Hemoglobin binding of arylamines and nitroarenes: molecular dosimetry and quantitative structure-activity relationships.

N-Oxidation and nitroreduction to yield N-hydroxyarylamines are metabolic steps that are crucial for the genotoxic properties of aromatic amines and nitroarenes, respectively. N-Hydroxyarylamines can form adducts with DNA, tissue proteins, and the blood proteins albumin and hemoglobin in a dose-dependent manner. The determination of hemoglobin adducts is a useful tool for biomonitoring exposed populations. We have established the hemoglobin binding index (HBI) [(mmole compound/mole Hb)/(mmole compound/kg body weight)] of several aromatic amines and nitroarenes in female Wistar rats. Incorporating values obtained by other researchers in the same rat strain, the logarithm of hemoglobin binding (log HBI) was plotted against several physicochemical parameters and against calculated electronic descriptors of nitroarenes and arylamines. Most arylamines and nitroarenes form hydrolyzable (e.g., sulfinamide) adducts with hemoglobin in rats. The amount of hemoglobin binding decreases with the oxidizability of the arylamines, except for compounds that are substituted with halogens in ortho or meta position. For halogen-substituted arylamines, the amount of hemoglobin binding is directly proportional to the pKa. Hemoglobin binding of nitroarenes increases with the reducibility of the nitro group. The structure activity relationships (SAR) for hemoglobin binding of nitroarenes and arylamines are comparable. The SAR found for hemoglobin binding were compared with the SAR found in the literature for mutagenicity, carcinogenicity, and cytotoxicity of arylamines and nitroarenes. In general, the mutagenicity or carcinogenicity of arylamines increases with their oxidizability. This first set of data suggests that the levels of hemoglobin binding, mutagenicity, and carcinogenicity of arylamines are not determined by the same electronic properties of the compounds, or not by these properties alone.(ABSTRACT TRUNCATED AT 250 WORDS)


Introduction
Aromatic amines and nitroarenes are very important intermediates in industrial manufacturing of dyes, pesticides, and plastics and are significant environmental pollutants. Ring oxidation, N-glucuronidation, N-acetylation, and N-oxidation are the major metabolic pathways of arylamines in mammals [review of metabolism in (1,2)]. N-Oxidation is a crucial step in the metabolism of arylamines and aromatic amides to toxic products. Arylamines are metabolized in the liver by monooxygenases to highly reactive N-hydroxyarylamines. Nitroarenes are reduced by microorganisms in the gut or by nitroreductases and aldehyde dehy-drogenase in hepatocytes to nitrosoarenes and N-hydroxyarylamines (3). N-Hydroxyarylamines can be further metabolized to N-sulfonyloxyarylamines, N-acetoxyarylamines or N-hydroxyarylamine N-glucuronide. These highly reactive intermediates are responsible for the genotoxic and cytotoxic effects (4)(5)(6)(7) of this class of compounds. They react with DNA and proteins (Scheme 1). For 4-aminobiphenyl (4ABP) (8), it has been shown that the N02 NO NH, Scheme same metabolite-the N-hydroxyarylamine-reacts with DNA of the bladder and with hemoglobin.
The main goal of the present work was to determine the bioavailability of the Nhydroxyarylamine for an array of substituted arylamines and nitroarenes by measuring the amount of hydrolyzable adducts formed with rat hemoglobin. The relationships between hemoglobin binding of nitroarenes and arylamines and their HENMOGLOBIN (10). 'New experimental data. The most stable conformer is with the methyl group anti to the proton on the nitrogen. gThe most stable conformer is with the methyl group syn to the proton on the nitrogen. electronic properties were studied. The structure-activity relationships (SAR) for hemoglobin binding were compared with those for their mutagenic and carcinogenic potency in order to test to which extent hemoglobin binding may be used as a predictor for genotoxicity.

Materials and Methods
The experimental details of the work with nitroarenes have been published elsewhere (9). The methods for the animal experiments, the isolation of hemoglobin, and the quantification of the arylamines bound to hemoglobin have been published recently (10). The aromatic amines and nitroarenes were given to female Wistar rats by gavage, and the rats were sacrificed 24 hr later. The hemoglobin was precipitated with ethanol, hydrolyzed in 0.1 M NaOH in the presence of recovery standards [e.g., 4-chloroaniline (4CA), d5-aniline] and extracted with hexane. The hexane fraction was analyzed by gas chromatography-mass spectrometry (GC-MS) with electron impact ionization in the single ion mode. Structure identification was based on the retention time and on the mass spectrum or the ratio of the main mass fragments. Arylamines with low hemoglobin binding were derivatized with pentafluoropropionic acid anhydride and analyzed by GC-MS.
To establish whether the aromatic amines recovered from the alkaline hydrolysis were covalently bound, all samples were extracted with hexane at neutral pH and analyzed by GC-MS. The electronic properties of the arylamines and the nitroarenes were calculated using the programs Modified Neglect of Differential Overlap (MNDO), Austin Model 1 (AM 1), and Parametric Method number 3 (PM3), which are part of MOPAC 6.0 (Quantum Chemistry Program Exchange, Indiana University, Bloomington, IN) (11). The oxidizability of arylamines was determined experimentally by HPLC equipped with an electrochemical detector (10). The electrode potential was decreased stepwise (0.05 V) from 1 to 0.4 V. The peak integrals obtained (average of two injections) were plotted against the electrode potential.
The half wave oxidation potential (E1/2) was obtained from the resulting hydrodynamic voltammograms. The values obtained are listed in Table 1.

Hemoglobin Binding ofArylamines and Nitroarenes
Rats were dosed with arylamines or nitroarenes and sacrificed after 24 hr. Hemoglobin was hydrolyzed with NaOH. The released arylamine was extracted in hexane and analyzed by GC-MS. The results are summarized in Figures 1A,B, and 3. The following structure activity relationships were found: The highest hemoglobin binding was obtained with compounds with a halogen in para position. A chlorine atom in ortho position reduces the formation of hemoglobin adducts drastically (1000-fold, for 2CA compared to 4CA). An additional ortho chlorine atom, as in 2,6-dichloroaniline (26DCA) or 2,3,4,5,6-pentachloroaniline (PCA), abolishes hemoglobin binding totally. All alkyl substituted amines have lower hemoglobin binding index (HBI) [(mmole compound/mole Hb)/(mmole compound/kg body weight) than aniline. The HBI of 3-ethylaniline (3EA) is higher than that of 2EA or 4EA. This might be explained by the fact that the oxidation of alkyl groups in ortho or para position to an amino group is facilitated compared with that of alkyl groups in meta position. Two methyl groups in ortho position, as in 2,6dimethylaniline (26DMA) or 2,4,6trimethylaniline (246TMA), almost abolish hemoglobin binding.
In general, lower hydrolyzable hemoglobin-adduct levels were found in rats that were given nitroarenes than in rats that were dosed with an equimolar amount of the corresponding arylamines The log HBI of aromatic amines for MNDO  stable conformation was obtained starting with a dihedral angle of 00. The energy differences between the two conformations are up to 1.7 kcal/mole for AMI and h nitrobenzenes with elec-MNDO, and 0.5 kcal/mole for PM3 calcusubstituents (e.g., 4-methyllations. The most stable geometry obtained 2,4-Dichloronitrobenzene for 2EA, 3EA, and 4EA is with the second 10) and 2,3,4,5,6-pentacarbon of the ethyl group out of the benizene (PCNB) (12)  order to bind to hemoglostable structure for the nitrenium ion of first have to be oxidized to 2EA has a dihedral angle Cl-C2-CH2rlamines. In the liver this CH3 of 00. Care must be taken with the tly catalyzed by cytochrome initial geometry of 4-methylmercaptoaniduct distribution of this oxi-line (4MSA) for MNDO calculations. increases the stability of this amine by 1.8 kcal/mole. The stability of the two possible rotamers of unsymmetrically substituted nitrenium ions were compared. In the previous publication (10) only the anti-conformer with the single proton on the nitrogen on the less substituted side was studied; H-N-C1-C2 = 1800. This was found not to represent the most stable conformer in all cases. Therefore, for the present study we have calculated stability of the syn conformer, with the proton on the nitrogen on the more substituted side with a dihedral angle H-N-C1-C2 = 00. We found that compounds with ethyl or methyl groups in ortho position the anti conformers are up to 0.9 kcal/mole more stable. Nitrenium ions with an ortho chloro group are more stable in the syn conformation as calculated by MNDO    lated by MNDO. For amines in which nitrenium ions are less stable than the one of aniline, the enthalpy change is <0 kcal/mole. The half wave oxidation potentials of arylamines correlate inversely with the stability of their nitrenium ions, calculated by MNDO, AM1, and PM3 with r = -0.93, -0.95, and 0.77, respectively. Arylamines that form a stable nitrenium ion have a smaller oxidation potential than arylamines that form more unstable nitrenium ions (e.g., 246TMA compared to 4CA). The oxidizability of the arylamines is directly proportional to the stability (AMNDOHF) of the corresponding nitrenium ions. X-Ph-NH+ + PhNH2 -* X-Ph-NH2 + PhNH+ [1] The log HBI of all arylamines was plotted against AMNDOHF, AAM1HF, and APM3HF. The best correlation (r= -0.92) was found for hemoglobin binding ofparasubstituted and alkyl-substituted arylamines and the stability of their nitrenium ions calculated with MNDO ( Figure 1A). Poorer calculations are found when log HBI was plotted against the stability of nitrenium ions calculated by AM 1, PM3, and the half wave oxidation potential with r = -0.80, -0.70, and 0.78, respectively. For AM 1 and the half wave oxidation potential the correlations to log HBI increase to 0.91 and 0.86 if 4MSA is excluded. The compounds with halogens in ortho or meta postion, 3-(trifluoromethyl)-aniline (3TFA), 3-cyanoaniline (3CNA), 4-(trifluoromethyl)-aniline (4TFA), and 4ABP do not fit the curve. From these SARs found in rats, it appears that the pharmacokinetics or the metabo- Nitroarenes. Nitroarenes have to be reduced to nitrosoarenes or to N-hydroxyarylamine to yield the same sulfinamide adducts as do arylamines. Therefore, hemoglobin binding of nitroarenes should depend on the ease of reduction of the nitro group. The energy level of the lowest unoccupied molecular orbital (ELUMO) is a good parameter for predicting the reducibility of nitroarenes (25,26). Thirteen nitroarenes were tested for hemoglobin binding. The logarithm of the hemoglobin binding index was plotted against ELUMO (Figure 3). The para-and alkyl-substituted arylamines fit the regression curve very well. As in the case of arylamines, compounds with halogens in ortho or meta position are outliers. Analysis of a larger group of nitroarenes has been recently published (9).
Except for two arylamines (2,6-dichloroaniline [26DCA] and PCA) and two nitroarenes (24DCNB and PCNB), all arylamines and nitroarenes given to female Wistar rats formed hydrolyzable hemoglobin adducts. Therefore, for most compounds the potentially genotoxic intermediate-Nhydroxyarylamine-is bioavailable. Hemoglobin binding can be predicted with the electronic properties of the arylamines and nitroarenes. In a further analysis we determined if the same electronic properties are predictive for the carcinogenic, mutagenic, and cytotoxic potency of these arylamines and nitroarenes.

Mutagenicity ofArylamines and
Nitroarenes Mutagenicity ofArylamines. Are the electronic properties responsible for high HBI values the same as those responsible for high mutagenic and carcinogenic potency? For several arylamines, it has been shown that the mutagenic potency increases with the stability of the corresponding nitrenium ions (14,15,27). However, several compounds that are not mutagenic should be mutagenic according to their electronic properties, for example, aniline, 3CA, 2C4MA, and 4C2MA in Salmonella typhimurium TA98. Even with the inclusion of additional parameters (partition coefficients, the energy levels of the LUMO, and highest occupied molecular orbital of the arylamines) in a predictive equation, the mutagenicity of several arylamines is not predicted correctly (28).
In the present work, the data available for the mutagenic potency (28-32) of arylamines, expressed as logarithm of revertants per nmole compound (log MUT), were plotted against log HBI of the arylamines and the stability of the corresponding nitrenium ions. The mutagenic potency is directly proportional to the oxidizability of the arylamines (Figure 4) (e.g., 2,4,5trimethylaniline [245TMA] is more mutagenic than 4CA), but inversely proportional to the amount of hemoglobin binding in rats ( Figure 5). In addition, several arylamines (e.g., aniline, 3-chloroaniline [3CA], 2C4MA, and 4C2MA) that are not mutagenic, bind to hemoglobin.
Mutagenicity ofNitrobenzenes. The logarithm of the mutagenicity (34,36) of the nitroarenes which had been tested for homoglobin binding were plotted against the reducibility of the nitro group (ELUMO). The mutagenic potency and the E LUMO fit on a linear regression line ( Figure 6). The mutagenicity of nitroarenes increases with the reducibility of the nitro group. Except for NB all compounds tested which bind to hemoblobin are mutagenic. Conversely, 24DCNB is mutagenic but does not bind to hemoglobin. Although hemoglobin binding increases with the reducibility of the nitro group, the correlation of mutagenicity with hemoglobin binding is very poor. This may be a function of insufficient data points, thus further analyses are necessary.

Carcinogenicity ofArylamines and Nitroarenes
Since the late 1970s, there has been a great deal of interest to elucidate the chemical properties of aromatic amines Figure 7. Carcinogenicity of arylamines in rats. Log 1/TD50 [mmolel was plotted against the relative stability of the nitrenium ions (AMNDOHF): r= 0.63. The TD50 values were obtained from Gold et al. (38). and nitroarenes that are responsible for the genotoxicity of this class of compounds (16)(17)(18)37). Much emphasis has been put on the structural features of the ultimate carcinogen. Although a bimolecular mechanism has been postulated in certain cases, most authors interpreted the carcinogenic potencies or the reactions with DNA with a nitrenium ion as an intermediate (16)(17)(18)37). For a comparison of hemoglobin binding with the carcinogenicity of arylamines in rats, the TD50 data compiled by Gold et al. (38) were used. TD50-values (mmole) of arylamines tested in rats were found for only five monocyclic arylamines. Carcinogenicity increases with the oxidizability of the arylamines (Figure 7). Carcinogenicity is inversely proportional to hemoglobin binding for these compounds (Figure 8). Only in the case of the bifunctional arylamines 3,3'-dichlorobenzidine, 4,4'methylenedianiline, 4,4'-methylenebis(2chloroaniline), 4,4'-oxydianiline, and benzidine do carcinogenicity and hemoglobin binding correlate positively (log(1/TD50 [mmole])=0.7+0.80 log HBI; r=0.85, data not shown). For the monocyclic nitroarenes investigated here, not enough data are available to study the correlation of carcinogenicity with hemoglobin binding.

Conclusions
Except for four compounds, it could be shown that after treating rats with nitroarenes or arylamines, hydrolyzable hemoglobin adducts are formed as a result of the formation of the potentially genotoxic intermediate N-hydroxyarylamine. Hemoglobin adducts are therefore a good dosimeter to biomonitor people exposed to a large array of arylamines and nitro- arenes. The amount of hemoglobin binding decreases with the oxidizability of the arylamines, except for compounds that are substituted with halogens in ortho or meta position. For halogen-substituted arylamines, the amount of hemoglobin binding is directly proportional to the pK. The level of hemoglobin binding and mutagenicity is directly proportional to the reducibility of the nitroarenes, but hemoglobin binding and mutagenicity do not correlate. For arylamines, the electronic properties that are important for mutagenicity or carcinogenicity are not the same as those important for hemoglobin binding. Moreover, the correlation of carcinogenicity or mutagenicity of arylamines with the electronic properties of the corresponding nitrenium ions is not as good as that for hemoglobin binding. For an equation that better predicts carcinogenicity, other parameters that are important in the process of carcinogenesis may have to be included (e.g., cytotoxicity, Km and V values of phase I and phase II max enzymes, partition coefficient octanolwater).
Experiments to test for cytotoxic effects of metabolites of arylamines or nitroarenes have been performed with hepatocytes by O'Brien et al. (5) and de Silva et al. (4). For nitroarenes, it was shown that cytotoxicity is increased by electron-withdrawing groups in para position to the nitro group.
For arylamines, a new mechanism for cytotoxic effects of N-hydroxy-2-aminofluorene and N-hydroxy-2-aminophenanthrene has been proposed by Neumann et al. (6,7). In vitro experiments with mitochondria showed that N-hydroxy-2aminofluorene or 2-nitrosofluorene Volume 102, Supplement 6, October 1994 caused cyanide-resistant oxygen consumption and calcium release. The formation of superoxide anion radical was demonstrated. However, several monocyclic arylamines (nitrosobenzene, 2-nitrosotoluene, 4-nitrosotoluene, N,N-dimethyl-4-nitrosotoluene, and 4-nitrosophenol) did not cause any oxidative stress; some of them (nitrosobenzene and 4-nitrosotoluene) induced calcium release. Further work has to be performed to see which structural parameters determine the cytotoxic properties of arylamines and their metabolites.