Evidence for direct-acting oxidative genotoxicity by reduction products of azo dyes.

The intestinal flora forms a complex ecosystem that metabolizes dietary and endogenous nutrients under primarily anaerobic conditions. The ingestion of azo dyes has been proposed as one source of potential genotoxic agents. Many intestinal bacteria are able to reduce the azo bond (termed azofission), which liberates the substituted naphthol compounds. The standard Ames test has not demonstrated mutagenicity either by various common food colorings or by their reduced end products in Salmonella typhimurium strains TA98 and TA100. In contrast, genetic toxicity was demonstrated in the Escherichia coli differential kill assay and in S. typhimurium TA102 for the reduced dyes. The superoxide free radical was produced by the azo dyes only after reduction by the intestinal bacteria Enterococcus faecalis and Bacteroides thetaiotaomicron.


Introduction
Azo compounds are the most common synthetic colorings used in the food, pharmaceutical, and cosmetic industry. Also known as coal tar dyes, they contain an aromatic ring linked by an azo bond to a second naphthalene or benzene ring. Coloring matter entering the intestinal tract is subjected to the action of acid, digestive enzymes, and microflora. Azo compounds may reach the intestine directly after oral ingestion or through the bile after parenteral administration. They are reduced by azo reductases from intestinal bacteria and, to a lesser extent, by enzymes of the cytosolic and microsomal fractions of the liver. The first catabolic step in the reduction of azo dyes, which is accompanied by a decrease in the visible light absorbance and then decoloration of the dye, is the reduction of the azo bond to produce aromatic amines ( Figure 1). Aromatic amines, some of which are known carcinogens, have been found in the urine of dyestuff workers and test animals following administration of azo dyes (1).
Although a number of commonly used dyes are not mutagenic in Salmonella typhimurium strains TA98 and TA100 even after azoreduction, the production of reactive oxygen species from o-hydroxy aromatic amine products has been suggested (2,3). This article reports the generation of superoxide anions from reduced azo dyes and aminonaphthols and the genetic toxicity of these products in Escherichia coli and a Salmonella strain (TA102), which is sensitive to oxidants.

Bacterial Reduction ofAzo Dyes
The standard assay mix consisted of 4.2 ml degassed potassium phosphate buffer (50 mM, pH 7.4), 0.4 ml washed cell suspension with or without glucose (0.2 ml, 10% wt/vol) and electron carriers in narrow assay tubes (18 x 142 mm). The various electron carriers were prepared anaerobically at 0.025 mM. The reaction mix was incubated at 37°C prior to starting the reaction by the addition of 0.1 ml of the appropriate azo dye (10 mM). Decoloration rate was determined spectrophotometrically at the wavelength of the maximal absorbance of the azo dye.

Chemical Reduction Assay
Dyes were reduced either with sodium dithionite according to the method of Zbaida and Levine (4), or with stannous chloride as described by Gasparic (5).

Detection ofSuperoxide Anion Radical
Superoxide anion (O°2) production was detected by the reduction of nitroblue tetrazolium (NBT) and its inhibition by superoxide dismutase (SOD), according to the method of Oberley et al. (6). The sample (100 pl) to be tested was added to 0.8 ml potassium phosphate buffer (50 mM, pH 7.8) containing 0.056 mM NBT, 0.1 mM EDTA, 0.1 mM xanthine, 0.06% wt/vol Triton X100, and 0.33 mg/ml gelatin, ± 100 pl SOD. Immediately after mixing, the absorbance at 560 nm was measured with respect to time. SOD activity was assayed using a xanthine/xanthine oxidase technique (6). SOD was inactivated by boiling for 15 min.

Differential Killing Assay
The E. coli differential cytotoxicity assay was used to determine the genotoxic activity of the azo dyes according to the method described by Tweats et al. (7). The tester strains were E. coli WP2, (repair proficient) and E. coli CM871 (repair deficient). E. coli CM871 is a uvrA recA lexA triple mutant that combines extreme repair deficiency with near wild-type growth. Compounds were accepted as genotoxic if survival of the repair deficient strain was at least 4-fold lower than survival of the repair proficient strain. All compounds were tested for genotoxicity with and without prior incubation with either E. faecalis or B. thetaiotaomicron cell suspensions. Incubation was for 1 hr with E. faecalis or overnight with B. thetaiotaomicron. Final concentrations of agents used were azo dyes, 5 mM; 1aminonaphthol HCl, 0.1 mM; 1-aminonaphthol-4-sulfonic acid, 10 mM; 4-aminonaphthalene HCl, 1 mM; and 4aminonaphthalene-sulfonic acid sodium salt, 5 mM.
Genotoxic activity was assessed after removing cells by centrifugation in a microfuge (8000g, 5 min), followed by filtration (0.2 pm Dynaguard filters, New Brunswick). An aliquot (100 pl) of the filtrate was added to the diluted E. coli tester strain (200 pl), and the genotoxic activity was determined. Inhibition by UV was used as a positive control for CM871.
The Rapid Automated Bacterial Impedance Technique (RABIT), (Don Whitley Scientific Ltd, Shipley, UK) and plate counts were used to measure microbial survival (8).

Salmonel Mutagenicity Test
The protocol used was essentially that of Ames et al. (9). The strains of S. typhimurium used were TA98, TA100, and TA102. Liver postmitochondrial supernatant was not incorporated in this test. Bleomycin was used as a positive control for TA102 (9).

Results
The reduction of the three dyes by E. faecalis and B. thetaiotaomicron is summarized in Table 1. Azo reductase activity was stimu-lated by the addition of extracellular electron acceptors such as FMN, FAD, and RF. The presence of glucose on reduction of azo dyes was inhibitory for B. thetaiotaomicron, but not for E. faecalis. This may be because of the different electron transport systems of these organisms.
After incubation with E. faecalis, the cleavage products of sunset yellow and amaranth were found to be genotoxic in the bacterial differential assay (Figure 2). Carmoisine was marginally genotoxic after azo reduction, but survival of E. coli CM871 was not consistently 4-fold less than survival of E. coli WP2. Similar results were obtained on incubation with B. thetaiotaomicron. Unreduced dyes were not genotoxic in this assay. Of the four coupling components tested, 1-amino-2-naphthol HCl and 1-amino-2-naphthol-4-sulfonate are considered genotoxic.
Mutagenicity studies on the same samples using the standard Ames test with S. typhimurium TA98 and TA100 showed no mutagenicity. When compounds were retested using S. typhimurium TA102, reduced amaranth and reduced sunset yellow gave a doubling of the spontaneous reversion rate (Figure 3). This strain detects a variety of oxidants as mutagens that are not detected by strains TA98 or TA100 (10).
Generation of°2 was observed for reduced dyes and some aminonaphthols but not for unreduced dyes (Figures 4,5). The increase in absorbance at 540 nm (NBT assay) was inhibited by SOD, while inactivated SOD showed no inhibition. Table 2 summarizes the rates of°2 production from unreduced and reduced azo dyes and their coupling components. Results were further confirmed by an ability to reduce cytochrome c (data not shown). Negligible

Discussion
It has been demonstrated that there are no cytochromes present in cell-free extracts of E. faecalis and that the transfer of electrons from reduced pyridine nucleotides to electron acceptors was mediated by a flavoprotein (11).
In view of their charged nature, it is uncertain whether FAD and FMN could readily shuttle across bacterial cell walls and membranes under physiologic conditions. In addition, it is unlikely that highly charged sulfonated dyes could pass through the cell wall of these organisms. An electron shuttle could affect the reduction with the dyes remaining extracellular. It may also serve as a partial explanation for the oxygen sensitivity of these reduction reactions, 1.47 With Inactivated SOD 1.2a reduced flavin having a greater affinity for oxygen than it has for the azo dye. In this article, we have tested three azo dyes for genotoxicity following bacterial reduction of the dye. Both amaranth and sunset yellow, when reduced, induced cytotoxicity indicating DNA damage in repairdeficient E. coli in the absence of hepatic enzymes, but they failed to mutate S. typhimurium strains TA98 and TAIOO. In contrast, strain TA102, which detects oxidative mutagens, (10,12) was mutated by reduced amaranth and reduced sunset yellow.
The findings suggest that various azo dye products are genotoxic, not through Nhydroxylation and esterification, which is characteristic of many aromatic amines (2,13), but rather through a mechanism involving oxygen radicals (3,14). In further support, the production of O2 radical from reduced azo dyes was detected by the  Figure 6. Possible mechanism for production of superoxide anions from reduced sunset yellow. Adapted from Nakayama (3).
Volume 102, Supplement 6, October 1994 reduction of nitrotetrazolium blue and confirmed using cytochrome c. The mechanism by which these species are formed is not clear, but they may be specific for o-hydroxy aromatic amines and involve the reactions postulated by Nakayama (3) and summarized in Figure 6. There may be a requirement for iron, as in Fenton chemistry, where 0°is converted via H202 to the highly reactive OH species that is known to damage DNA (14). All of the compounds shown to generate active oxygen species have a hydroxyl substituent ortho to the amino function. However, this does not explain the weakly mutagenic results for carmoisine. 02was detected for 1amino-2-naphthol HCI and 1-amino-2naphthol sulfonate, while no active oxygen was detected for 4-aminonaphthalene HCI and its sodium salt.
We are currently studying the ability of o-hydroxy aromatic amines to produce deoxynucleotide oxidation in DNA. The identification of this type of DNA damage hitherto not detected by many conventional genotoxicity assays may have important implications regarding the continued use of azo dyes in foodstuffs.