Reactions of aqueous chlorine and chlorine dioxide with model food compounds.

Chlorine and chlorine dioxide (ClO2), common disinfecting and bleaching chemicals used in the food industry, are potent oxidizing and chlorinating agents. Unfortunately, little is known about the nature of the reactions of chlorine with organic food constituents. This presentation reviews published information concerning the reactions of chlorine gas (Cl2[g]), aqueous chlorine, and ClO2 with model food compounds, the fate of chlorine during the chlorination of specific food products, and the potential toxicity of the reaction products. Fatty acids and their methyl esters react with chlorine with the degree of incorporation corresponding to their degree of unsaturation. Aqueous chlorine oxidizes and chlorinates lipids and amino acids much more readily than ClO2. Several amino acids are highly susceptible to oxidation and chlorination by chlorine compounds. Reactions of chlorine and ClO2 with several food products, including flour and shrimp, have also been characterized. In one model system, 99% of Cl2(g) either reacted with components of flour or was consumed by oxidation/chlorination reactions. The lipids extracted from the chlorinated flour contained significant amounts of chlorine. Exposure of shrimp to hypochlorous acid (HOCl) solution resulted in significant incorporation of chlorine into the edible portion. Although significant quantities of chlorine can be incorporated into specific model compounds and food products, the health risks associated with exposure to chlorinated organic products are unknown. Preliminary studies using the Ames Salmonella/microsome mutagenicity assay indicate that the reaction products from mixtures of aqueous chlorine and various lipids or tryptophan are nonmutagenic. Nevertheless, additional studies are warranted, so that the toxicological significance of these reaction products can be understood more fully.


Introduction
Chlorine has been used for many years by the food industry as the principal sanitizing and disinfecting agent (1) ( Table 1). The history of the discovery and use of chlorine in the food industry has been reviewed by Mercer and Somers (2) and Dychdala (3). Aqueous chlorine solutions are used to sanitize food processing equipment, to cool food containers, and to rinse various food products. Chlorine gas is also used in the flour industry as an oxidizing and bleaching agent and to improve the quality of flour. In the United States, chlorine and hypochlorites are considered "generally recognized as safe" (GRAS) for specific uses in food processing and for bottled water. In some cases, chlorine is considered a food additive, and therefore use levels have been established by the Food and Drug Administration.
This report is a review of the reactions of aqueous chlorine and chlorine dioxide with organic compounds *Food Science and Human Nutrition Department, University of Florida, Gainesville, FL 32611. tAuthor to whom correspondence and requests for reprints should be addressed.  found in foods and with specific food products. To understand these reactions more fully, the chemistry of aqueous chlorine and chlorine dioxide is first briefly reviewed. Unfortunately, most of the literature on reactions of chlorine and food chemicals has concentrated on net changes in gross chemical fractions of foQds such as lipids, glutens, and fat or lean portions rather than on the formation of potentially hazardous reaction prod-ucts. It is hoped that the information presented review will help in evaluating the possible healt associated with chlorination in food processing tainly, more information is needed concerning th and reactivity of chlorine used in various process idenity and toxicity of the reaction products, a levels of these compounds to which consumers , posed before we can fully understand the risks o chlorine in the food industry.
The reactivity of HOCI is based on the electr nature of the molecule at either the oxygen or thE ine atom (4). Reactions occurring at the oxygei produce a chloride ion by displacement. In re; with organic compounds, the chlorine atom may t electrophilic and may combine with an electron the substrate. This behavior accounts for the re; of HOCI with ammonia and amines, with phen( other aromatic substances, and in the formation roform from organic substrates. Chlorine dioxide (CI02) is a polar gas that i dissolves in, but does not react with, water (7 compound has a larger oxidation capacity than HOCI because it can accept 2.5 times more elE than HOCI, but its oxidizing potential is less, a cated by its redox potential (7): C102 + e > C102-C102 + 2H20 + 4e-> C1-+ 40H-In addition, there are no appreciable differences bactericidal activity of C102 as a function of pH Chlorine dioxide will oxidize and chlorinate many organic compounds, but at a much slower rate than that associated with HOCI reactions. In most cases, oxidation reactions predominate. Among organic compounds studied thus far, C102 is most reactive in aqueous solutions with aliphatic tertiary amines, aromatic amines, and phenolic compounds (7). In addition, C102 will oxidize aldehydes and ketones to form carboxylic acids. Chlorine dioxide does not react with saturated aliphatic hydrocarbons and aliphatic side chains, although the latter may be split from aromatic rings or other functional groups (7). The chlorination of aromatic and unsaturated aliphatic hydrocarbons by C102 has been reported (5,7); however, no trihalomethanes have been detected as reaction products of C102 with organic materials. Each of these reactions will be discussed in greater detail as it pertains to reactions with lipids, carbohydrates, and proteins.
The mechanisms of chlorination and oxidation of organic compounds by C102 are not known. Chlorination in aqueous solutions may occur indirectly through a progressive reduction of C102, which passes through the HOCI stage (7): C102 + e-* C102-C102-+ 2e-+ 3H+ -HOCI + H20 However, the mechanisms of chlorination by C102 and the above transformations are far from being resolved. +1.49V Reactions with Carbohydrates + 1.36 V Reactions of C12(g), aqueous chlorine, or C102 with + 0.90 v carbohydrates generally result in oxidation products.
In reactions with C12(g) or aqueous chlorine, glucose is gent in oxidized at carbons C-2 and C-3 to form ketones. Furtimes ther oxidation results in cleavage of the C-2 to C-3 bond and formation of carboxylic acids (9)(10)(11). Aqueous C102 ophilic will also react with glucose; the principal reactions inchlorvolve oxidation of the hydroxyl groups to form carbonyl n atom derivatives, which are subsequently oxidized to caractions boxylic acids (12). Reaction products of C102 and glucose ecome have not been identified. ,-D-Glucose reacts more rappair in idly with C102 than a-D-glucose (13). actions The reactions of C12(g), aqueous chlorine, or C102 with )ls and polysaccharides are primarily oxidative. The actions of Df chlo-C12(g) and aqueous chlorine on starch were extensively studied by Whistler and co-workers (9,(14)(15)(16)(17)(18)(19). These readily investigators found that the major effect of C12(g) on ). This starch is depolymerization induced by cleavage at the that of glycosidic bond between C-1 and the glycosidic oxygen ictrons atom. Depolymerization also occurs in starch slurries s inditreated with aqueous chlorine. The mechanism of the opening of the glycosidic bond may involve formation of a carbonium ion at carbon C-1 or a chloro-oxonium + 0.95 V complex at the glycosidic oxygen (19). The chloro-ox-+0.78V onium cation can be displaced with either hydroxyl anions from water to form normal sugar end units, or with in the the hydroxyl group at C-6 to form 1,6 anhydro rings, (8).
or levoglucosan units. The extent of levoglucose for-mation depends on the moisture content of the flour; a higher moisture content results in reduced formation. Oxidation of glucose units at the C-2 and C-3 carbons also occurs, but to a much lesser extent than depolymerization. Huang et al. (11) reported that depolymerization occurs mainly in the amylopectin fraction of starch. The improved baking properties of treated starch are attributed to depolymerization of the glycosidic bond and cleavage between the C-2 and C-3 carbons of glucose (10,11,20,21). The oxidation and depolymerization of starch by C12(g) or aqueous chlorine also occurs with the polysaccharide cellullose (19).
Oligosaccharides have been used as model compounds of cellulose to study C102 bleaching (22). As with reactions of C12(g) and aqueous chlorine, two principal reactions of C102 with the oligosaccharides occur, namely, oxidation of the glycosidic bond and oxidative cleavage of the C-2 and C-3 carbons to form carboxylic acids. The latter reaction proceeds through 2-ketoglucose and 2,3-diketoglucose intermediates. Oxidation may also occur in the cellulose molecules to produce gluconic acid units in the cellulose chains. Hemicellulose, which coexists with cellulose in natural products, reacts much more with C102 than does cellulose (7). However, the reaction products have not been identified.
Although oxidation is the predominant reaction of carbohydrates, Cranet et al. (23) demonstrated that low levels of volatile halocarbons may be produced when Dmannitol is mixed with aqueous chlorine. The mechanism of this reaction is believed to proceed via formation of a ketolactone, which can undergo ring opening by chloride, followed by the classical haloform process. Before this study, carbohydrates were not considered to be significant precursors of haloform formation (24).
The toxicological significance of chlorine-modified carbohydrates is not known. The potential formation of halocarbons from carbohydrates raises some concern about their production in treated flour and other food products. There is no evidence to determine whether other oxidation products are toxic after chronic exposure.

Reactions with Lipids
Lipids constitute a class of organic compounds in foods that readily reacts with chlorine. Leopold and Mutton (25) found that aqueous chlorine completely saturates triolein, the major reactions being addition of chlorine and hydroxyl groups to the double bonds. Chlorine dioxide will also react with triolein, modifying 25% to 50% of the double bonds present. Gilles et al. (26) found that C12(g) decreases total unsaturation in flour lipids. Although significantly modified, the lipid fraction of flour does not play a significant role in the cake-baking potential of flour (27,28).
Using HO36C1 and 36C102, Ghanbari et al. (5) measured the extent of incorporation of chlorine into free fatty acids and their corresponding methyl esters. The lipids were suspended in a buffered solution, and the radiolabeled aqueous chlorine or C102 solutions were then added. After 60 min, the modified lipids were extracted with chloroform:methanol (3:1), and the amount of radiolabel in the organic solvent fraction was determined. The amount of label present in the organic extract represented the amount of chlorine incorporated into the lipid. In reactions with either HO36CI or 36CIO2, the percentage of chlorine incorporation correlated with the number of double bonds in the lipid ( Table 2). Obviously, the increased number of reaction sites in the more unsaturated fatty acids allowed for greater interaction with chlorine. The free fatty acids incorporated significantly more 36C1 than the methyl esters with the decrease in reactivity of the methyl esters caused in part by their increased hydrophobic nature. After suspending the lipids in the buffered solution, the methyl esters separated out of solution and formed lipid droplets more rapidly than their corresponding fatty acids. Chlorine incorporation was more prevalent with aqueous chlorine than with C102. As previously mentioned, the major reaction of C102 is oxidation rather than chlorination. These results confirm the results of other investigators (25,29) that HOCI and C102 react with unsaturated fatty acids at their double bonds. In another study (30), chloroform extracts from aqueous chlorine-oleic acid reaction mixtures were methylated with diazomethane and analyzed by gas chromatography-mass spectrometry (GC-MS). The chromatogram suggested that six compounds were present in the solvent extractable fraction. The only chlorinated compound identified was 9-chloro-10-hydroxymethyl stearate. The other products were identified as hydroxylated products of oleic acid. The chlorinated derivative was the major product formed, representing greater than 95% of the modified lipids. The mechanism of reaction most likely involved addition of chlorine across the double bond to form a carbonium ion, which then combined with a hydroxyl group to form the chlorohydrin (30).
Reactions of C102 with methyl oleate, like aqueous chlorine, will result in oxidation and chlorination across the double bonds (25,29). However, the levels of chlorine incorporation into methyl oleate were much less than that with aqueous chlorine. In addition, unsaturated ketones and chloroketones are the result of a different reaction mechanism than that of aqueous chlorine. These products result from the formation of allyl radicals rather than from direct addition to the double bonds (29). Autoxidation also proceeds through this type of radical formation. The carbon-carbon double bond in the unsaturated hydroperoxide formed by autoxidation of methyl oleate has a trans configuration. The trans configuration of the unsaturated ketones obtained by C102 oxidation agrees with formation through allyl radicals.
Cunningham and Lawrence (31-36) conducted a series of experiments to determine the biological fate and toxicity of chlorinated fatty acids in rats. Chlorinated lipids are moderately toxic; the acute oral LD50 for chlorinated oleic acid was approximately 20 g/kg body weight (32). In addition, chlorination reduced the absorption and deposition of all lipids in tissues, the reduction being less severe with chlorinated oleic acid than with chlorinated linolenic or linoleic acid (31,33). Radiolabeled chlorinated fatty acids administered orally to pregnant and lactating rats may be transferred to the fetus and to suckling rats (34). Exposure to chlorinated flour lipids in the diet for 2 weeks reduced the growth rate and increased relative liver weights in rats (36). A decrease in body weight and a corresponding increase in relative organ weights are considered deleterious effects (37). Based on these results, the risks associated with exposure to chlorinated lipids are still not clear. The doses of chlorinated lipids used in these studies were much higher than would be expected from our food supply. However, the biological and possibly toxic effects of chlorinated lipids in laboratory animals should raise some concern about the chronic effects of chlorinated food commodities in consumers.
We conducted some preliminary experiments to determine the mutagenic potential of chloroform:methanol (3:1) extracts of reaction mixtures of aqueous chlorine and fatty acids using the Ames Salmonella/microsome mutagenicity assay (38). Approximately 5.0 mM aqueous chlorine plus 3.4 mM fatty acid in 0.1 M phosphate buffer, pH 6.0, were mixed for a period of time necessary for the levels of available chlorine, measured by iodometric titration, to drop to 25% of the original levels. The reactions were stopped by addition of an equal volume of 0.1 M sodium thiosulfate. The modified lipids in the mixture were extracted three times with equal volumes of chloroform:methanol (3:1). The organic solvent fractions were combined, the solvents were removed using a rotary evaporator, and the lipid residues were weighed and dissolved in dimethyl sulfoxide. None of the solvent-extractable lipids (up to 1.0 mg/plate) possessed mutagenic activity using strains TA 98 or TA 100 with and without metabolic activation. At the higher levels tested, cytotoxicity was observed. These results do not preclude the formation of mutagenic products. Highly volatile and aqueous reaction products and the products from C102 and lipid reactions have not been tested for their mutagenic potential. Also, the bacterial strains used in the mutagenicity assay may not be sensitive to the mutagenic effects of the reaction products. The formation of carbonyl compounds from reactions of C102 and lipids suggests a potential hazard since carbonyl compounds constitute a potentially important source of human mutagens (39). A new Salmonella tester strain (TA 104) has recently been developed for detecting these carbonyl mutagens (39). Studies are currently underway to determine the mutagenic potential of the chlorinated and oxidized lipid products using this new strain.

Reactions with Amino Acids, Peptides and Proteins
Numerous amino acids, peptides and proteins in biological systems can readily undergo oxidation or chlorination by chlorine compounds. Reactions of various amino acids with aqueous chlorine and C102 have been observed by many other investigators.
Margerum et al. (40) and Isaac and Morris (41) have studied the reactions of HOCl with several amino acids in an aqueous solution at neutral pH. All the amino acids tested reacted rapidly with HOCl. The formation of Nchloro compounds from the reaction mixtures was also characterized. Kirk and Mitchell (1) observed rapid changes and formation of an insoluble precipitate when they mixed aqueous chlorine with tryptophan. Using structurally similar indole compounds as tryptophan models, they concluded that the heterocyclic ring is involved in the reactions of the amino acid with aqueous chlorine. Although the reaction products were not conclusively identified, a monochlorinated product was isolated and partially characterized.
Although most amino acids do not readily react with C102, several amino acids including tyrosine, tryptophan, and cysteine will react rapidly (42)(43)(44)(45)(46). Tyrosine forms dopaquinone and dopachrome when oxidized by C102 (43). This reaction has been used as the basis for a colorimetric method to determine the concentration of C102 in water (42). Phenolic compounds such as tyrosine are especially susceptible to oxidation and chlorination by C102 (7). Chlorine dioxide oxidizes tryptophan to form insatin, indoxyl, indigo red, and unidentified yellow-orange substances (43). Oxidation of cystine by ClO2 results in the formation of cysteic acid through the disulfoxide intermediate (7).
Other investigators have studied the reactions ofC102 with amino acids under varying conditions. Schmidt and Braunsdorf (47) found that the least reactive amino acids were glycine, leucine, serine, alanine, phenylalanine, valine, hydroxyproline, and aspartic and glutamic acids. When a mixture of tryptophan, tyrosine, proline, threonine, methionine, valine, and lysine was reacted with diaphanol (a 50% solution of glacial acetic acid saturated with C102), the aromatic amino acids disappeared after 18 hr (48). These results indicate that aromatic amino acids tend to be more susceptible than aliphatic amino acids to oxidation by C102.
When flour was treated with ten times the commercial level of C12(g), the levels of methionine, cysteine, tyrosine, and histidine decreased by 68.3, 40.8, 31.6, and 16.8%, respectively (44). Tsen and Kulp (53) observed that the amount ofwater-extractable proteins in treated flour increased as chlorine concentration increased. The authors suggested that proteins were progressively cleaved with increasing chlorine concentration. In addition, dipeptides will form N,N-dichlorodipeptides in a concentrated aqueous chlorine solution (57). However, the amino group involved in-the peptide bond is not attacked by aqueous chlorine at room temperature.
Meredith et al. (58) treated wheat flour with high levels of C102 and did not find any protein or amino acid chloroderivatives. However, the levels of tryptophan and cystine were reduced to about 92 and 75%, respectively, of their original levels. In addition, the oxidation of cysteine to cysteic acid and of methionine to various oxides has been observed in flour treated with C102 (59). Other changes observed when wheat flour is exposed to high levels of C102 include an increase in water solubility of the flour proteins (44). When solutions of wheat proteins are exposed to C102, a pink-yellow color develops as a result of oxidation of tyrosine and tryptophan (44).
Very little is known about the toxicity of the oxidized or chlorinated derivatives of amino acids and proteins. Certain amino acids, such as aspartic acid, tryptophan, and tyrosine, as well as proteinaceous material may undergo halogenation to form dihaloacetonitriles and trihalomethanes (60), some of which have been shown to be genotoxic (61,62). In addition to tryptophan, Morris and Baum have identified proline and hydroxyproline as potential precursors of chloroform formation during water chlorination (24). The toxicities of the nonvolatile, halogenated, or oxidized products, however, are not known.
Sussmuth (63) tested the genotoxicity of the products forned from the chlorination of amino acids. Chlorination of methionine, tyrosine, phenylalanine, cysteine, and glycine resulted in the formation of mutagenic products in the Ames assay; however, the products were not identified. Horth et al. (64) also tested the mutagenic activity of extracts of chlorinated amino acids using the Ames assay. With the exception of glycine, mutagenic activity was observed in all of the extracts tested with the most potent precursors being methionine, tyrosine, phenylalanine, and a mixture of heterocyclic amino acids. They identified several compounds in the extracts by gas GC-MS. Only one of the identified products, dichloracetonitrile, is a known mutagen and was considered unlikely to account for a significant proportion of b20 g of ubbleached flour, pH 5.8, was chlorinated to pH 5.25 using radiolabeled chlorine gas with a specific activity of 1.26 x 106 dpm/ mg. the observed mutagenic activity. Nonvolatile products of chlorination were not tested for their mutagenic activity.
We are currently investigating the potential genotoxicity of reaction products of amino acids and chlorine compounds. In a preliminary study, using the Ames mutagenicity assay, we tested the mutagenic activity of the nonvolatile precipitate that forms immediately when tryptophan and aqueous chlorine in 0.1 M phosphate buffer, pH 6.0, are mixed. The precipitate was not mutagenic to TA 98 (up to 0.5 mg/plate) with and without metabolic activation. However, these preliminary results do not prove conclusively that the reaction products of tryptophan are not toxic. For example, antagonistic effects of two or more reaction products present in the precipitate may cancel or mask any mutagenic response. In addition, reaction conditions may not accurately reflect those found in the food industry.

Reactions with Food Products
Flour is commonly treated with 1,200 to 2,500 ppm C12(g) as a bleaching agent and to enhance its baking properties. The ability of chlorine to improve the baking properties of flour prompted many investigations into the reactions of chlorine with flour components.
Early studies demonstrated that the gluten and lipid fractions of flour were susceptible to chlorination (20,21). After fractionating bleached flours, approximately half the chlorine can be recovered in the water soluble fraction, presumably in the form of the chloride ion. Gilles et al. (26) analyzed chlorine content in bleached flour using X-ray spectroscopy. The lipid and water contents, which represent approximately 5% of the total flour, were found to contain greater than 90% of the chlorine introduced during the bleaching process. Using  Table 3). The highest levels of the 3Cl in the flour were present as chloride ion or were incorporated into the lipids.
The presence of the chloride ion indicates a reaction occuring in which chlorine is reduced to the ion. A likely target of chlorine oxidation is the starch fraction, which can be readily oxidized and depolymerized. Uchino and Whistler (14) deternined that, under normal chlorina- tion treatment of flour, 42% of the chlorine that is consumed is required for the resulting starch depolymerization. Interestingly, Wei et al. (65) found 47% of chlorine in their model system to be in the form of the chloride ion. Other studies have indicated that the actions of chlorine on the starch fraction are responsible for the improved baking qualities of bleached flour (10,11,20,21).
Considerable research has been conducted to determine the toxicity of chlorinated flour. Cunningham et al. (66) reported that weanling rats fed flour chlorinated at 2,000 and 10,000 ppm showed reduced growth rates and increased liver and kidney weights after 2 weeks of exposure. As mentioned previously, similar effects were noted in rats fed chlorinated proteins or lipids derived from bleached flour (36). In a 4-week feeding study, Fisher et al. (67) exposed rats to diets containing chlorinated (1257 or 2506 ppm chlorine) flour. Although a slight dose-related increase in relative kidney and liver weights was found in the treated animals, no pathological tissue changes were observed. In a chronic feeding study, rats fed flour chlorinated at levels used commercially (1200 to 2500 ppm), did not develop any pathological lesions, and their body and organ weights were the same as those of controls (68). In a similar study with mice, a dose-related increase in heart and kidney weights, and a decrease in ovary weights were observed (69). However, no evidence of carcinogenicity was apparent following treatment with chlorinated flour. The evidence indicates that chlorination of flour at levels used commercially does not pose a significant human health risk.
Chlorine incorporation into shrimp immersed in a 150 ppm solution of HO36Cl was recently determined (Table  4) (50). Approximately 2% of the chlorine was incorporated into shrimp, with 75% of this amount found in the edible portion. Most of the chlorine associated with the edible portion was present as chloride ion. Lipid analysis indicated that the more highly saturated fatty acids (i.e., cholesterol esters) incorporated less chlorine. In addition, phospholipids incorporated more chlorine than neutral lipids. The hydrophilic moiety of the phospholipid increased its solubility in the aqueous chlorine solution, allowing for the greater incorporation of chlorine.
Cunningham and Lawrence measured the incorporation of Cl into pieces of beef, pork, and poultry imersed in a 200 ppm aqueous 36Cl solution (51). Within 2 hr, over 50% of the chlorine reacted with both the lean and fat fractions of the meats. Most of the chlorine was reduced to chloride ion, which leached into the water. Nevertheless, significant amounts of chlorine reacted with tissue lipids and remained in the meat.
Robinson et al. (70) investigated the formation of chloroform in freshly eviscerated poultry carcasses exposed to water containing aqueous chlorine or chlorine dioxide. Treatment conditions simulated commercial broiler chicken processing. Low levels of chloroform were detected in carcasses that were treated with aqueous chlorine or C102 solutions. The highest concentration of chloroform (447 ppb) was found in carcasses immersed in water containing 50 mg/L of aqueous chlorine. The chloroform concentration was higher in depot fat than in the other tissues examined. The source of the chloroform is unknown, although the chloroform may have been present in the aqueous chlorine or C102 solutions before the chickens were immersed. Whether the immersion water was analyzed for chloroform content before the carcasses were treated was not mentioned. Nevertheless, the possibility that chloroform was formed from the reaction of chlorine or C102 with organic chemicals found naturally in the food product cannot be ruled out.

Summary
This report has focused on the reactions of chlorine compounds with carbohydrates, lipids, and proteins. Not mentioned are many other organic chemicals present in food that may react with chlorine, including the nucleic acids, vitamins and vitamin precursors, complex lipids, steroids, and other chemicals and essential trace elements. In most cases, the reactions of chlorine compounds with these food components have not been studied.
To determine adequately the risks of using chlorine compounds in the food industry, the amounts of chlorine used and the reactivities of the chlorine compounds with food chemicals must be examined. Also, the reaction products must be identified and their toxicities assessed. Finally, the amounts of these reaction products to which consumers are exposed should be determined. At this time, it is too early to assess the risks of using chlorine compounds in foods. It is hoped that we may have an answer in the near future.