Metabolism of the food-borne carcinogens 2-amino-3-methylimidazo-[4,5-f]quinoline and 2-amino-3,8- dimethylimidazo[4,5-f]-quinoxaline in the rat as a model for human biomonitoring.

Metabolism of 2-amino-3-methylimidazo[4,5-f]quinoline (IQ) and 2-amino-3,8-dimethylimidazol[4,5-f]quinoxaline (MeIQx) and their binding to blood proteins were examined in the rat to develop methods of human biomonitoring. Hemoglobin and serum albumin were among many blood proteins modified. Approximately 0.01% of the dose for both compounds was bound to these proteins, and induction of cytochrome P-450 with polychlorobiphenyls resulted in decreased levels of adduction. Hemoglobin sulfinic acid amide adducts could not be detected for either amine, however, as much as 10% of the IQ bound to albumin was characterized as an N2-cysteine(34)sulfinyl-IQ linkage. Human dosimetry of these carcinogens through such adducts may prove difficult due to the low levels of protein binding. Major routes of detoxification of both contaminants included cytochrome P-450-mediated ring hydroxylation at the C-5 position followed by conjugation to glucuronic or sulfuric acid. Direct conjugation to the exocyclic amine group through N-glucoronidation and sulfamate formation were other important routes of inactivation, but N-acetylation was a minor pathway. The N-glucoronide conjugate of the mutagenic metabolite N-hydroxy-MeIQx was also detected in urine. Rats given MeIQx at 10 micrograms/kg excreted 20% of the dose in urine within 24 hr and the remainder was recovered in feces. The N2-glucuronide was the major metabolite found in urine and accounted for 4% of the total dose. The other metabolites cited above also were excreted in urine at amounts ranging from 0.5 to 3% of the dose, whereas 0.5 to 2% was detected as unmetabolized MeIQx.(ABSTRACT TRUNCATED AT 250 WORDS)


Introduction
Methods have been developed to rapidly quantitate carcinogenic heterocyclic amines formed in cooked foods such as meat and fish (1)(2)(3)(4). However, measurement of these contaminants in food only provides a crude estimate ofexposure and does not account for absorption and metabolism ofthese procarcinogens to biologically active species or to detoxified products. Many ofthese mutagens contain an aminoimidazole group as a common structural feature. 2-Amino-3-methylimidazo[4,5-A quinoline (IQ) and 2-amino-3,8-dimethylimidazo[4,5-Jquinoxaline (MeIQx) are two important representatives ofthis class of contaminants. We have used the rat as an animal model to elucidate routes ofmetabolism and disposition ofthese two compounds. Particular emphasis has been placed on the analysis of acid-labile sulfinamide protein adducts derived from reactive Nhydroxy metabolites and on the analysis ofnontoxic metabolites excreted in urine for developing noninvasive methods ofhuman biomonitoring. We have observed that human liver tissue also transforms heterocyclic amines to reactive mutagens as well as to detoxified products. Our recent data on the quantification of heterocyclic amines in cooked foods, the use ofthe rodent model for developing methods of human biomonitoring, and preliminary data from human studies are presented in this article.
Male Sprague Dawley rats were used for protein adduct and metabolism studies (5,6). In brief, hemoglobin was purified by lysis oferythrocytes, followed by gel filtration chromatography. Serum albumin was purified by affinity chromatgraphy using Cibracon Blue conjugated to sepharose. Acid-labile sulfinamide protein adducts were quantitated by HPLC. Metabolites were isolated from biological fluids and reference standards were prepared biosynthetically using rat hepatocyte suspensions or through chemical syntheses (5,7).
Analysis of MeIQx excreted in urine of humans following consumption of fried beefwas done by the method of Murray et al. (8) using negative-ion chemical ionization GC-MS except that an immunoaffinity purification step was included following the acid/base partitioning step (3). Three subjects ingested 1 lb equivalent cooked beef prepared under typical household cooking practices and then collected urine for 24 hr. Trideuteriolabeled (N-Cd3)-MeIQx was used as an internal standard. MeIQx and d3-MeIQx were monitored at the ions m/z 438 and m/z 441 [corresponding to the M-227 ions of the di(3,5-bistrifluoromethylbenzyl)derivatives]. Human hepatic microsomes and cytosols were used to measure the apparent rates of Noxidation and N-acetylation ofheterocyclic amines (9). S-9 fortified with adenosine 3 '-phosphate-5 '-phosphosulfate (PAPS) was used to examine sulfamate formation.

Results
Heterocyclic amines can be detected at the low parts per billion levels in meat and fish prepared under typical household cooking practices. The production ofthese genotoxins in foods varies greatly and depends on the meat, temperature, and manner of preparation ( Table 1).
The major metabolites of IQ and MeIQx that have been identified in urine, bile, and feces of rodents are displayed in Figure  1 (5,(10)(11)(12). All these metabolites are detoxification products with the exception of the N-glucuronide conjugate of Nhydroxy-MeIQx, which is genotoxic in the presence of j-glucuronidase (7).
We examined the metabolism ofMeIQx at doses ranging from 0.01 to 20 mg/kg in noninduced rats and in rats pretreated with polychlorinated biphenyls (PCB) (5) (Fig. 2). At high dose exposure to MeIQx (20 mg/kg), the sulfamate and N2-glucuronide were the major metabolites excreted in urine, whereas PCBpretreated animals excreted greater amounts of conjugates of 5-hydroxy-MeIQx. Cytochrome P-450 induction had no influence on metabolism at the 0.01 mg/kg dose, indicating that under high exposure to MeIQx the cytochrome P-450 is limiting in the noninduced rat and phase II conjugation reactions make a larger contribution to metabolism. Notably, the formation and excretion of the metastable N-glucuronide conjugate of Nhydroxy-MeIQx was relatively more important at low-dose exposure.
The blood protein binding ofIQ was examined at doses from 2 to 150 itmole/kg (6). Among many proteins modified, hemoglobin and albumin were modified in a dose-dependent fashion. Albumin bound three to five times more IQ than hemoglobin per mole of protein. The amount of IQ bound to the total albumin pool ranged from 1.4 to 4.3 x 10-2% of the dose (Fig. 3).
Analysis ofthe enzymatically digested peptide fragments revealed many adducts. One adduct was identified as a tripeptide containing an N2-cysteine sulfinamide-IQ linkage and accounted for as much as 10% of the IQ bound to serum albumin ( Fig. 4). A chemically identical adduct was formed in vitro when Nhydroxy-IQ was incubated with serum albumin. The adduct was labile to acid with quantitative recovery of the parent amine.  (2,3), Gross (4), and Gross and Gruter (24  Adducts bound to hemoglobin were briefly examined for the presence of a sulfinamide adduct. Acid hydrolysis released approximately 30% of the bound IQ. Analysis by HPLC revealed that the product(s) were more polar than IQ and excluded the presence of a sulfinamide linkage with hemoglobin. The chemical structures of the polar products have not been elucidated.
The binding of MeIQx to albumin and hemoglobin was several fold lower than that observed for IQ. Hepatic cytochrome P-450 induction with PCB increased the rate of formation of the biologically reactive N-hydroxy metabolite by 20-fold in microsomal assays, but resulted in a 10-fold decrease in blood protein binding in vivo (5). Thus, in contrast to other aromatic amines that are converted to the hydroxylamine and adduct to hemoglobin as a sulfinamide linkage through the 93,B cysteine residue in vivo (13), IQ and MeIQx apparently do not form this adduct at appreciable levels.
The ability of human liver tissue to metabolize heterocyclic amines was examined. The apparent rates of N-oxidation were compared with that of4-aminobiphenyl (ABP), which is regarded as the most potent ofarylamine carcinogens. The levels ofIQ, MeIQx, and Glu-P-l N-oxidation were about half of that observed for ABP, while that of PhIP was slightly higher ( Table  2). Heterocyclic amines were poor substrates for cytosolic Nacetyltransferases which was in contrast to the arylamine carcinogens ABP or 2-aminofluorene (14), (Table 2). However, human liver S-9 fraction fortified with PAPS was found to readily detoxify IQ and MeIQx through sulfamate formation (R. Turesky, unpublished observations).
The urine of three subjects was examined for MeIQx before and after consumption of 1 lb ofcooked beef. MeIQx could not be detected in urine collections before consumption of meat, but the mutagen was detected in all three subjects after the meal (Fig.  5). The amounts ofMeIQx recovered ranged in values from 6 to 10 to 18 ng in a 24 hr urine collection. These values are similar to those ofMurray et al. (8) and indicate that MeIQx is absorbed and extensively biotransformed by humans.

Discussion
Heterocyclic amines formed in cooked foods at the low part per billion level are easily isolated by immunoaffinity chromatography or by solid-phase tandem extraction and then quantitated by HPLC. The class and the amounts of heterocyclic amines produced depend on several parameters including temperature, creatinine content, and meat preparation (15). PhIP, followed by AaC and MeIQx, were the most abundant heterocyclic amines found in grilled steak, fried beef, and salmon as well as in barbecued salmon. IQ and MeIQx were the predominant mutagens formed in meat extracts. These amounts are comparable to those reported by other investigators who used far more laborious methods ofpurification (15,16). Based upon these analyses, the daily exposure to each ofthese amines through the diet may be estimated at approximately 100 ng to 10 4tg per day.  The metabolism and blood protein binding of IQ and MeIQx in the rodent were examined to develop strategies for human biomonitoring. Measurement ofblood protein adducts has been successfully used to assess human exposure and metabolic activation for several different carcinogens, including 4-aminobiphenyl, aflatoxin, and several polycyclic aromatic hydrocarbons (13,17). The rodent model has been found to be a good surrogate for blood protein adduct formation by several ofthese carcinogens including the arylamine ABP, where over 5 % ofan administered dose is bound to hemoglobin as a sulfinamide linkage. This adduct is the result of a series of reactions between the hemoprotein and the carcinogenic N-hydroxy metabolite. The adduct is stable in vivo, but it can be cleaved in vitro with quantitative regeneration ofthe parent amine. Thus, measurement of this adduct is an indirect measure of metabolic activation and the biologically effective dose. Relative to ABP, the binding of IQ and MeIQx to hemoglobin was quite low, accounting for approximately 0.01% ofthe dose, and sulfinamide adducts could not be detected (5,6). Induction of hepatic cytochrome P-450 by PCB, which increased rates of formation ofthe hydroxylamine metabolites, actually resulted in as much as a 10-fold decrease in protein binding (5,6). Incubation ofthe microsomally generated hydroxylamines or the synthetic N-hydroxy derivatives ofIQ and MeIQx in vitro with erythrocytes generated methemoglobinemia and sulfinamide adduct formation. Thus, the hydroxylamines can penetrate the erythrocyte and react with the hemoprotein. The absence of such an adduct in vivo suggests that either very low levels of the N-hydroxy metabolites are excreted by the liver into the blood stream or that other routes ofbiotransformation are of far greater importance than N-hydroxylation. Based on the rodent model, formation ofsulfinamide adducts in humans would not be expected at appreciable levels, and human dosimetry may prove difficult. In support ofthis conclusion, a preliminary study assaying for hemoglobin sulfinamide adducts of MeIQx in hu-  aRates of N-oxidation were determined with hepatic microsomes from the same individual (n = 3,4). Similar results were obtained with microsomes from two other individuals (9,19).
Aproximately 0.001-0.004% of an administered dose of IQ bound to serum albumin in the rat as a sulfinamide adduct. In man, with a half-life of20 days for albumin turnover, the adduct level resulting from chronic exposure is approximately 30 times greater than that produced by a I-day exposure (13). Assuming an human exposure of 1 jig/day with the same adduct binding efficiency as the rat, the amount of serum albumin modfied from chronic exposure in a 70-kg individual containing 130 g of albumin would be approximately 2-8 pg of IQ bound/g of albumin. With this level ofmodification in humans, methods of enrichment would be required to measure covalent binding of sulfinamide albumin adducts by GC-MS.
We undertook urinary metabolism studies in the rodent with MeIQx at doses from 20 mg/kg to 10 jig/kg, a dose that is only several hundred-fold greater than the daily human exposure. MeIQx was extensively metabolized at the lowest dose examined with only 0.5-2% of the total dose recovered in urine found as unchanged MeIQx. In humans, the amount of MeIQx recovered in urine represented only several percent of the ingested dose and indicates that this procarcinogen is also absorbed and extensively metabolized by humans (8).  GC-MS analysis is sufficiently sensitive to detect MeIQx in urine following consumption of fried meat. Biotransformation of heterocyclic amines by human tissues resembles that of rodents in several instances. Metabolic activation through N-hydroxylation by hepatic cytochrome P-450 IA2 is comparable for both species (19)(20)(21). Both rodent and human liver O-acetyltransferases catalyze binding of the N-hydroxy metabolites to DNA-bound products, yet the parent amines are poor substrates for hepatic N-acetyltransferases (9,21 ). The Nhydroxy metabolites have also been shown to be substrates for human and rodent hepatic glucuronyltransferases (5,7,10,22), and the resulting metastable N-glucuronide conjugates may be implicated in colorectal carcinogenesis (22). A major route of detoxification of heterocyclic amines in rodents, in particular for IQ and MeIQx, is through sulfamate formation. This route of metabolism for aromatic amines is relatively uncommon, owing in part to the labililty of the arylsulfamate bond. We have observed that human liver tissue also converts these two amines to their respective sulfamate derivatives (R. Turesky, unpublished observations).
The major oxidative pathways of detoxification of IQ and MeIQx found in the rodent (5,(10)(11)(12), have also been identified in the monkey (23; R.Turesky and E. Snyderwine, unpublished observations). Other important routes of detoxification were through sulfamate and N2-glucuronide formation. Thus, there is a strong possibility that humans may also transform these compounds in similar fashions. Investigations are underway for detecting some of these polar metabolites of MeIQx that may be present in human urine, in particular, the N-hydroxy-N-glucuronide, which is an indirect measurement of metabolic activation. The development ofsuch biomarkers may enable us to better evaluate the health risk ofchronic dietary consumption oflow amounts of heterocyclic amines.