Characterization of Hemoglobin Adducts from a 4, 4

Department of Environmental Chemistry, Wallenberg Laboratory, and Department of Radiobiology,. Stockholm University, S-106 91 Stockholm, Sweden...
0 downloads 0 Views 82KB Size
614

Chem. Res. Toxicol. 1998, 11, 614-621

Articles Characterization of Hemoglobin Adducts from a 4,4′-Methylenedianiline Metabolite Evidently Produced by Peroxidative Oxidation in Vivo Antti Kautiainen,* Carl Axel Wachtmeister, and Lars Ehrenberg† Department of Environmental Chemistry, Wallenberg Laboratory, and Department of Radiobiology, Stockholm University, S-106 91 Stockholm, Sweden Received August 19, 1997

4,4′-Methylenedianiline (MDA) is a widely used mutagenic and carcinogenic industrial chemical. It is also a metabolite of 4,4′-methylenediphenyl diisocyanate (MDI), which is used in the manufacturing of polyurethane foams. Biomonitoring of MDA, like other aromatic amines, is mainly carried out by GC/MS measurement of cysteine adducts in Hb from the nitroso metabolite, released by alkaline hydrolysis. In the present study it was investigated whether the formation of Hb adducts from non-nitroso metabolites of MDA can be used for the dosimetry of MDA. The study was carried out by treatment of mice with MDA and tritiated MDA or deuterated MDA and by identification of their products of reaction with Hb, after enzymatic hydrolysis of the globin and enrichment of the adducts. The main adduct, about 50% of the total amount of MDA associated with Hb, was characterized by MS and was shown to be a reaction product of MDA and the amino group of N-terminal valine in Hb, the derived structure being 1-[(4-imino-2,5-cyclohexadien-1-ylidene)methyl]benzene-4-azo-2-isovaleric acid. It is likely that this quinonoid MDA imine adduct to valine was formed by an attack of a metabolite formed through peroxidative oxidation of MDA, in analogy with earlier observed oxidation of some other aromatic amines, e.g., benzidine. The reactive intermediate is suggested to be [(4-imino-2,5-cyclohexadien-1-ylidene)methyl]-4-aminobenzene. The formation of the adduct was confirmed by incubating MDA with valine methyl ester in vitro in the presence of H2O2 and lactoperoxidase. Further, the same adduct was detected in MDI-exposed and control rats, the level in the exposed animals being about 60 times higher than in the controls. This study indicates that, at least in the mouse, extrahepatic peroxidative metabolism is an important pathway for the bioactivation of MDA, possibly leading to a genotoxic reactive intermediate. This study also demonstrates the usefulness of Hb adduct analysis for the identification of reactive intermediates in vivo.

Introduction (MDA)1

4,4′-Methylenedianiline is an industrial chemical used in the production of 4,4′-methylenediphenyl diisocyanate (MDI) and as a component in epoxy resins. Furthermore, MDA is used to manufacture polyamineimide, in the preparation of azo-dyes, and as a curative for neoprene rubber (1). MDI, which is an important component in the production of polyurethane foams, is partly metabolized by hydrolysis and decarboxylation to MDA. Exposure to MDA and MDI occurs mainly during their production and further manufacturing of insulation * Corresponding author. Tel: +46-8-16 20 18. Fax: +46-8-15 25 61. E-mail: [email protected]. † Department of Radiobiology. 1 Abbreviations: MDA, 4,4′-methylenedianiline; MDI, 4,4′-methylenediphenyl diisocyanate; PFPA, pentafluoropropionic anhydride; 3HMDA, 4,4′-[3H]methylenedianiline; MDA-d8, methylenedi[2,3,5,6-2H4]aniline; CI, chemical ionization, CID, collision-induced dissociation; NICI, negative ion chemical ionization; SIM, selected ion monitoring; EI-MS, electron ionization mass spectrometry; PICI, positive ion chemical ionization; P450, cytochrome P450; PHS, prostaglandin H synthetase; 2-AF, 2-aminofluorene.

materials. Furthermore, the leakage of MDA and MDI from ubiquitous polyurethane materials is a possible source of exposure. MDA is mutagenic in the Ames test in the presence of S9 mix [Salmonella strains TA98 and TA100 (2)]. Induction of DNA damage caused by MDA treatment has been shown in Chinese hamster V79 cells (3) and in SpragueDawley rats (4) and has been shown to give rise to sister chromatid exchanges in BALB/c mice (5). Following oral administration to B6C3F1 mice and to Fischer 344/N rats, increased incidences of hepatocellular carcinomas and thyroid follicular-cell carcinomas, respectively, were observed (6). MDA has been classified as an animal carcinogen and as a suspected human carcinogen (1). Protein modifications can be utilized for the identification of electrophilically reactive compounds and intermediates in humans and animals. For in vivo dosimetry, sensitive methods based on MS techniques have been developed to analyze stable reaction products, so-called adducts, from several electrophilic compounds to, above

S0893-228x(97)00148-3 CCC: $15.00 © 1998 American Chemical Society Published on Web 05/23/1998

Hb Adducts from a Methylenedianiline Metabolite

all, the blood proteins Hb and serum albumin. Biomonitoring of MDA has been carried out by GC/MS measurement of cysteine adducts in Hb from the nitroso metabolite of MDA (7, 8) or by measuring MDA and N-acetylMDA in urine (9). In addition, its products of reaction with guanine in DNA, N-(deoxyguanosin-8-yl)-MDA and N′-acetyl-N-(deoxyguanosin-8-yl)-MDA, have been analyzed by the 32P-postlabeling method (10). MDA released by alkaline hydrolysis of DNA from MDI- and MDAtreated rats has been measured by GC/MS (11). The nitrosoarenes which give rise to the cysteine adducts, often measured for the purpose of exposure monitoring of aromatic amines, are formed through oxidation by oxy-Hb locally in the erythrocytes (cf. ref 12) and are not necessarily identical with the genotoxic metabolite(s). Besides that, the rates of formation of nitroso metabolites of aromatic amines do not necessarily reflect the rates of formation of mutagenic/carcinogenic metabolites. Furthermore, by analyzing hydrolyzable adducts from nitroso metabolites, differences between individuals in the metabolism with respect to mutagenic metabolites cannot be detected. Interindividual variation seems to be an important factor for cancer susceptibility, fast acetylators of aromatic amines being associated with higher risk of colon cancer and slow acetylators with higher risk of bladder cancer (13). It is therefore obvious that the dosimetry to be used for risk assessment should comprise a determination of adducts from metabolites responsible for the mutagenic and carcinogenic action. The present study aimed at the identification of MDA and MDI adducts to Hb that occur besides the cysteine adducts and that might reflect doses of genotoxic intermediates. Hb adducts were investigated in mice treated by ip injection of MDA and radiolabeled MDA or deuterated MDA and in rats which had received MDI by inhalation. Certain in vitro experiments were carried out for the verification of chemical structure.

Materials and Methods Caution: Methylenedianiline (MDA) is classified as an animal carcinogen and as a suspected human carcinogen and should be handled carefully. Chemicals. MDA, formaldehyde, and aniline were purchased from Aldrich Chemical (Milwaukee, WI). L-Valine methyl ester, lactoperoxidase (EC 1.11.1.7), trypsin (EC 3.4.21.4), and pentafluoropropionic anhydride (PFPA) were purchased from Sigma Chemical Co. (St Louis, MO), and pronase was from Boehringer Mannheim, Germany. All other chemicals were of best quality available. Syntheses. 4,4′-[3H]Methylenedianiline (3H-MDA) was synthesized from [3H]formaldehyde (DuPont NEN, Boston, MA) and aniline. Methylenedi[2,3,5,6-2H4]aniline (MDA-d8) was synthesized by incubating [2,3,4,5,6-2H5]aniline (Aldrich Chemical, Milwaukee, WI) and formaldehyde. Both of these syntheses were carried out according to Bailey et al. (7). The products were purified on a silica column eluted with chloroform and MDA-d8 was further recrystallized from aqueous methanol. The purities of 3H-MDA and MDA-d8 were checked by HPLC. The EI mass spectrum of MDA-d8 gave a base peak at m/z 206, which is consistent with the molecular mass, and fragment ions at m/z 190 and 110, due to loss of -NH2 and cleavage of the methylene bridge, respectively. Animal Treatments. Male C57Bl/6 mice, 8-10 weeks old, were treated by ip injection with 25 mg/kg of body wt of MDA or MDA-d8, dissolved in DMSO, four times with 1-day interval between injections. The last injection of MDA-treated mice was made with 3H-MDA (s.a. 19 mCi/mmol). Control animals were treated with DMSO only. The mice were sacrificed 2 days after

Chem. Res. Toxicol., Vol. 11, No. 6, 1998 615 the last injection. Blood was collected into heparinized tubes, the isolated erythrocytes were washed twice with H2O, hemolysates were dialyzed overnight, and globin was precipitated with acetone/HCl according to Anson and Mirsky (14). MDI Exposure of Rats. Female Wistar rats were exposed for 17 h/day, 5 days/week over 3 months, and the concentration of MDI in the inhalation chambers was 2.06 mg/m3. The conditions for the exposure have been published in detail elsewhere (15). Erythrocytes from two pooled blood samples obtained from four exposed and four control rats, respectively, were washed, and globin was precipitated with acetone/HCl. Enzymatic Hydrolysis of Globin. Globin (200 mg) from MDA- and MDA-d8-treated mice and from control mice, respectively, was dissolved in 40 mL of 0.1 M phosphate buffer, pH 7.4, and hydrolyzed by trypsin and pronase at 37 °C for 24 h. Two samples of 200 mg of globin from MDI-treated rats and from control rats, respectively, were also hydrolyzed by the same procedure. Hydrolyzed samples were evaporated, dissolved in a minimal amount of water, and applied on a Sep-Pak C-18 column (2 g) which was eluted with 8 mL of H2O, 6 mL of 10% (v/v) of MeOH in H2O, and finally 8 mL of 80% (v/v) MeOH/ H2O. The last eluate was evaporated and esterified by MeOH/ HCl, and the samples were fractionated (1 fraction/min) on a C-18 HPLC column (Hichrom, Reading, U.K.; 10 mm × 250 mm, flow 3 mL/min), eluted with H2O and MeOH on a linear gradient from 40% to 100% MeOH for 30 min. Aliquots of fractions from samples from 3H-MDA-treated mice were counted by a liquid scintillation counter, and the fractions with the main peak of radioactivity (fractions 17-23) were pooled and evaporated for GC/MS analysis. These fractions corresponded to approximately 45% of the total radioactivity in the globin. Corresponding fractions were also pooled from other hydrolyzed samples. Reaction of MDA with Valine Methyl Ester in Vitro. To 30 mL of 50 mM phosphate buffer, pH 7.0, were added 1 µmol of MDA, 5 µmol of valine methyl ester, 5 µL of 30% H2O2, and 100 µg of lactoperoxidase, and the mixture was incubated for 2 h at 37 °C. One reaction mixture was incubated without the addition of lactoperoxidase and one mixture without H2O2 and lactoperoxidase. The samples were extracted twice with 30 mL of H2O-saturated ethyl acetate. Evaporated organic phases were separated on an HPLC column; fractions were pooled and derivatized with PFPA and analyzed by GC/MS (cf. Enzymatic Hydrolysis of Globin, above). GC/MS Analysis. Aliquots of the fractions 17-23 from HPLC separations (animal samples) were derivatized with PFPA. Underivatized and derivatized samples were analyzed using a Varian 3400 gas chromatograph (GC) linked to a Finnigan TSQ 700 mass spectrometer. The operating conditions for the GC were as follows: helium was used as carrier gas at constant pressure of 8 psi, the samples were injected with a septum-equipped on-column injector (Varian) programmed from 70 to 300 °C, 175 °C/min. The GC oven was programmed from 100 °C (held 1 min) to 240 °C, 20 °C/min, then to 350 °C, 10 °C/min, and held for 5 min. A 30-m DB-17ht (0.33-mm i.d., 0.25µm phase thickness) fused silica column (J&W Scientific Inc., Rancho Cordova, CA) coupled to a phenylmethyl-deactivated retention gap (2.5 m, 0.53-mm i.d.) was used. The operating parameters for the MS were as follows: the electron energy was 70 eV, ion source temperature was 150 °C, for chemical ionization (CI) mode methane was used as the reagent gas. Collision-induced dissociation (CID) spectra were produced using argon as collision gas (pressure 1 mTorr) with the collision energy of 10-30 eV. The MS analysis of the PFPA-derivatized samples from MDI-exposed rats and corresponding control samples was carried out in the negative ion CI (NICI) mode and selected ion monitoring (SIM) of the most abundant fragment, m/z 449 (M - HF)-.

Results Analysis of the Major Globin Adduct in MDATreated Mice. Underivatized and PFPA-derivatized

616 Chem. Res. Toxicol., Vol. 11, No. 6, 1998

Kautiainen et al.

Figure 1. Full-scan EI mass spectra of esterified samples from (a) MDA- and (b) MDA-d8-treated mice.

pooled fractions containing the main peak of radioactivity (45% of the total) from HPLC separations of hydrolyzed globins were characterized by GC/MS. Full-scan MS analysis, in the electron impact (EI) ionization mode (Figure 1) of the samples obtained from MDA- and MDAd8-treated mice, exhibited similarities in the fragmentation pattern, the fragment ions with two or one benzene rings retained in MDA-d8-globin being generally 8 or 4 mass units heavier, respectively. Thus, the analysis by MS of samples from MDA-treated mice gave the base peak at m/z 323, interpreted as the molecular ion of the compound. In the corresponding sample from MDA-d8dosed animals, the base peak at m/z 331 was recorded. PFPA-derivatized samples (Figure 2) revealed the base peaks at m/z 469 and 477 (MDA- and MDA-d8-treated mice, respectively), interpreted as molecular ions. Promi-

nent fragments and their relative intensities from nonacylated and acylated samples are shown in Figure 3. The corresponding compounds could not be detected in the nonacylated or acylated samples from control mice. CID spectra of the fragment ions m/z 224, 223, 182, and 180 were recorded in the nonacylated sample from MDA-treated mice (cf. Figure 1) using EI ionization. The fragment ion of m/z 224 revealed the product ion spectrum with the base peak at m/z 106 and prominent fragments at m/z 223 (83% relative intensity), 195 (68), 182 (25), 180 (92), and 165 (46). The corresponding spectrum of the precursor ion of m/z 223 gave the base peak at m/z 180 with one major fragment at m/z 206 (36). For the fragment ion of m/z 182, the major product ions were 165 (base peak) and 180 (71), and for the precursor

Hb Adducts from a Methylenedianiline Metabolite

Chem. Res. Toxicol., Vol. 11, No. 6, 1998 617

Figure 2. Full-scan EI mass spectra of esterified and PFPA-derivatized samples from (a) MDA- and (b) MDA-d8-treated mice.

ion of m/z 180, the ion m/z 152 was recorded as the base peak. Full-scan MS analysis of nonacylated samples using positive ion chemical ionization (PICI) revealed fragments at m/z 324, 352, and 364 (MDA-globin) and m/z 332, 360, and 372 (MDA-d8 globin). These fragments correspond to M + 1, M + 29, and M + 41. For corresponding PFPA-derivatized samples, fragment ions at m/z 470, 498, and 510 (MDA-globin) and m/z 478, 506, and 518 (MDA-d8-globin) were recorded. Using NICI mode, these samples gave the base peaks at m/z 449 and 457 (MDA- and MDA-d8-globin, respectively), interpreted as (M - HF)- ions. Characterization of the in Vitro Reaction Product of MDA with Valine Methyl Ester. The samples obtained from in vitro incubations were derivatized by

PFPA and analyzed by GC/MS in the NICI mode. The full-scan MS analysis of the sample incubated with both H2O2 and lactoperoxidase revealed the same spectrum, with the base peak at m/z 449, and with identical retention time as the sample obtained from MDA-globin. MS/MS analysis of daughter ions of m/z 449 showed strong similarity with the analysis of the sample obtained from MDA-treated mice. Using SIM analysis of the ion m/z 449, the same product, but in lower yield, was monitored in the sample from the incubation mixture with only H2O2 added. In the sample incubated in the absence of both H2O2 and lactoperoxidase, no reaction product could be detected. In addition, when MDA and valine methyl ester were incubated in the presence of P4501A2 and a NADPH-generating system, no reaction product could be seen using the same preparation of the

618 Chem. Res. Toxicol., Vol. 11, No. 6, 1998

Kautiainen et al.

Figure 3. Suggested fragments and their relative intensities of the nonacylated and acylated reaction products upon EI ionization. Acyl groups of PFPA-derivatized samples not shown.

sample as used for the in vitro samples in this study (unpublished results). GC/MS Analysis of Globin Samples from MDIExposed Rats. Enzymatically hydrolyzed globin samples (200 mg) from control and MDI-exposed rats (two samples, respectively) were analyzed by GC/MS (NICI mode) and SIM of the ion m/z 449. The same reaction product and with the same retention time as determined in the corresponding material from MDA-treated mice and in the samples prepared in vitro by incubation of MDA and valine methyl ester in the presence of H2O2 could be detected in animal samples. In MDI-exposed rats the relative mean adduct level was, however, about 60-fold higher compared to the adduct level in control animals, confirming that MDI is hydrolyzed to MDA in vivo. Suggested Structure of the Reaction Product. The molecular masses, 323 and 331, of the adducts measured in enzymatically hydrolyzed globin samples from mice treated with MDA and MDA-d8, respectively, differ by 8 amu. Since all eight ring-bound hydrogens are retained in the adduct, it is concluded that MDA has reacted by one of its amino nitrogens. The proposed structure and some fragments of the reaction products and the PFPA-derivatized reaction products, respectively, are shown in Figure 3. The retention of all deuterium atoms in essential 2H-substituted fragments excludes the four carbon atoms ortho to the amino groups of MDA as otherwise expected electrophilic sites in reactive intermediates derived from an N-hydroxylated metabolite. The reaction of one of the amino groups is also in agreement with the EI-MS analyses of acylated samples which revealed base peaks at m/z 469 and 477, i.e.,

increments with 146 mass units when compared with nonacylated samples and, thus, admitting acylation of only one of the amino groups. Informative fragments are also given by the ions m/z 223 and 231 and m/z 369 and 377 (nonacylated and acylated samples, respectively), interpreted as formed via an intramolecular methyl transfer from the methoxycarbonyl group with loss of 3,3dimethylacrylic acid. Such a reaction in the gas phase would be highly facilitated by a transition state involving a five-membered ring. The characteristic valine fragment, although at low abundance, i.e., the m/z 72, interpreted as the immonium ion of the valine residue, was present in both MDA and MDA-d8-samples, as well as the fragment ion m/z 130, interpreted as a valine methyl ester fragment. In the CID spectra of the suggested molecular ions m/z 323 and 331, the ion of m/z 72 was prominent in both samples (12% and 14% in MDA and MDA-d8 samples, respectively; data not shown). The proposed structure is further supported by the recorded CID spectra of some fragment ions (cf. above and Figure 3). Thus, the base peak of the product ions of m/z 224, i.e., m/z 106, interpretetad as p-aminobenzyl ion, has most likely been formed by a direct loss of C7H6N2 from the precursor ion. The product ion of m/z 195, obtained from m/z 224, is apparently formed by the loss of a methylnitrene moiety, and another prominent product ion of m/z 182 has most likely been formed by the loss of diazomethane from the precursor ion. The fragmentation of the radical ion m/z 223 revealed the base peak at m/z 180, supposed to be formed by a concurrent loss of N2 and a methyl radical. The base peak of the precursor ion m/z 182, i.e., m/z 165, which is

Hb Adducts from a Methylenedianiline Metabolite

also present in the CID spectrum of m/z 224, has probably been formed by the loss of NH3 from m/z 182 followed by cyclization to a fluorenyl ion. The product ion m/z 152, recorded as the base peak in the CID spectrum of m/z 180, is suggested to be formed by the loss of H2CN and cyclization to a biphenylene radical ion. Unfortunately, attemps to verify the tentative structures of product ions by CID of the deuterated compound were not unambiguous because of scrambling of the hydrogen and deuterium atoms. NICI analysis of the PFPA-derivatized samples showed base peaks at m/z 449 and 457, respectively, interpreted as [M - HF]- ions, common fragments of fluorinated compounds in NICI-MS analysis. Confirmative evidence for the suggested molecular masses of the reaction products is given by the PICI analyses, whereby characteristic fragments M + 1, M + 29, and M + 41 were recorded for both nonacylated and acylated samples. Taken together, the structural information from this reaction product is consistent with a coupling of activated MDA to the nitrogen of N-terminal valine in Hb with formation of an azo-MDA-monoimine adduct (Figure 3). The recorded molecular mass would be obtained with oxidative loss of four hydrogens in the formation of the adduct. Further support for the proposed structure is given by the fact that the same reaction product that was obtained from MDA-treated mice, with identical mass spectrum and the same retention time in GC analysis, could be prepared in vitro by incubating MDA with valine methyl ester in the presence of H2O2 and, with higher yield, when lactoperoxidase was added to the incubation mixture. It should be admitted that the proposed structure (cf. Figure 3), containing an azo group bound to a methine group in the R-position to a carbonyl function, could undergo a well-known tautomerization to a hydrazone structure. However, at present we consider the available data to favor the azo structure of the compound. A first argument follows from the probably higher resonance stabilization of the azo derivative as compared to the hydrazo form, decreasing the rate of isomerization and shifting the equilibrium position toward the azo form. The second argument in favor of an azo structure is that, in the present case, the methine group of the azo adduct to valine methyl ester is activated by only one methoxycarbonyl group, possibly not sufficient to induce rapid tautomerization under nearly neutral conditions. The full verification of this proposed structure has to await NMR studies on synthetic material. Quantitative Aspects. In the formation of the studied major adduct from [methylene-3H]MDA, one of the methylene (-CH2-) hydrogens is lost (cf. Figure 3). If the so far unknown isotope effect for this reaction is low (∼1) or high (.1), the observation that 45% of the globinbound 3H is associated with the studied major adduct would mean that this adduct constitutes 62% or 45%, respectively, of the total adduct level. Against this background it has been concluded that the studied adduct constitutes “about 50%” of the total adduct level in mouse globin. The total radioactivity from 3H-MDA associated with globin amounts to about 0.8% and that of the studied valine adduct to about 0.4% of the injected material, based on the assumption that the globin has been hydrolyzed completely into single amino acids. These figures are relatively high compared with experience from

Chem. Res. Toxicol., Vol. 11, No. 6, 1998 619 Scheme 1. Suggested Formation of the Reactive Metabolite (b) from MDA (a)

other reactive intermediates. A similar figure, 0.4% of total globin-bound radioactivity, was obtained in the study of Bailey et al. (7) of 14C-MDA in rats. In the latter study about 40% of the globin-associated radioactivity could be released hydrolytically as free amine. In the present study the relative amount of this cysteine adduct was a few percent only (data not shown). This difference between rats and mice may, apart from differences in metabolism, be associated with the high reactivity of cysteine β125 in rat globin (16).

Discussion This study shows the formation of a major Hb adduct from MDA in rodents, with a structure deviating from that of adducts formed via the cytochrome P450 (P450) pathway. In the mouse this adduct constitutes about 50% (cf. Quantitative aspects, above) of the total level of globin adducts. Further, it is shown that the adduct identified is formed via a peroxidative pathway. Aspects of the metabolism and of the chemistry of the formation of the adduct are discussed below. Viewpoints on Metabolism. Aromatic amines can be metabolized by two different pathways, mediated either by the P450 system or by peroxidases. The initial step of the P450 metabolism involves N-oxidation to N-hydroxyarylamines, mainly by the hepatic 1A2 isoenzyme (17). Further metabolic steps comprise, e.g., N- and O-acetylation by acetyltranferases and conjugation by UDP-glucuronyltransferases and sulfotransferases. DNA adducts identified so far are, above all, reaction products of the amino moiety of aromatic amines and C8 of guanine or adenine and the reaction products of the carbon adjacent to the amino group in the aromatic amine and the exocyclic amino group of guanine (18). In erythrocytes, N-hydroxyarylamines can be further oxidized by oxy-Hb to nitrosoarylamines, which rapidly react with thiols, such as cysteine residues in Hb, to yield sulfinamide adducts (12). Various peroxidases utilize arylamines as cosubstrates and oxidize them to reactive macromolecule-binding species. Thus, benzidine has been shown to be metabolized by extrahepatic peroxidases, such as prostaglandin H synthetase (PHS) and lactoperoxidase, via free-radical intermediates, to benzidinediimine followed by the binding to proteins and DNA (19-21). Other aromatic amines shown to be bioactivated by peroxidase pathways comprise 2-naphthylamine, 2-aminofluorene, and 4,4′methylenedi(2-chloroaniline) (22, 20, 23). Tentative Pathway for the Adduct Formation. Like benzidine but unlike most aromatic monoamines, MDA can formally undergo two consecutive one-electron oxidations with formation of a neutral, oxygen-free quinonoid dehydrogenation product (Scheme 1). The electrophilic character of this monoimine is born out by

620 Chem. Res. Toxicol., Vol. 11, No. 6, 1998

the reported rapid and quantitative oxidative dimerization of MDA at high concentration to 4,4′-bis(4-aminobenzyl)azoxybenzene by treatment under alkaline conditions at room temperature with a 3-fold exess of potassium hexacyanoferrate(III) (24). The imine b in Scheme 1, in our view a probable reactive intermediate in the above oxidative dimerization, has been prepared in quite another context and analyzed as its resonance-stabilized hexafluorophosphate by Marji et al. (25). The reaction of the monoimine b with the nitrogen of N-terminal valine would lead to the formation of a hydrazine. This hydrazine may rapidly autoxidize to the corresponding azo compound (26), and further oxidation of the second aromatic amino group would lead to the formation of a quinonoid imine (cf. Figure 3). The driving force for the formation of the quinonoid imine is most likely the considerable resonance stabilization of the reaction product which therefore can be expected to be favored over the amine form. Indications of the formation of DNA adducts with a hydrazine configuration involving the amino group of the aromatic amine and cyclic or exocyclic nitrogens of guanine lend support to the suggested structure. Thus, evidence for the initial formation of a 2-aminofluorene (2-AF) nitrogen adduct at the N7 position of guanine, followed by rearrangement to the C8 adduct, has been reported (27). Further, the reaction of the exocyclic nitrogen of 4-(hydroxyamino)quinoline 1-oxide and N7 of guanine has been demonstrated (28). When benzidine was incubated in the presence of PHS and DNA, the reaction products were assumed to comprise, besides N-(deoxyguanosin-8-yl)benzidine and N,3-(deoxyguanosinN7,C8-yl)benzidine, also N-(deoxyguanosin-N2-yl)benzidine, i.e., a compound with a hydrazo configuration (21). Interestingly, when Schu¨tze et al. (11) analyzed DNA adducts in liver from MDA-treated rats, the major adduct found using HPLC analysis and 32P-postlabeling did not correspond to the synthetic guanine adducts N-(deoxyguanosin-8-yl)-MDA and N′-acetyl-N-(deoxyguanosin-8yl)-MDA, the adducts expected to be formed via P450mediated metabolism. The major adduct was not characterized, however. This also concerns a deoxyguanosine adduct formed in PHS activation of 2-AF, differing from synthetic standards of guanine C8 adducts (29). It is thus possible that N-hydroxylation is not a necessary step in bioactivation of MDA and other aromatic amines. This is also supported by the study of McGregor et al. (30) who showed that rat liver S9 is not a requirement for the significant response to MDA in a mouse lymphoma cell forward mutation assay. Heme-catalyzed oxidation of N-hydroxy-MDA might give rise to 4-amino-4′-nitrosodiphenylmethane, which might be assumed to react with terminal valine amino groups to give the azo-type derivative detected in the present study (cf. Figure 3). However, while primary arylamines are well-known to condense with nitrosoarenes, corresponding reactions involving alkylamines have to our knowledge not been described in the literature. A recent study by Eyer and Ascher (31) concerning the reactivity of nitrosobenzenes showed no reactions with nucleophilic sites in Hb when the SH groups were blocked with N-ethylmaleimide. It can thus be concluded that the reaction product determined in the present study is unlikely to be formed from the reaction of a nitroso derivative of MDA.

Kautiainen et al.

Analysis of Hb from MDI-Exposed Rats. The results of the analyses of globin samples from MDIexposed rats are consistent with the study of Sepai et al. (15) who measured MDA and N-acetyl-MDA, released by basic hydrolysis, in Hb from the same rats as were analyzed in this study. In control rats, up to 3 pmol/g of Hb was measured, whereas in MDI-exposed rats the amount of MDA was approximately 48 pmol/g of Hb. MDA was detected also in urine from control rats. Using the same method as the one of Sepai et al., MDA has also been detected in Hb from knowingly unexposed humans (8). The sources of the observed background adduct levels are not known, but exposure to MDI and MDA due to leakage from polyurethanes present in several manufactured products is possible.

Conclusion This study shows that MDI and MDA give rise to stable adducts to N-terminal valine in Hb which can be measured by GC/MS after enzymatic hydrolysis of globin and also demonstrates that MS analysis of Hb adducts is a useful tool for the identification of reactive in vivo intermediates. Further studies are needed to evaluate the role of the peroxidative activation pathway of MDA in humans, in comparison to the P450-mediated metabolism.

Acknowledgment. The authors thank Prof. D. Henschler, University of Wu¨rzburg, Germany, and Dr. E. Bailey, Surrey, U.K., for valuable viewpoints and Fraunhofer-Institut fu¨r Toxikologie und Aerosolforschung, Hannover, Germany, for the blood samples from MDIexposed rats. The study was supported financially by the Ko¨rber Foundation, Hamburg, Germany, and the Swedish Environmental Protection Agency.

References (1) International Agency for Research on Cancer (IARC). (1986) 4,4′Methylenedianiline and its dihydrochloride. IARC Monographs on the Evaluation of the Carcinogenic Risk of Chemicals to Humans 39, 347-365. (2) Shimizu, H., Suzuki, T., Suzuki, T., Akiyama, I., Sakitani, T., and Takemura, T. (1982) Mutagenicity of epoxy resin hardeners. Jpn. J. Ind. Health 33, 426-427. (3) Swenberg, J. A. (1981) Utilization of the alkaline elution assay as a short-term test for chemical carcinogenesis. In Short-Term Tests for Chemical Carcinogenesis (Stich, H. F., and San, R. H. C., Eds.) pp 48-58, Springer-Verlag, New York. (4) Parodi, S., Taningher, M., Russo, P., Pala, M., Tamaro, M., and Monti-Bragadin, C. (1981) DNA-damaging activity in vivo and bacterial mutagenicity of sixteen aromatic amines and azoderivatives, as related quantitatively to their carcinogenicity. Carcinogenesis 2, 1317-1326. (5) Go´recka-Turska, D., Mekler, U., and Go´rski, T. (1983) Sister chromatid exchanges in BALB/c mice bone-marrow induced by carcinogenic and noncarcinogenic products and intermediate products from the dyestoff industry (in Polish). Bromatol. Chem. Toksykol. 16, 37-42. (6) National Toxicology Program. (1983) Carcinogenesis Studies of 4,4′-Methylenedianiline Dihydrochloride (CAS No. 13552-44-8) in F334/N Rats and B6C3F1 Mice (Drinking Water Studies). Technical Report No. 248, Research Triangle Park, NC. (7) Bailey, E., Brooks, A. G., Bird, I., Farmer, P. B., and Street, B. (1990) Monitoring exposure to 4,4′-methylenedianiline by the gas chromatography-mass spectrometry determination of adducts to hemoglobin. Anal. Biochem. 190, 175-181. (8) Schu¨tze, D., Sepai, O., Lewalter, J., Miksche, L., Henschler, D., and Sabbioni, G. (1995) Biomonitoring of workers exposed to 4,4′methylenedianiline or 4,4′-methylenediphenyl diisocyanate. Carcinogenesis 16, 573-582. (9) Cocker, J., Boobis, A. R., and Davies, D. S. (1988) Determination of the N-acetyl metabolites of 4,4′-methylenedianiline and 4,4′-

Hb Adducts from a Methylenedianiline Metabolite

(10)

(11) (12) (13) (14) (15)

(16)

(17) (18)

(19)

(20)

(21)

methylene-bis(2-chloroaniline) in urine. Biomed. Environ. Mass Spectrom. 17, 161-167. Vock, E. H., Hoymann, H.-G., Heinrich, U., and Lutz, W. K. (1996) 32P-Postlabeling of a DNA adduct derived from 4,4′-methylenedianiline in the olfactory epithelium of rats exposed by inhalation to 4,4′-methylenediphenyl diisocyanate. Carcinogenesis 17, 10691073. Schu¨tze, D., Sagelsdorff, P., Sepai, O., and Sabbioni, G. (1996) Synthesis and quantification of DNA adducts of 4,4′-methylenedianiline. Chem. Res. Toxicol. 9, 1103-1112. Albrecht, W., and Neumann, H.-G. (1985) Biomonitoring of aniline and nitrosobenzene. Arch. Toxicol. 57, 1-5. Hein, D. W. (1988) Acetylator genotype and arylamine-induced carcinogenesis. Biochim. Biophys. Acta 948, 37-66. Anson, M. L., and Mirsky, A. E. J. (1930) Protein coagulation and its reversal. The preparation of insoluble globin, soluble globin and heme. J. Gen. Physiol. 13, 469-476. Sepai, O., Schu¨tze, D., Heinrich, U., Hoymann, H. G., Henschler, D., and Sabbioni, G. (1995) Hemoglobin adducts and urine metabolites of 4,4′-methylenedianiline after 4,4′-methylenediphenyl diisocyanate exposure of rats. Chem.-Biol. Interact. 97, 185-198. Hamboek, H., Fischer, R. W., Di Iorio, E. E., and Winterhalter, K. H. (1981) The binding of s-triazine metabolites to rodent hemoglobin appears irrelevant to other species. Mol. Pharmacol. 20, 579-584. Guengerich, F. P., and Shimada, T. (1991) Oxidation of toxic and carcinogenic chemicals by human cytochrome P-450 enzymes. Chem. Res. Toxicol. 4, 391-407. Beland, F. A., and Kadlubar, F. F. (1990) Metabolic activation and DNA adducts of aromatic amines. In Chemical Carcinogenesis and Mutagenesis I (Cooper, C. S., and Grover, P. L., Eds.) pp 267325, Springer-Verlag, Berlin. Zenser, T., Mattammal, M. B., Armbrecht, H. J., and Davis, B. B. (1980) Benzidine binding to nucleic acids mediated by the peroxidative activity of prostaglandin endoperoxide synthetase. Cancer Res. 40, 2839-2845. Kadlubar, F. F., Frederic, C. B., Weis, C. C., and Zenser, T. V. (1982) Prostaglandin endoperoxide synthase-mediated metabolism of carcinogenic aromatic amines and their binding to DNA and protein. Biochem. Biophys. Res. Commun. 108, 253-258. Yamazoe, Y., Zenser, T. V., Miller, D. W., and Kadlubar, F. F. (1988) Mechanism of formation and structural characterization

Chem. Res. Toxicol., Vol. 11, No. 6, 1998 621

(22)

(23)

(24)

(25) (26)

(27)

(28)

(29)

(30)

(31)

of DNA adducts derived from peroxidative activation of benzidine. Carcinogenesis 9, 1635-1641. Flammang, T. J., Yamazoe, Y., Benson, R. W., Roberts, D. W., Potter, D. W., Chu, D. Z. J., Lang, N. P., and Kadlubar, F. F. (1989) Arachidonic acid-dependent peroxidative activation of carcinogenic arylamines by extrahepatic human tissue microsomes. Cancer Res. 49, 1977-1982. Kadlubar, F. F., Butler, M. A., Kaderlik, K. R., Chou, H.-C., and Lang, N. (1992) Polymorphism for aromatic amine metabolism in humans: Relevance for human carcinogenesis. Environ. Health Perspect. 98, 69-74. Reitzenstein, F. (1910) Verfahren zur Darstellung von Azoxyverbindungen. (Procedures for preparation of azoxy compounds.) J. Prakt. Chem. 82 (2), 252-270. Marji, D., Tashtoush, H., and Ibrahim, J. (1986) A novel synthesis of quinone-immonium dyes. Tetrahedron 42, 4517-4521. Newbold, B. T. (1975) Oxidation and synthetic uses of hydrazo, azo and azoxy compounds. In The Chemistry of the Hydrazo, Azo and Azoxy Groups, Part 1 (Patai, S., Ed.) pp 541-597, John Wiley & Sons, London, New York, Sydney, Toronto. Humphreys, W. G., Kadlubar, F. F., and Guengerich, F. P. (1992) Mechanism of C8 alkylation of guanine residues by activated arylamines: Evidence for initial adduct formation at the N7 position. Proc. Natl. Acad. Sci. U.S.A. 89, 8278-8282. Kohda, K., Kawazoe, Y., Minoura, Y., and Tada, M. (1991) Separation and identification of N4-guanosin-7-yl)-4-aminoquinoline 1-oxide, a novel nucleic acid adduct of carcinogen 4-nitroquinoline 1-oxide. Carcinogenesis 12, 1523-1525. Boyd, J. A., and Eling, T. E. (1984) Evidence for a one-electron mechanism of 2-aminofluorene oxidation by prostaglandin H synthase and horseradish peroxidase. J. Biol. Chem. 259, 1388513896. McGregor, D. B., Brown, A., Cattanach, P., Edwards, I., McBride, D., Riach, C., and Caspary, W. J. (1988) Responses of the L5178Y tk+/tk- mouse lymphoma cell forward mutation assay: III. 72 Coded chemicals. Environ. Mol. Mutagen. 12, 85-154. Eyer, P., and Ascher, M. (1987) Reactions of para-substituted nitrosobenzenes with human hemoglobin. Biol. Chem. HoppeSeyler 368, 285-294.

TX9701485