Characterization of the Reactivity, Regioselectivity, and

N -terminal valine adduct from the anti-HIV drug abacavir in rat haemoglobin as evidence for abacavir metabolism to a reactive aldehyde in vivo. C Cha...
0 downloads 0 Views 219KB Size
Chem. Res. Toxicol. 1999, 12, 679-689

679

Characterization of the Reactivity, Regioselectivity, and Stereoselectivity of the Reactions of Butadiene Monoxide with Valinamide and the N-Terminal Valine of Mouse and Rat Hemoglobin Thomas S. Moll and Adnan A. Elfarra* Department of Comparative Biosciences and Center for Environmental Toxicology, University of Wisconsin, Madison, Wisconsin 53706-1102 Received March 10, 1999

Occupational exposure to 1,3-butadiene (BD) has been monitored by measuring the level of hemoglobin N-terminal valine adduct formation with the primary reactive metabolite, butadiene monoxide (BMO). However, mechanistic details concerning the relative reactivity, regioselectivity, and stereospecificity of BMO with the N-terminal valine of hemoglobin are lacking. In the studies presented here, L-valinamide was used as a model for the N-terminal valine of hemoglobin to compare the nucleophilic reactivity, regioselectivity, and stereoselectivity of the reaction both in aqueous solution and within a protein microenvironment. Four products produced by the reaction of L-valinamide with racemic BMO (two pairs of diastereomers produced by reactions at C-1 and C-2 of the epoxide moiety) were synthesized, purified, and characterized by 1H NMR and GC/MS. These four reaction products were used as analytical standards for kinetic studies of the reaction of valinamide with BMO at physiological pH (7.4) and temperature (37 °C). The results show that the adducts formed by reaction at C-2 were formed at a ratio of approximately 2:1 compared to the adducts formed by reaction at C-1. The stereoisomers of each respective regioisomer were produced with similar rates of formation. The reaction of BMO with the N-terminal valine of hemoglobin was also studied in vitro using intact erythrocytes from Sprague-Dawley rats and B6C3F1 mice. After cleavage of the N-modified valine by the N-alkyl Edman degradation procedure using pentafluorophenylisothiocyanate (PFPITC), a novel procedure was developed that allowed GC/MS detection and quantitation of the four expected products by silylation of the PFPTH-valine-BMO derivatives. The hemoglobin results contrast with the valinamide results in that the reaction of BMO with the N-terminal valine residue in both rat and mouse hemoglobin produced mostly C-1 adducts. The rates obtained with rat hemoglobin were much slower than the rates obtained with mouse hemoglobin or with valinamide. These results, and the finding that the reaction with rat hemoglobin produced a higher ratio of C1:C2 adducts in comparison with the reaction with mouse hemoglobin, indicate the importance of measuring all four adducts when comparing the relative rates of adduct formation both with model compounds and among different species.

Introduction 1,3-Butadiene (BD)1 is a carcinogenic and genotoxic chemical used primarily as a chemical intermediate and polymer component in the manufacture of synthetic rubber and plastics, and is also listed by the EPA as a hazardous air pollutant; environmental sources include cigarette smoke and automobile exhaust. BD is bioactivated by cytochrome P450-mediated oxidation to produce racemic 1,2-epoxy-3-butene [butadiene monoxide (BMO)] (reviewed in ref 1). This primary reactive metabolite has been shown to undergo nucleophilic substitution reac* To whom correspondence should be addressed: Department of Comparative Biosciences, School of Veterinary Medicine, University of WisconsinsMadison, 2015 Linden Dr. W., Madison, WI 53706-1102. Phone: (608) 262-6518. Fax: (608) 263-3926. E-mail: elfarraa@ svm.vetmed.wisc.edu. 1 Abbreviations: PFPTH, pentafluorophenylthiohydantoin; BD, 1,3butadiene; BMO, butadiene monoxide; EI, electron-impact ionization; BSTFA, bis(trimethylsilyl)trifluoroacetamide; TMCS, trimethylchlorosilane; PFPITC, pentafluorophenyl isothiocyanate; Val-ME, valine methyl ester; TMS, trimethylsilyl.

tions with numerous cellular macromolecules, including DNA (2) and proteins (3). Recently, several investigators have attempted to develop highly specific and sensitive bioassays involving the detection and quantitation of BMO adducts of the N-terminal valine residue of hemoglobin after exposure to BD. These assays are all based on a modified Edman degradation procedure developed by To¨rnqvist et al. (4), which has been termed the N-alkyl Edman method. The derivatives of the alkylvaline adducts produced by this method, the pentafluorophenylthiohydantoins (PFPTHs), are quantitated using high-resolution GC/MS. Although there has been great progress in the ability to detect and quantitate specific adducts at low levels (97%. (2) Characterization of Valinamide-BMO Adducts by 1H NMR. Purified valinamide-BMO adducts were dissolved (5 mg/mL) in D2O, and proton NMR spectra were obtained on a Bruker spectrometer (Karslruhe, Germany) at 500 MHz at room temperature. Chemical shifts are reported in parts per million with the H2O peak as an internal standard; J values are reported in hertz. P-1A: 1.01 (d, 3H, CδH3, Jβδ ) 9.0), 1.06 (d, 3H, CγH3, Jβγ ) 8.5), 2.23 (d, hep, 1H, CβH), 3.69 (m, 1H, C2H), 3.80 (d, 1H, CRH, JRβ ) 7.5), 3.88 (m, 2H, C1H2), 5.52 (d, 1H, C4Hcis, Jcis ) 10.0), 5.58 (d, 1H, C4Htrans, Jtrans ) 20.0), 5.61 (m, 1H, C3H). P-1B: 1.00 (d, 3H, CδH3, Jβδ ) 9.0), 1.06 (d, 3H, CγH3, Jβγ ) 8.6), 2.30 (d, hep, 1H, CβH), 3.82 (d, 1H, CRH, JRβ ) 7.5), 3.86 (m, 1H, C2H), 3.89 (m, 2H, C1H2), 5.52 (d, 1H, C4Hcis, Jcis ) 10.0), 5.57 (d, 1H, C4Htrans, Jtrans ) 20.0), 5.77 (m, 1H, C3H). P-2A/P-2B: 1.03 (dd, 6H, CδH3, Jβδ ) 9.0), 1.10 (dd, 6H, CγH3, Jβγ ) 8.5), 2.31 (d, hep, 2H, CβH), 3.06 (m, 2H, C1H2′′), 3.22 (m, 2H, C1H2′), 3.90 (d, 2H, CRH, JRβ ) 6.5), 4.50 (m, 2H, C2H), 5.32 (d, 2H, C4Hcis, Jcis ) 13.5), 5.41 (d, 2H, C4Htrans, Jtrans ) 21.5), 5.86 (m, 2H, C3H). (3) Characterization of Valinamide-BMO Adducts by GC/MS. Valinamide-BMO adducts were dissolved in distilled H2O, and aliquots were added to Teflon-capped vials along with internal standard HPV and brought to dryness by N2. BSTFA containing 1% TMCS as a catalyst was added (100 µL) and the mixture heated in a dry bath at 100 °C for 15 min. The silylated adducts were analyzed using a HP-6890 series mass selective detector interfaced with an HP 6890 series gas chromatograph (Hewlett-Packard, Palo Alto, CA). Separation of components was accomplished using a 30 m × 0.25 mm i.d. HP-5MS (0.25 µm) capillary column. The GC injection port temperature was maintained at 280 °C, and the inlet pressure of the ultrapure helium gas was maintained at 20 kPa during the run. All injections were made in the splitless mode. The initial oven setpoint was 100 °C with a hold for 1 min. A temperature gradient of 20 °C/min was used until the temperature reached 280 °C, followed by a hold of 5 min. The mass spectrometer was programmed for scanning mode with a scan range of 10-400 amu. (4) Kinetic Analysis of Adduct Formation under Physiological Conditions. The reactions of valinamide and BMO were carried out in 10 mL vials capped with Teflon septa. Valinamide (final concentration of 50 mM) and BMO (final

Hemoglobin Adducts of Butadiene Monoxide concentration of 50-500 mM) were added to 0.1 M phosphate buffer (pH 7.4, final volume of 5 mL), and the mixture was placed in a shaking water bath (37 °C). Aliquots (100 µL) were taken at timed intervals and extracted with 2 volumes (400 µL each) of toluene. Internal standard (HPV) was added to 25 µL aliquots of reaction mixture and brought to dryness under N2. The reaction products were silylated and analyzed by GC/MS as described above. (5) Synthesis and Purification of Valine-BMO Adducts. L-Valine methyl ester (4 mmol, 675 mg) was placed in a glass vial equipped with a Teflon septum. The compound was dissolved in distilled water (∼8 mL), and the solution was adjusted to pH 7.8 with 2 M NaOH. BMO (12 mmol, 1 mL) was added, and the mixture was incubated in a shaking water bath set at 37 °C for 24 h. The solution was extracted three times with 2 mL of toluene to remove excess BMO and byproducts. Aliquots were diluted 2-fold with 2 M NaOH and placed in a dry bath incubator set at 80 °C for 30 min to hydrolyze the methyl ester moiety. The crude valine-BMO adduct mixture was adjusted to pH 2.1 with dilute TFA, and the samples were injected onto an Ultraprep C18 (21.2 mm × 15 cm) HPLC column at a flow rate of 10 mL/min and monitored at A210 (0.01 AUFS). Mobile phase A was 0.1% TFA in H2O; mobile phase B was 40% ACN in H2O and 0.1% TFA. The samples were eluted isocratically with 5% B. Valine-BMO adducts (retention times between 15 and 20 min) were collected and lyophilized. The products were redissolved in distilled H2O and relyophilized. Yields were typically ∼60-70%. Purities of the adducts were greater than 95% as determined by HPLC. (6) Synthesis of PFPTH-Valine-BMO Adducts. The synthesis of PFPTH-valine-BMO adducts was carried out according to the procedure of To¨rnqvist et al. (4). The valineBMO adducts (5 mg, 27 µmol) were dissolved in 200 µL of 0.5 M NaHCO3 and 1-propanol (2:1 v/v) in a glass vial equipped with a Teflon septum. PFPITC was added (10 µL, 75 µmol), and the mixture was allowed to react in a shaking water bath at 45 °C for 3 h. The samples were extracted (3 × 500 µL) with heptane, and the combined extracts were dried under nitrogen at 50 °C. The samples were then redissolved in 100 µL of toluene and washed with 200 µL of 1 M HCl and 200 µL of distilled water. The samples were diluted with ACN and analyzed by GC/MS using an initial oven setpoint of 100 °C followed by a 1 min hold. The temperature was then increased at a rate of 30 °C/min to 250 °C, and then increased further at a rate of 10 °C/min to 280 °C. (7) Derivatization of PFPTH-Valine-BMO Adducts with BSTFA/TCMS. Aliquots of the PFPTH-valine-BMO adducts were evaporated to dryness under nitrogen at 50 °C. The samples were then silylated by the addition of 100 µL of BSTFA, containing 1% TCMS as a catalyst, and incubation in a dry bath heater at 100 °C for 15 min. The samples were allowed to cool, then diluted with ACN, and analyzed by GC/ MS. The initial oven setpoint was 100 °C followed by a hold of 1 min. The temperature was increased at a rate of 35 °C/min to 250 °C, and then increased at a rate of 3 °C/min to 270 °C with a 5 min hold. The final increase was carried out at a rate of 20 °C/min to 280 °C. The mass spectrometer was programmed to monitor the major fragment at m/z 142 as well as a qualifying ion at m/z 129. Quantitation of the valine-BMO adducts (0.5100 pmol/g of globin) was based on measurement of the internal reference standard (HPV). The response factor for the adducts in relation to HPV (mean ( standard deviation ) 0.97 ( 0.20) was determined using mixtures of the internal reference standard and the purified analytical standards. (8) In Vitro Reaction of Mouse and Rat Hemoglobin with BMO. Whole blood with heparin as the anticoagulant was obtained from freshly killed rats and mice and was processed within 24 h. Intact erythrocytes were initially separated from plasma proteins by centrifugation of whole mouse or rat blood for 5 min at 1500g. After removal of the plasma fraction, the red blood cells were washed three times with an equal volume of isotonic saline (0.9% w/v). The red blood cells were resus-

Chem. Res. Toxicol., Vol. 12, No. 8, 1999 681

Figure 1. Reaction scheme of L-valine (or L-valinamide) with BMO. The nucleophilic addition of the electrophile BMO to the primary amine group of L-valine (or L-valinamide) can occur at either the primary (C-1) or secondary (C-2) carbon of the epoxide moiety, resulting in two regioisomers, each regioisomer consisting of a pair of diastereomers. pended in an equal volume of phosphate-buffered saline (0.1 M, pH 7.4). After addition of BMO (0.1-20 µL/mL, 1.21-242 µmol/mL), the mixture was placed in a shaking water bath at 37 °C and the reaction stopped at designated time points. After being washed with cold saline, the cells were resuspended in cold distilled water and stored overnight at -10 °C. The thawed mixture was then centrifuged for 30 min at 30000g to remove cell membranes, and the supernatant was added dropwise with vigorous vortexing to cold acidified acetone (1% HCl) to precipitate globin. The sample was centrifuged for 15 min at 3000g, and the pellets were washed once with cold acidified acetone, once with cold acetone, and once with cold diethyl ether. The globin (10-50 mg) was dried under a gentle stream of nitrogen at 45 °C. The globin was then dissolved in 0.5 M NaHCO3 and 1-propanol (1 mL; 2:1 v/v), and the N-alkyl Edman procedure and subsequent silylation with BSTFA/TCMS were performed as described (vida infra).

Results The in vitro reaction between valinamide and racemic BMO results in the formation of four stereoisomeric adducts (two diastereomers of each regioisomer) as depicted in the general scheme of Figure 1. ValinamideBMO adducts were synthesized, and the reaction products were purified by reverse phase HPLC. Three main fractions, designated P-1A, P-1B, and P-2AB, were collected at 8.5, 9.5, and 12.6 min, respectively (data not shown). The products were then identified and characterized by 1H NMR and GC/MS. The 1H NMR spectra of P-1A (Figure 2A) and P-1B (spectrum not shown; for chemical shifts and J values, see Experimental Procedures) are very similar, indicative of diastereomers, and the chemical shifts of the C-1 and C-2 protons of the BMO moiety are consistent with the branched form (attack at C-2 of the epoxide) of the valinamide adduct. The only major difference in the spectra is found in the chemical shifts of the C-2 proton, which is 3.69 ppm in P-1A and 3.86 ppm in P-1B (∆ppm ) 0.17); the difference between the diastereomers is most likely due to the proximity of the C-2 proton to the chiral center of the valinamide moiety. The spectrum of the P-2A/P-2B mixture (Figure 2B), upon integration, indicates the presence of 28 protons (compared to 14 each with P-1A and P-1B); the appearance of closely related multiplets suggests a mixture of the other pair of diastereomers. The downfield shift of the C-2 proton, which

682 Chem. Res. Toxicol., Vol. 12, No. 8, 1999

Moll and Elfarra

Figure 2. 1H NMR of L-valinamide-BMO adducts. Valinamide-BMO adducts were initially separated by reverse phase HPLC, followed by lyophilization of the collected fractions. The samples were brought up in D2O at a concentration of ∼5 mg/mL and analyzed on a Bruker spectrometer art 500 MHz at room temperature. (A) 1H NMR of one of the purified C-2 diastereomers [N-(2-hydroxy3-buten-1-yl)valine]. (B) 1H NMR of a pure mixture of the two C-1 diastereomers [N-(1-hydroxy-3-buten-2-yl)valine].

is explained by the deshielding effect of the hydroxyl group, provides strong evidence that the linear form of the adduct (attack at the terminal C-1) is present. Again, the only major difference in the proton assignments is found in the chemical shifts of the C-1 protons, which is 3.06 ppm for one diastereomer and 3.22 ppm for the other (∆ppm ) 0.16). Similar to that for the P-1A and P-1B spectra, this difference between the diastereomers is directly related to the close proximity of these protons to the chiral center of valinamide.

Further evidence for the identification of the reaction products is provided by the results of the GC/MS experiments (Figure 3). After derivatization by BSTFA/TCMS, the four products were completely separated by GC and the mass spectra determined by identification of the characteristic fragment ions. The two peaks eluting at 8.31 and 8.61 min coincide with the P-1A and P-1B products, respectively, and the two later eluting peaks at 8.72 and 8.90 min represent the P-2A and P-2B adducts, respectively. The molecular ion (M+) of all four

Hemoglobin Adducts of Butadiene Monoxide

Chem. Res. Toxicol., Vol. 12, No. 8, 1999 683

Figure 3. Total ion chromatogram and major ion fragmentation scheme of L-valinamide-BMO standards. (A) The peaks eluting at 8.31 and 8.61 min are the diastereomers arising from the reaction at C-2 of BMO, and the peaks eluting at 8.72 and 8.90 min are the diastereomers obtained from reaction at C-1. (B) The m/z ratios of the characteristic fragment ions are listed above along with the major ion fragmentation patterns for P-1A/P-1B and P-2A/P-2B.

products is m/z 330, which corresponds to two trimethylsilyl moieties per molecule. Theoretically, there are three possible sites for derivatization (CONH2, NH, and OH), but the proximity of the hydroxyl group to the secondary amine group makes the simultaneous silyla-

tion of these two groups less likely due to steric hindrance. Since the hydroxyl group is significantly more reactive to silylating compounds than amines, it is postulated that the derivatization occurs primarily at the OH and CONH2 groups.

684 Chem. Res. Toxicol., Vol. 12, No. 8, 1999

Moll and Elfarra

Figure 5. Kinetic rates of formation of valinamide-BMO adducts. Valinamide-BMO adducts were analyzed by GC/MS using N-(hydroxypropyl)valine as an internal standard and the purified synthetic adducts as analytical reference standards. The reaction of BMO (100 mM) and L-valinamide (50 mM) was carried out under physiological conditions (0.1 M phosphate at pH 7.4 and 37 °C). Figure 4. Mass spectrum (EI-GC/MS) of valinamide-BMOTMS regioisomers. (A) Mass spectrum of one of the silylated C-2 diastereomers [N-(2-hydroxy-3-buten-1-yl)valine] of the purified valinamide-BMO adducts. (B) Mass spectrum of one of the silylated C-1 diastereomers [N-(1-hydroxy-3-buten-2-yl)valine]. In each case, the mass spectra of the other diastereomer of each regioisomer are identical (data not shown).

Despite the identical molecular weight and formula of the regioisomers, the fragmentation patterns are strikingly different (Figure 4). However, the diastereoisomers of both regioisomers have identical spectra (data not shown). The major similarities of P-1A and P-1B with P-2A and P-2B are the molecular (m/z 330), the M+ - 15 (methyl group loss from TMS), and the 214 ions (fragmentation at the carbonyl group). The characteristic fragments of P-1A/P-1B include the m/z 227 (TMSO-CH3 loss) and 143 (NH-valinamide loss) ions. With the P-2A/ P-2B adducts, the ion at m/z 201 (valinamide-C-1 loss) is a major characteristic fragment. The in vitro kinetics of valinamide-BMO adduct formation were determined under physiological conditions so the reaction rate constants could be correlated with those obtained in vitro using hemoglobin. The four reaction products under these conditions were monitored by GC/MS as a function of time (Figure 5). Quantitation of the adducts was based on the peak response ratio of an internal reference standard (HPV) with the synthetic adduct analytical standards. The data show that the level of the products arising from attack at C-1 of BMO (P-2A and P-2B) is significantly lower than the level of the products arising from attack at C-2 (P-1A and P-1B). The individual second-order rate constants are listed in Figure 6 and were calculated using a formula which corrects for the decrease in the concentration of BMO due to spontaneous hydrolysis (eq 1; Figure 6). For BMO in 0.1 M phosphate buffer (pH 7.4) at 37 °C, the first-order rate constant for hydrolysis (k′) that is used is 1.292 × 10-5 s-1 (14). The rate constants for P-1A and P-1B are significantly higher than the values obtained for P-2A

(1)

Figure 6. Rate constant parameters for the reaction of valinamide with BMO. The second-order rate constants for individual and total adduct formation are listed. The rate constants (k) were determined according to the equation described above using the specified constants. Values shown represent the means ( standard deviations (n ) 3). The initial BMO concentration represents the theoretical aqueous concentration (i.e., not corrected for headspace redistribution).

and P-2B, and the extents of reaction indicate that the C-2 products are produced at a molar ratio of nearly 2:1. In addition, although the C-1 product rate constants are similar to the rate constant for spontaneous hydrolysis at neutral pH, the combined rate constants for adduct formation (kP-1A + kP-1B + kP-2A + kP-2B) are nearly 5-fold greater than k′.

Hemoglobin Adducts of Butadiene Monoxide

Chem. Res. Toxicol., Vol. 12, No. 8, 1999 685

Figure 7. Synthesis of valine-BMO adducts and derivatives. The valine-BMO adducts were synthesized using valine methyl ester as the starting compound. The ester moiety was then hydrolyzed under basic conditions prior to purification. The adducts were initially analyzed after silylation by EI-GC/MS. Consequently, the PFPTH-valine-BMO compounds were synthesized, and were analyzed either with or without additional derivatization with BSTFA containing TCMS. For simplification, only the reaction occurring at C-1 is shown.

To study the in vitro reaction of BMO with the N-terminal valine of hemoglobin, the corresponding four products of free L-valine and BMO were synthesized and purified. However, since epoxides can also react with the carboxyl group in free valine, valine methyl ester (ValME) was used as the starting compound (Figure 7). The reaction of Val-ME with BMO was allowed to occur under slightly basic conditions (pH 7.8) at 37 °C for 24 h. Higher temperatures resulted in decreased yields, apparently due to the increased rate of spontaneous hydrolysis of BMO. After extraction with toluene to remove excess BMO and nonpolar byproducts, the mixture was adjusted to pH 11, and the basic hydrolysis of the ester moiety was carried out at 80 °C for 30 min. The valine-BMO adducts were then purified on a preparative C18 reversed phase column. The synthesis of the PFPTH-valine-BMO adducts was performed using the purified valine-BMO adducts as the starting material, and was carried out essentially by the procedure of To¨rnqvist et al. (4). The initial characterization of the PFPTH-valine-BMO adducts was accomplished by GC/MS in the electron-impact ionization (EI-GC/MS) mode (Figure 8). PFPTH amino acids are generally quite volatile at low temperatures, and the presence of the fluorine atoms [compared to the original phenyl isothiocyanate (PITC) used in the Edman degradation procedure] enhances the sensitivity of the compounds during mass selective detection. The molecular ion of the four compounds is calculated at 394 amu (Figure 8B,C), which corresponds to the peaks eluting between 12 and 13 min (Figure 8A). The presence of only two peaks in the total ion chromatogram indicated that the diastereomers of the two regioisomers were not resolved in this procedure, and was further evidenced by

the differences in the fragmentation patterns. The characteristic fragment that supports the regioisomer assignment is the M+ - 57 ion (337 amu; loss of CH2d CHCHOH) found in the early eluting peak (12.22 min), which could only be produced from the C-1 reaction product since the carbon chain of this adduct is not branched at the N-proximal carbon. The presence of a hydroxyl group in the adducts, and the results of the initial derivatization of these groups with silylating compounds, led us to attempt the derivatization of these groups as well in the PFPTH-valineBMO adducts. Although they have high thermal stability, the silyl derivatives tend to be chemically unstable; all samples were analyzed immediately after derivatization. The four silylated derivatives (PFPTH-valine-BMOTMS) were efficiently separated in this procedure (Figure 9A). Surprisingly, the first and last eluting peaks represent the C-1 reaction products, while the middle peaks comprise the C-2 compounds. The molecular ion of the silylated derivatives is calculated and found to be 466 amu (Figure 9B,C). Again, the main characteristic fragment in the C-1 compounds (Figure 9B) is the 129 amu ion, which corresponds to the 57 amu ion produced by the nonsilylated C-1 compound, although here it has a trimethylsilyl group attached (72 amu). All four compounds also produce a major fragment at 142 amu which corresponds to the BMO adduct with a TMS group attached at the hydroxyl moiety. Also produced is a minor fragment at 225 amu corresponding to C6F5NCS. This minor fragment is commonly found in the PFPTHs of N-substituted valines (15). It was also found that the silylation resulted in nearly a 10-fold increase in sensitivity compared with those of the nonsilylated compounds. In addition, measuring these adducts using

686 Chem. Res. Toxicol., Vol. 12, No. 8, 1999

Moll and Elfarra

Figure 8. Total ion chromatogram and mass spectrum of the PFPTH-valine-BMO derivatives. (A) The column was maintained at 200 °C for 20 min, and the temperature was then increased to 280 °C at a rate of 30 °C/min. The two peaks represent the C-1 and C-2 regioisomers; the respective diastereomers were not resolved under these conditions. (B) Mass spectrum of the C-2 regioisomers. (C) Mass spectrum of the C-1 regioisomers. The respective diastereomers were not resolved under the operating conditions. The internal standard PFPTHN-(hydroxylpropyl)valine elutes at 15.2 min (not shown).

Figure 9. Selected ion chromatograms and mass spectrum of the PFPTH-valine-BMO-TMS adducts produced by the reaction of hemoglobin with BMO. (A) Selected ion chromatogram of the in vitro reaction of mouse whole blood with BMO. The internal standard PFPTH-N-(hydroxylpropyl)valine-TMS elutes at 15.8 min (not shown). (B) Mass spectrum of one of the C-2 silylated PFPTH-valine-BMO adducts. (C) Mass spectrum of one of the diastereomers of the C-1 silylated PFPTH-valineBMO adducts. The mass spectrum of the other diastereomer of each respective regioisomer is identical (data not shown).

selected ion monitoring at 129 and 142 decreased the limit of detection from 0.5 nmol/g of globin to 0.5 pmol/g of globin. This level is low enough to allow monitoring of occupational exposure at levels less than 1 ppm BD (16, 17). The racemization of amino acids upon reaction with phenylthiohydantoins was previously observed during the development of the Edman degradation procedure (18). In our study, the products of the reactions of PFPITC with the purified valine-BMO adducts were rapidly racemized during the formation of the pentafluorophenylthiohydantoins (unpublished observation). This was evidenced by the observation that when purified samples of either of the two diastereomers of the C-1 valine adducts of BMO were reacted with PFPITC and silylated, both the first and last eluting peaks were detected by the GC/MS analysis (Figure 9). Thus, although the individual diastereomers of the two regioisomers could be detected and quantitated, whether they arise directly from the reaction of the epoxide with valine or during the racemization reaction with PFPITC cannot be determined using this method. Therefore, the validity of previously reported values of kinetic constants (6-8), obtained by the quantitation of a single regioisomer, should be interpreted with caution. The reactions of BMO with mouse (B6C3F1) and rat (Sprague-Dawley) erythrocytes were then studied under physiological conditions [0.1 M phosphate (pH 7.4) at 37 °C] so the intrinsic reactivity of the N-terminal amine

group of valinamide could be compared to that found in vitro with hemoglobin. The rates of formation of the various adducts were obtained from the second-order rate constants for the respective reactions. The rate constants were determined using the same relationship described earlier for valinamide-BMO adducts. In this instance, however, the initial nucleophile concentration pertains to hemoglobin, which is 137 mg/mL (2.1 mM) in mouse blood and 153 mg/mL (2.3 mM) in rat blood (19). Rate velocities were determined from the initial linear portion of the time course (Figure 10) and were used to determine the kinetic rate constants. The mouse and rat kinetic results, displayed in Figure 11, are strikingly different than those obtained using valinamide as a model, especially with regard to the regioselectivity of the products. It was noted previously that the C-2 regioisomers are the predominant adducts formed upon reaction of valinamide with BMO under physiological conditions. With rat and mouse blood, however, it is the C-1 diastereomers that are preferentially formed. In addition, there is a significant difference in the ratio of C-1 products to C-2 products between species, with the rat (C-1:C-2 ) 26.2) showing much less C-2 formation, relatively speaking, than the mouse (C-1:C-2 ) 4.6). Species-specific differences in reactivity were also seen, as evidenced by the second-order rate constants for adduct formation. The total rate constant, obtained by the addition of the individual rate constants, was roughly twice as much with mouse hemoglobin (4.99

Hemoglobin Adducts of Butadiene Monoxide

Chem. Res. Toxicol., Vol. 12, No. 8, 1999 687

Figure 11. Kinetic parameters (second-order rate constants) from the in vitro reaction of mouse and rat whole blood with BMO. Second-order rate constants for the individual and total N-terminal valine adduct formation from in vitro experiments with valinamide and rat and mouse whole blood. Values represent the mean ( standard deviation (n ) 6). Figure 10. Time course of the reaction of BMO with hemoglobin using intact erythrocytes from Sprague-Dawley rats and B6C3F1 mice. Erythrocytes from mice (A) and rats (B) were incubated with BMO under physiological conditions (0.1 M phosphate at pH 7.4 and 37 °C). Valine-BMO adducts were analyzed by GC/MS using HPV as internal standard and the purified synthetic adducts as analytical reference standards. The plotted values are the mean ( standard deviation (n ) 6).

× 10-5 M-1 s-1) as with rat hemoglobin (2.45 × 10-5 M-1 s-1); as a comparison, the mouse rate constant was only slightly lower than the value obtained with valinamide (5.39 × 10-5 M-1 s-1).

Discussion The formation of hemoglobin adducts upon exposure to BMO was first studied by Sun et al. (20) using [14C]BD. They showed that BMO could form adducts with mouse and rat hemoglobin, and that adduct formation was linearly related to administered doses up to 100 µmol of [14C]BD per kilogram of body weight. In addition, repeated daily injections of BD within this range correlated with a linear accumulation of hemoglobin adducts. In 1991, Osterman-Golkar et al. (3) measured the levels of hemoglobin-BMO adducts in rats using the N-alkyl Edman procedure and GC/MS of the thiohydantoin derivatives. These experiments were carried out using GC/MS in the negative ion chemical ionization mode (NICI). Although the chemical ionization process can enhance sensitivity due to the general absence of carboncarbon cleavage reactions, the resulting spectra provide little structural information due to the reduced level of fragmentation. Electron-impact ionization (EI) spectra contain abundant ions related to the molecular weight as well as numerous fragment ions which provide information for the absolute identification of compounds, and is especially suitable for the characterization of regioisomers. Modification of the GC/MS methodology employing EI-GC/MS allowed Albrecht et al. (14) to quantitate valine-BMO regioisomer adducts, but the diastereomers

of the regioisomers were not resolved under the conditions that were used. Further refinement of the GC/MS procedure allowed Richardson et al. (16) to detect the four expected products, although quantitation was based on the sum of the areas of the two major peaks due to low sensitivity. Unfortunately, very little information concerning the reactivity and stereoselectivity of the Nterminal valine of hemoglobin with BMO has been reported utilizing these methods. Initial experiments used valinamide (C-terminal amide of valine; Val-CONH2) as a model for the formation of epoxide adducts with the N-terminal valine of hemoglobin due to its favorable physical and chemical properties. To determine both the particular reactivity of the nucleophilic amine group of valine and the contribution of protein tertiary structure on preliminary association kinetics, the reaction of valinamide with BMO was studied under physiological conditions (pH 7.4 at 37 °C). The results indicated that the adducts formed by reaction at C-2 [N-(1-hydroxy-3-buten-2-yl)valinamide] are formed at an approximate ratio of 2:1 compared to the adducts formed by reaction at C-1 [N-(2-hydroxy-3-buten-1-yl)valinamide]. The stereoisomers of each respective regioisomer were produced with similar rates. This is in stark contrast to the results obtained with mouse and rat erythrocytes, which demonstrated the preferential formation of the C-1 adducts in in vitro studies of BMOhemoglobin adduct formation. The change in regioselectivity in the reaction of BMO with valinamide and the N-terminal valine could be due to steric hindrance in the intact hemoglobin molecule. In the R-chain of hemoglobin, for example, this steric hindrance could presumably be exerted by a salt bridge from the N-terminal valine to the C-terminal carboxylate group of arginine-141 of the opposite R-chain (21). There are also regioselective forces evident between the species that were studied. The higher ratio of C-1 and C-2 regioisomers seen in mouse blood could again be related to tertiary and/or quaternary influences that are a direct result of amino acid differ-

688 Chem. Res. Toxicol., Vol. 12, No. 8, 1999

ences in mouse and rat hemoglobin. Collectively, these results strongly suggest that for accurate biomonitoring of exposure to epoxides by the analysis of hemoglobin adducts, quantitation of both regioisomers and their diastereoisomers, when applicable, should be performed. Of special interest, as well as specific relevance to the current results, is the finding that distinct stereochemical differences in the biotransformation of BD to BMO exist between species. For example, the production of both (R)and (S)-BMO has been shown to occur in the metabolism of BD by rat and mouse liver microsomes (10). The oxidation of BD results in the formation of a chiral BMO metabolites, which may play a critical role in subsequent biotransformation pathways as well as during the nonenzymatic formation of macromolecular adducts. Thus, while racemic BMO has been employed in all of the experiments described here, further insight into the species-specific stereoselectivity and regioselectivity could be gained by measuring relative C-1 and C-2 adduct levels, as well as by monitoring configuration inversion, using specific BMO enantiomers. Valine-BMO standards were also synthesized and purified, which were used to identify and characterize the in vitro reaction products by GC/MS, as well as in the development of methods that would increase the separation and sensitivity of detection. Silylation of the PFPTH-valine-BMO adducts increased both the degree of separation of the four products and also increased the sensitivity nearly 10-fold over the nonsilylated products when quantitated by total ion GC/MS. Together, these studies allowed us to monitor the in vitro reaction of BMO with mouse and rat hemoglobin. Indeed, single-ion monitoring allowed the measurement of background levels of BMO adducts in commercially available human hemoglobin (0.7-0.9 pmol/g of globin; data not shown), which could be important in the accurate estimation of the internal dose associated with exposure and, ultimately, in adequately assessing human risk to either environmental or occupational BD exposure. Several key observations were apparent when examining the differences in the kinetic parameters both between the in vitro studies using hemoglobin and valinamide and when interspecies comparisons were made. It is of interest to note that the total rate constant for the N-terminal valine in mouse hemoglobin, obtained by summation of the individual rate constants, is only slightly lower than the value derived with the model compound valinamide. It should be noted, however, that the values calculated in this report are related to the molar concentration of globin, not hemoglobin, molecules, since the specific reaction under study is that with the N-terminal valine of each of the four globin chains per hemoglobin. Hence, if the intact hemoglobin molar concentration was used instead of that of globin, the reported rate constants would increase 4-fold. In addition, the second-order rate constants obtained with mouse hemoglobin were much higher than the corresponding constants derived from rat hemoglobin studies. The molecular basis for the lower rate of reaction with rat hemoglobin compared to that with mouse hemoglobin could be related to the presence of an additional free cysteine (β-Cys-125) in rat hemoglobin, which may be accessible for reaction (22). Consistent with these results, it was found that after exposure to BD mice have 5 times higher BMO-hemoglobin (N-terminal valine) adduct levels than rats (14). However, species

Moll and Elfarra

differences in BD and BMO metabolism may also contribute to the observed difference in N-terminal valine hemoglobin adducts in mice and rats in vivo (1). In conclusion, the results in these studies accentuate the caution that should be shown when measuring hemoglobin adducts as a marker of chemical exposure. Regioselectivity and stereoselectivity can vary greatly between model compounds such as valinamide and hemoglobin, as well as between species, as shown in the results between mouse and rat. In addition, the racemization of amino acids inherent in the alkyl Edman degradation step should be taken into consideration when measuring hemoglobin adducts, especially during quantitation. Finally, the finding that the reaction of BMO with rat hemoglobin produced a higher C-1:C-2 adduct ratio, in comparison with that from the reaction with mouse hemoglobin, stresses the importance of measuring all four adducts so the relative rates of adduct formation are accurately compared both with model compounds and among different species.

Acknowledgment. This research was made possible by Grant ES06841 from the National Institute of Environmental Health Sciences (NIEHS), NIH. The contents of this manuscript are solely the responsibility of the authors and do not necessarily represent the official views of the NIEHS, NIH. T.S.M. was supported by an NRSA Fellowship (NIEHS, ES05835).

References (1) Elfarra, A. A., Krause, R. J., and Selzer, R. R. (1996) Biochemistry of 1,3-butadiene metabolism and its relevance to 1,3-butadieneinduced carcinogenicity. Toxicology 113, 23-30. (2) Selzer, R. R., and Elfarra, A. A. (1999) In vitro reactions of butadiene monoxide with single and double stranded DNA: characterization and quantitation of several purine and pyrimidine adducts. Carcinogenesis 20, 285-292. (3) Osterman-Golkar, S., Kautiainen, A., Bergmark, E., Hakansson, K., and Maki-paakkanen, J. (1991) Hemoglobin adducts and urinary mercapturic acids in rats as biological indicators of butadiene exposure. Chem.-Biol. Interact. 80, 291-302. (4) To¨rnqvist, M., Mowrer, J., Jensen, S., and Ehrenberg, L. (1986) Monitoring of environmental cancer initiators through hemoglobin adducts by a modified Edman degradation method. Anal. Biochem. 154, 255-266. (5) Ehrenberg, L., Granath, F., and To¨rnqvist, M. (1996) Macromolecule adducts as biomarkers of exposure to environmental mutagens in human populations. Environ. Health Perspect. 104, 423-428. (6) Recio, L., Osterman-Golkar, S., Csana´dy, G. A., Turner, M. J., Myhr, B., Moss, O., and Bond, J. A. (1992) Determination of mutagenicity in tissues of transgenic mice following exposure to 1,3-butadiene and N-ethyl-N-nitrosourea. Toxicol. Appl. Pharmacol. 117, 58-64. (7) Osterman-Golkar, S. M., Bond, J. A., Ward, J. B., Jr., and Legator, M. S. (1993) Use of hemoglobin adducts for biomonitoring exposure to 1,3-butadiene. In Butadiene and Styrene: Assessment of Health Hazards (Sorsa, M., Peltonen, K., Vainio, H., and Hemminki, K., Eds.) pp 127-134, IARC Scientific Publications, Lyon, France. (8) Osterman-Golkar, S. M., Moss, O., James, A., Bryant, M. S., Turner, M., and Bond, J. A. (1998) Epoxybutene-hemoglobin adducts in rats and mice: Dose response for formation and persistence during and following long-term low-level exposure to butadiene. Toxicol. Appl. Pharmacol. 150, 166-173. (9) Bolt, H. M. (1993) Interspecies differences in metabolism and kinetics of 1,3-butadiene, isobutene and styrene. In Butadiene and Styrene: Assessment of Health Hazards (Sorsa, M., Peltonen, K., Vainio, H., and Hemminki, K., Eds.) pp 37-44, IARC Scientific Publications, Lyon, France. (10) Nieusma, J. L., Claffey, D. J., Maniglier-Poulet, C., Imiolczyk, T., Ross, D., and Ruth, J. A. (1997) Stereochemical aspects of 1,3butadiene metabolism and toxicity in rat and mouse liver microsomes and freshly isolated rat hepatocytes. Chem. Res. Toxicol. 10, 450-456.

Hemoglobin Adducts of Butadiene Monoxide (11) Poirier, V., and Calleman, C. J. (1983) The nucleophilic reactivity of valine methylamide. A model of the N-terminal valine residues of hemoglobin. Acta Chem. Scand. 37, 817-822. (12) Ellis, M. K., Golding, B. T., and Watson, W. P. (1984) Intrinsic reactivities in the alkylations of protected amino acids by (R)- and (S)-methyloxirane. J. Chem. Soc., Perkin Trans. 11, 1737-1743. (13) Cai, J., Myers, S. R., and Hurst, H. E. (1995) Measurement of the hemoglobin N-(2-oxoethyl)valine adduct in ethyl carbamatetreated mice. Toxicol. Appl. Pharmacol. 131, 73-79. (14) Albrecht, O. E., Filser, J. G., and Neumann, H. G. (1993) Biological monitoring of 1,3-butadiene: species differences in haemoglobin binding in rat and mouse. In Butadiene and Styrene: Assessment of Health Hazards (Sorsa, M., Peltonen, K., Vainio, H., and Hemminiki, K., Eds.) pp 135-142, IARC Scientific Publications, Lyon, France. (15) Rydberg, P., Luning, B., Wachtmeister, C. A., and To¨rnqvist, M. (1993) Synthesis and characterization of N-substituted valines and their phenyl- and pentafluorophenylthiohydantoins. Acta Chem. Scand. 47, 813-817. (16) Richardson, K. A., Megens, H. J., Webb, J. D., and van Sittert, N. J. (1996) Biological monitoring of butadiene exposure by measurement of haemoglobin adducts. Toxicology 113, 112-118. (17) Osterman-Golkar, S., Peltonen, K., Anttinen-Klemetti, T., Landin,

Chem. Res. Toxicol., Vol. 12, No. 8, 1999 689

(18) (19)

(20)

(21)

(22)

H. H., Zorcec, V., and Sorsa, M. (1996) Haemoglobin adducts as biomarkers of occupational exposure to 1,3-butadiene. Mutagenesis 11, 145-149. Edman, P. (1950) Preparation of phenyl thiohydantoins from natural amino acids. Acta Chem. Scand. 4, 277-282. Yeowell-O’Connell, K., Pauwels, W., Severi, M., Jin, Z., Walker, M. R., Rappaport, S. M., and Veulemans, H. (1997) Comparison of styrene-7,8-oxide adducts formed via reaction with cysteine, N-terminal valine and carboxylic acid residues in human, mouse and rat hemoglobin. Chem.-Biol. Interact. 106, 67-85. Sun, J. D., Dahl, A. R., Bond, J. A., Birnbaum, L. S., and Henderson, R. F. (1989) Characterization of hemoglobin adduct formation in mice and rats after administration of [14C]butadiene or [14C]isoprene. Toxicol. Appl. Pharmacol. 100, 86-95. Perutz, M. F., Fermi, G., Abraham, D. J., Poyart, C., and Bursaux, E. (1986) Hemoglobin as a receptor of drugs and peptides: X-ray studies of the stereochemistry of binding. J. Am. Chem. Soc. 108, 1064-1078. Garrick, L. M., Sharma, V. S., McDonald, M. J., and Ranney, H. M. (1975) Rat haemoglobin heterogeneity. Two structurally distinct alpha chains and functional behaviour of selected components. Biochem. J. 149, 245-258.

TX990043X