Some of them were of low abundance, and only peaks whose height was greater than 10 times background were considered further. Some peaks exhibited m/z values identical to those in corresponding regions of the chromatogram from control extracts and could be easily identified as tryptic fragments of trypsin itself. Finally, after applying these selection rules to the chromatograms of the digests of the 14, 25, and 30 kDa protein bands, we selected a total of 23, 31, and 20 digest-specific monoisotopic tryptic peptide masses, respectively, and used these sets of masses for database searching. The peptide mass search program MS-Fit was used to match LC/MS-derived masses of tryptic fragments to masses of predicted tryptic fragments of proteins in several data banks. The NCBI data bank was searched first in all cases, and the data that were obtained were then compared to those obtained by searching other databases (Genpept and SwissProt); no major differences in results were found. Using the search strategy described above, we identified the 14 kDa protein as liver fatty acid binding protein (LFABP) (Table 1). Even if a mass tolerance as high as 0.3% was assumed, there was only one viable candidate among the 113 rat-derived database entries having a molecular mass of 14-15 kDa (a few additional lowscoring protein candidates were found in the molecular mass range of 80-90 kDa). Thirteen MS-observed peptide masses were matched with 12 theoretical tryptic fragments at a mass tolerance of